fix:外协快速报工

Co-Authored-By: Claude <noreply@anthropic.com>

.agents/skills/release/SKILL.md

@@ -0,0 +1,23 @@
+---
+name: factory-release
+description: Use when the user explicitly asks to release the factory backend, run its release workflow, or bump its deployed version.
+---
+
+# 发布 Factory 后端
+
+仅在用户明确要求后端发版、release 或提升部署版本时执行本流程。普通代码修改、提交或推送不触发发版。
+
+严格按以下顺序执行：
+
+1. 确认目标是当前 `factory` 后端。若用户要求发布 `ehs_web` 前端，改读 `.codex/memory/reference_ehs_web_release.md`。
+2. 执行 `git status --short`，确认工作区状态。发布提交只能包含 `changelog.md` 和 `server/settings.py`；不得混入其他已修改或未跟踪文件。
+3. 执行 `bash update_changelog.sh`，从脚本输出读取形如 `3.1.YYYYMMDDHH` 的版本号，记为 `<VER>`。不得自行编造或改写版本号。
+4. 把 `server/settings.py` 中唯一权威版本字段 `SYS_VERSION` 更新为 `<VER>`。
+5. 读取 `changelog.md` 顶部约 20 行，确认标题是 `## <VER>`，且自上一个 tag 起的提交分类合理。出现重复段、明显漏项或无关内容时，先向用户确认，不得继续打 tag。
+6. 再次检查 diff，只暂存 `changelog.md` 和 `server/settings.py`。
+7. 提交信息使用 `release: <VER>`，不添加工具或模型专属署名。
+8. 创建无 `v` 前缀的 tag：`git tag <VER>`。
+9. 推送当前分支到 `origin`，成功后再执行 `git push origin <VER>`。逐步检查结果，不用命令链掩盖中间失败。
+10. 汇报 `<VER>`、changelog 包含的提交类别、commit hash、tag 名，以及分支和 tag 的推送结果。
+
+若提交、tag 或推送中途失败，先检查实际 Git 状态。任何会重写历史或删除远端 tag 的回退操作，都必须先取得用户明确授权。

.codex/memory/MEMORY.md

@@ -0,0 +1,15 @@
+# Factory 项目记忆索引
+
+- [前端路由由后端下发](feedback_routing.md)：新增 `ehs_web` 页面时不要修改静态路由表。
+- [scripts 目录脚本不入库](feedback_scripts_gitignore.md)：一次性回刷脚本不得强制加入 Git。
+- [发版只在明确要求时执行](feedback_version_bump.md)：普通改动不自动 bump 或 release。
+- [后端配套前端](project_frontend_ehs_web.md)：后端 API 改动时检查 `../ehs_web` 及统计页面惯例。
+- [生产数据只读验证](project_gx_db_access.md)：涉及真实数据形态时的安全查询原则。
+- [合批原料字段历史问题](project_material_ofrom_merge_bug.md)：`material_ofrom` 不一致的既有排查结论。
+- [前端独立发版流程](reference_ehs_web_release.md)：发布 `ehs_web` 时使用。
+- [项目 Python 虚拟环境](reference_python_venv.md)：运行 Django、pytest 和脚本时使用。
+- [项目测试数据库](reference_test_database.md)：测试与可切换的生产查询连接解耦，始终使用固定的 `test_ehs_develop`。
+- [两个 WebView 套壳 App](reference_wrapper_apps.md)：修改 h5x 与扫码、返回键交互时使用。
+- [前端验证时机](feedback_frontend_validation.md)：日常修改先跑 check，完整 build 留到 push 前执行。
+
+这些文件记录的是长期约定或历史上下文。执行任务前应结合当前代码和数据重新验证，尤其不要把历史缺陷结论直接当成当前故障原因。

.codex/memory/feedback_frontend_validation.md

@@ -0,0 +1,5 @@
+# 前端验证时机
+
+- 修改配套前端 `../ehs_web` 时，日常开发和中间验证优先运行项目已有的 `check`，不要每次修改后都运行完整 `build`。
+- 准备 push 前运行一次完整 `build`，用于发现生产构建阶段的问题。
+- 若当前前端尚未配置 `check` 脚本，应先说明现状，不得把其他命令擅自当作 `check`。

.codex/memory/feedback_routing.md

@@ -0,0 +1,11 @@
+# ehs_web 路由由后端下发
+
+`ehs_web` 使用后端下发的动态菜单和路由。新增管理页面时，不要向 `src/config/route.js` 添加条目；该文件中的既有静态路由属于历史或局部用法，不能作为新页面的默认范例。
+
+通常只需：
+
+- 创建 `src/views/<module>/<name>.vue`。
+- 补充 `src/api/model/<module>.js` 中的接口。
+- 在后端菜单与权限配置中设置组件相对路径和权限码。
+
+若页面未出现在菜单中，应检查后端菜单、权限与组件路径配置，不要用新增静态路由规避问题。

.codex/memory/feedback_scripts_gitignore.md

@@ -0,0 +1,11 @@
+# scripts 目录脚本不入库
+
+仓库 `.gitignore` 中的 `scripts/*.py` 是有意设置的：`correct_*` 等一次性回刷、修数脚本只在本地或服务器按需运行，不进入版本库。
+
+编写这类脚本后：
+
+- 正常保留在本地工作区。
+- 提交时不要执行 `git add -f`。
+- 不因仓库中存在少量历史已跟踪脚本而推断新脚本也应提交。
+
+只有用户明确要求某个脚本入库时，才可改变这一规则。

.codex/memory/feedback_version_bump.md

@@ -0,0 +1,7 @@
+# 发版只在用户明确要求时执行
+
+不要在每轮代码修改完成后自动提升版本号、生成 release commit 或打 tag。版本对应实际部署，多轮改动可以合并到一次发版。
+
+普通任务完成后可以正常提交代码，但只有用户明确说“发版”“走 release”或“bump 版本”时，才读取并执行 `.agents/skills/release/SKILL.md`。
+
+后端与前端有各自独立的发版流程；后端 release skill 不适用于 `../ehs_web`。

.codex/memory/project_frontend_ehs_web.md

@@ -0,0 +1,21 @@
+# Factory 配套前端 ehs_web
+
+当前仓库是 Django 后端，配套 Vue 前端位于 `../ehs_web`（通常解析为 `D:\projects\ehs_web`，也可从 `factory.code-workspace` 核对）。
+
+修改 serializer 字段、API 返回结构或接口行为后，应主动检查前端中消费该接口的：
+
+- `src/views/` 页面。
+- `src/api/model/` 接口定义。
+
+路由和菜单规则见 [前端路由由后端下发](feedback_routing.md)。
+
+统计页面现有惯例可参考 `src/views/statistics/good_check_gx2.vue` 和 `zt_batch_gx.vue`：
+
+- Options API 与 `el-container` 布局。
+- 复杂查询通过 `$API.wpm.batchstquery` 调用后端 `ComplexQueryMixin`。
+- `querys` 使用外层 OR、内层 AND 的 `{field, compare, value}` 条件。
+- JSONField 中文键常直接从 `scope.row.data` 读取。
+- 表格导出使用项目现有 `$XLSX` 封装。
+- JSONField 键的子串匹配使用 `icontains`；`contains` 表示 JSON 包含语义。
+
+这些是现有实现惯例，不替代对当前页面和接口代码的检查。

.codex/memory/project_gx_db_access.md

@@ -0,0 +1,13 @@
+# 生产数据只读验证
+
+各工厂 PostgreSQL 连接参数保存在本机被 Git 忽略的配置中。涉及批次谱系、统计口径或真实数据形态时，可以在任务确有需要且用户允许的范围内做只读抽样查询。
+
+安全要求：
+
+- 从本地忽略配置读取连接参数，不在文档、脚本、命令记录或提交中复制凭据。
+- 数据库会话显式设为只读。
+- 禁止执行 INSERT、UPDATE、DELETE、DDL、修数或其他写操作，除非用户针对该操作另行明确授权。
+- 临时查询脚本放在不会提交的本地位置。
+- Windows 中文输出需要时设置 UTF-8 编码。
+
+真实数据可能包含拆合批循环、自环边和期初断头批。修改统计或谱系算法前，优先用少量只读样本验证数据假设。

.codex/memory/project_material_ofrom_merge_bug.md

@@ -0,0 +1,29 @@
+# 合批原料 material_ofrom 历史排查
+
+## 历史现象
+
+曾出现合批或生产流转后，批次签带显示的原料类型与实物板段棒签不一致。同一工序可能存在多条并行原料输入，因此也可能发生真实混料。该问题当时决定暂缓，只有再次出现并取得可复现样本时再继续调查。
+
+## 已确认的代码风险
+
+`WMaterial.locked_get_or_create`（`apps/utils/models.py`）命中既有行时直接返回，不会应用 `defaults`。在 `apps/wpm/services.py` 的合批或流转逻辑中：
+
+- `material_ofrom`、`batch_ofrom`、`number_from` 只放在 `defaults`。
+- 定位既有行的键不包含 `material_ofrom`。
+
+因此，一旦命中相同业务键的旧行，后续不同原料流入时，`material_ofrom` 可能继续保留首次建行时的值。该字段又是单个外键，结构上无法表达混料。签带展示值来自 serializer 中的 `material_ofrom_name`。
+
+## 历史样本结论
+
+当时抽样数据中的相关 `WMaterial` 行已经归零，来源记录也指向原有材料；实物板段号在当前 WPM 相关表中没有找到。由此推测，实物棒签的材料源和板段号可能来自上游拉丝或滚圆子系统，单靠当前库无法闭环证明。
+
+## 下次接续
+
+再次出现时应先取得一个“实物棒签材料与批次签带材料不一致、且两者来源可在同一系统追踪”的批次样本，然后：
+
+1. 确认实物棒签由哪个系统和表生成。
+2. 用只读查询验证流转前后定位键是否命中旧行。
+3. 区分真实混料、上游标签来源和 `defaults` 未更新这三种原因。
+4. 在复现前不要直接修改字段结构或流转逻辑。
+
+查询时遵守 [生产数据只读验证](project_gx_db_access.md)。

.codex/memory/reference_ehs_web_release.md

@@ -0,0 +1,14 @@
+# ehs_web 前端独立发版流程
+
+`../ehs_web` 有独立于后端的发版流程。`.agents/skills/release/SKILL.md` 只适用于当前 `factory` 后端。
+
+前端发版要点：
+
+1. 权威版本号位于 `src/config/index.js` 的 `APP_VER`。
+2. 在前端仓库运行 `update_changelog.sh`，使用脚本生成的 `3.1.YYYYMMDDHH` 版本号。若上一个 tag 过旧，可向脚本传入合适的起点 ref，并检查提交分类。
+3. 把 `APP_VER` 更新为相同版本。
+4. 只暂存前端的 `changelog.md` 和 `src/config/index.js`。
+5. 提交信息使用 `release: <VER>`，tag 使用无 `v` 前缀的 `<VER>`。
+6. 推送到前端仓库配置的 `coding` 远端，而不是默认假设使用 `origin`。
+
+前端 `package.json` 中的旧 `version` 字段不是当前权威版本。仍须遵守“只有用户明确要求才发版”的约束。

.codex/memory/reference_python_venv.md

@@ -0,0 +1,15 @@
+# Factory 项目 Python 解释器
+
+项目根目录包含 Windows 虚拟环境 `.venv`。运行 `manage.py`、pytest 或项目 Python 脚本时使用：
+
+```powershell
+D:\projects\factory\.venv\Scripts\python.exe manage.py check
+```
+
+从 Git Bash 调用时可使用：
+
+```bash
+/d/projects/factory/.venv/Scripts/python.exe manage.py check
+```
+
+不要直接依赖系统 `python`，因为系统环境可能缺少项目依赖。

.codex/memory/reference_test_database.md

@@ -0,0 +1,15 @@
+# 项目测试数据库
+
+- 生产问题查询时，默认数据库连接可以根据工厂或环境切换到不同 IP 和业务库；这类连接只用于授权范围内的生产数据只读验证。
+- 所有 Django、pytest 及其他自动化测试必须与当前生产查询连接解耦，始终使用固定的 `test_ehs_develop` 测试数据库连接。
+- 不能仅依赖当前默认连接的 Django 自动 `test_<NAME>` 命名；即使当前业务库是 `bxerp` 或其他库，测试也不应转而使用 `test_bxerp` 或其他派生库。
+- 不得在生产数据库上运行会建表、迁移、写入或清理数据的测试。
+- 生产查询与固定测试库的连接参数均从本机已忽略配置中读取，不在项目记忆、源码或提交信息中记录凭据。
+
+当前实现：
+
+- 固定测试连接保存在已忽略的 `config/conf_test.py`。
+- `server/test_settings.py` 加载该连接，`manage.py test` 会自动选择测试 settings；其他 `manage.py` 命令仍使用当前业务库连接。
+- 测试 settings 中的基础 `NAME` 和 `TEST.NAME` 都必须是 `test_ehs_develop`，并在启动时校验，防止通过测试 settings 误操作 `ehs_develop` 或任何生产库。
+- pytest-django 通过项目根目录 `pytest.ini` 固定使用 `server.test_settings`。
+- 本地重复运行测试时优先使用 `.venv\\Scripts\\python.exe manage.py test --keepdb --noinput`。

.codex/memory/reference_wrapper_apps.md

@@ -0,0 +1,8 @@
+# Web 前端的两个套壳 App
+
+`ehs_web` 的 h5x 页面还由两个 WebView 壳工程加载：
+
+- 原生 Android 壳：`C:\Users\caoqi\AndroidStudioProjects\BxErp`，包名 `com.example.bxerp`。核心入口是 `app/src/main/java/com/example/bxerp/MainActivity.kt`，通过 `window.Android` 暴露扫码能力，常用交互为 `Android.openScanner()` 和页面回调 `onScanResult(data)`。
+- uni-app 壳：`D:\projects\mp_shell`，Vue 3/HBuilderX 工程，应用名“光芯ERP”。`pages/index/index.vue` 使用 `<web-view>` 加载 h5x，并处理返回键与状态栏适配。
+
+修改 h5x 的扫码、返回键或 WebView 消息交互时，应检查这两个壳的约定。工程位置和加载地址可能随部署变化，操作前以当前本地代码为准。

AGENTS.md

@@ -0,0 +1,28 @@
+# Factory 项目协作约定
+
+## 项目结构
+
+- 当前仓库是 Django 后端。
+- 配套 Vue 前端位于 `../ehs_web`。修改 serializer 字段、API 返回结构或接口行为时，检查并同步对应的前端页面和 `src/api/model/` 接口定义。
+- `ehs_web` 的菜单和路由由后端动态下发。新增页面时不要修改 `src/config/route.js`；页面组件、菜单和权限应按后端配置衔接。
+
+## 本地执行
+
+- 在 Windows 下运行 Django、pytest 或其他 Python 命令时，使用项目虚拟环境：`D:\projects\factory\.venv\Scripts\python.exe`。不要依赖系统 Python。
+- `scripts/*.py` 是有意忽略的一次性回刷或修数脚本。未经用户明确要求，不得用 `git add -f` 强制提交。
+
+## 数据安全
+
+- 涉及生产数据验证时，只允许执行只读查询；写入、修数或结构变更必须取得用户针对该操作的明确授权。
+- 数据库连接参数从本机已忽略的配置读取。不得把用户名、密码、连接串或可直接复用的带凭据命令写入文档、skill、提交信息或源码。
+
+## 发版
+
+- 不要在普通修改完成后自动提升版本号、打 tag 或发版。
+- 只有用户明确要求“发版”“release”或“bump 版本”时，才使用 `.agents/skills/release/SKILL.md`。
+- 前端 `ehs_web` 有独立发版流程，不要套用后端 release 流程。
+
+## 按需项目知识
+
+- 详细项目背景、历史反馈和排障记录见 `.codex/memory/MEMORY.md`。
+- 只在任务相关时读取对应 memory 文件，避免把历史问题误当成当前事实。

apps/bi/serializers.py

@@ -18,10 +18,35 @@ class DatasetCreateUpdateSerializer(CustomModelSerializer):
 
 
 class DatasetSerializer(CustomModelSerializer):
+    description = serializers.CharField(
+        label="适用场景与统计口径",
+        help_text="说明该数据集适合回答的问题、指标口径、参数格式和返回字段含义",
+        required=False,
+        allow_blank=True,
+    )
+    default_param = serializers.JSONField(
+        label="默认查询参数",
+        help_text="执行时可覆盖的参数及默认值；内部 SQL 片段参数应保留默认值",
+        required=False,
+    )
+    test_param = serializers.JSONField(
+        label="测试查询参数",
+        help_text="数据集维护时使用的示例参数，普通查询优先参考 description",
+        required=False,
+    )
+
     class Meta:
         model = Dataset
         fields = '__all__'
 
+
+class DatasetListResponseSerializer(serializers.Serializer):
+    count = serializers.IntegerField(label="数据集总数")
+    next = serializers.URLField(required=False, allow_null=True)
+    previous = serializers.URLField(required=False, allow_null=True)
+    results = DatasetSerializer(many=True)
+
+
 class DatasetRecordSerializer(CustomModelSerializer):
     class Meta:
         model = DatasetRecord
@@ -36,6 +61,14 @@ class DatasetRecordSerializer(CustomModelSerializer):
 
 class DataExecSerializer(serializers.Serializer):
     query = serializers.JSONField(
-        label="查询字典参数", required=False, allow_null=True)
+        label="查询字典参数",
-    is_test = serializers.BooleanField(label='是否测试', default=False)
+        help_text="按所选数据集 description/default_param 声明的业务参数填写",
-    raise_exception = serializers.BooleanField(label='是否直接报错', default=False)
+        required=False,
+        allow_null=True,
+    )
+    is_test = serializers.BooleanField(
+        label='是否测试', help_text="普通业务查询固定为 false", default=False
+    )
+    raise_exception = serializers.BooleanField(
+        label='是否直接报错', help_text="建议为 true，便于修正缺失或非法参数", default=True
+    )

apps/bi/services.py

@@ -1,11 +1,15 @@
-from rest_framework.exceptions import ParseError
+import concurrent.futures
 import json
-from jinja2 import Template
+import logging
+
+from rest_framework.exceptions import ParseError
+
 from apps.bi.models import Dataset
-import concurrent
 from apps.utils.sql import execute_raw_sql, format_sqldata
 from apps.utils.tools import MyJSONEncoder
 
+myLogger = logging.getLogger('log')
+
 forbidden_keywords = ["UPDATE", "DELETE", "DROP", "TRUNCATE", "INSERT", "CREATE", "ALTER", "GRANT", "REVOKE", "EXEC", "EXECUTE"]
 
 
@@ -32,31 +36,57 @@ def format_json_with_placeholders(json_str, **kwargs):
     return formatted_json
 
 
-def exec_dataset(dt: Dataset, xquery: dict = {}):
+def render_dataset_sql(dt: Dataset, xquery=None, *, is_test=False):
+    """根据数据集配置和调用参数生成经过安全检查的只读 SQL。"""
+    query = dict(dt.default_param or {})
+    query.update(dict(dt.test_param or {}) if is_test else dict(xquery or {}))
+    if not dt.sql_query:
+        return ''
+    try:
+        return check_sql_safe(dt.sql_query.format(**query))
+    except KeyError as exc:
+        raise ParseError(f'需指定查询参数_{str(exc)}') from exc
+
+
+def execute_rendered_dataset(dt: Dataset, full_sql: str, *, raise_exception=True):
+    """执行已经渲染和校验的 SQL，返回可合并到数据集响应的结果。"""
+    results = {}
+    results2 = {}
+    can_cache = True
+    sql_list = [sql for sql in full_sql.strip(';').split(';') if sql.strip()]
+    if sql_list:
+        with concurrent.futures.ThreadPoolExecutor(max_workers=6) as executor:
+            futures = {
+                executor.submit(execute_raw_sql, sql): (f'ds{index}', sql)
+                for index, sql in enumerate(sql_list)
+            }
+            for future in concurrent.futures.as_completed(futures):
+                name, sql = futures[future]
+                try:
+                    res = future.result()
+                    results[name], results2[name] = format_sqldata(res[0], res[1])
+                except Exception as exc:
+                    myLogger.error(f'bi查询异常：{str(exc)}-{dt.code}--{sql}')
+                    if raise_exception:
+                        raise ParseError(f'查询异常：{str(exc)}') from exc
+                    results[name] = 'error: ' + str(exc)
+                    can_cache = False
+
+    response_data = {'data': results, 'data2': results2}
+    if dt.echart_options and not dt.echart_options.startswith('function'):
+        for result in results.values():
+            if isinstance(result, str):
+                raise ParseError(result)
+        response_data['echart_options'] = format_json_with_placeholders(
+            dt.echart_options, **results
+        )
+    return response_data, can_cache
+
+
+def exec_dataset(dt: Dataset, xquery=None):
     """执行数据集
     返回 (sql语句, { rda})
     """
-    rdata = {}
+    full_sql = render_dataset_sql(dt, xquery)
-    results = {}
+    response_data, _ = execute_rendered_dataset(dt, full_sql)
-    results2 = {}
+    return full_sql, response_data
-    query = dt.default_param
-    if dt.sql_query:
-        query.update(xquery)
-        sql_f_ = check_sql_safe(dt.sql_query.format(**query))
-        sql_f_strip = sql_f_.strip(';')
-        sql_f_l = sql_f_strip.split(';')
-        # 多线程运行并返回字典结果
-        with concurrent.futures.ThreadPoolExecutor(max_workers=6) as executor:
-            fun_ps = []
-            for ind, val in enumerate(sql_f_l):
-                fun_ps.append((f'ds{ind}', execute_raw_sql, val))
-            # 生成执行函数
-            futures = {executor.submit(i[1], i[2]): i for i in fun_ps}
-            for future in concurrent.futures.as_completed(futures):
-                name, *_, sql_f = futures[future]  # 获取对应的键
-                res = future.result()
-                results[name],  results2[name] = format_sqldata(
-                    res[0], res[1])
-    rdata['data'] = results
-    rdata['data2'] = results2
-    return sql_f_, rdata

apps/bi/test_services.py

@@ -0,0 +1,105 @@
+from types import SimpleNamespace
+from unittest.mock import patch
+
+from django.test import SimpleTestCase
+from rest_framework.exceptions import ParseError
+
+from apps.bi.services import (
+    exec_dataset,
+    execute_rendered_dataset,
+    render_dataset_sql,
+)
+from apps.bi.views import DatasetViewSet
+
+
+def dataset(**overrides):
+    values = {
+        "code": "output_daily",
+        "sql_query": "select * from output where day = '{day}'",
+        "default_param": {"day": "2026-08-01"},
+        "test_param": {"day": "2026-08-02"},
+        "echart_options": "",
+    }
+    values.update(overrides)
+    return SimpleNamespace(**values)
+
+
+class DatasetExecutionServiceTests(SimpleTestCase):
+    def test_render_does_not_mutate_default_parameters(self):
+        item = dataset()
+
+        sql = render_dataset_sql(item, {"day": "2026-08-10"})
+
+        self.assertIn("2026-08-10", sql)
+        self.assertEqual(item.default_param, {"day": "2026-08-01"})
+
+    def test_render_reports_missing_parameters(self):
+        item = dataset(default_param={}, sql_query="select '{required}'")
+
+        with self.assertRaises(ParseError):
+            render_dataset_sql(item)
+
+    def test_execute_formats_each_statement(self):
+        item = dataset(echart_options='{"series": {ds0}}')
+        with (
+            patch("apps.bi.services.execute_raw_sql", return_value=([], [])),
+            patch(
+                "apps.bi.services.format_sqldata",
+                return_value=([{"count": 1}], {"count": [1]}),
+            ),
+        ):
+            response, can_cache = execute_rendered_dataset(
+                item, "select 1;select 2"
+            )
+
+        self.assertTrue(can_cache)
+        self.assertEqual(set(response["data"]), {"ds0", "ds1"})
+        self.assertIn('"count": 1', response["echart_options"])
+
+    def test_empty_dataset_has_stable_empty_result(self):
+        full_sql, response = exec_dataset(dataset(sql_query=""))
+
+        self.assertEqual(full_sql, "")
+        self.assertEqual(response, {"data": {}, "data2": {}})
+
+
+class DatasetViewExecutionTests(SimpleTestCase):
+    @patch("apps.bi.views.cache")
+    @patch("apps.bi.views.execute_rendered_dataset")
+    @patch("apps.bi.views.render_dataset_sql")
+    @patch("apps.bi.views.DatasetSerializer")
+    def test_api_reuses_shared_execution_service(
+        self,
+        serializer_mock,
+        render_mock,
+        execute_mock,
+        cache_mock,
+    ):
+        item = dataset(
+            enabled=True,
+            name="日产量",
+            cache_seconds=10,
+        )
+        serializer_mock.return_value.data = {
+            "code": item.code,
+            "echart_options": "",
+        }
+        render_mock.return_value = "select 1"
+        execute_mock.return_value = (
+            {"data": {"ds0": [{"count": 1}]}, "data2": {}},
+            True,
+        )
+        cache_mock.get.return_value = None
+        view = DatasetViewSet(basename="dataset")
+        view.get_object = lambda: item
+        request = SimpleNamespace(
+            data={"query": {"day": "2026-08-10"}},
+            user=SimpleNamespace(id=42, belong_dept=SimpleNamespace(id=7)),
+        )
+
+        response = view.exec(request)
+
+        render_query = render_mock.call_args.args[1]
+        self.assertEqual(render_query["r_user"], 42)
+        self.assertEqual(render_query["r_dept"], 7)
+        self.assertEqual(response.data["data"]["ds0"][0]["count"], 1)

apps/bi/tests.py

@@ -1,3 +1,25 @@
-from django.test import TestCase
+from django.test import SimpleTestCase
 
-# Create your tests here.
+from apps.bi.serializers import (
+    DataExecSerializer,
+    DatasetListResponseSerializer,
+    DatasetSerializer,
+)
+from apps.bi.views import DatasetViewSet
+
+
+class DatasetAgentDiscoveryTests(SimpleTestCase):
+    def test_dataset_catalog_searches_description(self):
+        self.assertEqual(
+            DatasetViewSet.search_fields,
+            ["name", "code", "description"],
+        )
+
+    def test_dataset_schema_explains_discovery_and_exec_parameters(self):
+        dataset = DatasetSerializer()
+        execute = DataExecSerializer()
+
+        self.assertIn("适合回答的问题", dataset.fields["description"].help_text)
+        self.assertIn("业务参数", execute.fields["query"].help_text)
+        self.assertTrue(execute.fields["raise_exception"].default)
+        self.assertIn("results", DatasetListResponseSerializer().fields)

apps/bi/views.py

@@ -3,17 +3,21 @@ from apps.utils.viewsets import CustomModelViewSet, CustomGenericViewSet
 from rest_framework.decorators import action
 from rest_framework.response import Response
 from apps.bi.models import Dataset, DatasetRecord
-from apps.bi.serializers import DatasetSerializer, DatasetCreateUpdateSerializer, DataExecSerializer, DatasetRecordSerializer
+from apps.bi.serializers import (
+    DataExecSerializer,
+    DatasetCreateUpdateSerializer,
+    DatasetListResponseSerializer,
+    DatasetRecordSerializer,
+    DatasetSerializer,
+)
 from django.apps import apps
-import concurrent.futures
 from django.core.cache import cache
-from apps.utils.sql import execute_raw_sql, format_sqldata
+from apps.bi.services import execute_rendered_dataset, render_dataset_sql
-from apps.bi.services import check_sql_safe, format_json_with_placeholders
 from rest_framework.exceptions import ParseError
 from rest_framework.generics import get_object_or_404
 from apps.utils.mixins import ListModelMixin
-import logging
+from drf_yasg import openapi
-myLogger = logging.getLogger('log')
+from drf_yasg.utils import swagger_auto_schema
 # Create your views here.
 
 
@@ -22,9 +26,54 @@ class DatasetViewSet(CustomModelViewSet):
     serializer_class = DatasetSerializer
     create_serializer_class = DatasetCreateUpdateSerializer
     update_serializer_class = DatasetCreateUpdateSerializer
-    search_fields = ['name', 'code']
+    search_fields = ['name', 'code', 'description']
     ordering = ['name', 'code', 'id']
 
+    @swagger_auto_schema(
+        operation_id="bi_dataset_list",
+        operation_summary="查询复杂统计报表的数据集目录",
+        operation_description=(
+            "产量、良率、缺陷、库存、绩效、趋势和按日/月汇总等统计聚合查询的统一入口。"
+            "先调用本接口，根据 name、description、default_param 和 test_param 选择数据集，"
+            "再调用 bi_dataset_exec。建议使用 query={id,name,code,description,default_param,"
+            "test_param,enabled} 裁剪字段，并设置 page_size=100 查看完整目录；"
+            "search 可按名称、code 或 description 检索。"
+        ),
+        manual_parameters=[
+            openapi.Parameter(
+                "search",
+                openapi.IN_QUERY,
+                description="按数据集名称、code 或适用场景关键词检索",
+                type=openapi.TYPE_STRING,
+            ),
+            openapi.Parameter(
+                "page",
+                openapi.IN_QUERY,
+                description="页码，从 1 开始",
+                type=openapi.TYPE_INTEGER,
+            ),
+            openapi.Parameter(
+                "page_size",
+                openapi.IN_QUERY,
+                description="每页数量；当前目录建议传 100",
+                type=openapi.TYPE_INTEGER,
+            ),
+            openapi.Parameter(
+                "query",
+                openapi.IN_QUERY,
+                description=(
+                    "django-restql 字段裁剪表达式，例如 "
+                    "{id,name,code,description,default_param,test_param,enabled}"
+                ),
+                type=openapi.TYPE_STRING,
+            ),
+        ],
+        responses={200: DatasetListResponseSerializer},
+        tags=["BI 数据集与报表"],
+    )
+    def list(self, request, *args, **kwargs):
+        return super().list(request, *args, **kwargs)
+
     def get_object(self):
         """
         Returns the object the view is displaying.
@@ -57,6 +106,18 @@ class DatasetViewSet(CustomModelViewSet):
 
         return obj
 
+    @swagger_auto_schema(
+        operation_id="bi_dataset_exec",
+        operation_summary="执行已配置的只读统计数据集",
+        operation_description=(
+            "使用 dataset list 返回的 id 或 code 执行数据集。body.query 只填写该数据集"
+            "description/default_param 声明的业务参数；正常查询设置 is_test=false。"
+            "统计聚合使用本接口，日志和业务明细列表用于逐条追溯。"
+        ),
+        request_body=DataExecSerializer,
+        responses={200: DatasetSerializer},
+        tags=["BI 数据集与报表"],
+    )
     @action(methods=['post'], detail=True, perms_map={'post': 'dataset.exec'}, serializer_class=DataExecSerializer, cache_seconds=0, logging_methods=[])
     def exec(self, request, pk=None):
         """执行sql查询
@@ -67,59 +128,24 @@ class DatasetViewSet(CustomModelViewSet):
         if not dt.enabled:
             raise ParseError(f'{dt.name}-该查询未启用')
         rdata = DatasetSerializer(instance=dt).data
-        xquery = request.data.get('query', {})
+        xquery = dict(request.data.get('query') or {})
         is_test = request.data.get('is_test', False)
         raise_exception = request.data.get('raise_exception', True)
         xquery['r_user'] = request.user.id
         xquery['r_dept'] = request.user.belong_dept.id if request.user.belong_dept else ''
-        can_cache = True
+        full_sql = render_dataset_sql(dt, xquery, is_test=is_test)
-        results = {}
+        hash_k = None
-        results2 = {}
+        if full_sql:
-        query = dt.default_param
+            sql_f_strip = full_sql.strip(';')
-        if dt.sql_query:
-            if is_test:
-                query.update(dt.test_param)
-            else:
-                query.update(xquery)
-            try:
-                sql_f_ = check_sql_safe(dt.sql_query.format(**query))
-            except KeyError as e:
-                raise ParseError(f'需指定查询参数_{str(e)}')
-            sql_f_strip = sql_f_.strip(';')
-            sql_f_l = sql_f_strip.split(';')
             hash_k = hash(sql_f_strip)
             hash_v = cache.get(hash_k, None)
             if hash_v:
                 return Response(hash_v)
-            # 多线程运行并返回字典结果
+        response_data, can_cache = execute_rendered_dataset(
-            with concurrent.futures.ThreadPoolExecutor(max_workers=6) as executor:
+            dt, full_sql, raise_exception=raise_exception
-                fun_ps = []
+        )
-                for ind, val in enumerate(sql_f_l):
+        rdata.update(response_data)
-                    fun_ps.append((f'ds{ind}', execute_raw_sql, val))
+        if response_data['data'] and can_cache and hash_k is not None:
-                # 生成执行函数
-                futures = {executor.submit(i[1], i[2]): i for i in fun_ps}
-                for future in concurrent.futures.as_completed(futures):
-                    name, *_, sql_f = futures[future]  # 获取对应的键
-                    try:
-                        res = future.result()
-                        results[name],  results2[name] = format_sqldata(
-                            res[0], res[1])
-                    except Exception as e:
-                        myLogger.error(f'bi查询异常：{str(e)}-{dt.code}--{sql_f}')
-                        if raise_exception:
-                            raise ParseError(f'查询异常：{str(e)}')
-                        else:
-                            results[name] = 'error: ' + str(e)
-                            can_cache = False
-        rdata['data'] = results
-        rdata['data2'] = results2
-        if rdata['echart_options'] and not rdata['echart_options'].startswith('function'):
-            for key in results:
-                if isinstance(results[key], str):
-                    raise ParseError(results[key])
-            rdata['echart_options'] = format_json_with_placeholders(
-                rdata['echart_options'], **results)
-        if results and can_cache:
             cache.set(hash_k, rdata, dt.cache_seconds)
         return Response(rdata)
 

apps/develop/tests.py

@@ -1,3 +1,32 @@
-from django.test import TestCase
+from django.test import SimpleTestCase
+from rest_framework.permissions import IsAdminUser
+from rest_framework.test import APIRequestFactory
 
-# Create your tests here.
+from apps.develop.views import ServerTime, TestViewSet
+
+
+class DevelopApiPermissionTests(SimpleTestCase):
+    def setUp(self):
+        self.factory = APIRequestFactory()
+
+    def test_test_endpoint_rejects_anonymous_requests(self):
+        request = self.factory.post(
+            '/api/develop/test/send_sms/',
+            {},
+            format='json',
+        )
+
+        response = TestViewSet.as_view({'post': 'send_sms'})(request)
+
+        self.assertIn(response.status_code, (401, 403))
+
+    def test_server_time_requires_admin(self):
+        request = self.factory.get('/api/develop/server_time/')
+
+        response = ServerTime.as_view()(request)
+
+        self.assertIn(response.status_code, (401, 403))
+
+    def test_develop_views_use_admin_permission(self):
+        self.assertEqual(TestViewSet.permission_classes, [IsAdminUser])
+        self.assertEqual(ServerTime.permission_classes, [IsAdminUser])

apps/develop/urls.py

@@ -1,4 +1,5 @@
-from django.urls import path, include
+from django.conf import settings
+from django.urls import include, path
 from apps.develop.views import (BackupDatabase, BackupMedia, ReloadClientGit, 
 ReloadServerGit, ReloadServerOnly, TestViewSet, CorrectViewSet, testScanHtml, ServerTime)
 from rest_framework.routers import DefaultRouter
@@ -6,9 +7,11 @@ from rest_framework.routers import DefaultRouter
 API_BASE_URL = 'api/develop/'
 HTML_BASE_URL = 'dhtml/develop/'
 router = DefaultRouter()
-router.register('test', TestViewSet, basename='api_test')
 router.register('correct', CorrectViewSet, basename='correct')
 
+if settings.DEBUG:
+    router.register('test', TestViewSet, basename='api_test')
+
 urlpatterns = [
     path(API_BASE_URL + 'reload_server_git/', ReloadServerGit.as_view()),
     # path(API_BASE_URL + 'reload_web_git/', ReloadClientGit.as_view()),
@@ -17,5 +20,7 @@ urlpatterns = [
     path(API_BASE_URL + 'backup_media/', BackupMedia.as_view()),
     path(API_BASE_URL + 'server_time/', ServerTime.as_view()),
     path(API_BASE_URL, include(router.urls)),
-    path(HTML_BASE_URL + "testscan/", testScanHtml)
 ]
+
+if settings.DEBUG:
+    urlpatterns.append(path(HTML_BASE_URL + "testscan/", testScanHtml))

apps/develop/views.py

@@ -40,11 +40,12 @@ from datetime import datetime
 # Create your views here.
 
 class ServerTime(APIView):
+    permission_classes = [IsAdminUser]
 
     def get_permissions(self):
-        if self.request.method == 'GET':
+        if self.request.method == "GET":
             return [AllowAny()]
-        return [IsAdminUser()]
+        return super().get_permissions()
 
     @swagger_auto_schema(responses={200: ServerTimeSerializer})
     def get(self, request):
@@ -62,9 +63,13 @@ class ServerTime(APIView):
 
         修改服务器时间
         """
-        command = f'date -s "{request.data["server_time"]}"'
+        serializer = ServerTimeSerializer(data=request.data)
+        serializer.is_valid(raise_exception=True)
+        server_time = serializer.validated_data['server_time'].strftime(
+            "%Y-%m-%d %H:%M:%S"
+        )
         completed = subprocess.run(
-            ["sudo", "-S", "sh", "-c", command],  # 添加 -S 参数
+            ["sudo", "-S", "date", "-s", server_time],
             input=SD_PWD + "\n",  # 注意要在密码后加换行符
             capture_output=True,
             text=True
@@ -269,10 +274,9 @@ class CorrectViewSet(CustomGenericViewSet):
 
 class TestViewSet(CustomGenericViewSet):
     perms_map = {}
-    authentication_classes = ()
+    permission_classes = [IsAdminUser]
-    permission_classes = ()
 
-    @action(methods=['post'], detail=False, serializer_class=SendSmsSerializer, authentication_classes=())
+    @action(methods=['post'], detail=False, serializer_class=SendSmsSerializer)
     def send_sms(self, request, pk=None):
         """发送短信测试
 
@@ -565,7 +569,7 @@ class TestViewSet(CustomGenericViewSet):
     #     correct_card_time()
     #     return Response()
 
-    @action(methods=['post'], detail=False, serializer_class=Serializer, permission_classes=[])
+    @action(methods=['post'], detail=False, serializer_class=Serializer)
     @transaction.atomic
     def correct_data(self, request, pk=None):
         """修正数据
@@ -674,7 +678,7 @@ class TestViewSet(CustomGenericViewSet):
             Ticket.objects.get_queryset(all=True).delete()
         return Response()
 
-    @action(methods=['post'], detail=False, serializer_class=Serializer, permission_classes=[])
+    @action(methods=['post'], detail=False, serializer_class=Serializer)
     def test_cal(self, request, pk=None):
         from apps.wpm.tasks import cal_exp_duration_sec
         cal_exp_duration_sec('3397169058570170368')

apps/edu/views.py

@@ -71,6 +71,8 @@ class ExamViewSet(CustomModelViewSet):
 
     def get_queryset(self):
         qs = super().get_queryset()
+        if getattr(self, 'swagger_fake_view', False):
+            return qs
         if has_perm(self.request.user, ["exam.view"]):
             return qs
         user:User = self.request.user
@@ -142,6 +144,8 @@ class ExamRecordViewSet(ListModelMixin, DestroyModelMixin, RetrieveModelMixin, C
 
     def get_queryset(self):
         qs = super().get_queryset()
+        if getattr(self, 'swagger_fake_view', False):
+            return qs
         if has_perm(self.request.user, ["examrecord.view"]):
             return qs
         return qs.filter(create_by=self.request.user)
@@ -207,6 +211,8 @@ class TrainRecordViewSet(CustomModelViewSet):
 
     def get_queryset(self):
         qs = super().get_queryset()
+        if getattr(self, 'swagger_fake_view', False):
+            return qs
         if has_perm(self.request.user, ["train.view"]):
             return qs
         return qs.filter(create_by=self.request.user)

apps/em/filters.py

@@ -6,6 +6,8 @@ from apps.utils.filters import MyJsonListFilter
 
 class EquipFilterSet(filters.FilterSet):
     tags = MyJsonListFilter(label='tags/json/list查询')
+    exclude_cate_name = filters.CharFilter(
+        field_name='cate__name', exclude=True, label='排除设备分类名称')
 
     class Meta:
         model = Equipment

apps/enm/urls.py

@@ -1,7 +1,8 @@
 from django.urls import path, include
 from rest_framework.routers import DefaultRouter
 from apps.enm.views import (MpointViewSet, MpointStatViewSet,
-                            EnStatViewSet, EnStat2ViewSet, XscriptViewSet, MpLogxAPIView)
+                            EnStatViewSet, EnStat2ViewSet, XscriptViewSet, MpLogxAPIView,
+                            WengfuMpLogxAPIView, WengfuMpointListAPIView)
 
 API_BASE_URL = 'api/enm/'
 HTML_BASE_URL = 'dhtml/enm/'
@@ -16,4 +17,6 @@ router.register('xscript', XscriptViewSet, basename='xscript')
 urlpatterns = [
     path(API_BASE_URL, include(router.urls)),
     path(f'{API_BASE_URL}mplogx/', MpLogxAPIView.as_view(), name='mplogx_list'),
+    path(f'{API_BASE_URL}wengfu_mplogx/', WengfuMpLogxAPIView.as_view(), name='wengfu_mplogx'),
+    path(f'{API_BASE_URL}wengfu_mpoints/', WengfuMpointListAPIView.as_view(), name='wengfu_mpoints'),
 ]

apps/enm/views.py

@@ -265,6 +265,134 @@ class MpLogxAPIView(APIView):
 
 
 
+class WengfuMpLogxAPIView(APIView):
+    perms_map = {"get": "*", "post": "*"}
+
+    def _get_conn(self):
+        from apps.utils.sql import DbConnection
+        return DbConnection(
+            host='10.0.11.52', user='postgres',
+            password='zcDsj@2024', database='hfnf', dbtype='pg'
+        )
+
+    def get(self, request, *args, **kwargs):
+        mpoint_ids = request.query_params.get("mpoint_ids", None)
+        timex_gte = request.query_params.get("timex__gte", None)
+        timex_lte = request.query_params.get("timex__lte", None)
+        page = int(request.query_params.get("page", 1))
+        page_size = int(request.query_params.get("page_size", 20))
+
+        conditions = []
+        params = []
+
+        if mpoint_ids:
+            ids = [x.strip() for x in mpoint_ids.split(",") if x.strip()]
+            placeholders = ",".join(["%s"] * len(ids))
+            conditions.append(f"m.mpoint_id IN ({placeholders})")
+            params.extend(ids)
+        if timex_gte:
+            conditions.append("m.timex >= %s")
+            params.append(timex_gte)
+        if timex_lte:
+            conditions.append("m.timex <= %s")
+            params.append(timex_lte)
+
+        where = ("WHERE " + " AND ".join(conditions)) if conditions else ""
+
+        with self._get_conn() as cur:
+            count_sql = f"""SELECT COUNT(*) FROM mplogx_wengfu m {where}"""
+            cur.execute(count_sql, params)
+            total = cur.fetchone()[0]
+
+            data_sql = f"""
+                SELECT m.mpoint_id, n.point_name AS mpoint_name, m.timex,
+                       m.val_float AS val, m.val_bool
+                FROM mplogx_wengfu m
+                LEFT JOIN cj_mpoint_name n ON n.mpoint_id = m.mpoint_id AND n.company = '瓮福'
+                {where}
+                ORDER BY m.timex DESC
+                LIMIT %s OFFSET %s
+            """
+            cur.execute(data_sql, params + [page_size, (page - 1) * page_size])
+            columns = [desc[0] for desc in cur.description]
+            results = []
+            for row in cur.fetchall():
+                d = dict(zip(columns, row))
+                if d.get("timex"):
+                    d["timex"] = d["timex"].strftime("%Y-%m-%d %H:%M:%S")
+                results.append(d)
+
+        return Response({"count": total, "results": results})
+
+    def post(self, request, *args, **kwargs):
+        """导出或图表数据（不分页）"""
+        mpoint_ids = request.data.get("mpoint_ids", [])
+        timex_gte = request.data.get("timex__gte")
+        timex_lte = request.data.get("timex__lte")
+
+        if not mpoint_ids or not timex_gte or not timex_lte:
+            raise ParseError("mpoint_ids, timex__gte, timex__lte are required")
+
+        conditions = []
+        params = []
+        placeholders = ",".join(["%s"] * len(mpoint_ids))
+        conditions.append(f"m.mpoint_id IN ({placeholders})")
+        params.extend(mpoint_ids)
+        conditions.append("m.timex >= %s")
+        params.append(timex_gte)
+        conditions.append("m.timex <= %s")
+        params.append(timex_lte)
+
+        where = "WHERE " + " AND ".join(conditions)
+
+        with self._get_conn() as cur:
+            sql = f"""
+                SELECT m.mpoint_id, n.point_name AS mpoint_name, m.timex, m.val_float AS val, m.val_bool
+                FROM mplogx_wengfu m
+                LEFT JOIN cj_mpoint_name n ON n.mpoint_id = m.mpoint_id AND n.company = '瓮福'
+                {where}
+                ORDER BY m.timex ASC
+            """
+            cur.execute(sql, params)
+            columns = [desc[0] for desc in cur.description]
+            results = []
+            for row in cur.fetchall():
+                d = dict(zip(columns, row))
+                if d.get("timex"):
+                    d["timex"] = d["timex"].strftime("%Y-%m-%d %H:%M:%S")
+                results.append(d)
+
+        return Response(results)
+
+
+class WengfuMpointListAPIView(APIView):
+    """瓮福测点列表（供下拉选择）"""
+    perms_map = {"get": "*"}
+
+    def get(self, request, *args, **kwargs):
+        keyword = request.query_params.get("keyword", "")
+        page = int(request.query_params.get("page", 1))
+        page_size = int(request.query_params.get("page_size", 50))
+        from apps.utils.sql import DbConnection
+        with DbConnection(host='10.0.11.52', user='postgres',
+                          password='zcDsj@2024', database='hfnf', dbtype='pg') as cur:
+            conditions = ["company = '瓮福'"]
+            params = []
+            if keyword:
+                conditions.append("(mpoint_id ILIKE %s OR point_name ILIKE %s)")
+                params.extend([f"%{keyword}%", f"%{keyword}%"])
+            where = "WHERE " + " AND ".join(conditions)
+            cur.execute(f"SELECT COUNT(*) FROM cj_mpoint_name {where}", params)
+            total = cur.fetchone()[0]
+            cur.execute(f"""
+                SELECT mpoint_id, point_name FROM cj_mpoint_name
+                {where} ORDER BY mpoint_id LIMIT %s OFFSET %s
+            """, params + [page_size, (page - 1) * page_size])
+            columns = [desc[0] for desc in cur.description]
+            results = [dict(zip(columns, row)) for row in cur.fetchall()]
+        return Response({"count": total, "results": results})
+
+
 class MpLogxViewSet(CustomListModelMixin, CustomGenericViewSet):
     """
     list: 测点采集数据

apps/hrm/views.py

@@ -391,6 +391,23 @@ class EmployeeViewSet(CustomModelViewSet):
         # 获取模型字段
         model_fields = {f.name: f for f in Employee._meta.fields}
 
+        from django.db.models import CharField, TextField
+        _ABSENT = object()   # 区分“列没出现”与“列为空”
+        _KEEP = object()     # 该字段不可清空时，保留原值
+
+        def empty_value(field_name):
+            """全量覆盖时，空单元格应写入的“清空值”；不可清空的字段返回 _KEEP。"""
+            if field_name == 'belong_dept_id':
+                return None  # 部门 FK 可空
+            mf = model_fields.get(field_name)
+            if mf is None:
+                return _KEEP
+            if getattr(mf, 'null', False):
+                return None
+            if isinstance(mf, (CharField, TextField)) and getattr(mf, 'blank', False):
+                return ''
+            return _KEEP  # 非空且不可为空（如布尔/性别/类别）-> 保留原值
+
         for row_num, row in enumerate(rows[1:], start=2):
             try:
                 data = {}
@@ -419,12 +436,15 @@ class EmployeeViewSet(CustomModelViewSet):
                         data['type'] = TYPE_MAPPING[excel_type]
                     else:
                         raise ParseError(f'第{row_num}行，人员类型"{excel_type}"无效，有效类型：{", ".join(TYPE_MAPPING.keys())}')
-                # 处理部门外键：填了就校验是否存在并赋值；为空时不动（新增场景的必填在下方创建处校验）
+                # 处理部门外键：填了校验并赋值；列存在但为空 -> 清空（新增场景的必填在创建处校验）
-                dept_name = data.pop('belong_dept', None)
+                dept_raw = data.pop('belong_dept', _ABSENT)
-                if dept_name:
+                if dept_raw is not _ABSENT:
-                    if dept_name not in dept_map:
+                    if dept_raw:
-                        raise ParseError(f'第{row_num}行，部门"{dept_name}"不存在')
+                        if dept_raw not in dept_map:
-                    data['belong_dept_id'] = dept_map[dept_name]
+                            raise ParseError(f'第{row_num}行，部门"{dept_raw}"不存在')
+                        data['belong_dept_id'] = dept_map[dept_raw]
+                    else:
+                        data['belong_dept_id'] = None
                 
                 # 数据验证
                 if data.get('phone'):
@@ -454,11 +474,14 @@ class EmployeeViewSet(CustomModelViewSet):
                                 (e for e in name_matches if e.id_number == id_number), None)
 
                         if existing:
-                            # 用 Excel 中填写了值的列覆盖数据库已有数据；空单元格保持原值不变
+                            # 全量覆盖：以本次 Excel 为准。填了值的列覆盖原值；
+                            # 空单元格清空原值（不可为空的字段如布尔/性别/类别则保留原值）。
                             updated_fields = []
                             for field_name, value in data.items():
                                 if value in [None, '']:
-                                    continue
+                                    value = empty_value(field_name)
+                                    if value is _KEEP:
+                                        continue
                                 if getattr(existing, field_name, None) != value:
                                     setattr(existing, field_name, value)
                                     updated_fields.append(field_name)
@@ -470,9 +493,11 @@ class EmployeeViewSet(CustomModelViewSet):
                             created = False
                         else:
                             # 新增人员时所属部门必填
-                            if 'belong_dept_id' not in data:
+                            if not data.get('belong_dept_id'):
                                 raise ParseError(f'第{row_num}行，新增人员时所属部门不能为空')
-                            Employee.objects.create(**data)
+                            # 新增时丢弃空值，交给模型默认值/允许为空
+                            create_data = {k: v for k, v in data.items() if v is not None}
+                            Employee.objects.create(**create_data)
                             created = True
                 except Exception as e:
                     raise
@@ -501,18 +526,23 @@ def convert_field_value(model_field, value, row_num):
 
     # ===== 日期字段 =====
     if isinstance(model_field, CharField):
+        sval = str(value)
         if isinstance(value, str):
             if 'e' in value.lower():
                 try:
-                    return str(int(float(value)))
+                    sval = str(int(float(value)))
                 except ValueError:
                     pass
             if '.' in value:
                 try:
-                    return str(int(float(value)))
+                    sval = str(int(float(value)))
                 except ValueError:
                     pass
-        return str(value)
+        # 超长字段给出明确提示，避免数据库报 value too long
+        if model_field.max_length and len(sval) > model_field.max_length:
+            raise ParseError(
+                f'第{row_num}行，字段 {model_field.verbose_name} 长度超过 {model_field.max_length} 个字符（当前 {len(sval)}）')
+        return sval
            
     if isinstance(model_field, DateField):
 

apps/inm/filters.py

@@ -1,10 +1,16 @@
 from django_filters import rest_framework as filters
 from apps.inm.models import MaterialBatch, MIO
 from django.db.models import Q, Subquery, OuterRef, F
+from apps.qm.defect_grades import effective_defect_grade_q
 
 class MaterialBatchFilter(filters.FilterSet):
     count_canmio__gt = filters.NumberFilter(
         method='filter_count_canmio__gt', label='可发数量大于')
+    defect_grade = filters.NumberFilter(
+        method='filter_defect_grade', label='有效缺陷等级')
+
+    def filter_defect_grade(self, queryset, name, value):
+        return queryset.filter(effective_defect_grade_q(value))
 
     class Meta:
         model = MaterialBatch

apps/inm/models.py

@@ -1,4 +1,6 @@
-from django.db import models
+import json
+
+from django.db import connection, models
 from apps.utils.models import BaseModel, CommonBModel, CommonBDModel, CommonADModel
 from apps.pum.models import Supplier, PuOrder
 from apps.sam.models import Customer, Order
@@ -38,6 +40,113 @@ class MaterialBatch(BaseModel):
     material_ofrom = models.ForeignKey(Material, verbose_name='原料物料', on_delete=models.SET_NULL, null=True, blank=True, related_name='mb_mofrom')
     defect = models.ForeignKey('qm.defect', verbose_name='缺陷', on_delete=models.PROTECT, null=True, blank=True)
 
+    INVENTORY_KEY_FIELDS = (
+        'material',
+        'batch',
+        'warehouse',
+        'state',
+        'defect',
+    )
+
+    @classmethod
+    def _normalize_inventory_lookup(cls, **kwargs):
+        """生成唯一、完整的仓库库存业务键。"""
+        unknown_fields = set(kwargs) - set(cls.INVENTORY_KEY_FIELDS)
+        if unknown_fields:
+            fields = ', '.join(sorted(unknown_fields))
+            raise TypeError(f'不支持的仓库库存定位字段: {fields}')
+
+        required_fields = ('material', 'batch', 'warehouse')
+        missing_fields = [
+            field for field in required_fields
+            if kwargs.get(field) is None
+        ]
+        if missing_fields:
+            fields = ', '.join(missing_fields)
+            raise ValueError(f'仓库库存业务键必须包含: {fields}')
+
+        lookup = {
+            field: kwargs.get(field)
+            for field in cls.INVENTORY_KEY_FIELDS
+        }
+        if lookup['state'] is None:
+            lookup['state'] = cls._meta.get_field('state').get_default()
+        return lookup
+
+    @classmethod
+    def _inventory_advisory_lock_payload(cls, lookup):
+        lock_values = {}
+        for name in cls.INVENTORY_KEY_FIELDS:
+            field = cls._meta.get_field(name)
+            value = lookup[name]
+            if field.is_relation and value is not None:
+                value = getattr(value, 'pk', value)
+            lock_values[field.attname] = value
+        return json.dumps(
+            {'model': cls._meta.label_lower, 'lookup': lock_values},
+            sort_keys=True,
+            ensure_ascii=False,
+            default=str,
+            separators=(',', ':'),
+        )
+
+    @classmethod
+    def locked_get_or_create_inventory(cls, defaults=None, **kwargs):
+        """
+        在事务中按完整业务键获取或创建仓库库存。
+
+        已存在记录使用行锁；首次创建使用 PostgreSQL 事务级 advisory lock，
+        并在取得锁后重新查询，避免并发创建重复库存。
+        """
+        if not connection.in_atomic_block:
+            raise RuntimeError(
+                'locked_get_or_create_inventory 必须在事务中调用'
+            )
+        if connection.vendor != 'postgresql':
+            raise RuntimeError(
+                'locked_get_or_create_inventory 仅支持 PostgreSQL'
+            )
+
+        defaults = defaults or {}
+        lookup = cls._normalize_inventory_lookup(**kwargs)
+        create_defaults = {
+            key: value
+            for key, value in defaults.items()
+            if key not in cls.INVENTORY_KEY_FIELDS
+        }
+
+        rows = list(
+            cls.objects.select_for_update().filter(**lookup)[:2]
+        )
+        if len(rows) > 1:
+            raise RuntimeError(
+                f'{cls.__name__} 数据异常：库存业务键 {lookup} 命中多条'
+            )
+        if rows:
+            return rows[0], False
+
+        lock_payload = cls._inventory_advisory_lock_payload(lookup)
+        with connection.cursor() as cursor:
+            cursor.execute(
+                'SELECT pg_advisory_xact_lock(hashtextextended(%s, 0))',
+                [lock_payload],
+            )
+
+        rows = list(
+            cls.objects.select_for_update().filter(**lookup)[:2]
+        )
+        if len(rows) > 1:
+            raise RuntimeError(
+                f'{cls.__name__} 数据异常：库存业务键 {lookup} 命中多条'
+            )
+        if rows:
+            return rows[0], False
+
+        return cls.objects.create(
+            **lookup,
+            **create_defaults,
+        ), True
+
 
     @property
     def count_mioing(self): 

apps/inm/serializers.py

@@ -14,6 +14,7 @@ from django.db.models import F, Sum, DecimalField
 from server.settings import get_sysconfig
 from apps.wpmw.models import Wpr
 from decimal import Decimal
+from apps.qm.defect_grades import DEFECT_GRADE_NAMES, effective_defect_grade
 
 
 class WareHourseSerializer(CustomModelSerializer):
@@ -49,6 +50,8 @@ class MaterialBatchSerializer(CustomModelSerializer):
         source='supplier', read_only=True)
     material_ = MaterialSerializer(source='material', read_only=True)
     defect_name = serializers.CharField(source="defect.name", read_only=True)
+    defect_grade = serializers.SerializerMethodField()
+    defect_grade_name = serializers.SerializerMethodField()
     count_mioing = serializers.SerializerMethodField(label='正在出入库数量')
 
     class Meta:
@@ -61,6 +64,12 @@ class MaterialBatchSerializer(CustomModelSerializer):
         # 保留 decimal 精度(原 IntegerField 会截断在途量, 导致可发量偏大)
         return instance.count_mioing_anno if hasattr(instance, 'count_mioing_anno') else instance.count_mioing
 
+    def get_defect_grade(self, instance):
+        return effective_defect_grade(instance)
+
+    def get_defect_grade_name(self, instance):
+        return DEFECT_GRADE_NAMES[self.get_defect_grade(instance)]
+
     def to_representation(self, instance):
         ret = super().to_representation(instance)
         if 'count' in ret:
@@ -86,6 +95,15 @@ class MaterialBatchDetailSerializer(CustomModelSerializer):
         source='a_mb', read_only=True, many=True)
     supplier_name = serializers.StringRelatedField(
         source='supplier', read_only=True)
+    defect_name = serializers.CharField(source="defect.name", read_only=True)
+    defect_grade = serializers.SerializerMethodField()
+    defect_grade_name = serializers.SerializerMethodField()
+
+    def get_defect_grade(self, instance):
+        return effective_defect_grade(instance)
+
+    def get_defect_grade_name(self, instance):
+        return DEFECT_GRADE_NAMES[self.get_defect_grade(instance)]
 
     class Meta:
         model = MaterialBatch

apps/inm/services.py

@@ -7,8 +7,10 @@ from apps.wpm.models import WMaterial, BatchSt, BatchLog
 from apps.wpm.services_2 import ana_batch_thread
 from apps.wpmw.models import Wpr
 from apps.qm.models import Ftest, Defect
+from django.db import transaction
 from django.db.models import Count, Q
 
+@transaction.atomic
 def do_out(item: MIOItem, is_reverse: bool = False):
     """
     生产领料到车间
@@ -45,7 +47,7 @@ def do_out(item: MIOItem, is_reverse: bool = False):
     
     if is_zhj:
         try:
-            mb = MaterialBatch.objects.get(
+            mb = MaterialBatch.objects.select_for_update().get(
                 material=item.material, 
                 warehouse=item.warehouse, 
                 batch=item.batch,
@@ -82,7 +84,7 @@ def do_out(item: MIOItem, is_reverse: bool = False):
         mb = None
         if not is_zhj:
             try:
-                mb = MaterialBatch.objects.get(
+                mb = MaterialBatch.objects.select_for_update().get(
                     material=xmaterial,
                     warehouse=item.warehouse,
                     batch=xbatch,
@@ -99,7 +101,7 @@ def do_out(item: MIOItem, is_reverse: bool = False):
 
         if xmaterial.into_wm:
             # 领到车间库存(或工段)
-            wm, new_create = WMaterial.objects.get_or_create(
+            wm, new_create = WMaterial.locked_get_or_create_inventory(
                 batch=xbatch, material=xmaterial,
                 belong_dept=belong_dept, mgroup=mgroup,
                 state=state, defect=defect)
@@ -107,7 +109,7 @@ def do_out(item: MIOItem, is_reverse: bool = False):
                 wm.create_by = do_user
                 wm.batch_ofrom = mb.batch if mb else None
                 wm.material_ofrom = mb.material if mb else None
-            wm.count = wm.count + item.count
+            wm.count = wm.count + xcount
             wm.update_by = do_user
             wm.save()
 
@@ -130,6 +132,7 @@ def do_out(item: MIOItem, is_reverse: bool = False):
     ana_batch_thread(xbatches)
 
 
+@transaction.atomic
 def do_in(item: MIOItem):
     """
     生产入库后更新车间物料
@@ -184,9 +187,9 @@ def do_in(item: MIOItem):
         xbatchs.append(xbatch)
         if xmaterial.into_wm:
             if xwm:
-                wm = xwm
+                wm = WMaterial.objects.select_for_update().get(pk=xwm.pk)
             else:
-                wm_qs = WMaterial.objects.filter(
+                wm_qs = WMaterial.objects.select_for_update().filter(
                     batch=xbatch,
                     material=xmaterial,
                     belong_dept=belong_dept,
@@ -220,7 +223,7 @@ def do_in(item: MIOItem):
         
         # 增加mb
         if not is_zhj:
-            mb, _ = MaterialBatch.objects.get_or_create(
+            mb, _ = MaterialBatch.locked_get_or_create_inventory(
                     material=xmaterial,
                     warehouse=item.warehouse,
                     batch=xbatch,
@@ -255,7 +258,7 @@ def do_in(item: MIOItem):
 
 
     if is_zhj: # 组合件单独处理并且不做追踪单个处理
-        mb, is_created = MaterialBatch.objects.get_or_create(
+        mb, is_created = MaterialBatch.locked_get_or_create_inventory(
                     material=item.material, 
                     warehouse=item.warehouse, 
                     batch=item.batch,
@@ -412,7 +415,7 @@ class InmService:
             state = WMaterial.WM_OK
             if defect and defect.okcate in [Defect.DEFECT_NOTOK]:
                 state = WMaterial.WM_NOTOK
-            mb, _ = MaterialBatch.objects.get_or_create(
+            mb, _ = MaterialBatch.locked_get_or_create_inventory(
                 material=material, 
                 warehouse=warehouse, 
                 batch=batch, 
@@ -433,11 +436,18 @@ class InmService:
                         raise ParseError("出入库与明细数量不一致,操作失败")
                     for mioitemw in mioitemws:
                         if mioitemw.wpr:
-                            Wpr.change_or_new(wpr=mioitemw.wpr, mb=mb, ftest=mioitemw.ftest)
+                            wpr = Wpr.change_or_new(
+                                wpr=mioitemw.wpr,
+                                mb=mb,
+                            )
                         else:
-                            wpr = Wpr.change_or_new(number=mioitemw.number, mb=mb, ftest=mioitemw.ftest)
+                            wpr = Wpr.change_or_new(
+                                number=mioitemw.number,
+                                mb=mb,
+                            )
                             mioitemw.wpr = wpr
                             mioitemw.save()
+                        wpr.apply_ftest(mioitemw.ftest)
 
             elif in_or_out == -1:
                 mb.count = mb.count - change_count
@@ -488,4 +498,3 @@ class InmService:
         # 若该出入库记录已无明细，自动删除
         if not MIOItem.objects.filter(mio=mio).exists():
             mio.delete()
-        

apps/inm/services_daoru.py

@@ -55,7 +55,7 @@ def daoru_mb(path: str):
                 process=process,
                 defaults={"type": type, "name": name, "specification": specification, "model": model, "process": process, "number": ranstr(6), "id": idWorker.get_id()},
             )
-            MaterialBatch.objects.get_or_create(
+            MaterialBatch.locked_get_or_create_inventory(
                 material=material, batch=batch, warehouse=warehouse, defaults={"material": material, "batch": batch, "warehouse": warehouse, "count": count, "id": idWorker.get_id()}
             )
             i = i + 1

apps/inm/tests.py

@@ -1,3 +1,183 @@
-from django.test import TestCase
+from concurrent.futures import ThreadPoolExecutor
+from decimal import Decimal
+from threading import Barrier
+from unittest import skipUnless
 
-# Create your tests here.
+from django.db import connection, connections, transaction
+from django.test import SimpleTestCase, TestCase, TransactionTestCase
+
+from apps.inm.filters import MaterialBatchFilter
+from apps.inm.models import MaterialBatch, WareHouse
+from apps.inm.serializers import MaterialBatchSerializer
+from apps.mtm.models import Material
+from apps.qm.models import Defect
+
+
+class MaterialBatchDefectGradeTests(TestCase):
+    @classmethod
+    def setUpTestData(cls):
+        cls.material = Material.objects.create(name='仓库缺陷等级测试物料')
+        cls.warehouse = WareHouse.objects.create(
+            number='GRADE',
+            name='等级测试仓库',
+            place='测试地点',
+        )
+        cls.defect_b = Defect.objects.create(
+            name='仓库B类缺陷',
+            cate=Defect.cate_list[0],
+            okcate=Defect.DEFECT_OK_B,
+        )
+        cls.notok_without_defect = MaterialBatch.objects.create(
+            material=cls.material,
+            warehouse=cls.warehouse,
+            batch='MB-NOTOK-NONE',
+            count=1,
+            state=20,
+        )
+        cls.normal_with_b_defect = MaterialBatch.objects.create(
+            material=cls.material,
+            warehouse=cls.warehouse,
+            batch='MB-NORMAL-B',
+            count=1,
+            state=10,
+            defect=cls.defect_b,
+        )
+
+    def test_serializer_uses_defect_or_defaults_to_ok_independent_of_state(self):
+        no_defect_data = MaterialBatchSerializer(
+            self.notok_without_defect
+        ).data
+        b_defect_data = MaterialBatchSerializer(
+            self.normal_with_b_defect
+        ).data
+
+        self.assertEqual(no_defect_data['defect_grade'], Defect.DEFECT_OK)
+        self.assertEqual(no_defect_data['defect_grade_name'], '合格')
+        self.assertEqual(b_defect_data['defect_grade'], Defect.DEFECT_OK_B)
+        self.assertEqual(b_defect_data['defect_grade_name'], '合格B类')
+
+    def test_effective_grade_filter_is_independent_of_state(self):
+        ok_items = MaterialBatchFilter(
+            {'defect_grade': Defect.DEFECT_OK},
+            queryset=MaterialBatch.objects.all(),
+        ).qs
+        b_items = MaterialBatchFilter(
+            {'defect_grade': Defect.DEFECT_OK_B},
+            queryset=MaterialBatch.objects.all(),
+        ).qs
+
+        self.assertQuerySetEqual(
+            ok_items,
+            [self.notok_without_defect],
+            transform=lambda item: item,
+        )
+        self.assertQuerySetEqual(
+            b_items,
+            [self.normal_with_b_defect],
+            transform=lambda item: item,
+        )
+
+
+class MaterialBatchInventoryKeyTests(SimpleTestCase):
+    def setUp(self):
+        self.material = Material(id='100', name='测试物料')
+        self.warehouse = WareHouse(
+            id='200',
+            number='TEST',
+            name='测试仓库',
+            place='测试地点',
+        )
+
+    def test_inventory_key_normalizes_omitted_optional_fields(self):
+        omitted = MaterialBatch._normalize_inventory_lookup(
+            material=self.material,
+            batch='BATCH-001',
+            warehouse=self.warehouse,
+        )
+        explicit = MaterialBatch._normalize_inventory_lookup(
+            material=self.material,
+            batch='BATCH-001',
+            warehouse=self.warehouse,
+            state=10,
+            defect=None,
+        )
+
+        self.assertEqual(omitted, explicit)
+        self.assertEqual(
+            MaterialBatch._inventory_advisory_lock_payload(omitted),
+            MaterialBatch._inventory_advisory_lock_payload(explicit),
+        )
+
+    def test_inventory_key_requires_material_batch_and_warehouse(self):
+        required_values = {
+            'material': self.material,
+            'batch': 'BATCH-001',
+            'warehouse': self.warehouse,
+        }
+
+        for field in required_values:
+            with self.subTest(field=field):
+                lookup = required_values.copy()
+                lookup[field] = None
+                with self.assertRaisesRegex(ValueError, field):
+                    MaterialBatch._normalize_inventory_lookup(**lookup)
+
+    def test_inventory_key_rejects_unknown_fields(self):
+        with self.assertRaisesRegex(TypeError, 'supplier'):
+            MaterialBatch._normalize_inventory_lookup(
+                material=self.material,
+                batch='BATCH-001',
+                warehouse=self.warehouse,
+                supplier=None,
+            )
+
+
+@skipUnless(
+    connection.vendor == 'postgresql',
+    'advisory lock concurrency test requires PostgreSQL',
+)
+class MaterialBatchConcurrencyTests(TransactionTestCase):
+    def test_concurrent_first_create_uses_one_inventory_record(self):
+        material = Material.objects.create(name='仓库并发测试物料')
+        warehouse = WareHouse.objects.create(
+            number='CONCURRENT',
+            name='并发测试仓库',
+            place='测试地点',
+        )
+        barrier = Barrier(2)
+
+        def create_inventory():
+            connections.close_all()
+            try:
+                barrier.wait()
+                with transaction.atomic():
+                    mb, created = (
+                        MaterialBatch.locked_get_or_create_inventory(
+                            material=material,
+                            batch='CONCURRENT-001',
+                            warehouse=warehouse,
+                            defaults={'count': Decimal('0')},
+                        )
+                    )
+                    mb.count += Decimal('1')
+                    mb.save(update_fields=['count'])
+                    return created
+            finally:
+                connections.close_all()
+
+        with ThreadPoolExecutor(max_workers=2) as executor:
+            created_results = list(executor.map(
+                lambda _: create_inventory(),
+                range(2),
+            ))
+
+        queryset = MaterialBatch.objects.filter(
+            material=material,
+            batch='CONCURRENT-001',
+            warehouse=warehouse,
+            state=10,
+            defect=None,
+        )
+        self.assertEqual(queryset.count(), 1)
+        self.assertEqual(queryset.get().count, Decimal('2'))
+        self.assertCountEqual(created_results, [True, False])

apps/inm/views.py

@@ -60,7 +60,7 @@ class MaterialBatchViewSet(ListModelMixin, CustomGenericViewSet):
     queryset = MaterialBatch.objects.filter(count__gt=0)
     serializer_class = MaterialBatchSerializer
     retrieve_serializer_class = MaterialBatchDetailSerializer
-    select_related_fields = ['warehouse', 'material', 'supplier']
+    select_related_fields = ['warehouse', 'material', 'supplier', 'defect']
     filterset_class = MaterialBatchFilter
     search_fields = ['material__name', 'material__number',
                      'material__model', 'material__specification', 'batch']
@@ -255,7 +255,8 @@ class MIOViewSet(CustomModelViewSet):
 
         提交
         """
-        ins:MIO = self.get_object()
+        current = self.get_object()
+        ins = MIO.objects.select_for_update().get(pk=current.pk)
         if ins.inout_date is None:
             raise ParseError('出入库日期未填写')
         if ins.state != MIO.MIO_CREATE:
@@ -276,7 +277,8 @@ class MIOViewSet(CustomModelViewSet):
 
         撤回
         """
-        ins = self.get_object()
+        current = self.get_object()
+        ins = MIO.objects.select_for_update().get(pk=current.pk)
         user = self.request.user
         if ins.state != MIO.MIO_SUBMITED:
             raise ParseError('记录状态异常')
@@ -586,4 +588,3 @@ class MIOItemwViewSet(CustomModelViewSet):
         if ftest:
             ftest.delete()
         self.cal_mioitem_count(mioitem)
-    

apps/mtm/admin.py

@@ -5,7 +5,7 @@ from apps.mtm.models import Material, Shift, Mgroup, Process
 
 @admin.register(Process)
 class ProcessAdmin(admin.ModelAdmin):
-    list_display = ('id', 'name', 'cate', 'sort', 'into_wm_mgroup', 'store_notok', 'batch_append_equip', 'mlog_need_ticket')
+    list_display = ('id', 'name', 'cate', 'sort', 'into_wm_scope', 'store_notok', 'batch_append_equip', 'mlog_need_ticket')
 
 
 @admin.register(Material)

apps/mtm/migrations/0068_process_into_wm_scope.py

@@ -0,0 +1,36 @@
+from django.db import migrations, models
+
+
+def migrate_scope_forward(apps, schema_editor):
+    Process = apps.get_model('mtm', 'Process')
+    Process.objects.filter(into_wm_mgroup=True).update(into_wm_scope=10)
+
+
+def migrate_scope_backward(apps, schema_editor):
+    Process = apps.get_model('mtm', 'Process')
+    Process.objects.filter(into_wm_scope=10).update(into_wm_mgroup=True)
+    Process.objects.exclude(into_wm_scope=10).update(into_wm_mgroup=False)
+
+
+class Migration(migrations.Migration):
+
+    dependencies = [
+        ('mtm', '0067_material_test_mode_in'),
+    ]
+
+    operations = [
+        migrations.AddField(
+            model_name='process',
+            name='into_wm_scope',
+            field=models.PositiveSmallIntegerField(
+                choices=[(10, '工段'), (20, '部门'), (30, '全局')],
+                default=20,
+                verbose_name='产出库存归属',
+            ),
+        ),
+        migrations.RunPython(migrate_scope_forward, migrate_scope_backward),
+        migrations.RemoveField(
+            model_name='process',
+            name='into_wm_mgroup',
+        ),
+    ]

apps/mtm/models.py

@@ -9,6 +9,28 @@ from django.utils import timezone
 from django.db.models import Sum, Q, Value, F, ExpressionWrapper, DecimalField
 from django.db.models.functions import Coalesce
 
+
+class WmScope(models.IntegerChoices):
+    MGROUP = 10, '工段'
+    DEPT = 20, '部门'
+    GLOBAL = 30, '全局'
+
+    @classmethod
+    def resolve_location(cls, scope, mgroup):
+        """将库存归属策略统一解析为 WMaterial 的位置字段。"""
+        scope = cls(scope)
+        if scope == cls.GLOBAL:
+            return {'mgroup': None, 'belong_dept': None}
+        if mgroup is None:
+            raise ParseError('工段或部门库存必须提供工段')
+        if scope == cls.MGROUP:
+            return {'mgroup': mgroup, 'belong_dept': mgroup.belong_dept}
+        return {'mgroup': None, 'belong_dept': mgroup.belong_dept}
+
+    @classmethod
+    def requires_mgroup(cls, scope):
+        return cls(scope) == cls.MGROUP
+
 class Process(CommonBModel):
     """
     TN:工序
@@ -29,7 +51,8 @@ class Process(CommonBModel):
     instruction = models.ForeignKey(
         File, verbose_name='指导书', on_delete=models.SET_NULL, null=True, blank=True)
     instruction_content = models.TextField('指导书内容', null=True, blank=True)
-    into_wm_mgroup = models.BooleanField('交接到工段', default=False)
+    into_wm_scope = models.PositiveSmallIntegerField(
+        '产出库存归属', choices=WmScope.choices, default=WmScope.DEPT)
     store_notok = models.BooleanField('不合格品是否入库', default=False)
     batch_append_equip = models.BooleanField('批号追加设备', default=False)
     mlog_need_ticket = models.BooleanField('日志提交是否需要审批', default=False)

apps/mtm/serializers.py

@@ -3,11 +3,12 @@ from apps.mtm.models import (Shift, Material, Mgroup, Team, Goal,
                              Process, Route, TeamMember, RoutePack, Srule, RouteMat)
 from apps.utils.constants import EXCLUDE_FIELDS, EXCLUDE_FIELDS_BASE, EXCLUDE_FIELDS_DEPT
 from rest_framework import serializers
-from rest_framework.exceptions import ValidationError, ParseError
+from rest_framework.exceptions import ValidationError, ParseError, PermissionDenied
 from apps.system.models import Dept, UserPost
 from django.db import transaction
 from apps.wf.serializers import TicketSimpleSerializer
 from apps.system.serializers import FileSerializer
+from apps.utils.permission import has_perm
 
 
 class ShiftSerializer(CustomModelSerializer):
@@ -177,6 +178,23 @@ class RoutePackCopySerializer(serializers.Serializer):
     material_in = serializers.CharField(label='原料ID')
     material_out = serializers.CharField(label='产品ID')
 
+
+class RouteMaterialOutCreateSerializer(serializers.Serializer):
+    """工艺路线内快捷新建半成品时允许录入的物料字段。"""
+    name = serializers.CharField(label='名称', max_length=50)
+    number = serializers.CharField(label='编号', max_length=100, required=False, allow_blank=True, allow_null=True)
+    cate = serializers.CharField(label='大类', max_length=20, required=False, allow_blank=True)
+    model = serializers.CharField(label='型号', max_length=100, required=False, allow_blank=True, allow_null=True)
+    specification = serializers.CharField(label='规格', max_length=100, required=False, allow_blank=True, allow_null=True)
+    unit = serializers.CharField(label='基准计量单位', max_length=10, required=False, default='个')
+    test_mode_in = serializers.ChoiceField(
+        label='入库检验方式',
+        choices=(Material.MA_TEST_EXEMPT, Material.MA_TEST_REQUIRED),
+        required=False,
+        default=Material.MA_TEST_REQUIRED,
+    )
+
+
 class RouteSerializer(CustomModelSerializer):
     name = serializers.CharField(source='__str__', read_only=True)
     material_ = MaterialSerializer(source='material', read_only=True)
@@ -194,6 +212,7 @@ class RouteSerializer(CustomModelSerializer):
     material_out_is_hidden = serializers.BooleanField(
         source='material_out.is_hidden', read_only=True)
     material_out_tracking = serializers.IntegerField(write_only=True,  required=False, allow_null=True)
+    material_out_create = RouteMaterialOutCreateSerializer(write_only=True, required=False)
 
     class Meta:
         model = Route
@@ -201,6 +220,10 @@ class RouteSerializer(CustomModelSerializer):
         read_only_fields = EXCLUDE_FIELDS
 
     def validate(self, attrs):
+        if attrs.get('material_out') and attrs.get('material_out_create'):
+            raise serializers.ValidationError(
+                '选择已有输出物料和新建半成品不可同时提交'
+            )
         if attrs.get('routepack', None):
             attrs['material'] = attrs['routepack'].material
         if 'mgroup' in attrs and attrs['mgroup']:
@@ -262,12 +285,30 @@ class RouteSerializer(CustomModelSerializer):
                                                                                           })
         return material_out
 
+    def create_material_out(self, material_data, process, tracking):
+        if self.request and not has_perm(self.request.user, ['material.create']):
+            raise PermissionDenied('无新建物料权限')
+        material_data.update({
+            'type': Material.MA_TYPE_HALFGOOD,
+            'process': process.id,
+            'tracking': tracking,
+            'is_hidden': False,
+        })
+        serializer = MaterialSerializer(data=material_data, context=self.context)
+        serializer.is_valid(raise_exception=True)
+        return serializer.save()
+    
+    @transaction.atomic
     def create(self, validated_data):
         process = validated_data['process']
         routepack = validated_data.get('routepack', None)
+        material_out_create = validated_data.pop('material_out_create', None)
         material_out_tracking = validated_data.pop("material_out_tracking", Material.MA_TRACKING_BATCH)
         if material_out_tracking is None:
             material_out_tracking = Material.MA_TRACKING_BATCH
+        if material_out_create:
+            validated_data['material_out'] = self.create_material_out(
+                material_out_create, process, material_out_tracking)
         if routepack:
             pass
             # if Route.objects.filter(routepack=routepack, process=process).exists():
@@ -308,12 +349,17 @@ class RouteSerializer(CustomModelSerializer):
             # raise ParseError(f"该工艺步骤已存在-{msg}")
         return instance
 
+    @transaction.atomic
     def update(self, instance, validated_data):
         validated_data.pop('material', None)
-        process = validated_data.pop('process', None)
+        process = validated_data.pop('process', None) or instance.process
+        material_out_create = validated_data.pop('material_out_create', None)
         material_out_tracking = validated_data.pop("material_out_tracking", Material.MA_TRACKING_BATCH)
         if material_out_tracking is None:
             material_out_tracking = Material.MA_TRACKING_BATCH
+        if material_out_create:
+            validated_data['material_out'] = self.create_material_out(
+                material_out_create, process, material_out_tracking)
 
         if not validated_data.get('material_out', instance.material_out) and not instance.material:
             raise ParseError('未指定输出物料，且工艺包未设置产品无法自动生成；请指定输出物料，或先在工艺包基本信息中选择产品')

apps/mtm/tests.py

@@ -1,3 +1,69 @@
 from django.test import TestCase
 
-# Create your tests here.
+from apps.mtm.models import Material, Process, Route, RoutePack
+from apps.mtm.serializers import RouteSerializer
+
+
+class RouteMaterialOutCreateSerializerTests(TestCase):
+    def setUp(self):
+        self.process = Process.objects.create(name='切割')
+        self.product = Material.objects.create(
+            name='产品A',
+            number='CP-A',
+            type=Material.MA_TYPE_GOOD,
+            process=self.process,
+        )
+        self.material_in = Material.objects.create(
+            name='原料A',
+            number='YL-A',
+            type=Material.MA_TYPE_MAINSO,
+        )
+        self.routepack = RoutePack.objects.create(name='产品A工艺', material=self.product)
+
+    def test_create_route_and_visible_halfgood_in_one_serializer_save(self):
+        serializer = RouteSerializer(data={
+            'routepack': self.routepack.id,
+            'process': self.process.id,
+            'material_in': self.material_in.id,
+            'material_out_tracking': Material.MA_TRACKING_SINGLE,
+            'material_out_create': {
+                'name': '分支半成品A',
+                'number': 'BCP-A',
+                'specification': 'S1',
+                'model': 'M1',
+                'unit': '件',
+            },
+        })
+
+        serializer.is_valid(raise_exception=True)
+        route = serializer.save()
+
+        material_out = route.material_out
+        self.assertEqual(material_out.type, Material.MA_TYPE_HALFGOOD)
+        self.assertEqual(material_out.process, self.process)
+        self.assertEqual(material_out.tracking, Material.MA_TRACKING_SINGLE)
+        self.assertFalse(material_out.is_hidden)
+        self.assertEqual(material_out.name, '分支半成品A')
+        self.assertEqual(material_out.unit, '件')
+        self.assertTrue(Route.objects.filter(pk=route.pk, material_out=material_out).exists())
+
+    def test_existing_and_new_material_out_are_mutually_exclusive(self):
+        existing = Material.objects.create(
+            name='已有半成品',
+            number='BCP-OLD',
+            type=Material.MA_TYPE_HALFGOOD,
+            process=self.process,
+        )
+        serializer = RouteSerializer(data={
+            'routepack': self.routepack.id,
+            'process': self.process.id,
+            'material_in': self.material_in.id,
+            'material_out': existing.id,
+            'material_out_create': {
+                'name': '新半成品',
+                'number': 'BCP-NEW',
+            },
+        })
+
+        self.assertFalse(serializer.is_valid())
+        self.assertIn('non_field_errors', serializer.errors)

apps/qm/defect_grades.py

@@ -0,0 +1,42 @@
+from django.db.models import Q
+
+
+DEFECT_OK = 10
+DEFECT_OK_B = 20
+DEFECT_NOTOK = 30
+
+DEFECT_GRADE_CHOICES = (
+    (DEFECT_OK, "合格"),
+    (DEFECT_OK_B, "合格B类"),
+    (DEFECT_NOTOK, "不合格"),
+)
+DEFECT_GRADE_NAMES = dict(DEFECT_GRADE_CHOICES)
+
+
+def effective_defect_grade(instance, notok_sign_field=None):
+    """Return the inventory grade without coupling it to inventory state."""
+    defect = getattr(instance, "defect", None)
+    if defect is not None:
+        return defect.okcate
+    if notok_sign_field and getattr(instance, notok_sign_field, None):
+        return DEFECT_NOTOK
+    return DEFECT_OK
+
+
+def effective_defect_grade_q(value, notok_sign_field=None):
+    """Build an index-friendly query matching ``effective_defect_grade``."""
+    explicit_grade = Q(defect__okcate=value)
+    without_defect = Q(defect__isnull=True)
+
+    if not notok_sign_field:
+        return explicit_grade | without_defect if value == DEFECT_OK else explicit_grade
+
+    has_legacy_sign = (
+        Q(**{f"{notok_sign_field}__isnull": False})
+        & ~Q(**{notok_sign_field: ""})
+    )
+    if value == DEFECT_OK:
+        return explicit_grade | (without_defect & ~has_legacy_sign)
+    if value == DEFECT_NOTOK:
+        return explicit_grade | (without_defect & has_legacy_sign)
+    return explicit_grade

apps/qm/filters.py

@@ -48,7 +48,7 @@ class FtestWorkFilter(filters.FilterSet):
             "material": ["exact"],
             "wm": ["exact", "isnull"],
             "mb": ["exact", "isnull"],
-            "batch": ["exact"],
+            "batch": ["exact", "contains"],
             "type": ["exact"],
             "type2": ["exact"],
             "shift": ["exact"]           

apps/qm/models.py

@@ -8,19 +8,25 @@ from django.utils.translation import gettext_lazy as _
 from django.db import transaction
 from django.db.models import Sum
 from rest_framework.exceptions import ParseError
+from apps.qm.defect_grades import (
+    DEFECT_GRADE_CHOICES,
+    DEFECT_NOTOK as GRADE_NOTOK,
+    DEFECT_OK as GRADE_OK,
+    DEFECT_OK_B as GRADE_OK_B,
+)
 
 class Defect(CommonAModel):
     """TN:缺陷项"""
-    DEFECT_OK = 10
+    DEFECT_OK = GRADE_OK
-    DEFECT_OK_B  = 20
+    DEFECT_OK_B = GRADE_OK_B
-    DEFECT_NOTOK = 30
+    DEFECT_NOTOK = GRADE_NOTOK
     cate_list = ["尺寸", "外观", "内质", "性能"]
     name = models.CharField(max_length=50, verbose_name="名称")
     code = models.CharField(max_length=50, verbose_name="标识", null=True, blank=True)
     cate = models.CharField(max_length=50, verbose_name="分类", help_text=str(cate_list))
     okcate= models.PositiveSmallIntegerField(verbose_name="不合格分类", 
-                                               choices=((DEFECT_OK, "合格"), (DEFECT_OK_B, "合格B类"), (DEFECT_NOTOK, "不合格")), 
+                                               choices=DEFECT_GRADE_CHOICES,
-                                               default=DEFECT_NOTOK)
+                                               default=GRADE_NOTOK)
     note = models.TextField('备注', null=True, blank=True)
     
     def __str__(self):

apps/qm/serializers.py

@@ -274,6 +274,7 @@ class FtestWorkSerializer(CustomModelSerializer):
     material_cate = serializers.CharField(source='material.cate', read_only=True)
     mb_ = MaterialBatchDetailSerializer(source='mb', read_only=True)
     test_user_name = serializers.CharField(source='test_user.name', read_only=True)
+    submit_user_name = serializers.CharField(source='submit_user.name', read_only=True)
 
     class Meta:
         model = FtestWork

apps/qm/services.py

@@ -7,6 +7,7 @@ from apps.wf.models import Ticket
 from apps.qm.models import NotOkOption, Defect
 from apps.wpm.services_2 import ana_batch_thread
 from apps.inm.models import MaterialBatch
+from django.db import transaction
 
 def ftestwork_submit_validate(ins: FtestWork):
     wm:WMaterial = ins.wm
@@ -21,8 +22,15 @@ def ftestwork_submit_validate(ins: FtestWork):
                 raise ParseError("不合格数不可大于批次数量")
 
 
+@transaction.atomic
 def ftestwork_submit(ins:FtestWork, user: User):
-    wm:WMaterial = ins.wm
+    ins = FtestWork.objects.select_for_update().get(pk=ins.pk)
+    if ins.submit_time is not None:
+        raise ParseError('该检验工作已提交')
+    wm = (
+        WMaterial.objects.select_for_update().get(pk=ins.wm_id)
+        if ins.wm_id else None
+    )
     fwd_qs = FtestworkDefect.objects.filter(ftestwork=ins)
     if wm and ins.need_update_wm:
         if ins.qct is None and not fwd_qs.exists():
@@ -46,7 +54,7 @@ def ftestwork_submit(ins:FtestWork, user: User):
                             need_move_count = need_move_count + v
                 count_ok = ins.count_ok - need_move_count
                 if count_ok > 0:
-                    wm, new_create = WMaterial.objects.get_or_create(
+                    wm, new_create = WMaterial.locked_get_or_create_inventory(
                         material=wm.material,
                         batch=wm.batch,
                         mgroup=wm.mgroup,
@@ -77,7 +85,7 @@ def ftestwork_submit(ins:FtestWork, user: User):
                         astate = WMaterial.WM_NOTOK
                         if NotOkOption.get_extra_info(notok_sign)['cate'] == 'ok_b':
                             astate = WMaterial.WM_OK
-                        wm_n, new_create = WMaterial.objects.get_or_create(
+                        wm_n, new_create = WMaterial.locked_get_or_create_inventory(
                             material=wm.material,
                             batch=wm.batch,
                             mgroup=wm.mgroup,
@@ -110,7 +118,7 @@ def ftestwork_submit(ins:FtestWork, user: User):
                     wmstate = WMaterial.WM_OK
                     if item.defect.okcate == Defect.DEFECT_NOTOK:
                         wmstate = WMaterial.WM_NOTOK
-                    wmx, new_create = WMaterial.objects.get_or_create(
+                    wmx, new_create = WMaterial.locked_get_or_create_inventory(
                         material=wm.material,
                         batch=wm.batch,
                         mgroup=wm.mgroup,
@@ -127,7 +135,7 @@ def ftestwork_submit(ins:FtestWork, user: User):
                         wmx.save()
     
     if ins.mb:
-        mb:MaterialBatch = ins.mb
+        mb = MaterialBatch.objects.select_for_update().get(pk=ins.mb_id)
         for item in fwd_qs:
             item:FtestworkDefect = item
             if item.count > 0:
@@ -138,7 +146,7 @@ def ftestwork_submit(ins:FtestWork, user: User):
                 mbstate = WMaterial.WM_OK
                 if item.defect.okcate == Defect.DEFECT_NOTOK:
                     mbstate = WMaterial.WM_NOTOK
-                mbx, new_create = MaterialBatch.objects.get_or_create(
+                mbx, new_create = MaterialBatch.locked_get_or_create_inventory(
                     material=mb.material, 
                     warehouse=mb.warehouse, 
                     batch=mb.batch, 
@@ -158,8 +166,15 @@ def ftestwork_submit(ins:FtestWork, user: User):
     ana_batch_thread(xbatchs=[ins.batch])
 
 
+@transaction.atomic
 def ftestwork_revert(ins: FtestWork):
-    wm:WMaterial = ins.wm
+    ins = FtestWork.objects.select_for_update().get(pk=ins.pk)
+    if ins.submit_time is None:
+        raise ParseError('该检验工作未提交')
+    wm = (
+        WMaterial.objects.select_for_update().get(pk=ins.wm_id)
+        if ins.wm_id else None
+    )
     if wm and ins.need_update_wm:
         fwd_qs = FtestworkDefect.objects.filter(ftestwork=ins)
         for item in fwd_qs:

apps/qm/views.py

@@ -20,6 +20,7 @@ from django.db import transaction
 from apps.qm.services import ftestwork_submit, ftestwork_revert
 from apps.wpm.services_2 import ana_batch_thread
 from apps.wf.models import State
+from django.db.models.functions import Length
 # Create your views here.
 
 class DefectViewSet(CustomModelViewSet):
@@ -43,12 +44,12 @@ class QctViewSet(CustomModelViewSet):
 
     检测模板
     """
-    queryset = Qct.objects.all()
+    queryset = Qct.objects.annotate(_number_len=Length("number")).order_by("_number_len", "number")
     serializer_class = QctSerializer
     retrieve_serializer_class = QctDetailSerializer
     filterset_class = QctFilter
     search_fields = ["name", "number"]
-    ordering = ["number", "-update_time"]
+    ordering = []
     ordering_fields = ["number", "create_time", "update_time"]
 
     @action(methods=['post'], detail=False, perms_map={'post': '*'}, serializer_class=QctGetSerializer)

apps/system/views.py

@@ -651,6 +651,7 @@ class FileViewSet(BulkCreateModelMixin, RetrieveModelMixin, CustomListModelMixin
 class ApkViewSet(MyLoggingMixin, CustomListModelMixin, BulkCreateModelMixin, GenericViewSet):
     perms_map = {'get': '*', 'post': 'apk.upload'}
     serializer_class = ApkSerializer
+    filter_backends = []
 
     def get_authenticators(self):
         if self.request.method == 'GET':

apps/third/views.py

@@ -69,6 +69,7 @@ class SpeakerViewSet(CustomGenericViewSet):
     """
     perms_map = {}
     serializer_class = serializers.Serializer
+    filter_backends = []
 
     @action(methods=['get'], detail=False,
             permission_classes=[IsAuthenticated])
@@ -125,6 +126,7 @@ class XxTestView(APIView):
 class XxCommonViewSet(CreateModelMixin, CustomGenericViewSet):
     perms_map = {'post': '*'}
     serializer_class = RequestCommonSerializer
+    filter_backends = []
 
     def create(self, request, *args, **kwargs):
         """
@@ -258,6 +260,7 @@ class KingCommonViewSet(CreateModelMixin, CustomGenericViewSet):
 class DhCommonViewSet(CreateModelMixin, CustomGenericViewSet):
     perms_map = {'post': '*'}
     serializer_class = RequestCommonSerializer
+    filter_backends = []
 
     def create(self, request, *args, **kwargs):
         """

apps/utils/management/commands/build_swagger.py

@@ -0,0 +1,71 @@
+import json
+import os
+from io import StringIO
+from pathlib import Path
+
+from django.conf import settings
+from django.core.management import BaseCommand, CommandError, call_command
+
+
+class Command(BaseCommand):
+    help = "生成供 Swagger UI 和 ReDoc 使用的静态 Swagger JSON"
+
+    def add_arguments(self, parser):
+        parser.add_argument(
+            "--output",
+            help="输出路径，默认使用 settings.SWAGGER_SCHEMA_PATH",
+        )
+        parser.add_argument(
+            "--url",
+            help="文档中的 API 根地址，默认使用 settings.BASE_URL",
+        )
+
+    def handle(self, *args, **options):
+        target = Path(options["output"] or settings.SWAGGER_SCHEMA_PATH)
+        if not target.is_absolute():
+            target = Path(settings.BASE_DIR) / target
+        target = target.resolve()
+        target.parent.mkdir(parents=True, exist_ok=True)
+
+        temporary = target.with_name(f".{target.name}.{os.getpid()}.tmp")
+        try:
+            output = StringIO()
+            call_command(
+                "generate_swagger",
+                "-",
+                format="json",
+                api_url=options["url"] or settings.BASE_URL,
+                mock=True,
+                verbosity=0,
+                stdout=output,
+            )
+            content = output.getvalue()
+            schema = json.loads(content)
+            if schema.get("swagger") != "2.0" or not schema.get("paths"):
+                raise CommandError("生成的 Swagger 文档缺少版本或接口路径")
+
+            content = json.dumps(
+                schema,
+                ensure_ascii=False,
+                separators=(",", ":"),
+            )
+            temporary.write_text(content, encoding="utf-8")
+            os.replace(temporary, target)
+        except Exception as exc:
+            if isinstance(exc, CommandError):
+                raise
+            raise CommandError(f"生成 Swagger 文档失败：{exc}") from exc
+        finally:
+            temporary.unlink(missing_ok=True)
+
+        operation_count = sum(
+            method.lower() in {"get", "post", "put", "patch", "delete"}
+            for path in schema["paths"].values()
+            for method in path
+        )
+        self.stdout.write(
+            self.style.SUCCESS(
+                f"Swagger文档已生成：{target} "
+                f"（{len(schema['paths'])}个路径，{operation_count}个操作）"
+            )
+        )

apps/utils/mixins.py

@@ -207,12 +207,28 @@ class CustomRetrieveModelMixin(RetrieveModelMixin):
 
 class CustomListModelMixin(ListModelMixin):
     
-    @swagger_auto_schema(manual_parameters=[
+    @swagger_auto_schema(
-        openapi.Parameter(name="query", in_=openapi.IN_QUERY, description="定制返回数据",
+        operation_description=(
-                          type=openapi.TYPE_STRING, required=False),
+            "通用列表接口用于记录或目录浏览以及逐条追溯。跨时间范围的产量、良率、缺陷、"
-        openapi.Parameter(name="with_children", in_=openapi.IN_QUERY, description="带有children(yes/no/count)",
+            "库存、绩效和趋势等统计聚合，优先查询 BI dataset 目录并执行匹配的数据集。"
-                          type=openapi.TYPE_STRING, required=False),
+        ),
-    ])
+        manual_parameters=[
+            openapi.Parameter(
+                name="query",
+                in_=openapi.IN_QUERY,
+                description="django-restql 返回字段裁剪表达式",
+                type=openapi.TYPE_STRING,
+                required=False,
+            ),
+            openapi.Parameter(
+                name="with_children",
+                in_=openapi.IN_QUERY,
+                description="带有children(yes/no/count)",
+                type=openapi.TYPE_STRING,
+                required=False,
+            ),
+        ],
+    )
     def list(self, request, *args, **kwargs):
 
         queryset = self.filter_queryset(self.get_queryset())

apps/utils/models.py

@@ -31,7 +31,27 @@ class SoftDeletableQuerySetMixin(object):
             return super(SoftDeletableQuerySetMixin, self).delete()
 
 
-class SoftDeletableQuerySet(SoftDeletableQuerySetMixin, QuerySet):
+class SnowflakeQuerySet(QuerySet):
+    """为 BaseModel 的批量创建统一准备雪花主键。"""
+
+    def bulk_create(self, objs, *args, **kwargs):
+        objs = list(objs)
+        for obj in objs:
+            if not obj.pk:
+                obj.pk = str(idWorker.get_id())
+            else:
+                obj.pk = str(obj.pk)
+        return super().bulk_create(objs, *args, **kwargs)
+
+
+class SnowflakeManager(models.Manager.from_queryset(SnowflakeQuerySet)):
+    pass
+
+
+class SoftDeletableQuerySet(
+    SoftDeletableQuerySetMixin,
+    SnowflakeQuerySet,
+):
     pass
 
 
@@ -111,6 +131,8 @@ class BaseModel(models.Model):
     is_deleted = models.BooleanField(
         default=False, verbose_name='删除标记', help_text='删除标记')
 
+    objects = SnowflakeManager()
+
     class Meta:
         abstract = True
 
@@ -154,8 +176,12 @@ class BaseModel(models.Model):
     @classmethod
     def locked_get_or_create(cls, defaults: dict, **kwargs):
         """
-        仅用于事务内
+        仅用于事务内锁定已存在的记录。
-        并发安全的 get_or_create
+
+        PostgreSQL 无法通过 select_for_update 锁定不存在的记录，因此该方法
+        不保证首次创建并发安全。需要防止首次重复创建的业务应提供稳定的业务
+        键并使用专用 advisory lock；车间库存使用
+        WMaterial.locked_get_or_create_inventory。
         """
         if not connection.in_atomic_block:
             raise RuntimeError("locked_get_or_create 必须在事务中调用")
@@ -185,7 +211,10 @@ class BaseModel(models.Model):
         is_create = False
         if not self.id:
             is_create = True
-            self.id = idWorker.get_id()
+            self.id = str(idWorker.get_id())
+        else:
+            # BaseModel 的主键是 CharField，保证新建后内存值与数据库回读值类型一致。
+            self.id = str(self.id)
 
         old_parent = None
         need_handle_parent = False
@@ -205,7 +234,7 @@ class BaseModel(models.Model):
         except IntegrityError as e:
             if is_create:
                 time.sleep(0.01)
-                self.id = idWorker.get_id()
+                self.id = str(idWorker.get_id())
                 ins = super().save(*args, **kwargs)
             raise e
         # 处理父级

apps/utils/swagger.py

@@ -0,0 +1,367 @@
+import re
+from pathlib import Path
+
+from django.apps import apps
+from django.conf import settings
+from django.http import FileResponse, JsonResponse
+from drf_yasg import openapi
+from drf_yasg.inspectors import FieldInspector, SwaggerAutoSchema
+from drf_yasg.inspectors.base import NotHandled
+
+
+CRUD_SUMMARIES = {
+    "list": "查询{resource}列表",
+    "retrieve": "查询{resource}详情",
+    "create": "新增{resource}",
+    "update": "更新{resource}",
+    "partial_update": "部分更新{resource}",
+    "destroy": "删除{resource}",
+}
+
+TAG_NAMES = {
+    "am": "区域与准入管理",
+    "asm": "资产管理",
+    "cm": "标签管理",
+    "cms": "内容管理",
+    "develop": "开发工具",
+    "ecm": "事件管理",
+    "edu": "培训考试",
+    "em": "设备管理",
+    "enm": "能源管理",
+    "inm": "库存管理",
+    "mpr": "物资申购与领用",
+    "mtm": "物料与工艺管理",
+    "ofm": "办公管理",
+    "opm": "作业许可",
+    "pm": "生产任务管理",
+    "pum": "采购管理",
+    "qm": "质量管理",
+    "rem": "研发项目管理",
+    "rpm": "相关方管理",
+    "sam": "销售管理",
+    "third": "第三方集成",
+    "utils": "通用工具",
+    "wpm": "生产管理",
+    "wpmw": "动态产品管理",
+    "file": "文件管理",
+}
+
+FIELD_NAMES = {
+    "id": "主键ID",
+    "ids": "主键ID列表",
+    "access": "访问令牌",
+    "refresh": "刷新令牌",
+    "password_check": "密码确认",
+    "base64": "Base64数据",
+    "server_time": "服务器时间",
+    "timezone": "时区",
+    "next": "下一页",
+    "previous": "上一页",
+    "results": "结果列表",
+    "detail": "详情",
+    "items": "明细列表",
+    "files": "附件列表",
+    "echart_options": "图表配置",
+    "tdata_list": "数据列表",
+    "page": "页码",
+    "page_size": "每页数量",
+    "ordering": "排序字段",
+    "querys": "查询条件列表",
+    "annotate_field_list": "聚合字段列表",
+}
+
+FIELD_TOKENS = {
+    "name": "名称",
+    "code": "编码",
+    "number": "编号",
+    "description": "说明",
+    "note": "备注",
+    "employee": "人员",
+    "user": "用户",
+    "leader": "负责人",
+    "manager": "负责人",
+    "keeper": "保管人",
+    "participant": "参与人",
+    "post": "岗位",
+    "dept": "部门",
+    "belong": "所属",
+    "create": "创建",
+    "update": "更新",
+    "submit": "提交",
+    "handle": "处理",
+    "test": "检验",
+    "material": "物料",
+    "supplier": "供应商",
+    "defect": "缺陷",
+    "equipment": "设备",
+    "warehouse": "仓库",
+    "process": "工序",
+    "operation": "操作",
+    "state": "状态",
+    "cate": "分类",
+    "type": "类型",
+    "area": "区域",
+    "team": "班组",
+    "shift": "班次",
+    "ticket": "工单",
+    "file": "文件",
+    "photo": "照片",
+    "image": "图片",
+    "origin": "来源",
+    "in": "入库",
+    "out": "出库",
+    "list": "列表",
+    "count": "数量",
+    "total": "总计",
+    "enabled": "是否启用",
+}
+
+QUERY_PARAMETERS = {
+    "page": "页码",
+    "page_size": "每页数量",
+    "search": "搜索关键字",
+    "ordering": "排序字段，字段名前加“-”表示倒序",
+    "format": "响应格式",
+}
+
+LOOKUP_NAMES = {
+    "in": "属于列表",
+    "contains": "包含",
+    "icontains": "包含（忽略大小写）",
+    "gte": "大于或等于",
+    "gt": "大于",
+    "lte": "小于或等于",
+    "lt": "小于",
+    "isnull": "是否为空",
+    "exact": "等于",
+}
+
+
+def _contains_chinese(value):
+    return bool(re.search(r"[\u4e00-\u9fff]", str(value or "")))
+
+
+def swagger_schema_file(request):
+    schema_path = Path(settings.SWAGGER_SCHEMA_PATH)
+    if not schema_path.is_file():
+        return JsonResponse(
+            {"detail": "Swagger文档尚未生成，请先运行 manage.py build_swagger"},
+            status=503,
+        )
+
+    response = FileResponse(
+        schema_path.open("rb"),
+        content_type="application/json; charset=utf-8",
+        filename="swagger.json",
+    )
+    response["Content-Disposition"] = 'inline; filename="swagger.json"'
+    response["Cache-Control"] = "no-cache"
+    return response
+
+
+def _serializer_model(field):
+    parent = getattr(field, "parent", None)
+    while parent is not None:
+        meta = getattr(parent, "Meta", None)
+        model = getattr(meta, "model", None)
+        if model is not None:
+            return model
+        parent = getattr(parent, "parent", None)
+    return None
+
+
+def _model_path_label(model, parts):
+    labels = []
+    for part in parts:
+        if model is None:
+            break
+        try:
+            model_field = model._meta.get_field(part)
+        except Exception:
+            break
+        verbose_name = getattr(model_field, "verbose_name", "")
+        if _contains_chinese(verbose_name):
+            labels.append(str(verbose_name))
+        model = getattr(model_field, "related_model", None)
+    return " / ".join(labels)
+
+
+def _field_name_label(field_name):
+    field_name = str(field_name or "").strip("_")
+    if field_name in FIELD_NAMES:
+        return FIELD_NAMES[field_name]
+    tokens = field_name.split("_")
+    if tokens and all(token in FIELD_TOKENS for token in tokens):
+        return "".join(FIELD_TOKENS[token] for token in tokens)
+    return ""
+
+
+class ChineseFieldInspector(FieldInspector):
+    """优先使用模型字段中文名称补全 serializer 字段标题。"""
+
+    def field_to_swagger_object(self, field, **kwargs):
+        return NotHandled
+
+    def process_result(self, result, method_name, obj, **kwargs):
+        if (
+            method_name != "field_to_swagger_object"
+            or not isinstance(result, openapi.SwaggerDict)
+            or "$ref" in result
+            or _contains_chinese(result.get("title"))
+        ):
+            return result
+
+        source_attrs = getattr(obj, "source_attrs", None) or []
+        model_label = _model_path_label(_serializer_model(obj), source_attrs)
+        label = model_label or _field_name_label(getattr(obj, "field_name", ""))
+        field_name = getattr(obj, "field_name", "")
+        if label:
+            result["title"] = label
+        elif field_name:
+            result["title"] = f"{field_name}（字段）"
+        return result
+
+
+class ChineseSwaggerAutoSchema(SwaggerAutoSchema):
+    """为未显式编写文档的接口补充稳定、可读的中文展示信息。"""
+
+    field_inspectors = [ChineseFieldInspector] + SwaggerAutoSchema.field_inspectors
+
+    def get_operation(self, operation_keys=None):
+        operation = super().get_operation(operation_keys)
+        model = getattr(getattr(self.view, "queryset", None), "model", None)
+        for parameter in operation.get("parameters", []):
+            current = parameter.get("description", "")
+            if _contains_chinese(current):
+                continue
+            location = parameter.get("in")
+            if location == openapi.IN_BODY:
+                description = "请求数据"
+            elif location == openapi.IN_PATH:
+                description = f"路径参数：{parameter.get('name', '')}"
+            else:
+                description = self._get_parameter_description(
+                    parameter.get("name", ""), model
+                )
+            if current:
+                description = f"{description}；{current}"
+            parameter["description"] = description
+        return operation
+
+    def get_summary_and_description(self):
+        summary, description = super().get_summary_and_description()
+        if summary:
+            if description and not _contains_chinese(description):
+                description = f"{summary}\n\n{description}"
+            return summary, description or summary
+
+        resource = self._get_resource_name()
+        action = getattr(self.view, "action", None)
+        template = CRUD_SUMMARIES.get(action)
+        if template:
+            summary = template.format(resource=resource)
+        elif resource:
+            action_name = str(action or self.method).replace("_", " ")
+            display_resource = resource
+            if not _contains_chinese(display_resource):
+                display_resource = f"{display_resource}接口"
+            summary = f"{display_resource}：{action_name}"
+
+        if description and not _contains_chinese(description):
+            description = f"{summary}\n\n{description}"
+        return summary, description or summary
+
+    def get_request_body_parameters(self, consumes):
+        parameters = super().get_request_body_parameters(consumes)
+        for parameter in parameters:
+            if parameter.get("in") == openapi.IN_BODY and not _contains_chinese(
+                parameter.get("description")
+            ):
+                parameter["description"] = "请求数据"
+        return parameters
+
+    def get_query_parameters(self):
+        parameters = super().get_query_parameters()
+        model = getattr(getattr(self.view, "queryset", None), "model", None)
+        for parameter in parameters:
+            current = parameter.get("description", "")
+            if _contains_chinese(current):
+                continue
+            description = self._get_parameter_description(
+                parameter.get("name", ""), model
+            )
+            if current:
+                description = f"{description}；{current}"
+            parameter["description"] = description
+        return parameters
+
+    def get_responses(self):
+        responses = super().get_responses()
+        descriptions = {
+            "200": "请求成功",
+            "201": "创建成功",
+            "202": "请求已接受",
+            "204": "操作成功，无响应内容",
+            "400": "请求参数错误",
+            "401": "未认证或认证已失效",
+            "403": "无权访问",
+            "404": "资源不存在",
+        }
+        for status, response in responses.items():
+            if not response.get("description"):
+                response["description"] = descriptions.get(str(status), "接口响应")
+        return responses
+
+    def _get_parameter_description(self, name, model):
+        if name in QUERY_PARAMETERS:
+            return QUERY_PARAMETERS[name]
+
+        parts = str(name).split("__")
+        lookup = LOOKUP_NAMES.get(parts[-1])
+        field_parts = parts[:-1] if lookup else parts
+        label = _model_path_label(model, field_parts)
+        if not label:
+            label = _field_name_label(field_parts[-1] if field_parts else name)
+        if not label:
+            label = f"查询参数：{name}"
+        if lookup:
+            label = f"{label}（{lookup}）"
+        return label
+
+    def get_tags(self, operation_keys=None):
+        tags = super().get_tags(operation_keys)
+        if self.overrides.get("tags") or not tags:
+            return tags
+
+        if tags[0] in TAG_NAMES:
+            return [TAG_NAMES[tags[0]]]
+
+        try:
+            app_config = apps.get_app_config(tags[0])
+        except LookupError:
+            return tags
+
+        if _contains_chinese(app_config.verbose_name):
+            return [str(app_config.verbose_name)]
+        return tags
+
+    def _get_resource_name(self):
+        queryset = getattr(self.view, "queryset", None)
+        model = getattr(queryset, "model", None)
+
+        if model is None:
+            serializer_class = getattr(self.view, "serializer_class", None)
+            meta = getattr(serializer_class, "Meta", None)
+            model = getattr(meta, "model", None)
+
+        if model is None:
+            return "接口"
+
+        match = re.search(r"TN\s*[:：]\s*([^\n\r]+)", model.__doc__ or "")
+        if match:
+            return match.group(1).strip()
+
+        verbose_name = str(model._meta.verbose_name)
+        if _contains_chinese(verbose_name):
+            return verbose_name
+        return model.__name__

apps/utils/test_swagger.py

@@ -0,0 +1,142 @@
+import json
+from pathlib import Path
+from tempfile import TemporaryDirectory
+from types import SimpleNamespace
+from unittest.mock import patch
+
+from django.conf import settings
+from django.core.management import call_command
+from django.test import SimpleTestCase, override_settings
+
+from apps.am.models import Area
+from apps.am.views import AreaViewSet
+from apps.utils.swagger import ChineseSwaggerAutoSchema, swagger_schema_file
+
+
+class ChineseSwaggerAutoSchemaTests(SimpleTestCase):
+    def make_schema(self, view, method="GET"):
+        schema = ChineseSwaggerAutoSchema.__new__(ChineseSwaggerAutoSchema)
+        schema.view = view
+        schema.method = method
+        schema.path = "/am/area/"
+        schema.overrides = {}
+        schema.operation_keys = ("am", "area", "list")
+        schema._sch = SimpleNamespace(get_description=lambda path, method: "")
+        return schema
+
+    def test_crud_summary_uses_model_chinese_name(self):
+        view = SimpleNamespace(queryset=Area.objects.all(), action="list")
+        schema = self.make_schema(view)
+
+        summary, description = schema.get_summary_and_description()
+
+        self.assertEqual(summary, "查询地图区域列表")
+        self.assertEqual(description, "查询地图区域列表")
+
+    def test_explicit_summary_takes_priority(self):
+        view = SimpleNamespace(queryset=Area.objects.all(), action="list")
+        schema = self.make_schema(view)
+        schema.overrides = {
+            "operation_summary": "区域自定义查询",
+            "operation_description": "自定义说明",
+        }
+
+        summary, description = schema.get_summary_and_description()
+
+        self.assertEqual(summary, "区域自定义查询")
+        self.assertEqual(description, "自定义说明")
+
+    def test_custom_action_with_english_model_name_has_chinese_hint(self):
+        model = SimpleNamespace(
+            __doc__="",
+            __name__="Dataset",
+            _meta=SimpleNamespace(verbose_name="dataset"),
+        )
+        queryset = SimpleNamespace(model=model)
+        view = SimpleNamespace(queryset=queryset, action="base")
+        schema = self.make_schema(view)
+
+        summary, _ = schema.get_summary_and_description()
+
+        self.assertEqual(summary, "Dataset接口：base")
+
+    def test_tag_uses_chinese_business_module_name(self):
+        view = SimpleNamespace(queryset=Area.objects.all(), action="list")
+        schema = self.make_schema(view)
+
+        self.assertEqual(schema.get_tags(("am", "area", "list")), ["区域与准入管理"])
+
+    def test_filter_parameter_uses_model_field_labels(self):
+        view = SimpleNamespace(queryset=Area.objects.all(), action="list")
+        schema = self.make_schema(view)
+
+        description = schema._get_parameter_description(
+            "manager__name__contains",
+            Area,
+        )
+
+        self.assertIn("区域负责人", description)
+        self.assertIn("包含", description)
+
+    def test_swagger_queryset_skips_permission_data_lookup(self):
+        view = AreaViewSet(basename="area")
+        view.action = "list"
+        view.swagger_fake_view = True
+
+        with patch("apps.utils.viewsets.get_user_perms_map") as permission_lookup:
+            queryset = view.get_queryset()
+
+        self.assertIs(queryset.model, Area)
+        permission_lookup.assert_not_called()
+
+
+class SwaggerSettingsTests(SimpleTestCase):
+    def test_swagger_supports_jwt_authorization_header(self):
+        from django.conf import settings
+
+        bearer = settings.SWAGGER_SETTINGS["SECURITY_DEFINITIONS"]["Bearer"]
+
+        self.assertEqual(bearer["type"], "apiKey")
+        self.assertEqual(bearer["name"], "Authorization")
+        self.assertEqual(bearer["in"], "header")
+
+    def test_swagger_ui_uses_static_schema(self):
+        from django.conf import settings
+
+        self.assertEqual(settings.SWAGGER_SETTINGS["SPEC_URL"], "schema-swagger-json")
+        self.assertEqual(settings.REDOC_SETTINGS["SPEC_URL"], "schema-swagger-json")
+
+
+class BuildSwaggerCommandTests(SimpleTestCase):
+    def test_command_writes_valid_utf8_schema(self):
+        schema = {
+            "swagger": "2.0",
+            "info": {"title": "中文文档"},
+            "paths": {"/demo/": {"get": {}}},
+        }
+
+        def generate_schema(command_name, output_file, **options):
+            self.assertEqual(command_name, "generate_swagger")
+            self.assertEqual(output_file, "-")
+            options["stdout"].write(json.dumps(schema, ensure_ascii=False))
+
+        with TemporaryDirectory(dir=settings.BASE_DIR) as directory:
+            target = Path(directory) / "swagger.json"
+            with override_settings(SWAGGER_SCHEMA_PATH=str(target)):
+                with patch(
+                    "apps.utils.management.commands.build_swagger.call_command",
+                    side_effect=generate_schema,
+                ):
+                    call_command("build_swagger", verbosity=0)
+
+                content = target.read_text(encoding="utf-8")
+                self.assertIn("中文文档", content)
+                self.assertEqual(json.loads(content), schema)
+
+                with override_settings(SWAGGER_SCHEMA_PATH=str(target)):
+                    response = swagger_schema_file(SimpleNamespace())
+                    body = b"".join(response.streaming_content)
+                    response.close()
+
+                self.assertEqual(response.status_code, 200)
+                self.assertEqual(json.loads(body), schema)

apps/utils/tests.py

@@ -0,0 +1,116 @@
+from unittest.mock import patch
+
+from django.db.models.query import QuerySet
+from django.test import SimpleTestCase, TestCase
+
+from apps.develop.models import Project
+from apps.mtm.models import Material
+from apps.system.models import User
+from apps.wpmw.models import WprDefect
+
+
+class SnowflakeBulkCreateTests(SimpleTestCase):
+    @patch.object(QuerySet, "bulk_create")
+    @patch(
+        "apps.utils.models.idWorker.get_id",
+        side_effect=[100, 101],
+    )
+    def test_base_model_bulk_create_assigns_missing_ids(
+        self,
+        get_id,
+        bulk_create,
+    ):
+        objs = [
+            WprDefect(defect_id="10", wpr_id="20"),
+            WprDefect(defect_id="11", wpr_id="20"),
+        ]
+
+        WprDefect.objects.bulk_create(objs)
+
+        self.assertEqual([obj.id for obj in objs], ["100", "101"])
+        self.assertEqual(get_id.call_count, 2)
+        bulk_create.assert_called_once()
+
+    @patch.object(QuerySet, "bulk_create")
+    @patch("apps.utils.models.idWorker.get_id")
+    def test_base_model_bulk_create_preserves_existing_ids(
+        self,
+        get_id,
+        bulk_create,
+    ):
+        objs = [
+            WprDefect(id=100, defect_id="10", wpr_id="20"),
+            WprDefect(id="101", defect_id="11", wpr_id="20"),
+        ]
+
+        WprDefect.objects.bulk_create(objs)
+
+        self.assertEqual([obj.id for obj in objs], ["100", "101"])
+        get_id.assert_not_called()
+        bulk_create.assert_called_once()
+
+    @patch.object(QuerySet, "bulk_create")
+    @patch(
+        "apps.utils.models.idWorker.get_id",
+        side_effect=[200, 201],
+    )
+    def test_soft_delete_manager_uses_snowflake_bulk_create(
+        self,
+        get_id,
+        bulk_create,
+    ):
+        objs = [
+            Material(name="物料1"),
+            Material(name="物料2"),
+        ]
+
+        Material.objects.bulk_create(objs)
+
+        self.assertEqual([obj.id for obj in objs], ["200", "201"])
+        self.assertEqual(get_id.call_count, 2)
+        bulk_create.assert_called_once()
+
+    @patch.object(QuerySet, "bulk_create")
+    @patch(
+        "apps.utils.models.idWorker.get_id",
+        side_effect=[300, 301],
+    )
+    def test_custom_user_manager_uses_snowflake_bulk_create(
+        self,
+        get_id,
+        bulk_create,
+    ):
+        objs = [
+            User(username="bulk-user-1"),
+            User(username="bulk-user-2"),
+        ]
+
+        User.objects.bulk_create(objs)
+
+        self.assertEqual([obj.id for obj in objs], ["300", "301"])
+        self.assertEqual(get_id.call_count, 2)
+        bulk_create.assert_called_once()
+
+
+class SnowflakeBulkCreateDatabaseTests(TestCase):
+    @patch(
+        "apps.utils.models.idWorker.get_id",
+        side_effect=[900000000000000001, 900000000000000002],
+    )
+    def test_bulk_create_persists_generated_ids(self, get_id):
+        objs = [
+            Project(name="批量项目1", code="bulk-project-1"),
+            Project(name="批量项目2", code="bulk-project-2"),
+        ]
+
+        Project.objects.bulk_create(objs)
+
+        self.assertEqual(
+            [obj.id for obj in objs],
+            ["900000000000000001", "900000000000000002"],
+        )
+        self.assertEqual(
+            set(Project.objects.values_list("id", flat=True)),
+            {"900000000000000001", "900000000000000002"},
+        )
+        self.assertEqual(get_id.call_count, 2)

apps/utils/viewsets.py

@@ -154,6 +154,9 @@ class CustomGenericViewSet(MyLoggingMixin, GenericViewSet):
     def get_queryset(self):
         queryset = super().get_queryset()
         queryset = self.get_queryset_custom(queryset)
+        # drf-yasg 生成文档时不应读取权限或业务数据。
+        if getattr(self, 'swagger_fake_view', False):
+            return queryset
         if self.data_filter:
             user = self.request.user
             if user.is_superuser:

apps/vm/tests.py

@@ -1,14 +1,38 @@
+from datetime import timedelta
+from unittest.mock import patch
+
 from django.test import TestCase
+from django.utils import timezone
+
 from apps.hrm.models import Employee
 from apps.hrm.services import HrmService
-
 from apps.vm.models import Visit
 
-# Create your tests here.
-
 
 class VisitTest(TestCase):
-    def test_audit_end(self):
+    @patch.object(HrmService, 'sync_dahua_employee')
-        v = Visit.objects.get(id='1547401054033940480')
+    def test_sync_employee_uses_visit_time_window(self, sync_employee):
-        ep = Employee.objects.get(id='1547406632055607296')
+        visit_time = timezone.now()
-        HrmService.sync_dahua_employee(ep, '', v.visit_time, v.leave_time)
+        leave_time = visit_time + timedelta(hours=2)
+        visit = Visit.objects.create(
+            purpose=Visit.V_PURPOSE_CHOICES[0][0],
+            level=Visit.V_LEVEL_CHOICES[0][0],
+            name='访客同步测试',
+            visit_time=visit_time,
+            leave_time=leave_time,
+        )
+        employee = Employee.objects.create(name='测试访客', type='visitor')
+
+        HrmService.sync_dahua_employee(
+            employee,
+            '',
+            visit.visit_time,
+            visit.leave_time,
+        )
+
+        sync_employee.assert_called_once_with(
+            employee,
+            '',
+            visit_time,
+            leave_time,
+        )

apps/wf/tests.py

@@ -1,3 +1,60 @@
-from django.test import TestCase
+from types import SimpleNamespace
+from unittest.mock import MagicMock, patch
 
-# Create your tests here.
+from django.test import SimpleTestCase
+from rest_framework.test import APIRequestFactory
+
+from apps.wf.views import TicketViewSet
+
+
+class TicketAvailableActionsTests(SimpleTestCase):
+    def setUp(self):
+        self.request = APIRequestFactory().get('/api/wf/ticket/1/available_actions/')
+        self.request.user = SimpleNamespace(id='user-1')
+        self.view = TicketViewSet()
+        self.ticket = SimpleNamespace(in_add_node=False)
+        self.view.get_object = MagicMock(return_value=self.ticket)
+
+    @patch('apps.wf.views.TransitionSerializer')
+    @patch('apps.wf.views.WfService.get_ticket_transitions')
+    @patch('apps.wf.views.WfService.ticket_handle_permission_check')
+    def test_returns_transitions_when_user_can_handle(
+            self, permission_check, get_transitions, serializer):
+        permission_check.return_value = {
+            'permission': True,
+            'need_accept': False,
+            'msg': '',
+        }
+        transitions = MagicMock()
+        transitions.select_related.return_value = transitions
+        get_transitions.return_value = transitions
+        serializer.return_value.data = [{'id': 'transition-1', 'name': '同意'}]
+
+        response = self.view.available_actions(self.request)
+
+        self.assertTrue(response.data['permission'])
+        self.assertFalse(response.data['need_accept'])
+        self.assertEqual(
+            response.data['transitions'],
+            [{'id': 'transition-1', 'name': '同意'}],
+        )
+
+    @patch('apps.wf.views.TransitionSerializer')
+    @patch('apps.wf.views.WfService.get_ticket_transitions')
+    @patch('apps.wf.views.WfService.ticket_handle_permission_check')
+    def test_hides_transitions_and_requests_accept_for_active_distribution(
+            self, permission_check, get_transitions, serializer):
+        permission_check.return_value = {
+            'permission': False,
+            'need_accept': True,
+            'msg': '需要先接单再处理',
+        }
+
+        response = self.view.available_actions(self.request)
+
+        self.assertFalse(response.data['permission'])
+        self.assertTrue(response.data['need_accept'])
+        self.assertEqual(response.data['message'], '需要先接单再处理')
+        self.assertEqual(response.data['transitions'], [])
+        get_transitions.assert_not_called()
+        serializer.assert_not_called()

apps/wf/views.py

@@ -363,6 +363,34 @@ class TicketViewSet(CreateUpdateCustomMixin, CreateModelMixin, ListModelMixin, R
         transitions = WfService.get_ticket_transitions(ticket)
         return Response(TransitionSerializer(instance=transitions.select_related('source_state', 'destination_state'), many=True).data)
 
+    @action(methods=['get'], detail=True, perms_map={'get': '*'})
+    def available_actions(self, request, pk=None):
+        """
+        获取当前用户对工单可执行的操作及接单状态。
+
+        前端应使用该接口决定是否展示处理按钮，避免自行解释 participant
+        后与后端实际处理权限不一致。原 transitions 接口保留用于兼容。
+        """
+        ticket = self.get_object()
+        permission_result = WfService.ticket_handle_permission_check(
+            ticket, request.user
+        )
+        transitions = []
+        if permission_result.get('permission', False):
+            transition_queryset = WfService.get_ticket_transitions(
+                ticket
+            ).select_related('source_state', 'destination_state')
+            transitions = TransitionSerializer(
+                instance=transition_queryset, many=True
+            ).data
+        return Response({
+            'permission': permission_result.get('permission', False),
+            'need_accept': permission_result.get('need_accept', False),
+            'message': permission_result.get('msg', ''),
+            'in_add_node': ticket.in_add_node,
+            'transitions': transitions,
+        })
+
     @action(methods=['post'], detail=True, perms_map={'post': '*'})
     def accpet(self, request, pk=None):
         """

apps/wpm/filters.py

@@ -5,6 +5,7 @@ from apps.mtm.models import Route, Material
 from django.db.models import Q, Exists, OuterRef
 from rest_framework.exceptions import ParseError
 from datetime import datetime
+from apps.qm.defect_grades import effective_defect_grade_q
 
 class SfLogFilter(filters.FilterSet):
     class Meta:
@@ -44,6 +45,10 @@ class WMaterialFilter(filters.FilterSet):
     mlog_date_start = filters.DateFilter(label="产出开始", method="filter_mlog_date_start")
     mlog_date_end = filters.DateFilter(label="产出结束", method="filter_mlog_date_end")
     current_merged = filters.BooleanFilter(label="是否本工段新合成的批", method="filter_current_merged")
+    defect_grade = filters.NumberFilter(label="有效缺陷等级", method="filter_defect_grade")
+
+    def filter_defect_grade(self, queryset, name, value):
+        return queryset.filter(effective_defect_grade_q(value, "notok_sign"))
 
     def filter_mlog_date_start(self, queryset, name, value):
         mgroupId = self.data.get("mgroup", None)
@@ -123,7 +128,7 @@ class WMaterialFilter(filters.FilterSet):
             "material__name": ["exact", "in", "contains"],
             "material__process": ["exact", "in"],
             "material__process__name": ["exact", "contains", "in"],
-            "belong_dept": ["exact"],
+            "belong_dept": ["exact", "isnull"],
             "belong_dept__name": ["exact", "in"],
             "batch": ["exact", "contains"],
             "mgroup": ["exact", "in", "isnull"],
@@ -132,7 +137,8 @@ class WMaterialFilter(filters.FilterSet):
             "notok_sign": ["exact", "in", "isnull"],
             "count_xtest": ["gte", "isnull"],
             "supplier": ["exact"],
-            "defect": ["exact", "isnull"]
+            "defect": ["exact", "isnull"],
+            "defect__okcate": ["exact", "in"],
         }
 
 class MlogFilter(filters.FilterSet):
@@ -171,15 +177,23 @@ class HandoverFilter(filters.FilterSet):
     dept = filters.CharFilter(label='DeptId', method='filter_dept')
 
     def filter_mgroup(self, queryset, name, value):
-        return queryset.filter(send_mgroup__id=value)|queryset.filter(recive_mgroup__id=value)
+        return queryset.filter(
+            Q(send_mgroup_id=value) | Q(recive_mgroup_id=value)
+        )
 
     def filter_mgroupx(self, queryset, name, value):
         dept = Mgroup.objects.get(id=value).belong_dept
-        return (queryset.filter(send_mgroup__id=value)|queryset.filter(recive_mgroup__id=value)|
+        return queryset.filter(
-        queryset.filter(send_dept=dept, send_mgroup__isnull=True)|queryset.filter(recive_dept=dept, recive_mgroup__isnull=True))
+            Q(send_mgroup_id=value)
+            | Q(recive_mgroup_id=value)
+            | Q(send_dept=dept, send_mgroup__isnull=True)
+            | Q(recive_dept=dept, recive_mgroup__isnull=True)
+        )
     
     def filter_dept(self, queryset, name, value):
-        return queryset.filter(send_dept__id=value)|queryset.filter(recive_dept__id=value)
+        return queryset.filter(
+            Q(send_dept_id=value) | Q(recive_dept_id=value)
+        )
     
     def filter_cbatch(self, queryset, name, value):
         qs1 = queryset.filter(Q(batch=value)|Q(new_batch=value)|Q(wm__batch=value))

apps/wpm/migrations/0135_mlog_stored_scope.py

@@ -0,0 +1,37 @@
+from django.db import migrations, models
+
+
+def migrate_scope_forward(apps, schema_editor):
+    Mlog = apps.get_model('wpm', 'Mlog')
+    Mlog.objects.filter(stored_mgroup=True).update(stored_scope=10)
+
+
+def migrate_scope_backward(apps, schema_editor):
+    Mlog = apps.get_model('wpm', 'Mlog')
+    Mlog.objects.filter(stored_scope=10).update(stored_mgroup=True)
+    Mlog.objects.exclude(stored_scope=10).update(stored_mgroup=False)
+
+
+class Migration(migrations.Migration):
+
+    dependencies = [
+        ('mtm', '0068_process_into_wm_scope'),
+        ('wpm', '0134_alter_mlog_clear_defect'),
+    ]
+
+    operations = [
+        migrations.AddField(
+            model_name='mlog',
+            name='stored_scope',
+            field=models.PositiveSmallIntegerField(
+                choices=[(10, '工段'), (20, '部门'), (30, '全局')],
+                default=20,
+                verbose_name='产出库存归属',
+            ),
+        ),
+        migrations.RunPython(migrate_scope_forward, migrate_scope_backward),
+        migrations.RemoveField(
+            model_name='mlog',
+            name='stored_mgroup',
+        ),
+    ]

apps/wpm/migrations/0136_alter_handover_global_scope_fields.py

@@ -0,0 +1,36 @@
+from django.db import migrations, models
+import django.db.models.deletion
+
+
+class Migration(migrations.Migration):
+
+    dependencies = [
+        ('wpm', '0135_mlog_stored_scope'),
+    ]
+
+    operations = [
+        migrations.AlterField(
+            model_name='handover',
+            name='send_dept',
+            field=models.ForeignKey(
+                blank=True,
+                null=True,
+                on_delete=django.db.models.deletion.CASCADE,
+                related_name='handover_send_dept',
+                to='system.dept',
+                verbose_name='送料部门',
+            ),
+        ),
+        migrations.AlterField(
+            model_name='handover',
+            name='recive_dept',
+            field=models.ForeignKey(
+                blank=True,
+                null=True,
+                on_delete=django.db.models.deletion.CASCADE,
+                related_name='handover_recive_dept',
+                to='system.dept',
+                verbose_name='接收部门',
+            ),
+        ),
+    ]

apps/wpm/migrations/0137_mlog_wpr_defect_snapshot.py

@@ -0,0 +1,18 @@
+# Generated by Django 4.2.27 on 2026-07-30 01:06
+
+from django.db import migrations, models
+
+
+class Migration(migrations.Migration):
+
+    dependencies = [
+        ('wpm', '0136_alter_handover_global_scope_fields'),
+    ]
+
+    operations = [
+        migrations.AddField(
+            model_name='mlog',
+            name='wpr_defect_snapshot',
+            field=models.JSONField(blank=True, editable=False, null=True, verbose_name='单件缺陷提交前快照'),
+        ),
+    ]

apps/wpm/migrations/0138_mlogbw_files.py

@@ -0,0 +1,24 @@
+# Generated by Django 4.2.27 on 2026-08-03
+
+from django.db import migrations, models
+
+
+class Migration(migrations.Migration):
+
+    dependencies = [
+        ("system", "0007_alter_dept_create_by_alter_dept_third_info_and_more"),
+        ("wpm", "0137_mlog_wpr_defect_snapshot"),
+    ]
+
+    operations = [
+        migrations.AddField(
+            model_name="mlogbw",
+            name="files",
+            field=models.ManyToManyField(
+                blank=True,
+                related_name="mlogbw_outputs",
+                to="system.file",
+                verbose_name="产出附件",
+            ),
+        ),
+    ]

apps/wpm/migrations/0139_mloguser_note.py

@@ -0,0 +1,16 @@
+from django.db import migrations, models
+
+
+class Migration(migrations.Migration):
+
+    dependencies = [
+        ('wpm', '0138_mlogbw_files'),
+    ]
+
+    operations = [
+        migrations.AddField(
+            model_name='mloguser',
+            name='note',
+            field=models.TextField(blank=True, default='', verbose_name='备注'),
+        ),
+    ]

apps/wpm/models.py

@@ -1,6 +1,6 @@
 from django.db import models
 from apps.utils.models import CommonADModel, CommonBDModel, BaseModel
-from apps.mtm.models import Mgroup, Team, Shift, Material, Route, Process
+from apps.mtm.models import Mgroup, Team, Shift, Material, Route, Process, WmScope
 from apps.pm.models import Mtask, Mtaskb
 from apps.system.models import User
 from django.utils.timezone import localtime
@@ -11,8 +11,9 @@ from django.db.models import Sum, Subquery
 from django.utils.translation import gettext_lazy as _
 from rest_framework.exceptions import ParseError
 from django.db.models import Count
-from django.db import transaction
+from django.db import connection, transaction
 from django.db.models import Max
+import json
 import re
 from django.db.models import Q, F
 import django.utils.timezone as timezone
@@ -103,6 +104,9 @@ class WmStateOption(models.IntegerChoices):
 class WMaterial(CommonBDModel):
     """
     TN: 车间库存
+
+    归属范围分三级：mgroup 指定工段，只有 belong_dept 指定部门，
+    两者都为空表示全局库存。
     """
     WM_OK = 10
     WM_NOTOK = 20
@@ -128,6 +132,122 @@ class WMaterial(CommonBDModel):
     number_from = models.TextField("来源于个号", null=True, blank=True)
     is_manual = models.BooleanField('手动创建', default=False)
 
+    INVENTORY_KEY_FIELDS = (
+        'material',
+        'batch',
+        'mgroup',
+        'belong_dept',
+        'state',
+        'defect',
+        'notok_sign',
+        'material_origin',
+        'state_origin',
+    )
+
+    @classmethod
+    def _normalize_inventory_lookup(cls, **kwargs):
+        """生成唯一、完整的库存业务键，避免省略 NULL 字段产生不同锁键。"""
+        unknown_fields = set(kwargs) - set(cls.INVENTORY_KEY_FIELDS)
+        if unknown_fields:
+            fields = ', '.join(sorted(unknown_fields))
+            raise TypeError(f'不支持的车间库存定位字段: {fields}')
+
+        if kwargs.get('material') is None or kwargs.get('batch') is None:
+            raise ValueError('车间库存业务键必须包含 material 和 batch')
+
+        lookup = {
+            field: kwargs.get(field)
+            for field in cls.INVENTORY_KEY_FIELDS
+        }
+        if lookup['state'] is None:
+            lookup['state'] = cls._meta.get_field('state').get_default()
+
+        mgroup = lookup['mgroup']
+        belong_dept = lookup['belong_dept']
+        if mgroup is not None:
+            mgroup_dept_id = getattr(mgroup, 'belong_dept_id', None)
+            belong_dept_id = getattr(belong_dept, 'pk', belong_dept)
+            if belong_dept is None:
+                lookup['belong_dept'] = mgroup.belong_dept
+            elif mgroup_dept_id != belong_dept_id:
+                raise ValueError('车间库存的工段与所属部门不匹配')
+
+        return lookup
+
+    @classmethod
+    def _inventory_advisory_lock_payload(cls, lookup):
+        lock_values = {}
+        for name in cls.INVENTORY_KEY_FIELDS:
+            field = cls._meta.get_field(name)
+            value = lookup[name]
+            if field.is_relation and value is not None:
+                value = getattr(value, 'pk', value)
+            lock_values[field.attname] = value
+        return json.dumps(
+            {'model': cls._meta.label_lower, 'lookup': lock_values},
+            sort_keys=True,
+            ensure_ascii=False,
+            default=str,
+            separators=(',', ':'),
+        )
+
+    @classmethod
+    def locked_get_or_create_inventory(cls, defaults=None, **kwargs):
+        """
+        在事务中按完整库存业务键获取或创建记录。
+
+        已存在记录使用行锁；首次创建使用 PostgreSQL 事务级 advisory lock，
+        并在取得锁后重新查询，避免两个事务同时创建相同库存。
+        """
+        if not connection.in_atomic_block:
+            raise RuntimeError(
+                'locked_get_or_create_inventory 必须在事务中调用'
+            )
+        if connection.vendor != 'postgresql':
+            raise RuntimeError(
+                'locked_get_or_create_inventory 仅支持 PostgreSQL'
+            )
+
+        defaults = defaults or {}
+        lookup = cls._normalize_inventory_lookup(**kwargs)
+        create_defaults = {
+            key: value
+            for key, value in defaults.items()
+            if key not in cls.INVENTORY_KEY_FIELDS
+        }
+
+        rows = list(
+            cls.objects.select_for_update().filter(**lookup)[:2]
+        )
+        if len(rows) > 1:
+            raise RuntimeError(
+                f'{cls.__name__} 数据异常：库存业务键 {lookup} 命中多条'
+            )
+        if rows:
+            return rows[0], False
+
+        lock_payload = cls._inventory_advisory_lock_payload(lookup)
+        with connection.cursor() as cursor:
+            cursor.execute(
+                'SELECT pg_advisory_xact_lock(hashtextextended(%s, 0))',
+                [lock_payload],
+            )
+
+        rows = list(
+            cls.objects.select_for_update().filter(**lookup)[:2]
+        )
+        if len(rows) > 1:
+            raise RuntimeError(
+                f'{cls.__name__} 数据异常：库存业务键 {lookup} 命中多条'
+            )
+        if rows:
+            return rows[0], False
+
+        return cls.objects.create(
+            **lookup,
+            **create_defaults,
+        ), True
+
     def delete(self, *args, **kwargs):
         if not self.is_manual:
             raise ParseError('只能删除手动创建的车间库存')
@@ -144,7 +264,11 @@ class WMaterial(CommonBDModel):
 
     @property
     def belong_dept_or_mgroup_id(self):
-        return self.mgroup.id if self.mgroup else self.belong_dept.id
+        if self.mgroup_id:
+            return ("mgroup", self.mgroup_id)
+        if self.belong_dept_id:
+            return ("belong_dept", self.belong_dept_id)
+        return ("global", None)
         
     @property
     def count_working(self):
@@ -171,7 +295,6 @@ class WMaterial(CommonBDModel):
         if qs is None:
             qs = cls.objects
         return WMaterial.ava_qs(mgroup=mtask.mgroup, qs=qs).filter(
-                mgroup=mtask.mgroup,
                 material=mtask.material_in,
                 batch__in=Subquery(
                     Mlogb.objects.filter(
@@ -240,9 +363,16 @@ class Mlog(CommonADModel):
     work_end_time = models.DateTimeField('生产结束时间', null=True, blank=True)
     hour_work = models.FloatField('预计工时', null=True, blank=True)
     reminder_interval_list = models.JSONField('提醒间隔', default=list, blank=True)
-    stored_mgroup = models.BooleanField('入库到工段', default=False)
+    stored_scope = models.PositiveSmallIntegerField(
+        '产出库存归属', choices=WmScope.choices, default=WmScope.DEPT)
     stored_notok = models.BooleanField('不合格品是否已入库', default=False)
     clear_defect = models.BooleanField('合格B类记为合格', default=False)
+    wpr_defect_snapshot = models.JSONField(
+        '单件缺陷提交前快照',
+        null=True,
+        blank=True,
+        editable=False,
+    )
     route = models.ForeignKey(Route, verbose_name='生产路线', on_delete=models.SET_NULL, null=True, blank=True)
     mtask = models.ForeignKey(
         Mtask, verbose_name='关联任务', on_delete=models.CASCADE, null=True, blank=True, related_name='mlog_mtask')
@@ -420,6 +550,7 @@ class MlogUser(BaseModel):
         Equipment, verbose_name='生产设备', on_delete=models.CASCADE, null=True, blank=True, related_name='mloguser_equipment')
     shift = models.ForeignKey(Shift, verbose_name='关联班次', on_delete=models.CASCADE)
     handle_date = models.DateField('操作日期')
+    note = models.TextField('备注', default='', blank=True)
 
 class Mlogb(BaseModel):
     """
@@ -610,6 +741,12 @@ class Mlogbw(BaseModel):
     """TN: 单个产品生产/检验日志
     """
     number = models.TextField('单个编号', db_index=True)
+    files = models.ManyToManyField(
+        "system.File",
+        verbose_name="产出附件",
+        blank=True,
+        related_name="mlogbw_outputs",
+    )
     mlogb = models.ForeignKey(Mlogb, verbose_name='生产记录', on_delete=models.CASCADE, related_name="w_mlogb")
     mlogbw_from = models.ForeignKey("self", verbose_name='来源个', on_delete=models.CASCADE, null=True, blank=True, related_name="w_mlogbw_from")
     wpr = models.ForeignKey("wpmw.wpr", verbose_name='关联产品', on_delete=models.SET_NULL
@@ -691,7 +828,8 @@ class Handover(CommonADModel):
     send_mgroup = models.ForeignKey(
         Mgroup, verbose_name='送料工段', on_delete=models.CASCADE, null=True, blank=True)
     send_dept = models.ForeignKey(
-        Dept, verbose_name='送料部门', on_delete=models.CASCADE, related_name='handover_send_dept')
+        Dept, verbose_name='送料部门', on_delete=models.CASCADE, related_name='handover_send_dept',
+        null=True, blank=True)
     batch = models.TextField('批次号', null=True, blank=True, db_index=True)
     material = models.ForeignKey(
         Material, verbose_name='物料', on_delete=models.CASCADE, related_name='h_ma')
@@ -700,7 +838,8 @@ class Handover(CommonADModel):
     count = models.DecimalField('送料数', default=0, max_digits=11, decimal_places=1)
     count_eweight = models.FloatField('单数重量', default=0)
     recive_dept = models.ForeignKey(
-        Dept, verbose_name='接收部门', on_delete=models.CASCADE, related_name='handover_recive_dept')
+        Dept, verbose_name='接收部门', on_delete=models.CASCADE, related_name='handover_recive_dept',
+        null=True, blank=True)
     recive_mgroup = models.ForeignKey(Mgroup, verbose_name='接收工段', on_delete=models.CASCADE, related_name='handover_recive_mgroup', null=True, blank=True)
     recive_user = models.ForeignKey(
         User, verbose_name='接收人', on_delete=models.CASCADE, related_name='handover_recive_user', null=True, blank=True)
@@ -738,7 +877,7 @@ class Handoverb(BaseModel):
 
     @property
     def handoverbw(self):
-        return Handoverbw.objects.filter(handoverb=self)
+        return self.w_handoverb.all()
 
 class Handoverbw(BaseModel):
     """TN: 单个产品交接记录

apps/wpm/serializers.py

@@ -8,11 +8,11 @@ from .models import (SfLog, StLog, SfLogExp, WMaterial, Mlog,
                     Handover, Handoverb, Mlogb, AttLog,
                      OtherLog, Fmlog, BatchSt, Mlogbw, Handoverbw, MlogbDefect, MlogUser, BatchLog)
 from apps.system.models import Dept, User
-from apps.system.serializers import UserSimpleSerializer
+from apps.system.serializers import FileSerializer, UserSimpleSerializer
 from apps.pm.models import Mtask, Mtaskb
 from apps.wpm.tasks import cal_enstat_when_pcoal_heat_change, cal_enstat_when_team_change, cal_exp_duration_sec
 from apps.wpm.services import get_sflog, find_material_can_change, generate_new_batch
-from apps.mtm.models import Mgroup, TeamMember, Shift, Material, Route, Process
+from apps.mtm.models import Mgroup, TeamMember, Shift, Material, Route, Process, WmScope
 from apps.mtm.serializers import MaterialSimpleSerializer
 from django.db import transaction
 from django.utils import timezone
@@ -24,12 +24,35 @@ from apps.wpmw.models import Wpr
 from apps.qm.serializers import FtestProcessSerializer, FtestProcessListSerializer
 import logging
 from apps.qm.models import Defect
+from apps.qm.defect_grades import DEFECT_GRADE_NAMES, effective_defect_grade
 from apps.utils.snowflake import idWorker
 from decimal import Decimal
 from apps.em.models import Equipment
 from django.db.models import Q
 mylogger = logging.getLogger("log")
 
+WM_STATE_NAMES = {
+    WMaterial.WM_OK: "合格",
+    WMaterial.WM_NOTOK: "不合格",
+    WMaterial.WM_REPAIR: "返修",
+    WMaterial.WM_REPAIRED: "返修完成",
+    WMaterial.WM_TEST: "检验",
+    WMaterial.WM_SCRAP: "报废",
+}
+
+
+class MlogEquipmentOptionSerializer(serializers.ModelSerializer):
+    mgroup_name = serializers.CharField(source="mgroup.name", read_only=True)
+    full_name = serializers.SerializerMethodField()
+
+    def get_full_name(self, obj):
+        return f"{obj.number}|{obj.name}|{obj.model}"
+
+    class Meta:
+        model = Equipment
+        fields = ["id", "name", "number", "model", "mgroup_name", "full_name"]
+
+
 class OtherLogSerializer(CustomModelSerializer):
     class Meta:
         model = OtherLog
@@ -189,12 +212,21 @@ class WMaterialSerializer(CustomModelSerializer):
     material_origin_name = serializers.StringRelatedField(source='material_origin', read_only=True)
     notok_sign_name = serializers.SerializerMethodField()
     defect_name = serializers.CharField(source="defect.name", read_only=True)
+    defect_okcate = serializers.IntegerField(source="defect.okcate", read_only=True)
+    defect_grade = serializers.SerializerMethodField()
+    defect_grade_name = serializers.SerializerMethodField()
     count_working = serializers.SerializerMethodField(label='在制数量')
     count_handovering = serializers.SerializerMethodField(label='正在交送的数量')
 
     def get_notok_sign_name(self, obj):
         return getattr(NotOkOption, obj.notok_sign, NotOkOption.qt).label if obj.notok_sign else None
 
+    def get_defect_grade(self, obj):
+        return effective_defect_grade(obj, "notok_sign")
+
+    def get_defect_grade_name(self, obj):
+        return DEFECT_GRADE_NAMES[self.get_defect_grade(obj)]
+
     def get_count_working(self, obj):
         # 列表接口 queryset 已注解(单次聚合); 嵌套等无注解场景回退模型属性
         # 保留 decimal 精度(原 IntegerField 会截断在途量, 导致可用/可交接量偏大)
@@ -232,18 +264,36 @@ class WMaterialSerializer(CustomModelSerializer):
 class WMaterialCreateSerializer(CustomModelSerializer):
     class Meta:
         model = WMaterial
-        fields = ['material', 'count', 'batch', 'mgroup']
+        fields = ['material', 'count', 'batch', 'mgroup', 'belong_dept']
         extra_kwargs = {
             'material': {'required': True},
             'count': {'required': True},
             'batch': {'required': True},
-            'mgroup': {'required': True, 'allow_null': False},
+            'mgroup': {'required': False, 'allow_null': True},
+            'belong_dept': {'required': False, 'allow_null': True},
         }
 
     def validate(self, attrs):
-        attrs['belong_dept'] = attrs['mgroup'].belong_dept
+        mgroup = attrs.get('mgroup')
+        if mgroup:
+            attrs['belong_dept'] = mgroup.belong_dept
         return attrs
 
+    @transaction.atomic
+    def create(self, validated_data):
+        lookup = {
+            field: validated_data.pop(field)
+            for field in WMaterial.INVENTORY_KEY_FIELDS
+            if field in validated_data
+        }
+        instance, created = WMaterial.locked_get_or_create_inventory(
+            **lookup,
+            defaults=validated_data,
+        )
+        if not created:
+            raise serializers.ValidationError('相同业务键的车间库存已存在')
+        return instance
+
 
 class MlogbDefectSerializer(CustomModelSerializer):
     defect_name = serializers.CharField(source="defect.name", read_only=True)
@@ -329,8 +379,7 @@ class MlogListSerializer(CustomModelSerializer):
 
     class Meta:
         model = Mlog
-        fields = '__all__'
+        exclude = ["wpr_defect_snapshot"]
-        # exclude = ["equipments","handle_users", "material_outs"]
 
 
 class MlogbDetailSerializer(CustomModelSerializer):
@@ -401,7 +450,7 @@ class MlogSerializer(CustomModelSerializer):
     wm_in_state = serializers.IntegerField(source='wm_in.state', read_only=True)
     class Meta:
         model = Mlog
-        fields = '__all__'
+        exclude = ["wpr_defect_snapshot"]
         read_only_fields = EXCLUDE_FIELDS + \
             ['submit_time', 'submit_user', 'material_outs']
         extra_kwargs = {
@@ -897,6 +946,8 @@ class MlogbInUpdateSerializer(CustomModelSerializer):
         return ins
     
 class MlogbwCreateUpdateSerializer(CustomModelSerializer):
+    files = serializers.PrimaryKeyRelatedField(many=True, read_only=True)
+    files_ = FileSerializer(source="files", many=True, read_only=True)
     ftest = FtestProcessSerializer(required=False, allow_null=True)
     equip_name = serializers.CharField(source='equip.name', read_only=True)
     equip_number = serializers.CharField(source='equip.number', read_only=True)
@@ -910,7 +961,8 @@ class MlogbwCreateUpdateSerializer(CustomModelSerializer):
         fields = ["id", "number", "wpr", "note",
                   "mlogb", "ftest", "equip", "tooling", "work_start_time",
                   "work_end_time", "mlogbw_from", "equip_name", "equip_number",
-                  "tooling_name", "tooling_number", "wpr_number_out", "wpr_pre_info", "mlogb__batch"]
+                  "tooling_name", "tooling_number", "wpr_number_out", "wpr_pre_info", "mlogb__batch",
+                  "files", "files_"]
         read_only_fields = ["mlogbw_from"]
 
     def validate(self, attrs):
@@ -954,10 +1006,18 @@ class MlogbwCreateUpdateSerializer(CustomModelSerializer):
             mlogbw = self.save_ftest(mlogbw, ftest_data)
         return mlogbw
 
+    @transaction.atomic
     def update(self, instance, validated_data):
+        old_number = instance.number
         validated_data.pop("mlogb")
         ftest_data = validated_data.pop("ftest", None)
         mlogbw:Mlogbw = super().update(instance, validated_data)
+        if (
+            mlogbw.number != old_number
+            and mlogbw.mlogb.material_out_id is not None
+            and mlogbw.wpr is not None
+        ):
+            mlogbw.wpr.change_number(mlogbw.number)
         if ftest_data:
             mlogbw = self.save_ftest(mlogbw, ftest_data)
         elif ftest_data is None:
@@ -972,6 +1032,15 @@ class MlogbwCreateUpdateSerializer(CustomModelSerializer):
 class MlogbwListSerializer(MlogbwCreateUpdateSerializer):
     ftest = FtestProcessListSerializer(read_only=True)
 
+
+class MlogbwFilesSerializer(CustomModelSerializer):
+    files_ = FileSerializer(source="files", many=True, read_only=True)
+
+    class Meta:
+        model = Mlogbw
+        fields = ["id", "files", "files_"]
+        read_only_fields = ["id", "files_"]
+
 class MlogbwStartTestSerializer(serializers.Serializer):
     mlogbw_ids = serializers.ListField(child=serializers.CharField(), label="mlogbwId列表")
     test_equip = serializers.CharField(label="测试设备", allow_null=True, required=False, allow_blank=True)
@@ -1218,10 +1287,56 @@ class Handoverbwserializer(CustomModelSerializer):
         read_only_fields = EXCLUDE_FIELDS_BASE + ["handoverb", "number"]
         extra_kwargs = {'wpr': {'required': True}}
 
+
+class CachedWMaterialPrimaryKeyRelatedField(serializers.PrimaryKeyRelatedField):
+    def to_internal_value(self, data):
+        cache = getattr(self.root, "_handover_wmaterial_cache", None)
+        if cache is None:
+            return super().to_internal_value(data)
+        if not isinstance(data, (str, int)):
+            self.fail("incorrect_type", data_type=type(data).__name__)
+        try:
+            return cache[str(data)]
+        except KeyError:
+            self.fail("does_not_exist", pk_value=data)
+
+
 class HandoverbSerializer(CustomModelSerializer):
+    wm = CachedWMaterialPrimaryKeyRelatedField(
+        queryset=WMaterial.objects.select_related(
+            "material", "defect", "mgroup", "belong_dept"
+        )
+    )
     notok_sign = serializers.CharField(source='wm.notok_sign', read_only=True)
+    notok_sign_name = serializers.SerializerMethodField()
     defect_name = serializers.CharField(source="wm.defect.name", read_only=True)
+    defect_grade = serializers.SerializerMethodField()
+    defect_grade_name = serializers.SerializerMethodField()
+    material_name = serializers.StringRelatedField(source="wm.material", read_only=True)
+    state_name = serializers.SerializerMethodField()
+    mgroup_name = serializers.CharField(source="wm.mgroup.name", read_only=True)
+    belong_dept_name = serializers.CharField(source="wm.belong_dept.name", read_only=True)
+    count_available = serializers.SerializerMethodField()
     handoverbw = Handoverbwserializer(many=True, required=False)
+
+    def get_notok_sign_name(self, obj):
+        return getattr(NotOkOption, obj.wm.notok_sign, NotOkOption.qt).label if obj.wm.notok_sign else None
+
+    def get_defect_grade(self, obj):
+        return effective_defect_grade(obj.wm, "notok_sign")
+
+    def get_defect_grade_name(self, obj):
+        return DEFECT_GRADE_NAMES[self.get_defect_grade(obj)]
+
+    def get_state_name(self, obj):
+        return WM_STATE_NAMES.get(obj.wm.state, str(obj.wm.state))
+
+    def get_count_available(self, obj):
+        # 编辑未提交交接时，当前明细占用的数量仍应允许重新填写。
+        if obj.handover.submit_time is not None:
+            return obj.count
+        return obj.wm.count - obj.wm.count_handovering + obj.count
+
     class Meta:
         model = Handoverb
         fields = "__all__"
@@ -1244,9 +1359,14 @@ class HandoverSerializer(CustomModelSerializer):
     recive_user_name = serializers.CharField(
         source='recive_user.name', read_only=True)
     recive_dept_name = serializers.CharField(
-        source='recive_dept', read_only=True)
+        source='recive_dept.name', read_only=True)
+    send_dept_name = serializers.CharField(source='send_dept.name', read_only=True)
     send_mgroup_name = serializers.CharField(source='send_mgroup.name', read_only=True)
     recive_mgroup_name = serializers.CharField(source='recive_mgroup.name', read_only=True)
+    submit_user_name = serializers.CharField(source='submit_user.name', read_only=True)
+    type_name = serializers.CharField(source='get_type_display', read_only=True)
+    mtype_name = serializers.CharField(source='get_mtype_display', read_only=True)
+    state_changed_name = serializers.SerializerMethodField()
     material_ = MaterialSimpleSerializer(source='material', read_only=True)
     material_name = serializers.StringRelatedField(
         source='material', read_only=True)
@@ -1255,6 +1375,30 @@ class HandoverSerializer(CustomModelSerializer):
     handoverb = HandoverbSerializer(many=True, required=False)
     ticket_ = TicketSimpleSerializer(source='ticket', read_only=True)
 
+    def get_state_changed_name(self, obj):
+        return WM_STATE_NAMES.get(obj.state_changed) if obj.state_changed is not None else None
+
+    def to_internal_value(self, data):
+        handoverb = data.get("handoverb", []) if hasattr(data, "get") else []
+        wm_ids = {
+            str(item["wm"])
+            for item in handoverb
+            if isinstance(item, dict) and item.get("wm") is not None
+        }
+        if not wm_ids:
+            return super().to_internal_value(data)
+
+        queryset = WMaterial.objects.select_related(
+            "material", "defect", "mgroup", "belong_dept"
+        )
+        self._handover_wmaterial_cache = {
+            str(pk): instance for pk, instance in queryset.in_bulk(wm_ids).items()
+        }
+        try:
+            return super().to_internal_value(data)
+        finally:
+            del self._handover_wmaterial_cache
+    
     def validate(self, attrs):
         if "mtype" not in attrs:
             attrs['mtype'] = Handover.H_NORMAL
@@ -1262,6 +1406,9 @@ class HandoverSerializer(CustomModelSerializer):
             attrs['type'] = Handover.H_NORMAL
         
         mtype = attrs["mtype"]
+        clear_defect = bool(attrs.get('clear_defect', False))
+        if clear_defect and mtype != Handover.H_MERGE:
+            raise ParseError('仅合批时可清除批次缺陷')
 
         new_wm:WMaterial = attrs.get("new_wm", None)
         if new_wm:
@@ -1297,16 +1444,68 @@ class HandoverSerializer(CustomModelSerializer):
         else:
             raise ParseError('必须指定车间库存')
         attrs['material'] = wm.material
-        attrs['send_dept'] = wm.belong_dept
+        is_global_source = wm.mgroup_id is None and wm.belong_dept_id is None
-        if wm.mgroup:
+        send_mgroup = attrs.get('send_mgroup')
+        if send_mgroup:
+            attrs['send_dept'] = send_mgroup.belong_dept
+        else:
+            attrs['send_dept'] = wm.belong_dept
+        if not send_mgroup and wm.mgroup:
             attrs['send_mgroup'] = wm.mgroup
-        if attrs['material'].process and attrs['material'].process.into_wm_mgroup and 'recive_mgroup' not in attrs:
+
+        if (
+            is_global_source
+            and attrs['type'] == Handover.H_NORMAL
+            and mtype == Handover.H_NORMAL
+        ):
+            raise ParseError('全局库存无需正常交接')
+
+        is_global_rebatch = (
+            is_global_source
+            and attrs['type'] == Handover.H_NORMAL
+            and mtype in [Handover.H_DIV, Handover.H_MERGE]
+        )
+        if is_global_rebatch:
+            # 全局库存的拆批、合批不改变库存归属。
+            attrs['recive_mgroup'] = None
+            attrs['recive_dept'] = None
+            if new_wm and (
+                new_wm.mgroup_id is not None
+                or new_wm.belong_dept_id is not None
+            ):
+                raise ParseError('全局库存合批目标必须是全局库存')
+        elif (
+            is_global_source
+            and attrs['type'] in [
+                Handover.H_REPAIR,
+                Handover.H_SCRAP,
+                Handover.H_CHANGE,
+            ]
+            and not attrs.get('recive_mgroup')
+        ):
+            raise ParseError('全局库存返修、报废或改版必须指定接收工段')
+
+        material_process = attrs['material'].process
+        if (
+            not is_global_rebatch
+            and material_process
+            and WmScope.requires_mgroup(material_process.into_wm_scope)
+            and not attrs.get('recive_mgroup')
+        ):
             raise ParseError('必须指定交接工段')
         if 'recive_mgroup' in attrs and attrs['recive_mgroup']:
             attrs['recive_dept'] = attrs['recive_mgroup'].belong_dept
-        if not attrs.get('recive_dept', None) and not attrs.get('recive_mgroup', None):
+        if (
+            not is_global_rebatch
+            and not attrs.get('recive_dept', None)
+            and not attrs.get('recive_mgroup', None)
+        ):
             raise ParseError('必须指定收料车间或收料工段')
-        if not attrs.get('send_dept', None) and not attrs.get('send_mgroup', None):
+        if (
+            not is_global_source
+            and not attrs.get('send_dept', None)
+            and not attrs.get('send_mgroup', None)
+        ):
             raise ParseError('必须指定送料车间或送料工段')
         
         # if attrs["mtype"] == Handover.H_NORMAL and attrs.get("recive_mgroup", None) == attrs.get("send_mgroup", None):
@@ -1334,11 +1533,14 @@ class HandoverSerializer(CustomModelSerializer):
             elif deptOrmgroupId != current_mdept_id:
                 raise ParseError(f'第{ind+1}行-交接物料所属工段/车间不一致')
             if mtype == Handover.H_MERGE:
-                clear_defect = attrs.get('clear_defect', False)
-                if clear_defect and wm.state != WMaterial.WM_OK:
-                    raise ParseError(f'第{ind+1}行-清除缺陷合批仅支持合格品')
                 if clear_defect and new_wm is not None and new_wm.defect is not None:
-                    raise ParseError('清除缺陷合批的目标批次不能带缺陷')
+                    raise ParseError('清除批次缺陷时目标批次不能带缺陷')
+                if clear_defect and tracking == Material.MA_TRACKING_BATCH:
+                    defect_grade = effective_defect_grade(wm, "notok_sign")
+                    if defect_grade not in [Defect.DEFECT_OK, Defect.DEFECT_OK_B]:
+                        raise ParseError(
+                            f'第{ind+1}行-批次追踪物料仅合格品和合格B类可清除批次缺陷'
+                        )
                 if next_mat is None:
                     next_mat = wm.material
                     next_state = wm.state
@@ -1463,7 +1665,7 @@ class HandoverSerializer(CustomModelSerializer):
                 elif count == wm.count:
                     wpr_qs = Wpr.get_qs_by_wm(wm)
                     for item in wpr_qs:
-                        Handoverbw.objects.get_or_create(wpr=item, handoverb=handoverb,
+                        Handoverbw.objects.get_or_create(wpr=item, handoverb=hb,
                                                             defaults={"number": item.number})
                 else:
                     raise ParseError(f'第{ind+1}行-请提供交接物料明细')
@@ -1641,6 +1843,7 @@ class MlogQuickSerializer(serializers.Serializer):
     route = serializers.CharField(label="工艺步骤ID", required=False)
     mtask = serializers.CharField(label="任务ID", required=False)
     handle_user = serializers.CharField(label="操作人员ID")
+    supplier = serializers.CharField(label="外协单位ID", required=False)
 
 
 class BatchChangeSerializer(serializers.Serializer):

apps/wpm/services.py

@@ -1,4 +1,5 @@
 import datetime
+from collections import defaultdict
 
 from django.core.cache import cache
 from django.db.models import Sum
@@ -21,11 +22,50 @@ from apps.wpmw.models import Wpr, WprDefect
 from ..qm.models import Defect, Ftest
 from django.db.models import Count, Q
 from apps.utils.tasks import ctask_run
-from apps.mtm.models import Process
+from apps.utils.snowflake import idWorker
+from apps.mtm.models import Process, WmScope
 from django.db.models import F
 
 myLogger = logging.getLogger('log')
 
+RECENT_EQUIPMENT_LOG_LIMIT = 50
+
+
+def get_recent_mgroup_equipment_ids(
+    mgroup_id, log_limit=RECENT_EQUIPMENT_LOG_LIMIT
+):
+    """按日志时间倒序返回工段最近使用过的设备 ID，空值和重复值忽略。"""
+    recent_logs = list(
+        Mlog.objects.filter(mgroup_id=mgroup_id)
+        .order_by("-create_time", "-id")
+        .values_list("id", "equipment_id", "equipment_2_id")[:log_limit]
+    )
+    if not recent_logs:
+        return []
+
+    log_ids = [log_id for log_id, _, _ in recent_logs]
+    multiple_equipment_ids = defaultdict(list)
+    for log_id, equipment_id in (
+        Mlog.equipments.through.objects.filter(mlog_id__in=log_ids)
+        .order_by("id")
+        .values_list("mlog_id", "equipment_id")
+    ):
+        multiple_equipment_ids[log_id].append(equipment_id)
+
+    result = []
+    seen = set()
+    for log_id, equipment_id, equipment_2_id in recent_logs:
+        candidate_ids = [
+            equipment_id,
+            equipment_2_id,
+            *multiple_equipment_ids[log_id],
+        ]
+        for candidate_id in candidate_ids:
+            if candidate_id and candidate_id not in seen:
+                seen.add(candidate_id)
+                result.append(candidate_id)
+    return result
+
 def inherit_zt_batch(source: BatchSt, target: BatchSt):
     """拆批/报工改号时目标批继承来源批的直通统计大批归属(纯继承, 不做判定)
 
@@ -165,6 +205,74 @@ def get_pcoal_heat(year_s: int, month_s: int, day_s: int):
             return 25000
 
 
+def snapshot_mlog_wpr_defects(mlog: Mlog):
+    """锁定并保存日志提交前已有单件的缺陷状态。"""
+    wpr_ids = list(Mlogbw.objects.filter(
+        mlogb__mlog=mlog,
+        wpr__isnull=False,
+    ).values_list("wpr_id", flat=True).distinct())
+    list(Wpr.objects.select_for_update().filter(
+        id__in=wpr_ids,
+    ).values_list("id", flat=True))
+
+    snapshot = {str(wpr_id): [] for wpr_id in wpr_ids}
+    for defect in WprDefect.objects.filter(wpr_id__in=wpr_ids).values(
+        "wpr_id",
+        "defect_id",
+        "is_main",
+    ):
+        snapshot[str(defect["wpr_id"])].append({
+            "defect_id": str(defect["defect_id"]),
+            "is_main": defect["is_main"],
+        })
+    mlog.wpr_defect_snapshot = snapshot
+    mlog.save(update_fields=["wpr_defect_snapshot", "update_time"])
+    return snapshot
+
+
+def restore_mlog_wpr_defects(mlog: Mlog):
+    """按快照精确恢复日志提交前的单件缺陷。"""
+    snapshot = mlog.wpr_defect_snapshot
+    if snapshot is None:
+        return False
+
+    wpr_ids = list(snapshot)
+    wprs = {
+        str(wpr.id): wpr
+        for wpr in Wpr.objects.select_for_update().filter(id__in=wpr_ids)
+    }
+    WprDefect.objects.filter(wpr_id__in=wpr_ids).delete()
+    restored = []
+    for wpr_id, defects in snapshot.items():
+        wpr = wprs.get(str(wpr_id))
+        if wpr is None:
+            myLogger.warning(
+                "生产日志%s测回时未找到快照中的单件%s",
+                mlog.id,
+                wpr_id,
+            )
+            continue
+        restored.extend(
+            WprDefect(
+                id=idWorker.get_id(),
+                wpr=wpr,
+                defect_id=item["defect_id"],
+                is_main=item.get("is_main", False),
+            )
+            for item in defects
+        )
+    WprDefect.objects.bulk_create(restored)
+    return True
+
+
+def restore_legacy_wpr_defects_from_wm(wpr: Wpr, wm: WMaterial):
+    """兼容无快照历史日志的旧测回行为。"""
+    defects = []
+    if wm.defect_id:
+        defects.append({"defect_id": wm.defect_id, "is_main": True})
+    wpr.replace_defects(defects)
+
+
 def mlog_submit(mlog: Mlog, user: User, now: Union[datetime.datetime, None]):
     """
     生产日志提交后需要执行的操作
@@ -187,7 +295,7 @@ def mlog_submit(mlog: Mlog, user: User, now: Union[datetime.datetime, None]):
     
     mgroup = mlog.mgroup
     process = mgroup.process
-    stored_mgroup = process.into_wm_mgroup
+    stored_scope = process.into_wm_scope
     stored_notok = process.store_notok
     belong_dept = mgroup.belong_dept
     material_out: Material = mlog.material_out
@@ -195,10 +303,13 @@ def mlog_submit(mlog: Mlog, user: User, now: Union[datetime.datetime, None]):
     supplier = mlog.supplier  # 外协
     is_fix = mlog.is_fix
     if is_fix: # 如果是返工，直接放到工段下
-        stored_mgroup = True
+        stored_scope = WmScope.MGROUP
+    stored_location = WmScope.resolve_location(stored_scope, mgroup)
     m_ins_list = []
     m_ins_bl_list = []
 
+    wpr_defect_snapshot = snapshot_mlog_wpr_defects(mlog)
+
     if is_fix:
         can_matoutIds = process.get_canout_mat_ids()
     # 建立关系链
@@ -237,15 +348,23 @@ def mlog_submit(mlog: Mlog, user: User, now: Union[datetime.datetime, None]):
             continue
         if mi_count <= 0:
             raise ParseError('存在非正数!')
-        # 需要判断领用数是否合理
+        # 优先使用日志已选库存；未选择时按工段、部门、全局依次匹配并记录实际来源
-        # 优先使用工段库存
+        wm_in = getattr(mlog_or_b, 'wm_in', None)
-        if isinstance(mlog_or_b, Mlogb) and mlog_or_b.wm_in:
+        if wm_in:
-            wm = WMaterial.objects.select_for_update().get(id=mlog_or_b.wm_in.id)
+            wm = WMaterial.objects.select_for_update().get(id=wm_in.id)
         else:
             wm_qs = WMaterial.objects.filter(batch=mi_batch, material=mi_ma, mgroup=mgroup, state=WMaterial.WM_OK)
             if not wm_qs.exists():
                 wm_qs = WMaterial.objects.filter(batch=mi_batch, material=mi_ma,
                                                     belong_dept=belong_dept, mgroup=None, state=WMaterial.WM_OK)
+            if not wm_qs.exists():
+                wm_qs = WMaterial.objects.filter(
+                    batch=mi_batch,
+                    material=mi_ma,
+                    belong_dept=None,
+                    mgroup=None,
+                    state=WMaterial.WM_OK,
+                )
             count_x = wm_qs.count()
             if count_x == 1:
                 wm = WMaterial.objects.select_for_update().get(id=wm_qs.first().id)
@@ -255,6 +374,8 @@ def mlog_submit(mlog: Mlog, user: User, now: Union[datetime.datetime, None]):
             else:
                 raise ParseError(
                     f'{str(mi_ma)}-{mi_batch}-存在多个相同批次!')
+            mlog_or_b.wm_in = wm
+            mlog_or_b.save(update_fields=['wm_in'])
         
         if mi_count > wm.count:
             raise ParseError(
@@ -268,7 +389,8 @@ def mlog_submit(mlog: Mlog, user: User, now: Union[datetime.datetime, None]):
                 if mlogbws.count() != mi_count:
                     raise ParseError("日志与实际使用明细数量不一致,操作失败")
                 for item in mlogbws:
-                    Wpr.change_or_new(wpr=item.wpr, old_wm=wm, ftest=item.ftest)
+                    wpr = Wpr.change_or_new(wpr=item.wpr, old_wm=wm)
+                    wpr.apply_ftest(item.ftest)
 
         # 针对加工前不良的暂时额外处理 
         # 合格B类的需要当合格品
@@ -280,8 +402,15 @@ def mlog_submit(mlog: Mlog, user: User, now: Union[datetime.datetime, None]):
                 c_state = WMaterial.WM_NOTOK
                 if isinstance(defect, Defect) and defect.okcate == Defect.DEFECT_OK_B:
                     c_state = WMaterial.WM_OK
-                lookup = {'batch': batch, 'material': material, 'mgroup': mgroup, 'defect': defect, 'state': c_state}
+                lookup = {
-                wm, is_create = WMaterial.locked_get_or_create(**lookup, defaults={"belong_dept": belong_dept})
+                    'batch': batch,
+                    'material': material,
+                    'defect': defect,
+                    'state': c_state,
+                    **stored_location,
+                }
+                wm, is_create = WMaterial.locked_get_or_create_inventory(
+                    **lookup, defaults={})
                 wm.count = wm.count + count
                 if is_create:
                     wm.create_by = user
@@ -294,7 +423,12 @@ def mlog_submit(mlog: Mlog, user: User, now: Union[datetime.datetime, None]):
                     if mlogbws.count() != count:
                         raise ParseError("日志与加工前不良明细数量不一致,操作失败")
                     for item in mlogbws:
-                        Wpr.change_or_new(wpr=item.wpr, wm=wm, old_wm=wm, ftest=item.ftest)
+                        wpr = Wpr.change_or_new(
+                            wpr=item.wpr,
+                            wm=wm,
+                            old_wm=wm,
+                        )
+                        wpr.apply_ftest(item.ftest)
                 
 
     mlogb_out_qs = Mlogb.objects.filter(mlog=mlog, material_out__isnull=False)
@@ -359,18 +493,20 @@ def mlog_submit(mlog: Mlog, user: User, now: Union[datetime.datetime, None]):
             wm_state = WMaterial.WM_OK if notok_sign_or_defect is None or (
                     isinstance(notok_sign_or_defect, Defect) and notok_sign_or_defect.okcate in [Defect.DEFECT_OK, Defect.DEFECT_OK_B]
                 ) else WMaterial.WM_NOTOK
-        lookup = {'batch': mo_batch, 'material': mo_ma, 'mgroup': None,
+        lookup = {
-                    'notok_sign': None, 'defect': None, 'state': wm_state}
+            'batch': mo_batch,
+            'material': mo_ma,
+            'notok_sign': None,
+            'defect': None,
+            'state': wm_state,
+            **stored_location,
+        }
         if isinstance(notok_sign_or_defect, Defect):
             lookup['defect'] = notok_sign_or_defect
         elif notok_sign_or_defect is not None:
             lookup['notok_sign'] = notok_sign_or_defect
-        if stored_mgroup:
+        wm, is_create2 = WMaterial.locked_get_or_create_inventory(
-            lookup['mgroup'] = mgroup
+            **lookup, defaults={})
-        else:
-            lookup['belong_dept'] = belong_dept
-
-        wm, is_create2 = WMaterial.locked_get_or_create(**lookup, defaults={"belong_dept": belong_dept})
         wm.count = wm.count + mo_count
         wm.count_eweight = mo_count_eweight
         wm.update_by = user
@@ -399,25 +535,43 @@ def mlog_submit(mlog: Mlog, user: User, now: Union[datetime.datetime, None]):
                 raise ParseError("日志与明细数量不一致,操作失败")
             for item in mlogbws:
                 if item.wpr:
-                    Wpr.change_or_new(wpr=item.wpr, wm=wm, ftest=item.ftest)
+                    wpr = Wpr.change_or_new(
+                        wpr=item.wpr,
+                        wm=wm,
+                    )
                 else:
                     if not check_wpr_number(item.number):
                         raise ParseError(f'{item.number} 该编号格式有误')
                     wpr_from = None
                     if item.mlogbw_from:
                         wpr_from = item.mlogbw_from.wpr
-                    wpr = Wpr.change_or_new(number=item.number, 
+                    wpr = Wpr.change_or_new(
-                                            wm=wm, ftest=item.ftest, 
+                        number=item.number,
-                                            wpr_from=wpr_from)
+                        wm=wm,
+                        wpr_from=wpr_from,
+                    )
                     item.wpr = wpr
                     item.save()
+                wpr_id = str(wpr.id)
+                if wpr_id not in wpr_defect_snapshot:
+                    wpr_defect_snapshot[wpr_id] = [
+                        {
+                            "defect_id": str(defect["defect_id"]),
+                            "is_main": defect["is_main"],
+                        }
+                        for defect in WprDefect.objects.filter(
+                            wpr=wpr,
+                        ).values("defect_id", "is_main")
+                    ]
+                wpr.apply_ftest(item.ftest)
         
     mlog.submit_time = now
     mlog.submit_user = user
     mlog.stored_notok = stored_notok
-    mlog.stored_mgroup = stored_mgroup
+    mlog.stored_scope = stored_scope
     if mlog.work_end_time is None and mlog.work_start_time is not None:
         mlog.work_end_time = now
+    mlog.wpr_defect_snapshot = wpr_defect_snapshot
     mlog.save()
 
     # 更新任务进度
@@ -448,8 +602,10 @@ def mlog_revert(mlog: Mlog, user: User, now: Union[datetime.datetime, None]):
     material_out:Material = mlog.material_out
     material_in:Material = mlog.material_in
     stored_notok = mlog.stored_notok
-    stored_mgroup = mlog.stored_mgroup
+    stored_scope = mlog.stored_scope
+    stored_location = WmScope.resolve_location(stored_scope, mgroup)
     is_fix = mlog.is_fix
+    has_wpr_defect_snapshot = mlog.wpr_defect_snapshot is not None
     
     if is_fix:
         can_matoutIds = process.get_canout_mat_ids()
@@ -523,15 +679,18 @@ def mlog_revert(mlog: Mlog, user: User, now: Union[datetime.datetime, None]):
             wm_state = WMaterial.WM_OK if notok_sign_or_defect is None or (
                     isinstance(notok_sign_or_defect, Defect) and notok_sign_or_defect.okcate in [Defect.DEFECT_OK, Defect.DEFECT_OK_B]
                 ) else WMaterial.WM_NOTOK
-        lookup = {'batch': mo_batch, 'material': mo_ma, 'mgroup': None, 'notok_sign': None, 'defect': None, 'state': wm_state}
+        lookup = {
+            'batch': mo_batch,
+            'material': mo_ma,
+            'notok_sign': None,
+            'defect': None,
+            'state': wm_state,
+            **stored_location,
+        }
         if isinstance(notok_sign_or_defect, Defect):
             lookup['defect'] = notok_sign_or_defect
         else:
             lookup['notok_sign'] = notok_sign_or_defect
-        if stored_mgroup:
-            lookup['mgroup'] = mgroup
-        else:
-            lookup['belong_dept'] = belong_dept
         wm_qs = WMaterial.objects.filter(**lookup)
         count_x = wm_qs.count()
         if count_x == 1:
@@ -561,7 +720,8 @@ def mlog_revert(mlog: Mlog, user: User, now: Union[datetime.datetime, None]):
                 if mlogbws.count() != mo_count:
                     raise ParseError("日志与明细数量不一致,操作失败")
                 for item in mlogbws:
-                    Wpr.change_or_new(wpr=item.wpr, old_wm=wm)
+                    wpr = Wpr.change_or_new(wpr=item.wpr, old_wm=wm)
+                    wpr.revert_ftest_info(item.ftest)
 
     # 再生成消耗
     m_ins_list = []
@@ -584,18 +744,19 @@ def mlog_revert(mlog: Mlog, user: User, now: Union[datetime.datetime, None]):
             continue
         if mi_count <= 0:
             raise ParseError('存在非正数!')
-        if isinstance(mlog_or_b, Mlogb) and mlog_or_b.wm_in:
+        wm_in = getattr(mlog_or_b, 'wm_in', None)
-            wm = WMaterial.objects.select_for_update().get(id=mlog_or_b.wm_in.id)
+        if wm_in:
+            wm = WMaterial.objects.select_for_update().get(id=wm_in.id)
         else:
             # 针对光子的情况，实际上必须需要wm_in
-            lookup = {'batch': mi_batch, 'material': mi_ma, 'mgroup': None, 'state': WMaterial.WM_OK}
+            lookup = {
-            if stored_mgroup:
+                'batch': mi_batch,
-                # 退回到本工段
+                'material': mi_ma,
-                lookup['mgroup'] = mgroup
+                'state': WMaterial.WM_OK,
-            else:
+                **stored_location,
-                lookup['belong_dept'] = belong_dept
+            }
-
+            wm, _ = WMaterial.locked_get_or_create_inventory(
-            wm, _ = WMaterial.locked_get_or_create(**lookup, defaults={"belong_dept": belong_dept})
+                **lookup, defaults={})
         wm.count = wm.count + mi_count
         wm.update_by = user
         wm.save()
@@ -604,7 +765,9 @@ def mlog_revert(mlog: Mlog, user: User, now: Union[datetime.datetime, None]):
             if mlogbws.count() != mi_count:
                 raise ParseError("日志与实际使用明细数量不一致,操作失败")
             for item in mlogbws:
-                Wpr.change_or_new(wpr=item.wpr, wm=wm)
+                wpr = Wpr.change_or_new(wpr=item.wpr, wm=wm)
+                if not has_wpr_defect_snapshot:
+                    restore_legacy_wpr_defects_from_wm(wpr, wm)
 
     # 针对加工前不良的暂时额外处理 
     if stored_notok:
@@ -615,12 +778,15 @@ def mlog_revert(mlog: Mlog, user: User, now: Union[datetime.datetime, None]):
             c_state = WMaterial.WM_NOTOK
             if isinstance(defect, Defect) and defect.okcate == Defect.DEFECT_OK_B:
                 c_state = WMaterial.WM_OK
-            lookup = {'batch': batch, 'material': material, 'mgroup': None, 'defect': defect, 'state': c_state}
+            lookup = {
-            if stored_mgroup:
+                'batch': batch,
-                lookup['mgroup'] = mgroup
+                'material': material,
-            else:
+                'defect': defect,
-                lookup['belong_dept'] = belong_dept
+                'state': c_state,
-            wm, is_create = WMaterial.locked_get_or_create(**lookup, defaults={"belong_dept": belong_dept})
+                **stored_location,
+            }
+            wm, is_create = WMaterial.locked_get_or_create_inventory(
+                **lookup, defaults={})
             wm.count = wm.count - count
             if wm.count < 0:
                 raise ParseError('加工前不良数量大于库存量')
@@ -634,7 +800,18 @@ def mlog_revert(mlog: Mlog, user: User, now: Union[datetime.datetime, None]):
                 if mlogbws.count() != count:
                     raise ParseError("日志与加工前不良明细数量不一致,操作失败")
                 for item in mlogbws:
-                    Wpr.change_or_new(wpr=item.wpr, wm=wm)
+                    wpr = Wpr.change_or_new(wpr=item.wpr, wm=wm)
+                    if not has_wpr_defect_snapshot:
+                        restore_legacy_wpr_defects_from_wm(wpr, wm)
+
+    if has_wpr_defect_snapshot:
+        restore_mlog_wpr_defects(mlog)
+        mlog.wpr_defect_snapshot = None
+    else:
+        myLogger.warning(
+            "生产日志%s无单件缺陷快照，测回按历史兼容逻辑处理，无法精确恢复提交前缺陷",
+            mlog.id,
+        )
 
     mlog.submit_time = None
     mlog.submit_user = None
@@ -845,24 +1022,30 @@ def handover_submit(handover:Handover, user: User, now: Union[datetime.datetime,
         WMaterial.decrease(wm_id=wm_from.id, user=user, count=xcount)
 
         if need_add:
+            target_defect = (
+                merge_target_defect
+                if mtype == Handover.H_MERGE
+                else wm_from.defect
+            )
             # 开始变动
             if handover.type == Handover.H_NORMAL:
                 if mtype == Handover.H_MERGE and handover.new_wm:
                     wm_to = WMaterial.objects.select_for_update().get(id=handover.new_wm.id)
-                    defect_ok = wm_to.defect == wm_from.defect or (
+                    if (
-                        merge_clear_defect and wm_to.defect is None and wm_from.state == WMaterial.WM_OK
+                        wm_to.state != wm_from.state
-                    )
+                        or wm_to.material != wm_from.material
-                    if wm_to.state != wm_from.state or wm_to.material != wm_from.material or not defect_ok:
+                        or wm_to.defect != target_defect
+                    ):
                         raise ParseError("正常合并到的车间库存状态或物料异常")
                 else:
-                    wm_to, _ = WMaterial.locked_get_or_create(
+                    wm_to, _ = WMaterial.locked_get_or_create_inventory(
                         batch=batch,
                         material=material,
                         mgroup=recive_mgroup,
                         belong_dept=recive_dept,
                         state=wm_from.state,
                         notok_sign=wm_from.notok_sign,
-                        defect=merge_target_defect if mtype == Handover.H_MERGE else wm_from.defect,
+                        defect=target_defect,
                         defaults={
                             "batch_ofrom": wm_from.batch_ofrom,
                             "material_ofrom": wm_from.material_ofrom,
@@ -876,16 +1059,20 @@ def handover_submit(handover:Handover, user: User, now: Union[datetime.datetime,
                 wm_state = WMaterial.WM_REPAIR
                 if mtype == Handover.H_MERGE and handover.new_wm:
                     wm_to = WMaterial.objects.select_for_update().get(id=handover.new_wm.id)
-                    if wm_to.state != WMaterial.WM_REPAIR or wm_to.material != wm_from.material or wm_to.defect != wm_from.defect:
+                    if (
+                        wm_to.state != WMaterial.WM_REPAIR
+                        or wm_to.material != wm_from.material
+                        or wm_to.defect != target_defect
+                    ):
                         raise ParseError("返修合并到的车间库存状态或物料异常")
                 elif recive_mgroup:
-                    wm_to, _ = WMaterial.locked_get_or_create(
+                    wm_to, _ = WMaterial.locked_get_or_create_inventory(
                         batch=batch,
                         material=material,
                         mgroup=recive_mgroup,
                         belong_dept=recive_dept,
                         notok_sign=wm_from.notok_sign,
-                        defect=wm_from.defect,
+                        defect=target_defect,
                         material_origin=material,
                         state=wm_state,
                         defaults={
@@ -900,16 +1087,20 @@ def handover_submit(handover:Handover, user: User, now: Union[datetime.datetime,
             elif handover.type == Handover.H_SCRAP:
                 if mtype == Handover.H_MERGE and handover.new_wm:
                     wm_to = WMaterial.objects.select_for_update().get(id=handover.new_wm.id)
-                    if wm_to.state != WMaterial.WM_SCRAP or wm_to.material != wm_from.material or wm_to.defect != wm_from.defect:
+                    if (
+                        wm_to.state != WMaterial.WM_SCRAP
+                        or wm_to.material != wm_from.material
+                        or wm_to.defect != target_defect
+                    ):
                         raise ParseError("报废合并到的车间库存状态或物料异常")
                 elif recive_mgroup:
-                    wm_to, _ = WMaterial.locked_get_or_create(
+                    wm_to, _ = WMaterial.locked_get_or_create_inventory(
                         batch=batch,
                         material=material,
                         mgroup=recive_mgroup,
                         belong_dept=recive_dept,
                         notok_sign=wm_from.notok_sign,
-                        defect=wm_from.defect,
+                        defect=target_defect,
                         state=WMaterial.WM_SCRAP,
                         defaults={
                         "batch_ofrom": wm_from.batch_ofrom,
@@ -923,17 +1114,21 @@ def handover_submit(handover:Handover, user: User, now: Union[datetime.datetime,
             elif handover.type == Handover.H_CHANGE:
                 if mtype == Handover.H_MERGE and handover.new_wm:
                     wm_to = WMaterial.objects.select_for_update().get(id=handover.new_wm.id)
-                    if wm_to.material != handover.material_changed or wm_to.state != handover.state_changed:
+                    if (
+                        wm_to.material != handover.material_changed
+                        or wm_to.state != handover.state_changed
+                        or wm_to.defect != target_defect
+                    ):
                         raise ParseError("改版合并到的车间库存状态或物料异常")
                 elif handover.recive_mgroup:
-                    wm_to, _ = WMaterial.locked_get_or_create(
+                    wm_to, _ = WMaterial.locked_get_or_create_inventory(
                         batch=batch,
                         material=handover.material_changed,
                         state=handover.state_changed,
                         mgroup=recive_mgroup,
                         belong_dept=recive_dept,
                         notok_sign=wm_from.notok_sign,
-                        defect=wm_from.defect,
+                        defect=target_defect,
                         material_origin=None if handover.material_changed == material else material,
                         state_origin=None if handover.state_changed == wm_from.state else wm_from.state,
                         defaults={
@@ -950,15 +1145,21 @@ def handover_submit(handover:Handover, user: User, now: Union[datetime.datetime,
                     raise ParseError("仅合格品支持退回")
                 if mtype == Handover.H_MERGE and handover.new_wm:
                     wm_to = WMaterial.objects.select_for_update().get(id=handover.new_wm.id)
+                    if (
+                        wm_to.state != wm_from.state
+                        or wm_to.material != wm_from.material
+                        or wm_to.defect != target_defect
+                    ):
+                        raise ParseError("退料合并到的车间库存状态或物料异常")
                 else:
-                    wm_to, _ = WMaterial.locked_get_or_create(
+                    wm_to, _ = WMaterial.locked_get_or_create_inventory(
                         batch=batch,
                         material=material,
                         mgroup=recive_mgroup,
                         belong_dept=recive_dept,
                         state=wm_from.state,
                         notok_sign=wm_from.notok_sign,
-                        defect=merge_target_defect if mtype == Handover.H_MERGE else wm_from.defect,
+                        defect=target_defect,
                         defaults={
                             "batch_ofrom": wm_from.batch_ofrom,
                             "material_ofrom": wm_from.material_ofrom,
@@ -983,7 +1184,12 @@ def handover_submit(handover:Handover, user: User, now: Union[datetime.datetime,
                     raise ParseError(f'{batch}物料明细中存在{len(wm_ids)}个不同物料批次')
                 for item in handoverbws:
                     wpr:Wpr = item.wpr
-                    Wpr.change_or_new(wpr=wpr, wm=wm_to, old_wm=wpr.wm, old_mb=wpr.mb)
+                    Wpr.change_or_new(
+                        wpr=wpr,
+                        wm=wm_to,
+                        old_wm=wpr.wm,
+                        old_mb=wpr.mb,
+                    )
                 db_count = WMaterial.objects.filter(id=wm_to.id).values_list("count", flat=True).get()
                 if db_count != Wpr.objects.filter(wm=wm_to).count():
                     raise ParseError("交接与明细数量不一致2,操作失败")
@@ -1038,7 +1244,13 @@ def handover_revert(handover:Handover, handler:User=None):
                 wm = WMaterial.objects.get(id=wm.id)
                 for item in handoverbws:
                     wpr:Wpr = item.wpr
-                    Wpr.change_or_new(wpr=wpr, wm=wm, old_wm=wpr.wm, old_mb=wpr.mb, add_version=False)
+                    Wpr.change_or_new(
+                        wpr=wpr,
+                        wm=wm,
+                        old_wm=wpr.wm,
+                        old_mb=wpr.mb,
+                        add_version=False,
+                    )
 
     handover.submit_time = None
     handover.submit_user = None

apps/wpm/tests.py

@@ -1,3 +1,854 @@
-from django.test import TestCase
+from concurrent.futures import ThreadPoolExecutor
+from decimal import Decimal
+from threading import Barrier
+from types import SimpleNamespace
+from unittest.mock import MagicMock, patch
 
-# Create your tests here.
+from django.db import connection, connections, transaction
+from django.test import SimpleTestCase, TestCase, TransactionTestCase
+from unittest import skipUnless
+
+from apps.mtm.models import Material, Mgroup, WmScope
+from apps.qm.models import Defect
+from apps.system.models import Dept, User
+from apps.wpm.filters import HandoverFilter, WMaterialFilter
+from apps.wpm.models import Handover, WMaterial
+from apps.wpm.serializers import (
+    HandoverSerializer,
+    WMaterialCreateSerializer,
+    WMaterialSerializer,
+)
+from apps.wpm.services import restore_mlog_wpr_defects, snapshot_mlog_wpr_defects
+from apps.wpm.views import MlogbwViewSet
+from apps.wpmw.models import Wpr, WprDefect
+from rest_framework.exceptions import ParseError
+
+
+class HandoverFilterTests(TestCase):
+    @classmethod
+    def setUpTestData(cls):
+        cls.dept = Dept.objects.create(name="交接筛选测试车间")
+        cls.other_dept = Dept.objects.create(name="其他测试车间")
+        cls.mgroup = Mgroup.objects.create(
+            name="交接筛选测试工段",
+            belong_dept=cls.dept,
+        )
+        cls.peer_mgroup = Mgroup.objects.create(
+            name="同车间其他工段",
+            belong_dept=cls.dept,
+        )
+        cls.other_mgroup = Mgroup.objects.create(
+            name="其他车间工段",
+            belong_dept=cls.other_dept,
+        )
+        cls.user = User.objects.create_user(username="handover-filter-user")
+        cls.material = Material.objects.create(name="交接筛选测试物料")
+
+    @classmethod
+    def create_handover(cls, **kwargs):
+        defaults = {
+            "send_date": "2026-07-28",
+            "send_user": cls.user,
+            "material": cls.material,
+        }
+        defaults.update(kwargs)
+        return Handover.objects.create(**defaults)
+
+    def test_mgroup_only_returns_records_involving_the_selected_mgroup(self):
+        sent = self.create_handover(
+            send_mgroup=self.mgroup,
+            recive_mgroup=self.peer_mgroup,
+        )
+        received = self.create_handover(
+            send_mgroup=self.peer_mgroup,
+            recive_mgroup=self.mgroup,
+        )
+        self.create_handover(
+            send_mgroup=self.peer_mgroup,
+            recive_mgroup=self.other_mgroup,
+        )
+
+        queryset = HandoverFilter(
+            {"mgroup": self.mgroup.id},
+            queryset=Handover.objects.all(),
+        ).qs
+
+        self.assertCountEqual(queryset.values_list("id", flat=True), [
+            sent.id,
+            received.id,
+        ])
+
+    def test_mgroupx_adds_only_department_level_records_for_same_dept(self):
+        direct = self.create_handover(send_mgroup=self.mgroup)
+        dept_level = self.create_handover(
+            send_dept=self.dept,
+            send_mgroup=None,
+        )
+        self.create_handover(
+            send_mgroup=self.peer_mgroup,
+            recive_mgroup=self.other_mgroup,
+        )
+        self.create_handover(
+            send_dept=self.other_dept,
+            send_mgroup=None,
+        )
+
+        queryset = HandoverFilter(
+            {"mgroupx": self.mgroup.id},
+            queryset=Handover.objects.all(),
+        ).qs
+
+        self.assertCountEqual(queryset.values_list("id", flat=True), [
+            direct.id,
+            dept_level.id,
+        ])
+
+
+class WMaterialDefectGradeTests(TestCase):
+    @classmethod
+    def setUpTestData(cls):
+        cls.material = Material.objects.create(name="缺陷等级测试物料")
+        cls.defect_b = Defect.objects.create(
+            name="B类缺陷",
+            cate=Defect.cate_list[0],
+            okcate=Defect.DEFECT_OK_B,
+        )
+        cls.defect_notok = Defect.objects.create(
+            name="不合格缺陷",
+            cate=Defect.cate_list[0],
+            okcate=Defect.DEFECT_NOTOK,
+        )
+        cls.normal_with_notok_defect = WMaterial.objects.create(
+            material=cls.material,
+            batch="NORMAL-NOTOK",
+            count=1,
+            state=WMaterial.WM_OK,
+            defect=cls.defect_notok,
+        )
+        cls.notok_with_b_defect = WMaterial.objects.create(
+            material=cls.material,
+            batch="NOTOK-B",
+            count=1,
+            state=WMaterial.WM_NOTOK,
+            defect=cls.defect_b,
+        )
+        cls.normal_without_defect = WMaterial.objects.create(
+            material=cls.material,
+            batch="NORMAL-NONE",
+            count=1,
+            state=WMaterial.WM_OK,
+        )
+        cls.repair_without_defect = WMaterial.objects.create(
+            material=cls.material,
+            batch="REPAIR-NONE",
+            count=1,
+            state=WMaterial.WM_REPAIR,
+        )
+        cls.notok_with_legacy_sign = WMaterial.objects.create(
+            material=cls.material,
+            batch="NOTOK-LEGACY",
+            count=1,
+            state=WMaterial.WM_NOTOK,
+            notok_sign="zw",
+        )
+
+    def test_serializer_exposes_effective_defect_grade_without_using_state(self):
+        normal_notok_data = WMaterialSerializer(
+            self.normal_with_notok_defect
+        ).data
+        notok_b_data = WMaterialSerializer(self.notok_with_b_defect).data
+        no_defect_data = WMaterialSerializer(self.normal_without_defect).data
+        repair_no_defect_data = WMaterialSerializer(
+            self.repair_without_defect
+        ).data
+        legacy_data = WMaterialSerializer(self.notok_with_legacy_sign).data
+
+        self.assertEqual(
+            normal_notok_data["defect_grade"], Defect.DEFECT_NOTOK
+        )
+        self.assertEqual(normal_notok_data["defect_grade_name"], "不合格")
+        self.assertEqual(
+            notok_b_data["defect_grade"], Defect.DEFECT_OK_B
+        )
+        self.assertEqual(notok_b_data["defect_grade_name"], "合格B类")
+        self.assertEqual(no_defect_data["defect_grade"], Defect.DEFECT_OK)
+        self.assertEqual(no_defect_data["defect_grade_name"], "合格")
+        self.assertEqual(
+            repair_no_defect_data["defect_grade"], Defect.DEFECT_OK
+        )
+        self.assertEqual(repair_no_defect_data["defect_grade_name"], "合格")
+        self.assertEqual(legacy_data["defect_grade"], Defect.DEFECT_NOTOK)
+        self.assertEqual(legacy_data["defect_grade_name"], "不合格")
+
+    def test_filtering_state_and_defect_grade_are_independent(self):
+        normal_notok = WMaterialFilter(
+            {
+                "state": WMaterial.WM_OK,
+                "defect__okcate": Defect.DEFECT_NOTOK,
+            },
+            queryset=WMaterial.objects.all(),
+        ).qs
+        notok_b = WMaterialFilter(
+            {
+                "state": WMaterial.WM_NOTOK,
+                "defect__okcate": Defect.DEFECT_OK_B,
+            },
+            queryset=WMaterial.objects.all(),
+        ).qs
+
+        self.assertQuerySetEqual(
+            normal_notok,
+            [self.normal_with_notok_defect],
+            transform=lambda item: item,
+        )
+        self.assertQuerySetEqual(
+            notok_b,
+            [self.notok_with_b_defect],
+            transform=lambda item: item,
+        )
+
+    def test_effective_grade_filter_includes_defaults_and_legacy_signs(self):
+        ok_items = WMaterialFilter(
+            {"defect_grade": Defect.DEFECT_OK},
+            queryset=WMaterial.objects.all(),
+        ).qs
+        notok_items = WMaterialFilter(
+            {"defect_grade": Defect.DEFECT_NOTOK},
+            queryset=WMaterial.objects.all(),
+        ).qs
+
+        self.assertCountEqual(
+            ok_items.values_list("id", flat=True),
+            [self.normal_without_defect.id, self.repair_without_defect.id],
+        )
+        self.assertCountEqual(
+            notok_items.values_list("id", flat=True),
+            [
+                self.normal_with_notok_defect.id,
+                self.notok_with_legacy_sign.id,
+            ],
+        )
+
+
+class MlogbwViewSetTests(SimpleTestCase):
+    @patch("apps.wpm.views.MlogViewSet.lock_and_check_can_update")
+    @patch("apps.wpm.views.Mlogbw.cal_count_notok")
+    @patch("apps.wpm.views.Mlogbw.objects.get_or_create")
+    @patch("apps.wpm.views.Mlogb.objects.filter")
+    def test_perform_create_syncs_fix_output_without_route(
+        self,
+        mlogb_filter,
+        mlogbw_get_or_create,
+        cal_count_notok,
+        lock_and_check_can_update,
+    ):
+        wm = object()
+        wpr = SimpleNamespace(wm=wm)
+        material = SimpleNamespace(tracking=Material.MA_TRACKING_SINGLE)
+        mlog = SimpleNamespace(
+            route=None,
+            is_fix=True,
+            cal_mlog_count_from_mlogb=MagicMock(),
+        )
+        mlogb_in = SimpleNamespace(
+            mlog=mlog,
+            route=None,
+            wm_in=wm,
+            material_in=material,
+        )
+        mlogb_out = object()
+        mlogb_qs = MagicMock()
+        mlogb_qs.exists.return_value = True
+        mlogb_qs.__iter__.return_value = iter([mlogb_out])
+        mlogb_filter.return_value = mlogb_qs
+        ins = SimpleNamespace(
+            mlogb=mlogb_in,
+            wpr=wpr,
+            number="2606P1095-3",
+        )
+        serializer = MagicMock()
+        serializer.save.return_value = ins
+        lock_and_check_can_update.return_value = mlog
+
+        MlogbwViewSet().perform_create(serializer)
+
+        mlogbw_get_or_create.assert_called_once_with(
+            mlogb=mlogb_out,
+            wpr=wpr,
+            defaults={
+                "number": "2606P1095-3",
+                "mlogbw_from": ins,
+            },
+        )
+        self.assertEqual(cal_count_notok.call_count, 2)
+        mlog.cal_mlog_count_from_mlogb.assert_called_once_with()
+
+
+class WMaterialScopeTests(SimpleTestCase):
+    def test_inventory_key_normalizes_omitted_nullable_fields(self):
+        material = Material(id="100", name="测试物料")
+
+        omitted = WMaterial._normalize_inventory_lookup(
+            material=material,
+            batch="BATCH-001",
+            state=WMaterial.WM_OK,
+        )
+        explicit = WMaterial._normalize_inventory_lookup(
+            material=material,
+            batch="BATCH-001",
+            mgroup=None,
+            belong_dept=None,
+            state=WMaterial.WM_OK,
+            defect=None,
+            notok_sign=None,
+            material_origin=None,
+            state_origin=None,
+        )
+
+        self.assertEqual(omitted, explicit)
+        self.assertEqual(
+            WMaterial._inventory_advisory_lock_payload(omitted),
+            WMaterial._inventory_advisory_lock_payload(explicit),
+        )
+
+    def test_scope_resolver_returns_consistent_location_fields(self):
+        dept = object()
+        mgroup = SimpleNamespace(belong_dept=dept)
+
+        self.assertEqual(
+            WmScope.resolve_location(WmScope.MGROUP, mgroup),
+            {"mgroup": mgroup, "belong_dept": dept},
+        )
+        self.assertEqual(
+            WmScope.resolve_location(WmScope.DEPT, mgroup),
+            {"mgroup": None, "belong_dept": dept},
+        )
+        self.assertEqual(
+            WmScope.resolve_location(WmScope.GLOBAL, mgroup),
+            {"mgroup": None, "belong_dept": None},
+        )
+
+    def test_only_mgroup_scope_requires_handover_mgroup(self):
+        self.assertTrue(WmScope.requires_mgroup(WmScope.MGROUP))
+        self.assertFalse(WmScope.requires_mgroup(WmScope.DEPT))
+        self.assertFalse(WmScope.requires_mgroup(WmScope.GLOBAL))
+
+    def test_global_scope_key_does_not_access_missing_relations(self):
+        wm = WMaterial()
+
+        self.assertEqual(wm.belong_dept_or_mgroup_id, ("global", None))
+
+    def test_department_and_mgroup_scope_keys_cannot_collide(self):
+        dept_wm = WMaterial(belong_dept_id=10)
+        mgroup_wm = WMaterial(mgroup_id=10)
+
+        self.assertNotEqual(
+            dept_wm.belong_dept_or_mgroup_id,
+            mgroup_wm.belong_dept_or_mgroup_id,
+        )
+
+    def test_manual_create_allows_global_scope(self):
+        serializer = WMaterialCreateSerializer()
+        attrs = {
+            "material": Material(),
+            "count": 1,
+            "batch": "GLOBAL-001",
+        }
+
+        validated = serializer.validate(attrs)
+
+        self.assertFalse(serializer.fields["mgroup"].required)
+        self.assertFalse(serializer.fields["belong_dept"].required)
+        self.assertNotIn("mgroup", validated)
+        self.assertNotIn("belong_dept", validated)
+
+    def test_global_scope_can_be_filtered_explicitly(self):
+        self.assertIn(
+            "isnull",
+            WMaterialFilter.Meta.fields["belong_dept"],
+        )
+        self.assertIn(
+            "isnull",
+            WMaterialFilter.Meta.fields["mgroup"],
+        )
+
+    def test_manual_create_derives_department_from_mgroup(self):
+        dept = SimpleNamespace(id=20)
+        mgroup = SimpleNamespace(id=10, belong_dept=dept)
+        attrs = {
+            "material": Material(),
+            "count": 1,
+            "batch": "MGROUP-001",
+            "mgroup": mgroup,
+        }
+
+        validated = WMaterialCreateSerializer().validate(attrs)
+
+        self.assertIs(validated["belong_dept"], dept)
+
+    def test_global_inventory_cannot_enter_normal_handover(self):
+        wm = WMaterial(
+            material=Material(tracking=Material.MA_TRACKING_BATCH),
+            batch="GLOBAL-001",
+            count=1,
+        )
+
+        with self.assertRaisesMessage(ParseError, "全局库存无需正常交接"):
+            HandoverSerializer().validate({
+                "wm": wm,
+                "count": 1,
+                "type": Handover.H_NORMAL,
+                "mtype": Handover.H_NORMAL,
+            })
+
+    def test_global_inventory_split_stays_global(self):
+        wm = WMaterial(
+            material=Material(tracking=Material.MA_TRACKING_BATCH),
+            batch="GLOBAL-001",
+            count=1,
+        )
+
+        validated = HandoverSerializer().validate({
+            "wm": wm,
+            "count": 1,
+            "type": Handover.H_NORMAL,
+            "mtype": Handover.H_DIV,
+        })
+
+        self.assertIsNone(validated["send_dept"])
+        self.assertIsNone(validated["recive_dept"])
+        self.assertIsNone(validated["recive_mgroup"])
+
+    def test_global_inventory_cannot_merge_into_scoped_inventory(self):
+        material = Material(tracking=Material.MA_TRACKING_BATCH)
+        wm = WMaterial(material=material, batch="GLOBAL-001", count=1)
+        target = WMaterial(
+            material=material,
+            batch="GLOBAL-MERGED",
+            count=0,
+            belong_dept_id=20,
+        )
+
+        with self.assertRaisesMessage(ParseError, "全局库存合批目标必须是全局库存"):
+            HandoverSerializer().validate({
+                "wm": wm,
+                "count": 1,
+                "new_wm": target,
+                "type": Handover.H_NORMAL,
+                "mtype": Handover.H_MERGE,
+            })
+
+    def test_global_inventory_scrap_requires_receiving_mgroup(self):
+        wm = WMaterial(
+            material=Material(tracking=Material.MA_TRACKING_BATCH),
+            batch="GLOBAL-001",
+            count=1,
+        )
+
+        with self.assertRaisesMessage(
+            ParseError,
+            "全局库存返修、报废或改版必须指定接收工段",
+        ):
+            HandoverSerializer().validate({
+                "wm": wm,
+                "count": 1,
+                "type": Handover.H_SCRAP,
+                "mtype": Handover.H_NORMAL,
+            })
+
+    def test_handover_keeps_explicit_sending_mgroup(self):
+        inventory_dept = Dept(id="20", name="库存车间")
+        sending_dept = Dept(id="30", name="送料车间")
+        mgroup = Mgroup(id="10", name="测试工段", belong_dept=sending_dept)
+        wm = WMaterial(
+            material=Material(tracking=Material.MA_TRACKING_BATCH),
+            batch="DEPT-001",
+            count=1,
+            belong_dept=inventory_dept,
+        )
+
+        validated = HandoverSerializer().validate({
+            "wm": wm,
+            "count": 1,
+            "send_mgroup": mgroup,
+            "recive_dept": object(),
+            "type": Handover.H_NORMAL,
+            "mtype": Handover.H_NORMAL,
+        })
+
+        self.assertIs(validated["send_mgroup"], mgroup)
+        self.assertIs(validated["send_dept"], sending_dept)
+
+    def test_batch_tracking_merge_can_clear_ok_b_batch_defects(self):
+        material = Material(tracking=Material.MA_TRACKING_BATCH)
+        defect_a = Defect(id="1", okcate=Defect.DEFECT_OK_B)
+        defect_b = Defect(id="2", okcate=Defect.DEFECT_OK_B)
+        wm_a = WMaterial(
+            id="10", material=material, batch="B-001", count=1,
+            state=WMaterial.WM_OK, defect=defect_a,
+        )
+        wm_b = WMaterial(
+            id="20", material=material, batch="B-002", count=1,
+            state=WMaterial.WM_OK, defect=defect_b,
+        )
+
+        validated = HandoverSerializer().validate({
+            "wm": wm_a,
+            "handoverb": [
+                {"wm": wm_a, "count": 1},
+                {"wm": wm_b, "count": 1},
+            ],
+            "new_batch": "B-MERGED",
+            "clear_defect": True,
+            "type": Handover.H_NORMAL,
+            "mtype": Handover.H_MERGE,
+        })
+
+        self.assertTrue(validated["clear_defect"])
+        self.assertEqual(validated["count"], 2)
+
+    def test_batch_tracking_merge_can_clear_ok_and_ok_b_defects(self):
+        material = Material(tracking=Material.MA_TRACKING_BATCH)
+        defect_b = Defect(id="1", okcate=Defect.DEFECT_OK_B)
+        wm_ok = WMaterial(
+            id="10", material=material, batch="OK-001", count=1,
+            state=WMaterial.WM_OK, defect=None,
+        )
+        wm_b = WMaterial(
+            id="20", material=material, batch="B-001", count=1,
+            state=WMaterial.WM_OK, defect=defect_b,
+        )
+
+        validated = HandoverSerializer().validate({
+            "wm": wm_ok,
+            "handoverb": [
+                {"wm": wm_ok, "count": 1},
+                {"wm": wm_b, "count": 1},
+            ],
+            "new_batch": "OK-MERGED",
+            "clear_defect": True,
+            "type": Handover.H_NORMAL,
+            "mtype": Handover.H_MERGE,
+        })
+
+        self.assertTrue(validated["clear_defect"])
+        self.assertEqual(validated["count"], 2)
+
+    def test_batch_tracking_merge_cannot_clear_same_grade_notok_defects(self):
+        material = Material(tracking=Material.MA_TRACKING_BATCH)
+        defect_a = Defect(id="1", okcate=Defect.DEFECT_NOTOK)
+        defect_b = Defect(id="2", okcate=Defect.DEFECT_NOTOK)
+        wm_a = WMaterial(
+            id="10", material=material, batch="N-001", count=1,
+            state=WMaterial.WM_NOTOK, defect=defect_a,
+        )
+        wm_b = WMaterial(
+            id="20", material=material, batch="N-002", count=1,
+            state=WMaterial.WM_NOTOK, defect=defect_b,
+        )
+
+        with self.assertRaisesMessage(
+            ParseError,
+            "批次追踪物料仅合格品和合格B类可清除批次缺陷",
+        ):
+            HandoverSerializer().validate({
+                "wm": wm_a,
+                "handoverb": [
+                    {"wm": wm_a, "count": 1},
+                    {"wm": wm_b, "count": 1},
+                ],
+                "new_batch": "N-MERGED",
+                "clear_defect": True,
+                "type": Handover.H_NORMAL,
+                "mtype": Handover.H_MERGE,
+            })
+
+    def test_batch_tracking_merge_cannot_clear_mixed_defect_grades(self):
+        material = Material(tracking=Material.MA_TRACKING_BATCH)
+        defect_b = Defect(id="1", okcate=Defect.DEFECT_OK_B)
+        defect_notok = Defect(id="2", okcate=Defect.DEFECT_NOTOK)
+        wm_a = WMaterial(
+            id="10", material=material, batch="B-001", count=1,
+            state=WMaterial.WM_OK, defect=defect_b,
+        )
+        wm_b = WMaterial(
+            id="20", material=material, batch="N-001", count=1,
+            state=WMaterial.WM_OK, defect=defect_notok,
+        )
+
+        with self.assertRaisesMessage(
+            ParseError,
+            "批次追踪物料仅合格品和合格B类可清除批次缺陷",
+        ):
+            HandoverSerializer().validate({
+                "wm": wm_a,
+                "handoverb": [
+                    {"wm": wm_a, "count": 1},
+                    {"wm": wm_b, "count": 1},
+                ],
+                "new_batch": "MIXED-MERGED",
+                "clear_defect": True,
+                "type": Handover.H_NORMAL,
+                "mtype": Handover.H_MERGE,
+            })
+
+    def test_single_tracking_merge_can_clear_notok_batch_defect(self):
+        material = Material(tracking=Material.MA_TRACKING_SINGLE)
+        defect = Defect(id="1", okcate=Defect.DEFECT_NOTOK)
+        wm = WMaterial(
+            id="10", material=material, batch="S-001", count=1,
+            state=WMaterial.WM_NOTOK, defect=defect,
+        )
+        wm_b = WMaterial(
+            id="20", material=material, batch="S-002", count=1,
+            state=WMaterial.WM_NOTOK, defect=defect,
+        )
+
+        validated = HandoverSerializer().validate({
+            "wm": wm,
+            "handoverb": [
+                {"wm": wm, "count": 1},
+                {"wm": wm_b, "count": 1},
+            ],
+            "new_batch": "S-MERGED",
+            "clear_defect": True,
+            "type": Handover.H_NORMAL,
+            "mtype": Handover.H_MERGE,
+        })
+
+        self.assertTrue(validated["clear_defect"])
+        self.assertEqual(validated["count"], 2)
+
+
+class WprDefectLifecycleTests(SimpleTestCase):
+    @patch("apps.wpmw.models.WprDefect.objects.bulk_create")
+    @patch("apps.wpmw.models.WprDefect.objects.filter")
+    @patch(
+        "apps.wpmw.models.idWorker.get_id",
+        side_effect=["100", "101"],
+    )
+    def test_replace_defects_assigns_unique_ids_before_bulk_create(
+        self,
+        get_id,
+        defect_filter,
+        bulk_create,
+    ):
+        wpr = Wpr(id="1")
+
+        wpr.replace_defects([
+            {"defect_id": "10", "is_main": True},
+            {"defect_id": "11", "is_main": False},
+        ])
+
+        defect_filter.return_value.delete.assert_called_once_with()
+        created = bulk_create.call_args.args[0]
+        self.assertEqual([str(item.id) for item in created], ["100", "101"])
+        self.assertEqual(get_id.call_count, 2)
+
+    @patch("apps.wpmw.models.WprDefect.objects.filter")
+    def test_change_inventory_never_changes_individual_defects(self, defect_filter):
+        material = Material(id="1", tracking=Material.MA_TRACKING_SINGLE)
+        old_wm = WMaterial(
+            id="10", material=material, state=WMaterial.WM_OK,
+        )
+        target_wm = WMaterial(
+            id="20", material=material, state=WMaterial.WM_OK,
+            defect=None,
+        )
+        wpr = SimpleNamespace(
+            wm=old_wm,
+            mb=None,
+            material=material,
+            state=WMaterial.WM_OK,
+            version=1,
+            save=MagicMock(),
+        )
+
+        Wpr.change_or_new(
+            wpr=wpr,
+            wm=target_wm,
+            old_wm=old_wm,
+        )
+
+        defect_filter.assert_not_called()
+        self.assertIs(wpr.wm, target_wm)
+
+    @patch.object(Wpr, "replace_defects")
+    @patch("apps.wpmw.models.FtestDefect.objects.filter")
+    @patch("apps.wpmw.models.FtestItem.objects.filter")
+    def test_apply_ftest_replaces_defects_and_carries_info(
+        self,
+        ftest_item_filter,
+        ftest_defect_filter,
+        replace_defects,
+    ):
+        ftest = object()
+        ftest_defect_filter.return_value.values.return_value = [
+            {"defect_id": "30", "is_main": False},
+            {"defect_id": "31", "is_main": True},
+        ]
+        ftest_item_filter.return_value.select_related.return_value = [
+            SimpleNamespace(
+                id="40",
+                test_val_json={"value": 1},
+                testitem=SimpleNamespace(name="尺寸"),
+            ),
+        ]
+        wpr = Wpr(
+            oinfo={"existing": {"value": 2}},
+        )
+        wpr.save = MagicMock()
+
+        wpr.apply_ftest(ftest)
+
+        ftest_defect_filter.assert_called_once_with(ftest=ftest, has=True)
+        replace_defects.assert_called_once_with([
+            {"defect_id": "30", "is_main": False},
+            {"defect_id": "31", "is_main": True},
+        ])
+        self.assertEqual(
+            wpr.oinfo["40"],
+            {"name": "尺寸", "val": {"value": 1}},
+        )
+
+    @patch.object(Wpr, "replace_defects")
+    def test_apply_no_ftest_clears_defects(self, replace_defects):
+        wpr = Wpr(oinfo={"existing": True})
+
+        wpr.apply_ftest(None)
+
+        replace_defects.assert_called_once_with([])
+        self.assertEqual(wpr.oinfo, {"existing": True})
+
+    @patch("apps.wpmw.models.FtestItem.objects.filter")
+    def test_revert_ftest_info_only_removes_matching_keys(self, item_filter):
+        item_filter.return_value.values_list.return_value = ["10", "11"]
+        wpr = Wpr(oinfo={"10": {"v": 1}, "other": {"v": 2}})
+        wpr.save = MagicMock()
+
+        wpr.revert_ftest_info(object())
+
+        self.assertEqual(wpr.oinfo, {"other": {"v": 2}})
+        wpr.save.assert_called_once()
+
+
+class MlogWprDefectSnapshotTests(SimpleTestCase):
+    @patch("apps.wpm.services.WprDefect.objects.filter")
+    @patch("apps.wpm.services.Wpr.objects.select_for_update")
+    @patch("apps.wpm.services.Mlogbw.objects.filter")
+    def test_snapshot_keeps_multiple_defects_and_empty_wpr(
+        self,
+        mlogbw_filter,
+        select_for_update,
+        defect_filter,
+    ):
+        mlogbw_filter.return_value.values_list.return_value.distinct.return_value = [
+            "1",
+            "2",
+        ]
+        select_for_update.return_value.filter.return_value.values_list.return_value = [
+            "1",
+            "2",
+        ]
+        defect_filter.return_value.values.return_value = [
+            {"wpr_id": "1", "defect_id": "10", "is_main": True},
+            {"wpr_id": "1", "defect_id": "11", "is_main": False},
+        ]
+        mlog = SimpleNamespace(id="100", save=MagicMock())
+
+        snapshot = snapshot_mlog_wpr_defects(mlog)
+
+        self.assertEqual(snapshot["2"], [])
+        self.assertEqual(len(snapshot["1"]), 2)
+        select_for_update.return_value.filter.return_value.values_list.assert_called_once_with(
+            "id",
+            flat=True,
+        )
+
+    @patch("apps.wpm.services.WprDefect.objects.bulk_create")
+    @patch("apps.wpm.services.WprDefect.objects.filter")
+    @patch("apps.wpm.services.Wpr.objects.select_for_update")
+    @patch(
+        "apps.wpm.services.idWorker.get_id",
+        side_effect=["100", "101"],
+    )
+    def test_restore_replaces_current_defects_exactly(
+        self,
+        get_id,
+        select_for_update,
+        defect_filter,
+        bulk_create,
+    ):
+        wpr = Wpr(id="1")
+        select_for_update.return_value.filter.return_value = [wpr]
+        mlog = SimpleNamespace(
+            id="100",
+            wpr_defect_snapshot={
+                "1": [
+                    {"defect_id": "10", "is_main": True},
+                    {"defect_id": "11", "is_main": False},
+                ],
+            },
+        )
+
+        restored = restore_mlog_wpr_defects(mlog)
+
+        self.assertTrue(restored)
+        defect_filter.return_value.delete.assert_called_once_with()
+        created = bulk_create.call_args.args[0]
+        self.assertEqual(
+            [(str(item.defect_id), item.is_main) for item in created],
+            [("10", True), ("11", False)],
+        )
+        self.assertEqual([str(item.id) for item in created], ["100", "101"])
+        self.assertEqual(get_id.call_count, 2)
+
+
+@skipUnless(
+    connection.vendor == "postgresql",
+    "advisory lock concurrency test requires PostgreSQL",
+)
+class WMaterialConcurrencyTests(TransactionTestCase):
+    def test_concurrent_first_create_uses_one_inventory_record(self):
+        material = Material.objects.create(name="并发库存测试物料")
+        barrier = Barrier(2)
+
+        def create_inventory():
+            connections.close_all()
+            try:
+                barrier.wait()
+                with transaction.atomic():
+                    wm, created = (
+                        WMaterial.locked_get_or_create_inventory(
+                            material=material,
+                            batch="CONCURRENT-001",
+                            state=WMaterial.WM_OK,
+                            defaults={"count": Decimal("0")},
+                        )
+                    )
+                    wm.count += Decimal("1")
+                    wm.save(update_fields=["count"])
+                    return created
+            finally:
+                connections.close_all()
+
+        with ThreadPoolExecutor(max_workers=2) as executor:
+            created_results = list(executor.map(
+                lambda _: create_inventory(),
+                range(2),
+            ))
+
+        queryset = WMaterial.objects.filter(
+            material=material,
+            batch="CONCURRENT-001",
+            mgroup=None,
+            belong_dept=None,
+            state=WMaterial.WM_OK,
+            defect=None,
+            notok_sign=None,
+            material_origin=None,
+            state_origin=None,
+        )
+        self.assertEqual(queryset.count(), 1)
+        self.assertEqual(queryset.get().count, Decimal("2"))
+        self.assertCountEqual(created_results, [True, False])

apps/wpm/tests/__init__.py

@@ -0,0 +1 @@
+

apps/wpm/tests/test_defect_grade.py

@@ -0,0 +1,86 @@
+from django.test import TestCase
+
+from apps.mtm.models import Material
+from apps.qm.models import Defect
+from apps.wpm.filters import WMaterialFilter
+from apps.wpm.models import WMaterial
+from apps.wpm.serializers import WMaterialSerializer
+
+
+class WMaterialDefectGradeTests(TestCase):
+    @classmethod
+    def setUpTestData(cls):
+        cls.material = Material.objects.create(name="缺陷等级测试物料")
+        cls.defect_b = Defect.objects.create(
+            name="B类缺陷",
+            cate=Defect.cate_list[0],
+            okcate=Defect.DEFECT_OK_B,
+        )
+        cls.defect_notok = Defect.objects.create(
+            name="不合格缺陷",
+            cate=Defect.cate_list[0],
+            okcate=Defect.DEFECT_NOTOK,
+        )
+        cls.repair_without_defect = WMaterial.objects.create(
+            material=cls.material,
+            batch="REPAIR-NONE",
+            count=1,
+            state=WMaterial.WM_REPAIR,
+        )
+        cls.normal_with_notok_defect = WMaterial.objects.create(
+            material=cls.material,
+            batch="NORMAL-NOTOK",
+            count=1,
+            state=WMaterial.WM_OK,
+            defect=cls.defect_notok,
+        )
+        cls.notok_with_b_defect = WMaterial.objects.create(
+            material=cls.material,
+            batch="NOTOK-B",
+            count=1,
+            state=WMaterial.WM_NOTOK,
+            defect=cls.defect_b,
+        )
+        cls.notok_with_legacy_sign = WMaterial.objects.create(
+            material=cls.material,
+            batch="NOTOK-LEGACY",
+            count=1,
+            state=WMaterial.WM_NOTOK,
+            notok_sign="zw",
+        )
+
+    def test_serializer_uses_effective_grade_independent_of_state(self):
+        repair_data = WMaterialSerializer(self.repair_without_defect).data
+        notok_data = WMaterialSerializer(
+            self.normal_with_notok_defect
+        ).data
+        b_data = WMaterialSerializer(self.notok_with_b_defect).data
+        legacy_data = WMaterialSerializer(self.notok_with_legacy_sign).data
+
+        self.assertEqual(repair_data["defect_grade"], Defect.DEFECT_OK)
+        self.assertEqual(repair_data["defect_grade_name"], "合格")
+        self.assertEqual(notok_data["defect_grade"], Defect.DEFECT_NOTOK)
+        self.assertEqual(b_data["defect_grade"], Defect.DEFECT_OK_B)
+        self.assertEqual(legacy_data["defect_grade"], Defect.DEFECT_NOTOK)
+
+    def test_effective_grade_filter_matches_serializer_rules(self):
+        ok_items = WMaterialFilter(
+            {"defect_grade": Defect.DEFECT_OK},
+            queryset=WMaterial.objects.all(),
+        ).qs
+        notok_items = WMaterialFilter(
+            {"defect_grade": Defect.DEFECT_NOTOK},
+            queryset=WMaterial.objects.all(),
+        ).qs
+
+        self.assertCountEqual(
+            ok_items.values_list("id", flat=True),
+            [self.repair_without_defect.id],
+        )
+        self.assertCountEqual(
+            notok_items.values_list("id", flat=True),
+            [
+                self.normal_with_notok_defect.id,
+                self.notok_with_legacy_sign.id,
+            ],
+        )

apps/wpm/tests/test_equipment_log_options.py

@@ -0,0 +1,52 @@
+from unittest.mock import MagicMock, patch
+
+from django.test import SimpleTestCase
+
+from apps.wpm.services import get_recent_mgroup_equipment_ids
+
+
+class RecentMgroupEquipmentTests(SimpleTestCase):
+    @patch("apps.wpm.services.Mlog")
+    def test_collects_all_equipment_fields_in_log_order_without_duplicates(
+        self, mlog
+    ):
+        log_queryset = MagicMock()
+        mlog.objects.filter.return_value = log_queryset
+        log_queryset.order_by.return_value.values_list.return_value.__getitem__.return_value = [
+            ("log-new", "equipment-a", None),
+            ("log-middle", "equipment-b", "equipment-a"),
+            ("log-empty", None, None),
+        ]
+        through_queryset = MagicMock()
+        mlog.equipments.through.objects.filter.return_value = through_queryset
+        through_queryset.order_by.return_value.values_list.return_value = [
+            ("log-new", "equipment-c"),
+            ("log-middle", "equipment-c"),
+            ("log-middle", "equipment-d"),
+        ]
+
+        result = get_recent_mgroup_equipment_ids("mgroup-1", log_limit=50)
+
+        self.assertEqual(
+            result,
+            ["equipment-a", "equipment-c", "equipment-b", "equipment-d"],
+        )
+        log_queryset.order_by.return_value.values_list.return_value.__getitem__.assert_called_once_with(
+            slice(None, 50, None)
+        )
+
+    @patch("apps.wpm.services.Mlog")
+    def test_returns_empty_when_recent_logs_have_no_equipment(self, mlog):
+        log_queryset = MagicMock()
+        mlog.objects.filter.return_value = log_queryset
+        log_queryset.order_by.return_value.values_list.return_value.__getitem__.return_value = [
+            ("log-1", None, None),
+            ("log-2", None, None),
+        ]
+        through_queryset = MagicMock()
+        mlog.equipments.through.objects.filter.return_value = through_queryset
+        through_queryset.order_by.return_value.values_list.return_value = []
+
+        result = get_recent_mgroup_equipment_ids("mgroup-1")
+
+        self.assertEqual(result, [])

apps/wpm/tests/test_handover_serializer.py

@@ -0,0 +1,87 @@
+from django.db import connection
+from django.test import TestCase
+from django.test.utils import CaptureQueriesContext
+
+from apps.mtm.models import Material
+from apps.qm.models import Defect
+from apps.system.models import Dept, User
+from apps.wpm.models import Handover, Handoverb, WMaterial
+from apps.wpm.serializers import HandoverSerializer
+
+
+class HandoverSerializerQueryTests(TestCase):
+    @classmethod
+    def setUpTestData(cls):
+        cls.dept = Dept.objects.create(name="合批查询测试车间")
+        cls.user = User.objects.create_user(username="handover-query-user")
+        cls.material = Material.objects.create(name="合批查询测试物料")
+        cls.defect_b = Defect.objects.create(
+            name="合批查询测试B类缺陷",
+            cate=Defect.cate_list[0],
+            okcate=Defect.DEFECT_OK_B,
+        )
+        cls.inventories = [
+            WMaterial.objects.create(
+                material=cls.material,
+                batch=f"QUERY-{index}",
+                count=1,
+                state=WMaterial.WM_OK,
+                defect=cls.defect_b if index == 2 else None,
+                belong_dept=cls.dept,
+            )
+            for index in range(3)
+        ]
+
+    def test_handover_inventory_is_loaded_in_one_query(self):
+        serializer = HandoverSerializer(data={
+            "send_date": "2026-08-07",
+            "send_user": self.user.id,
+            "send_dept": self.dept.id,
+            "recive_dept": self.dept.id,
+            "handoverb": [
+                {"wm": inventory.id, "count": 1}
+                for inventory in self.inventories
+            ],
+            "new_batch": "QUERY-MERGED",
+            "clear_defect": True,
+            "type": Handover.H_NORMAL,
+            "mtype": Handover.H_MERGE,
+        })
+
+        with CaptureQueriesContext(connection) as queries:
+            self.assertTrue(serializer.is_valid(), serializer.errors)
+
+        inventory_queries = [
+            query["sql"] for query in queries
+            if 'FROM "wpm_wmaterial"' in query["sql"]
+        ]
+        self.assertEqual(len(inventory_queries), 1, inventory_queries)
+
+    def test_detail_contains_display_fields_for_handover_and_items(self):
+        handover = Handover.objects.create(
+            send_date="2026-08-07",
+            send_user=self.user,
+            send_dept=self.dept,
+            recive_dept=self.dept,
+            material=self.material,
+            wm=self.inventories[0],
+            count=1,
+            type=Handover.H_NORMAL,
+            mtype=Handover.H_NORMAL,
+        )
+        Handoverb.objects.create(
+            handover=handover,
+            wm=self.inventories[0],
+            batch=self.inventories[0].batch,
+            count=1,
+        )
+
+        data = HandoverSerializer(handover).data
+
+        self.assertEqual(data["send_dept_name"], self.dept.name)
+        self.assertEqual(data["recive_dept_name"], self.dept.name)
+        self.assertEqual(data["type_name"], "正常交接")
+        self.assertEqual(data["mtype_name"], "正常")
+        self.assertEqual(data["handoverb"][0]["material_name"], str(self.material))
+        self.assertEqual(data["handoverb"][0]["state_name"], "合格")
+        self.assertEqual(data["handoverb"][0]["count_available"], 1)

apps/wpm/tests/test_mlogbw_files.py

@@ -0,0 +1,36 @@
+from django.test import SimpleTestCase
+from django.urls import reverse
+
+from apps.system.models import File
+from apps.wpm.models import Mlogbw
+from apps.wpm.serializers import MlogbwFilesSerializer, MlogbwListSerializer
+from apps.wpm.views import MlogbwViewSet
+
+
+class MlogbwFilesDefinitionTests(SimpleTestCase):
+    def test_files_field_links_system_file(self):
+        field = Mlogbw._meta.get_field("files")
+
+        self.assertTrue(field.blank)
+        self.assertIs(field.remote_field.model, File)
+        self.assertEqual(field.remote_field.related_name, "mlogbw_outputs")
+
+    def test_list_serializer_exposes_file_ids_and_details(self):
+        fields = MlogbwListSerializer().fields
+
+        self.assertTrue(fields["files"].read_only)
+        self.assertTrue(fields["files_"].read_only)
+
+    def test_files_serializer_only_writes_file_relations(self):
+        fields = MlogbwFilesSerializer().fields
+
+        self.assertEqual(set(fields), {"id", "files", "files_"})
+        self.assertFalse(fields["files"].read_only)
+
+    def test_files_action_supports_read_and_update(self):
+        self.assertEqual(MlogbwViewSet.files.mapping["get"], "files")
+        self.assertEqual(MlogbwViewSet.files.mapping["put"], "files")
+        self.assertEqual(
+            reverse("mlogbw-files", kwargs={"pk": "single-id"}),
+            "/api/wpm/mlogbw/single-id/files/",
+        )

apps/wpm/tests/test_mlogbw_number.py

@@ -0,0 +1,80 @@
+from types import SimpleNamespace
+from unittest.mock import MagicMock, patch
+
+from django.test import SimpleTestCase
+
+from apps.wpm.serializers import MlogbwCreateUpdateSerializer
+from apps.wpmw.models import Wpr
+
+
+class MlogbwNumberUpdateTests(SimpleTestCase):
+    @staticmethod
+    def _apply_update(instance, validated_data):
+        instance.number = validated_data["number"]
+        return instance
+
+    @patch("apps.wpm.serializers.CustomModelSerializer.update")
+    def test_output_number_update_changes_linked_wpr_number(self, base_update):
+        base_update.side_effect = self._apply_update
+        wpr = MagicMock()
+        instance = SimpleNamespace(
+            number="OLD-001",
+            mlogb=SimpleNamespace(material_out_id="material-out"),
+            wpr=wpr,
+            ftest=None,
+        )
+
+        MlogbwCreateUpdateSerializer.update.__wrapped__(
+            MlogbwCreateUpdateSerializer(),
+            instance,
+            {"mlogb": instance.mlogb, "number": "NEW-001"},
+        )
+
+        wpr.change_number.assert_called_once_with("NEW-001")
+
+    @patch("apps.wpm.serializers.CustomModelSerializer.update")
+    def test_input_number_update_does_not_change_wpr_number(self, base_update):
+        base_update.side_effect = self._apply_update
+        wpr = MagicMock()
+        instance = SimpleNamespace(
+            number="OLD-001",
+            mlogb=SimpleNamespace(material_out_id=None),
+            wpr=wpr,
+            ftest=None,
+        )
+
+        MlogbwCreateUpdateSerializer.update.__wrapped__(
+            MlogbwCreateUpdateSerializer(),
+            instance,
+            {"mlogb": instance.mlogb, "number": "NEW-001"},
+        )
+
+        wpr.change_number.assert_not_called()
+
+    @patch("apps.wpmw.models.MIOItemw.objects.filter")
+    @patch("apps.wpmw.models.Handoverbw.objects.filter")
+    @patch("apps.wpmw.models.Mlogbw.objects.filter")
+    @patch("apps.wpmw.models.Wpr.objects.filter")
+    def test_wpr_number_change_updates_all_number_copies(
+        self,
+        wpr_filter,
+        mlogbw_filter,
+        handoverbw_filter,
+        mioitemw_filter,
+    ):
+        conflict_qs = MagicMock()
+        conflict_qs.exists.return_value = False
+        current_qs = MagicMock()
+        wpr_filter.side_effect = [conflict_qs, current_qs]
+        mlogbw_qs = mlogbw_filter.return_value
+        handoverbw_qs = handoverbw_filter.return_value
+        mioitemw_qs = mioitemw_filter.return_value
+        wpr = Wpr(id="wpr-id", number="OLD-001")
+
+        wpr.change_number("NEW-001")
+
+        current_qs.update.assert_called_once_with(number="NEW-001")
+        mlogbw_qs.update.assert_called_once_with(number="NEW-001")
+        handoverbw_qs.update.assert_called_once_with(number="NEW-001")
+        mioitemw_qs.update.assert_called_once_with(number="NEW-001")
+        self.assertEqual(wpr.number, "NEW-001")

apps/wpm/tests/test_number_rule.py

@@ -0,0 +1,193 @@
+from datetime import date
+from types import SimpleNamespace
+from unittest.mock import MagicMock, patch
+
+from django.test import SimpleTestCase
+
+from apps.wpm.views import MlogbInViewSet
+
+
+class GenNumberWithRuleFilterTests(SimpleTestCase):
+    def test_date_filters_follow_rule_placeholders(self):
+        cases = [
+            ("SN-{n_count:04d}", {}),
+            ("{c_year}-{n_count:04d}", {"year": 2026}),
+            ("{c_year2}-{n_count:04d}", {"year": 2026}),
+            ("{c_month:02d}-{n_count:04d}", {"month": 8}),
+            ("{c_day:02d}-{n_count:04d}", {"day": 4}),
+            (
+                "{c_year}{c_month:02d}{c_day:02d}-{n_count:04d}",
+                {"year": 2026, "month": 8, "day": 4},
+            ),
+        ]
+        material = SimpleNamespace(model=None)
+        mlog = SimpleNamespace(
+            handle_date=date(2026, 8, 4),
+            mgroup=SimpleNamespace(process=SimpleNamespace(id=123)),
+        )
+
+        for rule, expected_dates in cases:
+            with self.subTest(rule=rule):
+                queryset = MagicMock()
+                queryset.values_list.return_value.distinct.return_value.iterator.return_value = []
+
+                with patch("apps.wpmw.models.Wpr.objects.filter", return_value=queryset) as mock_filter:
+                    MlogbInViewSet.gen_number_with_rule(rule, material, mlog)
+
+                filters = mock_filter.call_args.kwargs
+                date_prefix = "wpr_mlogbw__mlogb__mlog__handle_date__"
+                actual_dates = {
+                    key.removeprefix(date_prefix): value
+                    for key, value in filters.items()
+                    if key.startswith(date_prefix)
+                }
+                self.assertEqual(actual_dates, expected_dates)
+
+    def test_escaped_date_placeholder_text_does_not_add_filter(self):
+        queryset = MagicMock()
+        queryset.values_list.return_value.distinct.return_value.iterator.return_value = []
+        material = SimpleNamespace(model=None)
+        mlog = SimpleNamespace(
+            handle_date=date(2026, 8, 4),
+            mgroup=SimpleNamespace(process=SimpleNamespace(id=123)),
+        )
+
+        with patch("apps.wpmw.models.Wpr.objects.filter", return_value=queryset) as mock_filter:
+            MlogbInViewSet.gen_number_with_rule("{{c_year}}-{n_count:04d}", material, mlog)
+
+        self.assertFalse(
+            any("handle_date" in key for key in mock_filter.call_args.kwargs)
+        )
+
+    def test_previous_sequence_width_is_compatible(self):
+        queryset = MagicMock()
+        queryset.values_list.return_value.distinct.return_value.iterator.return_value = [
+            "2608P0001",
+            "2608P0003",
+            "2608P0002",
+        ]
+        material = SimpleNamespace(model="P")
+        mlog = SimpleNamespace(
+            handle_date=date(2026, 8, 6),
+            mgroup=SimpleNamespace(process=SimpleNamespace(id=123)),
+        )
+
+        with patch("apps.wpmw.models.Wpr.objects.filter", return_value=queryset):
+            number = MlogbInViewSet.gen_number_with_rule(
+                "{c_year2}{c_month:02d}{m_model}{n_count:05d}",
+                material,
+                mlog,
+            )
+
+        self.assertEqual(number, "2608P00004")
+
+    def test_mixed_sequence_widths_use_numeric_maximum(self):
+        queryset = MagicMock()
+        queryset.values_list.return_value.distinct.return_value.iterator.return_value = [
+            "2608P9999",
+            "2608P10000",
+            "历史异常编号",
+        ]
+        material = SimpleNamespace(model="P")
+        mlog = SimpleNamespace(
+            handle_date=date(2026, 8, 6),
+            mgroup=SimpleNamespace(process=SimpleNamespace(id=123)),
+        )
+
+        with patch("apps.wpmw.models.Wpr.objects.filter", return_value=queryset):
+            number = MlogbInViewSet.gen_number_with_rule(
+                "{c_year2}{c_month:02d}{m_model}{n_count:05d}",
+                material,
+                mlog,
+            )
+
+        self.assertEqual(number, "2608P10001")
+
+    def test_unrelated_historical_formats_do_not_participate(self):
+        queryset = MagicMock()
+        queryset.values_list.return_value.distinct.return_value.iterator.return_value = [
+            "202505508001",
+            "3p05013",
+            "05002",
+            "3pb003",
+        ]
+        material = SimpleNamespace(model="P")
+        mlog = SimpleNamespace(
+            handle_date=date(2026, 8, 7),
+            mgroup=SimpleNamespace(process=SimpleNamespace(id=123)),
+        )
+
+        with patch("apps.wpmw.models.Wpr.objects.filter", return_value=queryset):
+            number = MlogbInViewSet.gen_number_with_rule(
+                "{m_model}{n_count:04d}",
+                material,
+                mlog,
+            )
+
+        self.assertEqual(number, "P0001")
+
+    def test_current_rule_uses_only_matching_model_numbers(self):
+        queryset = MagicMock()
+        queryset.values_list.return_value.distinct.return_value.iterator.return_value = [
+            "P0002",
+            "P0010",
+            "B0099",
+            "3pb100",
+        ]
+        material = SimpleNamespace(model="P")
+        mlog = SimpleNamespace(
+            handle_date=date(2026, 8, 7),
+            mgroup=SimpleNamespace(process=SimpleNamespace(id=123)),
+        )
+
+        with patch("apps.wpmw.models.Wpr.objects.filter", return_value=queryset):
+            number = MlogbInViewSet.gen_number_with_rule(
+                "{m_model}{n_count:04d}",
+                material,
+                mlog,
+            )
+
+        self.assertEqual(number, "P0011")
+
+    def test_empty_historical_number_is_ignored(self):
+        queryset = MagicMock()
+        queryset.values_list.return_value.distinct.return_value.iterator.return_value = [
+            None,
+            "",
+            "P0002",
+        ]
+        material = SimpleNamespace(model="P")
+        mlog = SimpleNamespace(
+            handle_date=date(2026, 8, 7),
+            mgroup=SimpleNamespace(process=SimpleNamespace(id=123)),
+        )
+
+        with patch("apps.wpmw.models.Wpr.objects.filter", return_value=queryset):
+            number = MlogbInViewSet.gen_number_with_rule(
+                "{m_model}{n_count:04d}",
+                material,
+                mlog,
+            )
+
+        self.assertEqual(number, "P0003")
+
+    def test_repeated_sequence_placeholder_does_not_break_matching(self):
+        queryset = MagicMock()
+        queryset.values_list.return_value.distinct.return_value.iterator.return_value = [
+            "P02-0002",
+            "P03-0004",
+        ]
+        material = SimpleNamespace(model="P")
+        mlog = SimpleNamespace(
+            handle_date=date(2026, 8, 7),
+            mgroup=SimpleNamespace(process=SimpleNamespace(id=123)),
+        )
+
+        with patch("apps.wpmw.models.Wpr.objects.filter", return_value=queryset):
+            number = MlogbInViewSet.gen_number_with_rule(
+                "{m_model}{n_count:02d}-{n_count:04d}",
+                material,
+                mlog,
+            )
+
+        self.assertEqual(number, "P03-0003")

apps/wpm/views.py

@@ -1,12 +1,13 @@
 import math
 import re
+from string import Formatter
 
 from django.db import transaction
 from rest_framework.decorators import action
 from rest_framework.exceptions import ParseError
 from rest_framework.response import Response
 from rest_framework.serializers import Serializer
-from django.db.models import Sum
+from django.db.models import Case, IntegerField, Sum, When
 from django.utils import timezone
 from apps.system.models import User
 
@@ -46,19 +47,30 @@ from .serializers import (
     BatchStSerializer,
     MlogbwCreateUpdateSerializer,
     MlogbwListSerializer,
+    MlogbwFilesSerializer,
     HandoverMgroupSerializer,
     MlogListSerializer,
     MlogbSerializer,
     MlogUserSerializer,
     BatchLogSerializer,
     MlogQuickSerializer,
+    MlogEquipmentOptionSerializer,
     MlogbwStartTestSerializer,
     HandoverListSerializer,
     BatchChangeSerializer,
     MlogbOutPatchUpdateSerializer
 )
-from .services import mlog_submit, handover_submit, mlog_revert, get_batch_dag, handover_revert
+from .services import (
-from apps.wpm.services import mlog_submit_validate, generate_new_batch
+    RECENT_EQUIPMENT_LOG_LIMIT,
+    generate_new_batch,
+    get_batch_dag,
+    get_recent_mgroup_equipment_ids,
+    handover_revert,
+    handover_submit,
+    mlog_revert,
+    mlog_submit,
+    mlog_submit_validate,
+)
 from apps.wf.models import State, Ticket
 from apps.wpmw.models import Wpr
 from apps.qm.models import Qct, Ftest, TestItem
@@ -72,7 +84,6 @@ from django.db.models import Prefetch
 from drf_yasg.utils import swagger_auto_schema
 from drf_yasg import openapi
 from django.db import connection
-from django.db.models.functions import Substr, Length
 from apps.qm.models import FtestDefect, FtestItem
 
 # Create your views here.
@@ -331,6 +342,92 @@ class MlogViewSet(CustomModelViewSet):
     ]
     ordering_fields = ["create_time", "update_time"]
 
+    @swagger_auto_schema(
+        manual_parameters=[
+            openapi.Parameter(
+                name="mgroup",
+                in_=openapi.IN_QUERY,
+                description="日志所属工段",
+                type=openapi.TYPE_STRING,
+                required=True,
+            ),
+            openapi.Parameter(
+                name="search",
+                in_=openapi.IN_QUERY,
+                description="按设备名称或编号搜索全部生产设备",
+                type=openapi.TYPE_STRING,
+                required=False,
+            ),
+        ]
+    )
+    @action(
+        methods=["get"],
+        detail=False,
+        perms_map={"get": "*"},
+        serializer_class=MlogEquipmentOptionSerializer,
+    )
+    def equipment_options(self, request, *args, **kwargs):
+        """返回本工段设备、最近 50 条日志用过的设备或搜索结果。"""
+        mgroup_id = request.query_params.get("mgroup")
+        if not mgroup_id:
+            raise ParseError("请传入mgroup参数")
+
+        search = request.query_params.get("search", "").strip()
+        owned_ids = list(
+            Equipment.objects.filter(
+                type=Equipment.EQUIP_TYPE_PRO,
+                mgroup_id=mgroup_id,
+            )
+            .order_by("name", "number")
+            .values_list("id", flat=True)
+        )
+        owned_id_set = set(owned_ids)
+
+        if search:
+            queryset = (
+                Equipment.objects.filter(type=Equipment.EQUIP_TYPE_PRO)
+                .filter(Q(name__icontains=search) | Q(number__icontains=search))
+                .order_by("name", "number")
+            )
+            option_group = "搜索结果"
+        else:
+            recent_ids = get_recent_mgroup_equipment_ids(
+                mgroup_id, RECENT_EQUIPMENT_LOG_LIMIT
+            )
+            option_ids = list(dict.fromkeys([*owned_ids, *recent_ids]))
+            if option_ids:
+                order = Case(
+                    *[
+                        When(id=equipment_id, then=position)
+                        for position, equipment_id in enumerate(option_ids)
+                    ],
+                    output_field=IntegerField(),
+                )
+                queryset = Equipment.objects.filter(
+                    id__in=option_ids,
+                    type=Equipment.EQUIP_TYPE_PRO,
+                ).order_by(order)
+            else:
+                queryset = Equipment.objects.none()
+            option_group = None
+
+        queryset = queryset.select_related("mgroup")
+        page = self.paginate_queryset(queryset)
+        equipment_list = page if page is not None else queryset
+        data = MlogEquipmentOptionSerializer(
+            equipment_list,
+            many=True,
+            context=self.get_serializer_context(),
+        ).data
+        for item in data:
+            item["option_group"] = option_group or (
+                "本工段设备" if item["id"] in owned_id_set else "近期使用"
+            )
+
+        if page is not None:
+            return self.get_paginated_response(data)
+        return Response(data)
+
     def add_info_for_item(self, data):
         if data.get("oinfo_json", {}):
             czx_dict = dict(TestItem.objects.filter(id__in=data.get("oinfo_json", {}).keys()).values_list("id", "name"))
@@ -352,6 +449,7 @@ class MlogViewSet(CustomModelViewSet):
         return super().get_serializer_class()
 
     @swagger_auto_schema(
+        operation_summary="查询生产日志明细（逐条追溯）",
         manual_parameters=[
             openapi.Parameter(name="query", in_=openapi.IN_QUERY, description="定制返回数据", type=openapi.TYPE_STRING, required=False),
             openapi.Parameter(name="with_children", in_=openapi.IN_QUERY, description="带有children(yes/no/count)", type=openapi.TYPE_STRING, required=False),
@@ -560,6 +658,8 @@ class MlogViewSet(CustomModelViewSet):
             mloginit_data["equipment"] = vdata["equipment"]
         if "team" in vdata:
             mloginit_data["team"] = vdata["team"]
+        if "supplier" in vdata:
+            mloginit_data["supplier"] = vdata["supplier"]
         sr_1 = MlogInitSerializer(data=mloginit_data, request=request)
         sr_1.is_valid(raise_exception=True)
         mlog: Mlog = sr_1.save()
@@ -590,7 +690,20 @@ class HandoverViewSet(CustomModelViewSet):
     select_related_fields = ["send_user", "send_mgroup", "send_dept", "recive_user", "recive_mgroup", "recive_dept", "wm", "material_changed", "material", "material__process"]
     filterset_class = HandoverFilter
     search_fields = ["material__name", "material__number", "material__specification", "batch", "material__model", "b_handover__batch", "new_batch", "wm__batch"]
-    prefetch_related_fields = [Prefetch("b_handover", queryset=Handoverb.objects.select_related("wm__defect")), "ticket__state"]
+    prefetch_related_fields = ["ticket__state"]
+
+    def get_queryset_custom(self, queryset):
+        if self.action not in ["list", "retrieve"]:
+            return queryset
+
+        detail_queryset = Handoverb.objects.select_related(
+            "handover", "wm__defect", "wm__material", "wm__mgroup", "wm__belong_dept"
+        )
+        if self.action == "retrieve":
+            detail_queryset = detail_queryset.prefetch_related("w_handoverb")
+        return queryset.prefetch_related(
+            Prefetch("b_handover", queryset=detail_queryset)
+        )
 
     def perform_destroy(self, instance: Handover):
         user = self.request.user
@@ -678,7 +791,7 @@ class HandoverViewSet(CustomModelViewSet):
             m_qs = m_qs.filter(process__route_p__material_in__id=materialInId) | m_qs.filter(process__route_p__routemat_route__material__id=materialInId)
         elif type in [Handover.H_SCRAP]:
             m_qs = m_qs.filter(process=None)
-        return Response(list(m_qs.values("id", "name").distinct()))
+        return Response(list(m_qs.values("id", "name", "belong_dept").distinct()))
 
     @action(methods=["post"], detail=False, perms_map={"post": "handover.create"}, serializer_class=GenHandoverWmSerializer)
     @transaction.atomic
@@ -1009,13 +1122,20 @@ class MlogbInViewSet(BulkCreateModelMixin, BulkUpdateModelMixin, BulkDestroyMode
     def gen_number_with_rule(cls, rule, material_out: Material, mlog: Mlog, gen_count=1):
         from apps.wpmw.models import Wpr
 
+        formatter = Formatter()
+        rule_parts = list(formatter.parse(rule))
+        rule_fields = {
+            field_name
+            for _, field_name, _, _ in rule_parts
+            if field_name
+        }
         handle_date = mlog.handle_date
         c_year = handle_date.year
         c_year2 = str(c_year)[-2:]
         c_month = handle_date.month
         c_day = handle_date.day
         m_model = material_out.model
-        if 'm_model' in rule:
+        if "m_model" in rule_fields:
             if m_model is None:
                 raise ParseError("生成编号出错:产品型号不能为空")
             elif m_model and m_model.islower():
@@ -1028,29 +1148,64 @@ class MlogbInViewSet(BulkCreateModelMixin, BulkUpdateModelMixin, BulkDestroyMode
         if connection.vendor == "postgresql" and connection.in_atomic_block:
             with connection.cursor() as cursor:
                 cursor.execute("SELECT pg_advisory_xact_lock(hashtextextended(%s, 0))", [f"wpr_number_rule:{process.id}"])
-        # 按生产日志查询, 流水号归零周期跟随规则中最细的日期占位符
+        # 只按规则中实际使用的日期占位符筛选历史编号
         wpr_filter = {
             "wpr_mlogbw__mlogb__material_out__isnull": False,
             "wpr_mlogbw__mlogb__mlog__mgroup__process": process,
             "wpr_mlogbw__mlogb__mlog__is_fix": False,
             "wpr_mlogbw__mlogb__mlog__submit_time__isnull": False,
-            "wpr_mlogbw__mlogb__mlog__handle_date__year": c_year,
-            "wpr_mlogbw__mlogb__mlog__handle_date__month": c_month,
         }
-        if "c_day" in rule:
+        if rule_fields & {"c_year", "c_year2"}:
+            wpr_filter["wpr_mlogbw__mlogb__mlog__handle_date__year"] = c_year
+        if "c_month" in rule_fields:
+            wpr_filter["wpr_mlogbw__mlogb__mlog__handle_date__month"] = c_month
+        if "c_day" in rule_fields:
             wpr_filter["wpr_mlogbw__mlogb__mlog__handle_date__day"] = c_day
-        wpr = (
+        rule_values = {
-            Wpr.objects.filter(**wpr_filter)
+            "c_year": c_year,
-            .annotate(last_seq=Substr("number", Length("number") - (cq_w - 1)))
+            "c_year2": c_year2,
-            .order_by("last_seq")
+            "c_month": c_month,
-            .last()
+            "c_day": c_day,
-        )
+            "m_model": m_model,
-        n_count = 0
+        }
-        if wpr:
+        number_pattern_parts = ["^"]
+        sequence_group_names = []
+        for literal_text, field_name, format_spec, conversion in rule_parts:
+            number_pattern_parts.append(re.escape(literal_text))
+            if not field_name:
+                continue
+            if field_name == "n_count":
+                # 流水号宽度可以变化，规则中的其他部分必须与当前上下文一致。
+                group_name = f"n_count_{len(sequence_group_names)}"
+                sequence_group_names.append(group_name)
+                number_pattern_parts.append(fr"(?P<{group_name}>[0-9]+)")
+                continue
             try:
-                n_count = int(wpr.number[-cq_w:])
+                field_value = rule_values[field_name]
-            except Exception as e:
+                if conversion:
-                raise ParseError(f"获取该类产品最后编号错误: {str(e)}")
+                    field_value = formatter.convert_field(field_value, conversion)
+                formatted_value = formatter.format_field(field_value, format_spec)
+            except (KeyError, TypeError, ValueError) as e:
+                raise ParseError(f"个号生成错误: {e}")
+            number_pattern_parts.append(re.escape(formatted_value))
+        number_pattern_parts.append("$")
+        number_pattern = re.compile("".join(number_pattern_parts))
+        n_count = 0
+        # 只从符合当前规则固定部分的历史编号中提取流水号。例如当前规则为
+        # P{n_count:04d}时，3pb003等同工序的旧格式编号不能参与续号；同时
+        # 流水号使用数字匹配，以兼容04d调整为05d后的历史编号。
+        numbers = Wpr.objects.filter(**wpr_filter).values_list("number", flat=True).distinct()
+        for number in numbers.iterator():
+            if not isinstance(number, str):
+                continue
+            sequence_match = number_pattern.fullmatch(number)
+            if sequence_match and sequence_group_names:
+                sequence_values = {
+                    int(sequence_match.group(group_name))
+                    for group_name in sequence_group_names
+                }
+                if len(sequence_values) == 1:
+                    n_count = max(n_count, sequence_values.pop())
         if n_count + gen_count > 10 ** cq_w - 1:
             raise ParseError(f"流水号超出{cq_w}位上限, 请调整编号规则")
         try:
@@ -1147,6 +1302,7 @@ class MlogbwViewSet(CustomModelViewSet):
     filterset_class = MlogbwFilter
     select_related_fields = ["ftest", "equip", "tooling", "wpr", "mlogb"]
     prefetch_related_fields = [
+        "files",
         Prefetch(
             "ftest__items_ftest",
             queryset=FtestItem.objects.select_related(
@@ -1166,6 +1322,21 @@ class MlogbwViewSet(CustomModelViewSet):
     ordering = ["mlogb", "number", "create_time"]
     ordering_fields = ["number", "create_time", "mlogb"]
 
+    @action(detail=True, methods=["get", "put"], url_path="files")
+    @transaction.atomic
+    def files(self, request, *args, **kwargs):
+        instance = self.get_object()
+        if instance.mlogb.material_out_id is None:
+            raise ParseError("只能给输出单件上传附件")
+        if request.method == "PUT":
+            MlogViewSet.lock_and_check_can_update(instance.mlogb.mlog)
+            serializer = MlogbwFilesSerializer(instance, data=request.data)
+            serializer.is_valid(raise_exception=True)
+            serializer.save()
+        else:
+            serializer = MlogbwFilesSerializer(instance)
+        return Response(serializer.data)
+
     def perform_create(self, serializer):
         ins: Mlogbw = serializer.save()
         mlog: Mlog = None
@@ -1199,12 +1370,14 @@ class MlogbwViewSet(CustomModelViewSet):
                     material_out = material_in
                 else:
                     raise ParseError("获取产出物料错误")
-                mtype = route.process.mtype if route.process else None
+                process = route.process if route else None
+                # 返修不依赖工艺路线，按个到个的普通流转同步产出。
+                mtype = Process.PRO_NORMAL if mlog.is_fix else process.mtype if process else None
                 if mlogb_qs.exists() and material_out.tracking == Material.MA_TRACKING_SINGLE:
                     for mlogb in mlogb_qs:
-                        if route.process.mtype == Process.PRO_NORMAL:
+                        if mtype == Process.PRO_NORMAL:
                             Mlogbw.objects.get_or_create(mlogb=mlogb, wpr=ins.wpr, defaults={"number": ins.number, "mlogbw_from": ins})
-                        elif route.process.mtype == Process.PRO_DIV:
+                        elif mtype == Process.PRO_DIV:
                             if route.div_number == 1:
                                 Mlogbw.objects.get_or_create(wpr=ins.wpr, mlogb=mlogb, defaults={"number": ins.wpr.number, "mlogbw_from": ins})
                             else:

apps/wpmw/models.py

@@ -5,6 +5,7 @@ from apps.utils.models import BaseModel
 from apps.mtm.models import Material
 from rest_framework.exceptions import ParseError
 from apps.wpm.models import WmStateOption, Mlogbw, Handoverbw
+from apps.utils.snowflake import idWorker
 from apps.utils.tools import update_dict
 from apps.inm.models import MIOItemw
 from django.db.models import F, Value
@@ -32,8 +33,24 @@ class Wpr(BaseModel):
     data = models.JSONField(verbose_name="数据", default=dict, blank=True)
     pre_info = models.JSONField(verbose_name="预处理信息", default=dict, blank=True, null=True)
 
+    def change_number(self, new_number):
+        """修改产品编号，并同步所有保存了编号副本的关联明细。"""
+        if self.number == new_number:
+            return
+        if Wpr.objects.filter(number=new_number).exists():
+            raise ParseError("新编号已存在,不可使用")
+
+        Wpr.objects.filter(id=self.id).update(number=new_number)
+        Mlogbw.objects.filter(wpr=self).update(number=new_number)
+        Handoverbw.objects.filter(wpr=self).update(number=new_number)
+        MIOItemw.objects.filter(wpr=self).update(number=new_number)
+        self.number = new_number
+
     @classmethod
-    def change_or_new(cls, wpr=None, number=None, mb=None, wm=None, old_mb=None, old_wm=None, ftest=None, wpr_from=None, add_version=True, number_out=None):
+    def change_or_new(
+        cls, wpr=None, number=None, mb=None, wm=None, old_mb=None,
+        old_wm=None, wpr_from=None, add_version=True, number_out=None,
+    ):
         is_create = False
         if wpr is None and number is None:
             raise ParseError("id和number不能同时为空")
@@ -109,25 +126,9 @@ class Wpr(BaseModel):
             ins.state = wm.state
             if is_create:
                 ins.material_start = wm.material
-            WprDefect.objects.filter(wpr=ins).delete()
         if add_version:
             ins.version = ins.version + 1
         ins.save()
-        if wm:
-            if wm.defect:
-                WprDefect.objects.create(wpr=ins, defect=wm.defect, is_main=True)
-        if ftest: 
-            # 通过检验变更其缺陷项
-            WprDefect.objects.filter(wpr=ins).delete()
-            for ftestdefect in FtestDefect.objects.filter(ftest=ftest, has=True):
-                WprDefect.objects.create(wpr=ins, defect=ftestdefect.defect, is_main=ftestdefect.is_main)
-            # 携带某些检测信息
-            oinfo = {}
-            for ftestitem in FtestItem.objects.filter(ftest=ftest, addto_wpr=True):
-                if ftestitem.test_val_json:
-                    oinfo[ftestitem.id] = {"name": ftestitem.testitem.name, "val": ftestitem.test_val_json}
-            ins.oinfo = update_dict(ins.oinfo, oinfo)
-            ins.save()
         # if ins.mb and ins.wm:
         #     raise ParseError("所属仓库批次和车间批次不可同时存在2")
         if number_out:
@@ -138,6 +139,63 @@ class Wpr(BaseModel):
                 raise ParseError(f"{ins.number}-出库编号-{number_out}-已存在")
         return ins
 
+    def replace_defects(self, defects):
+        """用给定缺陷列表完整替换单件缺陷。"""
+        WprDefect.objects.filter(wpr=self).delete()
+        WprDefect.objects.bulk_create([
+            WprDefect(
+                id=idWorker.get_id(),
+                wpr=self,
+                defect_id=item["defect_id"],
+                is_main=item.get("is_main", False),
+            )
+            for item in defects
+        ])
+
+    def apply_ftest(self, ftest):
+        """应用一次明确的单件检验；无检验结果时缺陷为空。"""
+        defects = []
+        if ftest is not None:
+            defects = list(FtestDefect.objects.filter(
+                ftest=ftest,
+                has=True,
+            ).values("defect_id", "is_main"))
+        self.replace_defects(defects)
+
+        if ftest is None:
+            return
+
+        oinfo = {}
+        for ftestitem in FtestItem.objects.filter(
+            ftest=ftest,
+            addto_wpr=True,
+        ).select_related("testitem"):
+            if ftestitem.test_val_json:
+                oinfo[str(ftestitem.id)] = {
+                    "name": ftestitem.testitem.name,
+                    "val": ftestitem.test_val_json,
+                }
+        if oinfo:
+            self.oinfo = update_dict(self.oinfo, oinfo)
+            self.save(update_fields=["oinfo", "update_time"])
+
+    def revert_ftest_info(self, ftest):
+        """仅撤销指定检验写入单件携带信息的字段。"""
+        if ftest is None or not self.oinfo:
+            return
+        item_ids = FtestItem.objects.filter(
+            ftest=ftest,
+            addto_wpr=True,
+        ).values_list("id", flat=True)
+        changed = False
+        oinfo = dict(self.oinfo)
+        for item_id in item_ids:
+            if oinfo.pop(str(item_id), None) is not None:
+                changed = True
+        if changed:
+            self.oinfo = oinfo
+            self.save(update_fields=["oinfo", "update_time"])
+
     @classmethod
     def clear(cls, number_list):
         cls.objects.filter(number__in=number_list).update(mb=None, wm=None)

apps/wpmw/views.py

@@ -63,15 +63,8 @@ class WprViewSet(BulkUpdateModelMixin, CustomListModelMixin, CustomRetrieveModel
         vdata = sr.validated_data
         new_number = vdata["new_number"]
         old_number = vdata["old_number"]
-        if Wpr.objects.filter(number=new_number).exists():
-            raise ParseError("新编号已存在,不可使用")
         wpr = Wpr.objects.get(number=old_number)
-        from apps.wpm.models import Mlogbw, Handoverbw
+        wpr.change_number(new_number)
-        from apps.inm.models import MIOItemw
-        Wpr.objects.filter(id=wpr.id).update(number=new_number)
-        Mlogbw.objects.filter(wpr=wpr).update(number=new_number)
-        Handoverbw.objects.filter(wpr=wpr).update(number=new_number)
-        MIOItemw.objects.filter(wpr=wpr).update(number=new_number)
         return Response()
 
     @action(methods=["post"], detail=False, perms_map={"post": "*"}, serializer_class=WprNewSerializer)

changelog.md

@@ -1,3 +1,13 @@
+## 3.1.2026072713
+
+- fix: 问题修复
+  - unify workflow action permissions [caoqianming]
+- other: 其他变更
+  - docs: record Codex context migration plan [caoqianming]
+  - docs: migrate project memory for Codex [caoqianming]
+  - docs: add factory release skill [caoqianming]
+  - docs: add Codex project instructions [caoqianming]
+  - docs: design Claude to Codex context migration [caoqianming]
 ## 3.1.2026072316
 
 - feat: 新增功能

docs/mcp.md

@@ -0,0 +1,49 @@
+# Factory MCP 服务
+
+Factory MCP 是仓库顶层的独立服务，使用官方 Python SDK v2，通过 Streamable HTTP 暴露 Agent 工具。当前已提供基础工具和第一批 Dataset 领域工具。
+
+## 启动
+
+在项目根目录使用项目虚拟环境启动独立进程：
+
+```powershell
+.venv\Scripts\python.exe -m mcp_server
+```
+
+默认监听 `127.0.0.1:2260`，MCP 端点为 `/mcp`。客户端必须在每次请求中携带 Factory access token：
+
+```text
+Authorization: Bearer <Factory access token>
+```
+
+## 配置
+
+生产环境在本机已忽略的 `config/conf.py` 中覆盖以下配置：
+
+- `MCP_HOST`：监听地址。
+- `MCP_PORT`：监听端口。
+- `MCP_PATH`：Streamable HTTP 路径。
+- `MCP_ALLOWED_HOSTS`：允许的 HTTP Host，支持 `hostname:*` 端口通配形式。
+- `MCP_ALLOWED_ORIGINS`：允许的浏览器 Origin；非浏览器客户端通常不发送 Origin。
+- `MCP_MAX_REQUEST_BODY_SIZE`：单个 MCP 请求体上限，默认 1 MiB。
+- `MCP_MAX_RESULT_BYTES`：单次领域工具结果上限，默认 512 KiB。
+
+生产部署必须明确配置实际域名的 Host 白名单，不应直接复用 Django 当前的宽泛 `ALLOWED_HOSTS`。
+
+## 基础工具
+
+- `factory_server_info`：返回系统版本、MCP 协议版本和认证方式。
+- `factory_whoami`：返回当前 JWT 对应的 Factory 用户。
+- `search_datasets`：按名称、code 或描述搜索启用的数据集，不返回 SQL 配置。
+- `execute_dataset`：按 code 执行数据集，需要当前用户具有 `dataset.exec` 权限。
+- `search_wprs`：按编号、物料、批次、状态和当前位置搜索 WPR，只返回摘要。
+- `get_wpr`：按 ID、内部编号或对外编号读取 WPR 详情、缺陷和业务数据。
+
+WPR 当前沿用既有 API 的读取边界：有效登录用户可读，且该 ViewSet 未启用部门数据过滤。MCP 不开放修改编号、分配对外编号或更新预处理信息等写操作。
+
+- `search_batch_stats`：按批次、直通大批、起始物料和版本搜索批次统计摘要。
+- `get_batch_stat`：读取指定批次版本的完整统计数据，可附带直接拆批/合批关系。
+
+BatchSt 同样沿用既有 API 的 `get: *` 读取边界，不提供创建、重算或修改工具。完整统计结果仍受 `MCP_MAX_RESULT_BYTES` 限制。
+
+新增领域工具时必须从 MCP 请求身份获取用户，并复用 Factory 的权限码和数据范围过滤；不得直接使用固定管理员身份查询 ORM。

docs/superpowers/plans/2026-07-27-claude-to-codex-project-context.md

@@ -0,0 +1,217 @@
+# Claude to Codex Project Context Implementation Plan
+
+> **For agentic workers:** REQUIRED SUB-SKILL: Use superpowers:subagent-driven-development (recommended) or superpowers:executing-plans to implement this plan task-by-task. Steps use checkbox (`- [ ]`) syntax for tracking.
+
+**Goal:** Add a project-level Codex entry point, release skill, and sanitized memory set while leaving `.claude` unchanged.
+
+**Architecture:** `AGENTS.md` contains short, always-on rules. `.agents/skills/release/SKILL.md` contains the opt-in backend release workflow. `.codex/memory/` contains detailed project knowledge linked from a single index so Codex can load it only when relevant.
+
+**Tech Stack:** Markdown, Codex `AGENTS.md`, Codex project skills
+
+---
+
+### Task 1: Add the always-on Codex project instructions
+
+**Files:**
+- Create: `AGENTS.md`
+
+- [x] **Step 1: Create the instruction entry point**
+
+Add concise sections covering:
+
+- Repository shape: Django backend here and Vue frontend at `../ehs_web`.
+- Frontend route/menu behavior: routes are delivered by the backend; do not add new views to `src/config/route.js`.
+- Python execution: use `D:\projects\factory\.venv\Scripts\python.exe`.
+- Ignored scripts: never force-add `scripts/*.py` without explicit instruction.
+- Release policy: never bump or release unless explicitly requested; then use `.agents/skills/release/SKILL.md`.
+- Production data policy: read-only queries only when relevant, credentials from ignored local configuration, and no secrets in docs or commands.
+- Pointers to `.codex/memory/MEMORY.md` for detailed context.
+
+- [x] **Step 2: Verify the entry point**
+
+Run:
+
+```powershell
+rg -n "ehs_web|route.js|.venv|scripts/\\*\\.py|release|只读|memory" AGENTS.md
+```
+
+Expected: every rule category appears at least once.
+
+- [x] **Step 3: Commit**
+
+```powershell
+git add -- AGENTS.md
+git commit -m "docs: add Codex project instructions"
+```
+
+### Task 2: Add the opt-in backend release skill
+
+**Files:**
+- Create: `.agents/skills/release/SKILL.md`
+
+- [x] **Step 1: Create the skill**
+
+Add valid skill frontmatter:
+
+```yaml
+---
+name: factory-release
+description: Use only when the user explicitly asks to release the factory backend, bump its version, or run the release workflow.
+---
+```
+
+The body must require this exact order:
+
+1. Confirm the user explicitly requested a backend release.
+2. Check the working tree and prevent unrelated files from entering the release commit.
+3. Run `bash update_changelog.sh` and use its emitted version.
+4. Update `server/settings.py` `SYS_VERSION`.
+5. Inspect the top of `changelog.md`.
+6. Stage only `changelog.md` and `server/settings.py`.
+7. Commit as `release: <VER>` without Claude-specific attribution.
+8. Create tag `<VER>`, push the branch, then push the tag.
+9. Report version, changelog categories, commit hash, tag, and push results.
+
+The skill must state that frontend `ehs_web` releases use a separate workflow documented in `.codex/memory/reference_ehs_web_release.md`.
+
+- [x] **Step 2: Verify the skill**
+
+Run:
+
+```powershell
+rg -n "explicitly|update_changelog|SYS_VERSION|changelog.md|git tag|git push|ehs_web|Co-Authored-By" .agents/skills/release/SKILL.md
+```
+
+Expected: all required release steps appear and `Co-Authored-By` does not appear.
+
+- [x] **Step 3: Commit**
+
+```powershell
+git add -- .agents/skills/release/SKILL.md
+git commit -m "docs: add factory release skill"
+```
+
+### Task 3: Add the sanitized Codex project memory
+
+**Files:**
+- Create: `.codex/memory/MEMORY.md`
+- Create: `.codex/memory/feedback_routing.md`
+- Create: `.codex/memory/feedback_scripts_gitignore.md`
+- Create: `.codex/memory/feedback_version_bump.md`
+- Create: `.codex/memory/project_frontend_ehs_web.md`
+- Create: `.codex/memory/project_gx_db_access.md`
+- Create: `.codex/memory/project_material_ofrom_merge_bug.md`
+- Create: `.codex/memory/reference_ehs_web_release.md`
+- Create: `.codex/memory/reference_python_venv.md`
+- Create: `.codex/memory/reference_wrapper_apps.md`
+
+- [x] **Step 1: Create the memory index**
+
+Create one relative Markdown link for each of the nine detailed memory files. Each link must summarize when the file is relevant.
+
+- [x] **Step 2: Migrate stable feedback and references**
+
+Preserve the operational meaning of:
+
+- Backend-delivered frontend routes.
+- The intentional `scripts/*.py` ignore rule.
+- Explicit-only release/version bumps.
+- The project virtual environment.
+- The independent frontend release workflow.
+- The two wrapper applications.
+
+Remove Claude session IDs, Claude metadata, wiki-link syntax, and Claude-specific tool names.
+
+- [x] **Step 3: Migrate project background and historical diagnosis**
+
+Preserve:
+
+- Backend/frontend coupling and statistics page conventions.
+- Read-only production-data verification policy.
+- The `material_ofrom` historical diagnosis, affected code locations, known evidence, and next investigation point.
+
+Replace all database credentials with instructions to read connection parameters from ignored local configuration. Do not include usernames, passwords, `PGPASSWORD`, or ready-to-run production connection commands.
+
+- [x] **Step 4: Verify index targets**
+
+Run:
+
+```powershell
+$missing = @()
+Get-Content .codex/memory/MEMORY.md |
+  Select-String -AllMatches '\]\(([^)]+\.md)\)' |
+  ForEach-Object { $_.Matches } |
+  ForEach-Object {
+    $target = Join-Path '.codex/memory' $_.Groups[1].Value
+    if (-not (Test-Path -LiteralPath $target)) { $missing += $target }
+  }
+if ($missing.Count -gt 0) { $missing; exit 1 }
+"All memory links resolve"
+```
+
+Expected: `All memory links resolve`.
+
+- [x] **Step 5: Commit**
+
+```powershell
+git add -- .codex/memory
+git commit -m "docs: migrate project memory for Codex"
+```
+
+### Task 4: Validate migration safety and coverage
+
+**Files:**
+- Inspect: `AGENTS.md`
+- Inspect: `.agents/skills/release/SKILL.md`
+- Inspect: `.codex/memory/*.md`
+- Inspect unchanged: `.claude/settings.local.json`
+- Inspect unchanged: `.claude/commands/release.md`
+
+- [x] **Step 1: Scan for prohibited Claude artifacts and secrets**
+
+Run:
+
+```powershell
+rg -n -i "originSessionId|scratchpad|PGPASSWORD|Co-Authored-By: Claude|password\s*=|postgresql://" AGENTS.md .agents/skills .codex/memory
+```
+
+Expected: no matches.
+
+- [x] **Step 2: Check topic coverage**
+
+Run:
+
+```powershell
+Get-ChildItem .codex/memory -File | Sort-Object Name | Select-Object -ExpandProperty Name
+```
+
+Expected: `MEMORY.md` plus all nine detailed memory files are listed.
+
+- [x] **Step 3: Confirm Claude files are untouched**
+
+Run:
+
+```powershell
+git status --short -- .claude
+git diff -- .claude
+```
+
+Expected: no output.
+
+- [x] **Step 4: Check formatting and scoped changes**
+
+Run:
+
+```powershell
+git diff --check
+git status --short
+```
+
+Expected: no whitespace errors; only pre-existing user files and this migration's plan file remain uncommitted.
+
+- [x] **Step 5: Commit the implementation plan**
+
+```powershell
+git add -- docs/superpowers/plans/2026-07-27-claude-to-codex-project-context.md
+git commit -m "docs: record Codex context migration plan"
+```

docs/superpowers/specs/2026-07-27-claude-to-codex-project-context-design.md

@@ -0,0 +1,101 @@
+# Claude 项目上下文迁移到 Codex：设计说明
+
+## 目标
+
+在不删除或修改现有 `.claude` 配置的前提下，把 `factory` 项目中仍有效的 Claude 项目知识迁移成 Codex 可稳定使用的项目级上下文。
+
+迁移后的配置应做到：
+
+- Codex 每次进入仓库时都能读取关键项目约束。
+- 较长的背景资料和历史排障记录不会挤占常规任务上下文。
+- 发版流程只在用户明确要求时启用。
+- 不复制数据库口令等敏感信息。
+- Claude 与 Codex 可以继续并行使用。
+
+## 现状
+
+项目内现有 Claude 配置：
+
+- `.claude/settings.local.json`：Claude 工具授权历史，包含大量机器相关命令及明文数据库连接信息。
+- `.claude/commands/release.md`：后端发版命令。
+- Claude 用户目录中的项目 memory：包含前后端关系、动态路由、Python 虚拟环境、生产库只读查询约束、脚本忽略规则、发版约束、套壳 App 背景及一次历史缺陷排查。
+
+项目当前没有根目录 `CLAUDE.md`、`AGENTS.md` 或项目级 Codex memory。
+
+## 采用方案
+
+采用分层兼容结构：
+
+1. 根目录 `AGENTS.md`
+   - 放置每次工作都应遵守的稳定规则。
+   - 内容保持简短，避免把一次性历史排障细节注入所有任务。
+   - 指向更详细的 memory 和按需 skill。
+
+2. `.agents/skills/release/SKILL.md`
+   - 把 `.claude/commands/release.md` 转换为 Codex 项目 skill。
+   - 仅在用户明确要求“发版”“release”或“bump 版本”时使用。
+   - 保留版本生成、更新 `SYS_VERSION`、检查 changelog、提交、打 tag 和推送的顺序。
+   - 去掉 Claude 专属的 `Co-Authored-By` 署名。
+   - 保留远端写操作前的工作区检查和失败处理约束。
+
+3. `.codex/memory/`
+   - `MEMORY.md` 作为主题索引。
+   - 按主题保存项目背景、用户反馈和历史排障资料。
+   - 去除 Claude 的 session ID、Claude 专属元数据和 wiki 链接语法。
+   - 使用普通 Markdown 相对链接，便于人工和 Codex 按需读取。
+
+4. `.claude/`
+   - 原样保留，不删除、不重写。
+
+## 内容映射
+
+必须进入 `AGENTS.md` 的规则：
+
+- 配套前端位于 `../ehs_web`；后端 API 变化时检查对应前端调用。
+- `ehs_web` 菜单和路由由后端动态下发，新页面不要修改 `src/config/route.js`。
+- Django/Python 命令使用项目根目录 `.venv/Scripts/python.exe`。
+- `scripts/*.py` 是有意忽略的一次性脚本，未经明确要求不得 `git add -f`。
+- 不自动提升版本或发版，只有用户明确要求时才执行 release skill。
+- 生产数据库只允许只读查询；任何写操作需要用户另行明确授权。
+- 不在项目文档、skill 或命令中保存数据库口令。
+
+保存在 `.codex/memory/` 的资料：
+
+- 前后端工程关系与统计页面惯例。
+- 生产数据库只读验证方法，但连接参数只指向本地忽略配置，不记录凭据。
+- 前端独立发版流程。
+- 两个 WebView 套壳 App 的位置与交互约定。
+- `material_ofrom` 合批历史缺陷的排查结论和后续接续点。
+- 上述关键反馈规则的详细原因。
+
+不迁移的内容：
+
+- `.claude/settings.local.json` 中的 Claude 权限语法。
+- 临时 scratchpad 路径、历史会话 ID、一次性命令白名单。
+- 数据库用户名、密码及可直接复用的带密码命令。
+
+## 安全与错误处理
+
+- 新文件中扫描常见密码片段和 `PGPASSWORD`，确认没有凭据泄漏。
+- 不读取或修改被 `.gitignore` 排除的本地数据库配置。
+- 不连接生产数据库验证迁移，因为本任务只迁移文档和工作约束。
+- 不删除 `.claude`，迁移失败时现有 Claude 工作流不受影响。
+- 不执行 release skill；仅验证其结构与引用路径。
+
+## 验证
+
+完成迁移后执行：
+
+1. 检查 `AGENTS.md`、`.agents/skills/release/SKILL.md` 和 `.codex/memory/*.md` 均存在。
+2. 检查 `AGENTS.md` 中的索引链接均能解析到实际文件。
+3. 搜索新文件中的密码、`PGPASSWORD`、Claude session ID 和临时 scratchpad 路径。
+4. 对照原 memory 索引，确认所有仍有效主题均已覆盖。
+5. 检查 Git diff，确认 `.claude` 没有变化，且没有混入用户现有未跟踪文件。
+
+## 完成标准
+
+- Codex 项目级入口、按需发版 skill 和 memory 索引全部建立。
+- 原 Claude 配置保持不变。
+- 原有 9 个 memory 主题均被迁移或被更高层规则覆盖。
+- 新配置不含明文凭据或 Claude 专属运行痕迹。
+- 验证命令通过，Git diff 只包含本次迁移相关文件。

manage.py

@@ -5,7 +5,12 @@ import sys
 
 
 def main():
-    os.environ.setdefault('DJANGO_SETTINGS_MODULE', 'server.settings')
+    settings_module = (
+        'server.test_settings'
+        if sys.argv[1:2] == ['test']
+        else 'server.settings'
+    )
+    os.environ.setdefault('DJANGO_SETTINGS_MODULE', settings_module)
     try:
         from django.core.management import execute_from_command_line
     except ImportError as exc:

mcp_server/__init__.py

@@ -0,0 +1 @@
+"""Factory MCP v2 integration."""

mcp_server/__main__.py

@@ -0,0 +1,24 @@
+import os
+
+import django
+import uvicorn
+
+
+def main() -> None:
+    os.environ.setdefault("DJANGO_SETTINGS_MODULE", "server.settings")
+    django.setup()
+
+    from mcp_server.server import application
+
+    from django.conf import settings
+
+    uvicorn.run(
+        application,
+        host=settings.MCP_HOST,
+        port=settings.MCP_PORT,
+        log_level="info",
+    )
+
+
+if __name__ == "__main__":
+    main()

mcp_server/auth.py

@@ -0,0 +1,66 @@
+from asgiref.sync import sync_to_async
+from rest_framework.exceptions import APIException
+from rest_framework_simplejwt.authentication import JWTAuthentication
+from rest_framework_simplejwt.exceptions import TokenError
+
+from mcp.server.auth.provider import AccessToken
+from mcp.server.auth.middleware.bearer_auth import RequireAuthMiddleware
+from starlette.types import ASGIApp, Receive, Scope, Send
+
+
+def verify_factory_jwt(token: str) -> AccessToken | None:
+    """验证 Factory access token，并生成 MCP 的逐请求身份信息。"""
+    authentication = JWTAuthentication()
+    try:
+        validated_token = authentication.get_validated_token(token)
+        user = authentication.get_user(validated_token)
+    except (APIException, TokenError):
+        return None
+
+    return AccessToken(
+        token=token,
+        client_id="factory-mcp",
+        scopes=["factory:user"],
+        expires_at=validated_token.get("exp"),
+        subject=str(user.pk),
+        claims={
+            "factory_user": {
+                "id": str(user.pk),
+                "username": user.get_username(),
+                "name": user.name,
+                "is_superuser": user.is_superuser,
+            },
+        },
+    )
+
+
+class FactoryJWTVerifier:
+    """让 MCP SDK 复用 Factory SimpleJWT 的验证规则。"""
+
+    async def verify_token(self, token: str) -> AccessToken | None:
+        return await sync_to_async(
+            verify_factory_jwt,
+            thread_sensitive=True,
+        )(token)
+
+
+class RequireFactoryJWTMiddleware:
+    """仅保护 HTTP 请求，并把 ASGI lifespan 原样交给 MCP SDK。"""
+
+    def __init__(self, app: ASGIApp):
+        self.app = app
+        self.protected_app = RequireAuthMiddleware(
+            app,
+            required_scopes=[],
+        )
+
+    async def __call__(
+        self,
+        scope: Scope,
+        receive: Receive,
+        send: Send,
+    ) -> None:
+        if scope["type"] != "http":
+            await self.app(scope, receive, send)
+            return
+        await self.protected_app(scope, receive, send)

mcp_server/context.py

@@ -0,0 +1,32 @@
+from typing import Any
+
+from django.contrib.auth import get_user_model
+from mcp.server.auth.middleware.auth_context import get_access_token
+
+from apps.utils.permission import has_perm
+
+
+def authenticated_user_claims() -> dict[str, Any]:
+    """返回当前请求中经过 Factory JWT 校验的用户摘要。"""
+    access_token = get_access_token()
+    claims = access_token.claims if access_token else None
+    user = claims.get("factory_user") if claims else None
+    if not isinstance(user, dict):
+        raise RuntimeError("当前 MCP 请求缺少有效的 Factory 用户身份")
+    return user
+
+
+def authenticated_factory_user():
+    """加载当前 JWT 对应的 Django 用户，供权限和数据范围逻辑复用。"""
+    claims = authenticated_user_claims()
+    try:
+        return get_user_model().objects.select_related("belong_dept").get(
+            pk=claims["id"]
+        )
+    except (KeyError, get_user_model().DoesNotExist) as exc:
+        raise RuntimeError("当前 JWT 对应的 Factory 用户不存在") from exc
+
+
+def require_permission(user, permission_code: str) -> None:
+    if not has_perm(user, [permission_code]):
+        raise PermissionError(f"当前用户缺少权限：{permission_code}")

mcp_server/server.py

@@ -0,0 +1,89 @@
+from collections.abc import Sequence
+from typing import Any
+
+from django.conf import settings
+from mcp.server import MCPServer
+from mcp.server.auth.middleware.auth_context import AuthContextMiddleware
+from mcp.server.auth.middleware.bearer_auth import (
+    BearerAuthBackend,
+)
+from mcp.server.auth.provider import TokenVerifier
+from mcp.server.transport_security import TransportSecuritySettings
+from starlette.middleware.authentication import AuthenticationMiddleware
+from starlette.types import ASGIApp
+
+from mcp_server.auth import (
+    FactoryJWTVerifier,
+    RequireFactoryJWTMiddleware,
+)
+from mcp_server.context import authenticated_user_claims
+from mcp_server.tools.batch_stats import register_batch_stat_tools
+from mcp_server.tools.datasets import register_dataset_tools
+from mcp_server.tools.wprs import register_wpr_tools
+
+
+PROTOCOL_REVISION = "2026-07-28"
+
+mcp = MCPServer(
+    name="factory",
+    title="Factory MCP",
+    description="Factory 面向 Agent 的受控业务能力入口。",
+    instructions="所有工具均使用当前请求携带的 Factory JWT 身份执行。",
+    version=settings.SYS_VERSION,
+)
+
+
+@mcp.tool()
+def factory_server_info() -> dict[str, Any]:
+    """返回 Factory MCP 服务版本及协议基础信息。"""
+    return {
+        "name": "factory",
+        "system_version": settings.SYS_VERSION,
+        "protocol_revision": PROTOCOL_REVISION,
+        "authentication": "factory_jwt",
+        "domain_tools_ready": True,
+    }
+
+
+@mcp.tool()
+def factory_whoami() -> dict[str, Any]:
+    """返回当前 Factory JWT 对应的用户身份。"""
+    return authenticated_user_claims()
+
+
+register_dataset_tools(mcp)
+register_wpr_tools(mcp)
+register_batch_stat_tools(mcp)
+
+
+def create_app(
+    *,
+    allowed_hosts: Sequence[str] | None = None,
+    allowed_origins: Sequence[str] | None = None,
+    token_verifier: TokenVerifier | None = None,
+) -> ASGIApp:
+    """创建仅接受 Factory JWT 的 MCP v2 Streamable HTTP 应用。"""
+    transport_security = TransportSecuritySettings(
+        enable_dns_rebinding_protection=True,
+        allowed_hosts=list(settings.MCP_ALLOWED_HOSTS if allowed_hosts is None else allowed_hosts),
+        allowed_origins=list(settings.MCP_ALLOWED_ORIGINS if allowed_origins is None else allowed_origins),
+    )
+    app: ASGIApp = mcp.streamable_http_app(
+        streamable_http_path=settings.MCP_PATH,
+        json_response=True,
+        max_request_body_size=settings.MCP_MAX_REQUEST_BODY_SIZE,
+        transport_security=transport_security,
+        host=settings.MCP_HOST,
+    )
+
+    # 包装顺序保证先解析 Bearer JWT，再写入 MCP 请求上下文，最后强制认证。
+    app = RequireFactoryJWTMiddleware(app)
+    app = AuthContextMiddleware(app)
+    app = AuthenticationMiddleware(
+        app,
+        backend=BearerAuthBackend(token_verifier or FactoryJWTVerifier()),
+    )
+    return app
+
+
+application = create_app()

mcp_server/test_batch_stats.py

@@ -0,0 +1,105 @@
+from datetime import datetime
+from types import SimpleNamespace
+from unittest.mock import MagicMock, patch
+
+from django.test import SimpleTestCase
+
+from apps.wpm.models import BatchSt
+from mcp_server.tools.batch_stats import get_batch_stat, search_batch_stats
+
+
+def batch_stat(**overrides):
+    values = {
+        "id": "500",
+        "batch": "BATCH-001",
+        "version": 1,
+        "zt_batch": "ZT-001",
+        "first_time": datetime(2026, 8, 1, 8, 0),
+        "last_time": datetime(2026, 8, 2, 8, 0),
+        "material_start": SimpleNamespace(
+            id="100",
+            name="原料",
+            model="M-1",
+            specification="S-1",
+        ),
+        "data": {"output": {"count": 10}, "quality": {"ok": 9}},
+        "update_time": datetime(2026, 8, 10, 8, 0),
+    }
+    values.update(overrides)
+    return SimpleNamespace(**values)
+
+
+class BatchStatToolTests(SimpleTestCase):
+    @patch("mcp_server.tools.batch_stats._base_queryset")
+    @patch("mcp_server.tools.batch_stats.authenticated_factory_user")
+    def test_search_returns_summary_without_full_data(
+        self,
+        _user_mock,
+        queryset_mock,
+    ):
+        queryset = MagicMock()
+        queryset.filter.return_value = queryset
+        queryset.order_by.return_value = queryset
+        queryset.__getitem__.return_value = [batch_stat()]
+        queryset_mock.return_value = queryset
+
+        result = search_batch_stats(query="BATCH", limit=10)
+
+        self.assertEqual(result["items"][0]["batch"], "BATCH-001")
+        self.assertEqual(result["items"][0]["data_keys"], ["output", "quality"])
+        self.assertNotIn("data", result["items"][0])
+
+    @patch("mcp_server.tools.batch_stats.BatchLog.objects.filter")
+    @patch("mcp_server.tools.batch_stats._base_queryset")
+    @patch("mcp_server.tools.batch_stats.authenticated_factory_user")
+    def test_get_returns_data_and_direct_relations(
+        self,
+        _user_mock,
+        queryset_mock,
+        relation_filter_mock,
+    ):
+        item = batch_stat()
+        queryset_mock.return_value.get.return_value = item
+        relation_filter_mock.return_value.select_related.return_value.values.return_value = [
+            {
+                "id": "600",
+                "relation_type": "split",
+                "source_id": "500",
+                "source__batch": "BATCH-001",
+                "source__version": 1,
+                "target_id": "501",
+                "target__batch": "BATCH-001-1",
+                "target__version": 1,
+                "handover_id": "700",
+                "mlog_id": None,
+            }
+        ]
+
+        result = get_batch_stat("BATCH-001")
+
+        self.assertEqual(result["data"]["output"]["count"], 10)
+        self.assertEqual(result["relations"][0]["relation_type"], "split")
+
+    @patch("mcp_server.tools.batch_stats.BatchLog.objects.filter")
+    @patch("mcp_server.tools.batch_stats._base_queryset")
+    @patch("mcp_server.tools.batch_stats.authenticated_factory_user")
+    def test_get_can_omit_relations(
+        self,
+        _user_mock,
+        queryset_mock,
+        relation_filter_mock,
+    ):
+        queryset_mock.return_value.get.return_value = batch_stat()
+
+        result = get_batch_stat("BATCH-001", include_relations=False)
+
+        self.assertNotIn("relations", result)
+        relation_filter_mock.assert_not_called()
+
+    @patch("mcp_server.tools.batch_stats._base_queryset")
+    @patch("mcp_server.tools.batch_stats.authenticated_factory_user")
+    def test_get_reports_missing_batch(self, _user_mock, queryset_mock):
+        queryset_mock.return_value.get.side_effect = BatchSt.DoesNotExist
+
+        with self.assertRaisesRegex(ValueError, "未找到批次统计"):
+            get_batch_stat("missing")

mcp_server/test_datasets.py

@@ -0,0 +1,96 @@
+from types import SimpleNamespace
+from unittest.mock import MagicMock, patch
+
+from django.test import SimpleTestCase, override_settings
+
+from mcp_server.tools.datasets import execute_dataset, search_datasets
+
+
+class DatasetToolTests(SimpleTestCase):
+    @patch("mcp_server.tools.datasets.authenticated_factory_user")
+    @patch("mcp_server.tools.datasets.Dataset.objects.filter")
+    def test_search_returns_safe_catalog_fields(self, filter_mock, _user_mock):
+        queryset = MagicMock()
+        filter_mock.return_value = queryset
+        queryset.order_by.return_value.values.return_value.__getitem__.return_value = [
+            {
+                "code": "daily_output",
+                "name": "日产量",
+                "description": "按日统计产量",
+                "default_param": {"day": "2026-08-10"},
+                "test_param": {},
+            }
+        ]
+
+        result = search_datasets(limit=10)
+
+        self.assertEqual(result["items"][0]["code"], "daily_output")
+        self.assertNotIn("sql_query", result["items"][0])
+
+    @patch("mcp_server.tools.datasets.cache")
+    @patch("mcp_server.tools.datasets.execute_rendered_dataset")
+    @patch("mcp_server.tools.datasets.render_dataset_sql")
+    @patch("mcp_server.tools.datasets.require_permission")
+    @patch("mcp_server.tools.datasets.authenticated_factory_user")
+    @patch("mcp_server.tools.datasets.Dataset.objects.get")
+    def test_execute_reuses_identity_permission_and_service(
+        self,
+        get_mock,
+        user_mock,
+        permission_mock,
+        render_mock,
+        execute_mock,
+        cache_mock,
+    ):
+        item = SimpleNamespace(
+            code="daily_output",
+            name="日产量",
+            description="按日统计产量",
+            cache_seconds=10,
+        )
+        user = SimpleNamespace(id=42, belong_dept_id=7)
+        get_mock.return_value = item
+        user_mock.return_value = user
+        render_mock.return_value = "select 1"
+        cache_mock.get.return_value = None
+        execute_mock.return_value = (
+            {"data": {"ds0": [{"count": 1}]}, "data2": {}},
+            True,
+        )
+
+        result = execute_dataset("daily_output", {"day": "2026-08-10"})
+
+        permission_mock.assert_called_once_with(user, "dataset.exec")
+        render_query = render_mock.call_args.args[1]
+        self.assertEqual(render_query["r_user"], 42)
+        self.assertEqual(render_query["r_dept"], 7)
+        self.assertEqual(result["data"]["ds0"][0]["count"], 1)
+        self.assertNotIn("sql_query", result)
+
+    @override_settings(MCP_MAX_RESULT_BYTES=1)
+    @patch("mcp_server.tools.datasets.cache")
+    @patch("mcp_server.tools.datasets.execute_rendered_dataset")
+    @patch("mcp_server.tools.datasets.render_dataset_sql", return_value="")
+    @patch("mcp_server.tools.datasets.require_permission")
+    @patch("mcp_server.tools.datasets.authenticated_factory_user")
+    @patch("mcp_server.tools.datasets.Dataset.objects.get")
+    def test_execute_rejects_oversized_results(
+        self,
+        get_mock,
+        user_mock,
+        _permission_mock,
+        _render_mock,
+        execute_mock,
+        _cache_mock,
+    ):
+        get_mock.return_value = SimpleNamespace(
+            code="daily_output",
+            name="日产量",
+            description="",
+            cache_seconds=0,
+        )
+        user_mock.return_value = SimpleNamespace(id=42, belong_dept_id=None)
+        execute_mock.return_value = ({"data": {"ds0": [1]}, "data2": {}}, True)
+
+        with self.assertRaisesRegex(RuntimeError, "超过 MCP 响应上限"):
+            execute_dataset("daily_output")

mcp_server/test_wprs.py

@@ -0,0 +1,103 @@
+from datetime import datetime
+from types import SimpleNamespace
+from unittest.mock import MagicMock, patch
+
+from django.test import SimpleTestCase
+
+from mcp_server.tools.wprs import get_wpr, search_wprs
+
+
+def material(material_id="100", name="成品"):
+    return SimpleNamespace(
+        id=material_id,
+        name=name,
+        model="M-1",
+        specification="S-1",
+    )
+
+
+def wpr(**overrides):
+    values = {
+        "id": "200",
+        "number": "WPR-001",
+        "number_out": "OUT-001",
+        "version": 1,
+        "state": 10,
+        "get_state_display": lambda: "正常",
+        "material": material(),
+        "material_start": material("101", "原料"),
+        "wm_id": "300",
+        "wm": SimpleNamespace(batch="WP-001"),
+        "mb_id": None,
+        "mb": None,
+        "wpr_from_id": None,
+        "wpr_from": None,
+        "oinfo": {"test": "ok"},
+        "data": {"route": []},
+        "pre_info": {"tooling": "T-1"},
+        "create_time": datetime(2026, 8, 1, 8, 0),
+        "update_time": datetime(2026, 8, 10, 8, 0),
+    }
+    values.update(overrides)
+    return SimpleNamespace(**values)
+
+
+class WprToolTests(SimpleTestCase):
+    @patch("mcp_server.tools.wprs._base_queryset")
+    @patch("mcp_server.tools.wprs.authenticated_factory_user")
+    def test_search_returns_read_only_summary(self, _user_mock, queryset_mock):
+        queryset = MagicMock()
+        queryset.filter.return_value = queryset
+        queryset.distinct.return_value = queryset
+        queryset.order_by.return_value = queryset
+        queryset.__getitem__.return_value = [wpr()]
+        queryset_mock.return_value = queryset
+
+        result = search_wprs(query="WPR", location="workshop", limit=10)
+
+        self.assertEqual(result["items"][0]["number"], "WPR-001")
+        self.assertEqual(result["items"][0]["workshop_batch"], "WP-001")
+        self.assertNotIn("data", result["items"][0])
+        self.assertNotIn("pre_info", result["items"][0])
+
+    @patch("mcp_server.tools.wprs.WprDefect.objects.filter")
+    @patch("mcp_server.tools.wprs._base_queryset")
+    @patch("mcp_server.tools.wprs.authenticated_factory_user")
+    def test_get_returns_business_detail(
+        self,
+        _user_mock,
+        queryset_mock,
+        defect_filter_mock,
+    ):
+        item = wpr()
+        queryset = MagicMock()
+        queryset.filter.return_value.order_by.return_value.first.return_value = item
+        queryset_mock.return_value = queryset
+        defect_filter_mock.return_value.select_related.return_value.values.return_value = [
+            {
+                "defect_id": "400",
+                "defect__name": "划伤",
+                "is_main": True,
+            }
+        ]
+
+        result = get_wpr("WPR-001")
+
+        self.assertEqual(result["material"]["name"], "成品")
+        self.assertEqual(result["defects"][0]["defect__name"], "划伤")
+        self.assertEqual(result["pre_info"]["tooling"], "T-1")
+
+    @patch("mcp_server.tools.wprs._base_queryset")
+    @patch("mcp_server.tools.wprs.authenticated_factory_user")
+    def test_get_reports_missing_wpr(self, _user_mock, queryset_mock):
+        queryset = MagicMock()
+        queryset.filter.return_value.order_by.return_value.first.return_value = None
+        queryset_mock.return_value = queryset
+
+        with self.assertRaisesRegex(ValueError, "未找到 WPR"):
+            get_wpr("missing")
+
+    @patch("mcp_server.tools.wprs.authenticated_factory_user")
+    def test_search_rejects_unknown_location(self, _user_mock):
+        with self.assertRaisesRegex(ValueError, "不支持的 WPR 位置"):
+            search_wprs(location="invalid")

mcp_server/tests.py

@@ -0,0 +1,164 @@
+import asyncio
+from types import SimpleNamespace
+from unittest.mock import patch
+
+from django.test import SimpleTestCase
+from mcp.server.auth.middleware.auth_context import auth_context_var
+from mcp.server.auth.middleware.bearer_auth import AuthenticatedUser
+from mcp.server.auth.provider import AccessToken
+from rest_framework.exceptions import AuthenticationFailed
+from starlette.testclient import TestClient
+
+from mcp_server.auth import verify_factory_jwt
+from mcp_server.server import (
+    PROTOCOL_REVISION,
+    create_app,
+    factory_server_info,
+    factory_whoami,
+    mcp,
+)
+
+
+class FactoryJWTVerifierTests(SimpleTestCase):
+    def test_factory_access_token_is_accepted(self):
+        user = SimpleNamespace(
+            pk=42,
+            name="MCP用户",
+            is_superuser=False,
+            get_username=lambda: "mcp-user",
+        )
+        authentication = patch("mcp_server.auth.JWTAuthentication").start()
+        self.addCleanup(patch.stopall)
+        authentication.return_value.get_validated_token.return_value = {
+            "exp": 1234567890,
+        }
+        authentication.return_value.get_user.return_value = user
+
+        access_token = verify_factory_jwt("access-token")
+
+        self.assertIsNotNone(access_token)
+        self.assertEqual(access_token.subject, str(user.pk))
+        self.assertEqual(
+            access_token.claims["factory_user"]["username"],
+            user.get_username(),
+        )
+
+    def test_invalid_token_is_rejected(self):
+        with patch("mcp_server.auth.JWTAuthentication") as authentication:
+            authentication.return_value.get_validated_token.side_effect = AuthenticationFailed("invalid token")
+
+            self.assertIsNone(verify_factory_jwt("invalid-token"))
+
+
+class FactoryMCPServerTests(SimpleTestCase):
+    def test_base_tools_are_registered(self):
+        tools = asyncio.run(mcp.list_tools())
+        names = {tool.name for tool in tools}
+
+        self.assertEqual(
+            names,
+            {
+                "execute_dataset",
+                "factory_server_info",
+                "factory_whoami",
+                "get_batch_stat",
+                "get_wpr",
+                "search_batch_stats",
+                "search_datasets",
+                "search_wprs",
+            },
+        )
+
+    def test_server_info_targets_mcp_v2(self):
+        result = factory_server_info()
+
+        self.assertEqual(result["protocol_revision"], PROTOCOL_REVISION)
+        self.assertTrue(result["domain_tools_ready"])
+
+    def test_whoami_uses_authenticated_request_context(self):
+        user = {
+            "id": "42",
+            "username": "agent-user",
+            "name": "Agent用户",
+            "is_superuser": False,
+        }
+        authenticated = AuthenticatedUser(
+            AccessToken(
+                token="test-token",
+                client_id="factory-mcp",
+                scopes=["factory:user"],
+                subject=user["id"],
+                claims={"factory_user": user},
+            )
+        )
+        context_token = auth_context_var.set(authenticated)
+        try:
+            self.assertEqual(factory_whoami(), user)
+        finally:
+            auth_context_var.reset(context_token)
+
+    def test_http_endpoint_requires_bearer_token(self):
+        app = create_app(allowed_hosts=["testserver"])
+
+        with TestClient(app) as client:
+            response = client.post("/mcp", json={})
+
+        self.assertEqual(response.status_code, 401)
+        self.assertEqual(response.json()["error"], "invalid_token")
+
+    def test_mcp_v2_request_uses_bearer_identity(self):
+        user = {
+            "id": "42",
+            "username": "agent-user",
+            "name": "Agent用户",
+            "is_superuser": False,
+        }
+
+        class TestTokenVerifier:
+            async def verify_token(self, token):
+                if token != "valid-token":
+                    return None
+                return AccessToken(
+                    token=token,
+                    client_id="test-client",
+                    scopes=["factory:user"],
+                    subject=user["id"],
+                    claims={"factory_user": user},
+                )
+
+        app = create_app(
+            allowed_hosts=["testserver"],
+            token_verifier=TestTokenVerifier(),
+        )
+        request = {
+            "jsonrpc": "2.0",
+            "id": 1,
+            "method": "tools/call",
+            "params": {
+                "name": "factory_whoami",
+                "arguments": {},
+                "_meta": {
+                    "io.modelcontextprotocol/protocolVersion": (PROTOCOL_REVISION),
+                    "io.modelcontextprotocol/clientInfo": {
+                        "name": "factory-tests",
+                        "version": "1.0",
+                    },
+                    "io.modelcontextprotocol/clientCapabilities": {},
+                },
+            },
+        }
+        headers = {
+            "Authorization": "Bearer valid-token",
+            "MCP-Protocol-Version": PROTOCOL_REVISION,
+            "Mcp-Method": "tools/call",
+            "Mcp-Name": "factory_whoami",
+        }
+
+        with TestClient(app) as client:
+            response = client.post("/mcp", json=request, headers=headers)
+
+        self.assertEqual(response.status_code, 200)
+        self.assertEqual(
+            response.json()["result"]["structuredContent"],
+            user,
+        )

mcp_server/tools/__init__.py

@@ -0,0 +1 @@
+"""Factory MCP 领域工具，按业务域拆分并在 server 中显式注册。"""

mcp_server/tools/batch_stats.py

@@ -0,0 +1,104 @@
+from typing import Any
+
+from django.db.models import Q
+
+from apps.wpm.models import BatchLog, BatchSt
+from mcp_server.context import authenticated_factory_user
+from mcp_server.tools.common import json_safe_result, validate_result_size
+
+
+def _base_queryset():
+    return BatchSt.objects.select_related("material_start")
+
+
+def _batch_summary(batch_stat: BatchSt) -> dict[str, Any]:
+    material = batch_stat.material_start
+    return {
+        "id": str(batch_stat.id),
+        "batch": batch_stat.batch,
+        "version": batch_stat.version,
+        "zt_batch": batch_stat.zt_batch,
+        "first_time": batch_stat.first_time,
+        "last_time": batch_stat.last_time,
+        "material_start": (
+            {
+                "id": str(material.id),
+                "name": material.name,
+                "model": material.model,
+                "specification": material.specification,
+            }
+            if material
+            else None
+        ),
+        "data_keys": sorted((batch_stat.data or {}).keys()),
+        "update_time": batch_stat.update_time,
+    }
+
+
+def search_batch_stats(
+    query: str = "",
+    zt_batch: str = "",
+    material_id: str | None = None,
+    version: int | None = 1,
+    limit: int = 20,
+) -> dict[str, Any]:
+    """搜索批次统计；摘要仅返回数据分组名称，不返回完整统计数据。"""
+    authenticated_factory_user()
+    safe_limit = max(1, min(limit, 100))
+    queryset = _base_queryset()
+    if query.strip():
+        queryset = queryset.filter(batch__icontains=query.strip())
+    if zt_batch.strip():
+        queryset = queryset.filter(zt_batch=zt_batch.strip())
+    if material_id:
+        queryset = queryset.filter(material_start_id=material_id)
+    if version is not None:
+        queryset = queryset.filter(version=version)
+    items = [
+        _batch_summary(item)
+        for item in queryset.order_by("batch", "version")[:safe_limit]
+    ]
+    result = json_safe_result({"items": items, "limit": safe_limit})
+    validate_result_size(result)
+    return result
+
+
+def get_batch_stat(
+    batch: str,
+    version: int = 1,
+    include_relations: bool = True,
+) -> dict[str, Any]:
+    """读取指定批次版本的完整统计数据，并可附带直接拆合批关系。"""
+    authenticated_factory_user()
+    try:
+        batch_stat = _base_queryset().get(batch=batch, version=version)
+    except BatchSt.DoesNotExist as exc:
+        raise ValueError(f"未找到批次统计：{batch} v{version}") from exc
+
+    result = _batch_summary(batch_stat)
+    result["data"] = batch_stat.data
+    if include_relations:
+        result["relations"] = list(
+            BatchLog.objects.filter(Q(source=batch_stat) | Q(target=batch_stat))
+            .select_related("source", "target")
+            .values(
+                "id",
+                "relation_type",
+                "source_id",
+                "source__batch",
+                "source__version",
+                "target_id",
+                "target__batch",
+                "target__version",
+                "handover_id",
+                "mlog_id",
+            )
+        )
+    result = json_safe_result(result)
+    validate_result_size(result)
+    return result
+
+
+def register_batch_stat_tools(server) -> None:
+    server.tool()(search_batch_stats)
+    server.tool()(get_batch_stat)

mcp_server/tools/common.py

@@ -0,0 +1,18 @@
+import json
+from typing import Any
+
+from django.conf import settings
+
+from apps.utils.tools import MyJSONEncoder
+
+
+def json_safe_result(value: Any) -> Any:
+    return json.loads(json.dumps(value, cls=MyJSONEncoder, ensure_ascii=False))
+
+
+def validate_result_size(value: Any) -> None:
+    encoded = json.dumps(value, ensure_ascii=False, separators=(",", ":")).encode()
+    if len(encoded) > settings.MCP_MAX_RESULT_BYTES:
+        raise RuntimeError(
+            "工具结果超过 MCP 响应上限，请缩小查询范围或增加筛选参数"
+        )

mcp_server/tools/datasets.py

@@ -0,0 +1,79 @@
+import hashlib
+from typing import Any
+
+from django.core.cache import cache
+from django.db.models import Q
+
+from apps.bi.models import Dataset
+from apps.bi.services import execute_rendered_dataset, render_dataset_sql
+from mcp_server.context import authenticated_factory_user, require_permission
+from mcp_server.tools.common import json_safe_result, validate_result_size
+
+
+def search_datasets(query: str = "", limit: int = 20) -> dict[str, Any]:
+    """搜索可执行的数据集目录，不返回 SQL 等敏感配置。"""
+    authenticated_factory_user()
+    safe_limit = max(1, min(limit, 100))
+    queryset = Dataset.objects.filter(enabled=True)
+    if query.strip():
+        queryset = queryset.filter(
+            Q(name__icontains=query.strip())
+            | Q(code__icontains=query.strip())
+            | Q(description__icontains=query.strip())
+        )
+    rows = queryset.order_by("name", "code", "id").values(
+        "code",
+        "name",
+        "description",
+        "default_param",
+        "test_param",
+    )[:safe_limit]
+    return {"items": list(rows), "limit": safe_limit}
+
+
+def execute_dataset(
+    code: str,
+    parameters: dict[str, Any] | None = None,
+) -> dict[str, Any]:
+    """以当前 Factory 用户身份执行启用的数据集。需要 dataset.exec 权限。"""
+    user = authenticated_factory_user()
+    require_permission(user, "dataset.exec")
+    try:
+        dataset = Dataset.objects.get(code=code, enabled=True)
+    except Dataset.DoesNotExist as exc:
+        raise ValueError(f"未找到已启用的数据集：{code}") from exc
+    except Dataset.MultipleObjectsReturned as exc:
+        raise RuntimeError(f"数据集 code 不唯一，无法执行：{code}") from exc
+
+    query = dict(parameters or {})
+    query["r_user"] = user.id
+    query["r_dept"] = user.belong_dept_id or ""
+    full_sql = render_dataset_sql(dataset, query)
+
+    cache_key = None
+    response_data = None
+    if full_sql and dataset.cache_seconds:
+        digest = hashlib.sha256(full_sql.strip(";").encode()).hexdigest()
+        cache_key = f"mcp:dataset:{digest}"
+        response_data = cache.get(cache_key)
+
+    if response_data is None:
+        response_data, can_cache = execute_rendered_dataset(dataset, full_sql)
+        if cache_key and can_cache and response_data["data"]:
+            cache.set(cache_key, response_data, dataset.cache_seconds)
+
+    result = json_safe_result(
+        {
+            "code": dataset.code,
+            "name": dataset.name,
+            "description": dataset.description,
+            **response_data,
+        }
+    )
+    validate_result_size(result)
+    return result
+
+
+def register_dataset_tools(server) -> None:
+    server.tool()(search_datasets)
+    server.tool()(execute_dataset)

mcp_server/tools/wprs.py

@@ -0,0 +1,139 @@
+from typing import Any, Literal
+
+from django.db.models import Q
+
+from apps.wpmw.models import Wpr, WprDefect
+from mcp_server.context import authenticated_factory_user
+from mcp_server.tools.common import json_safe_result, validate_result_size
+
+
+WprLocation = Literal["all", "workshop", "warehouse", "unassigned"]
+
+
+def _base_queryset():
+    return Wpr.objects.select_related(
+        "material",
+        "material_start",
+        "wm",
+        "mb",
+        "wpr_from",
+    )
+
+
+def _wpr_summary(wpr: Wpr) -> dict[str, Any]:
+    material = wpr.material
+    return {
+        "id": str(wpr.id),
+        "number": wpr.number,
+        "number_out": wpr.number_out,
+        "version": wpr.version,
+        "state": wpr.state,
+        "state_name": wpr.get_state_display(),
+        "material": {
+            "id": str(material.id),
+            "name": material.name,
+            "model": material.model,
+            "specification": material.specification,
+        },
+        "workshop_batch": wpr.wm.batch if wpr.wm_id else None,
+        "warehouse_batch": wpr.mb.batch if wpr.mb_id else None,
+        "create_time": wpr.create_time,
+        "update_time": wpr.update_time,
+    }
+
+
+def search_wprs(
+    query: str = "",
+    state: int | None = None,
+    material_id: str | None = None,
+    batch: str = "",
+    location: WprLocation = "all",
+    limit: int = 20,
+) -> dict[str, Any]:
+    """按编号、物料或批次搜索单件产品；仅提供只读摘要。"""
+    authenticated_factory_user()
+    if location not in {"all", "workshop", "warehouse", "unassigned"}:
+        raise ValueError(f"不支持的 WPR 位置：{location}")
+    safe_limit = max(1, min(limit, 100))
+    queryset = _base_queryset()
+    if query.strip():
+        keyword = query.strip()
+        queryset = queryset.filter(
+            Q(number__icontains=keyword)
+            | Q(number_out__icontains=keyword)
+            | Q(material__name__icontains=keyword)
+            | Q(material__model__icontains=keyword)
+            | Q(material__specification__icontains=keyword)
+        )
+    if state is not None:
+        queryset = queryset.filter(state=state)
+    if material_id:
+        queryset = queryset.filter(material_id=material_id)
+    if batch.strip():
+        queryset = queryset.filter(
+            Q(wm__batch__icontains=batch.strip())
+            | Q(mb__batch__icontains=batch.strip())
+        )
+    if location == "workshop":
+        queryset = queryset.filter(wm__isnull=False)
+    elif location == "warehouse":
+        queryset = queryset.filter(mb__isnull=False)
+    elif location == "unassigned":
+        queryset = queryset.filter(wm__isnull=True, mb__isnull=True)
+
+    items = [
+        _wpr_summary(wpr)
+        for wpr in queryset.distinct().order_by("number", "create_time")[:safe_limit]
+    ]
+    result = json_safe_result({"items": items, "limit": safe_limit})
+    validate_result_size(result)
+    return result
+
+
+def get_wpr(identifier: str) -> dict[str, Any]:
+    """按 WPR ID、内部编号或对外编号读取单件详情。"""
+    authenticated_factory_user()
+    lookup = Q(number=identifier) | Q(number_out=identifier)
+    if identifier.isdigit():
+        lookup |= Q(pk=identifier)
+    wpr = _base_queryset().filter(lookup).order_by("-version", "-update_time").first()
+    if wpr is None:
+        raise ValueError(f"未找到 WPR：{identifier}")
+
+    result = _wpr_summary(wpr)
+    material_start = wpr.material_start
+    result.update(
+        {
+            "material_start": (
+                {
+                    "id": str(material_start.id),
+                    "name": material_start.name,
+                    "model": material_start.model,
+                    "specification": material_start.specification,
+                }
+                if material_start
+                else None
+            ),
+            "wpr_from": (
+                {"id": str(wpr.wpr_from.id), "number": wpr.wpr_from.number}
+                if wpr.wpr_from_id
+                else None
+            ),
+            "oinfo": wpr.oinfo,
+            "data": wpr.data,
+            "pre_info": wpr.pre_info,
+            "defects": list(
+                WprDefect.objects.filter(wpr=wpr)
+                .select_related("defect")
+                .values("defect_id", "defect__name", "is_main")
+            ),
+        }
+    )
+    result = json_safe_result(result)
+    validate_result_size(result)
+    return result
+
+
+def register_wpr_tools(server) -> None:
+    server.tool()(search_wprs)
+    server.tool()(get_wpr)

pytest.ini

@@ -0,0 +1,3 @@
+[pytest]
+DJANGO_SETTINGS_MODULE = server.test_settings
+python_files = tests.py test_*.py *_tests.py

requirements.txt

@@ -10,6 +10,11 @@ django-cors-headers==4.9.0
 djangorestframework-simplejwt==5.5.1
 django-restql==0.15.2
 
+# =======================
+# Agent Integration
+# =======================
+mcp==2.0.0
+
 # =======================
 # Celery
 # =======================