diff --git a/apps/resm/tasks.py b/apps/resm/tasks.py index da1ac8e..ef5a88e 100644 --- a/apps/resm/tasks.py +++ b/apps/resm/tasks.py @@ -21,6 +21,12 @@ config.retry_http_codes = [429, 500, 503] OPENALEX_KEY = "NPimoE2ecdWmfdhH8abxEp" config.api_key = OPENALEX_KEY +ELSEVIER_APIKEY = 'aa8868cac9e27d6153ab0a0acd7b50bf' +ELSEVIER_HEAEDERS = { + "X-ELS-APIKey": ELSEVIER_APIKEY, + "X-ELS-Insttoken": "135fa874aea9f0de11cad187ccb4878c " +} + @shared_task(base=CustomTask) def get_paper_meta_from_openalex(publication_year:int, keywords:str="", search:str="", end_year:int=None): cache_key = f"openalex_cursor_{publication_year}_{keywords}{search}" @@ -94,9 +100,6 @@ def get_paper_meta_from_openalex(publication_year:int, keywords:str="", search:s countdown=5 ) - -ELSEVIER_APIKEY = 'aa8868cac9e27d6153ab0a0acd7b50bf' - # 常用的 User-Agent 列表 USER_AGENTS = [ "Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36", @@ -191,7 +194,8 @@ def get_abstract_from_elsevier(number_of_task:int = 20): if has_fulltext: paper.has_fulltext = True paper.has_fulltext_xml = True - save_pdf_from_elsevier(paper) + if paper.has_fulltext_pdf is False: + save_pdf_from_elsevier(paper) count_fulltext += 1 paper.save_file_xml(xml_str)