change:org_id和org_name修改为list类型

science_article_add/science_article_add/spiders/wos_latest_increment.py

@@ -35,9 +35,9 @@ class WosLatestIncrementSpider(scrapy.Spider):
     def __init__(self, task_obj):
         scrapy.Spider.__init__(self)
         self.task_obj = task_obj
-        self.record_id = task_obj['id']
-        self.org_id = task_obj['org_id']
-        self.org_name = task_obj['org_name']
+        self.record_id = task_obj['task_id']
+        self.org_id = self.tolist(task_obj['org_id'])
+        self.org_name = self.tolist(task_obj['org_name'])
         self.query_id = task_obj['query_id']
         self.query_content = task_obj['content']
         self.query_condition = task_obj['task_condition']
@@ -45,6 +45,13 @@ class WosLatestIncrementSpider(scrapy.Spider):
         self.first_page = task_obj.get('first_page', 1)
         self._records_found = 0
 
+    @staticmethod
+    def tolist(datas) -> list:
+        if isinstance(datas, (list, tuple, set)):
+            return list(set(datas))
+        else:
+            raise TypeError("不支持的类型：%s" % (type(datas)))
+
     async def start(self):
         full_query = self.query_content
         if self.query_condition is not None:
@@ -62,9 +69,9 @@ class WosLatestIncrementSpider(scrapy.Spider):
     async def parse(self, response: JsonResponse, **kwargs):
         meta = response.meta
         request: scrapy.Request = response.request
-        task_query_id = self.query_id
-        task_org_id = self.org_id
-        task_record_id = self.record_id
+        task_query_id: int = self.query_id
+        task_org_id: list = self.org_id
+        task_record_id: int = self.record_id
 
         if response.status != 200:
             self.logger.warning("""
@@ -97,7 +104,7 @@ class WosLatestIncrementSpider(scrapy.Spider):
             relation_item = WosIdRelationItem()
             relation_item['third_id'] = record.get("uid")
             relation_item['query_ids'] = [task_query_id]
-            relation_item['school_ids'] = [task_org_id]
+            relation_item['school_ids'] = task_org_id
             relation_item['task_ids'] = [task_record_id]
             relation_item['updated_at'] = batch_time
             yield relation_item