wos:一键运行

science_article_wos/starter/crawl_article_latest.py

@@ -112,7 +112,31 @@ def starter_latest_all():
                 client.close()
 
     process = CrawlerProcess(get_project_settings())
-    f(True)
+    f(running=True)
+    process.start()
+    process.stop()
+
+
+def starter_latest_by_record(record_id: int):
+    @defer.inlineCallbacks
+    def f():
+        client: pymysql.Connection = get_connect()
+        cursor = client.cursor(cursors.DictCursor)
+        cursor.execute(
+            'select b.id as task_id, q.id as query_id, q.content as content, b.task_condition as task_condition, q.source_type as source_type, b.is_done as is_done from task_batch_record as b join task_search_strategy as q on b.query_id=q.id where b.id=%s and q.source_type=1 limit 1',
+            (record_id,))
+        result = cursor.fetchone()
+        query_id = result['query_id']
+        cursor.execute('select org_id, org_name from relation_org_query where query_id=%s', (query_id,))
+        org_results: List[dict] = cursor.fetchall()
+        result['org_id'] = [org_result['org_id'] for org_result in org_results]
+        result['org_name'] = [org_result['org_name'] for org_result in org_results]
+
+        init_params = result
+        yield process.crawl(WosLatestIncrementSpider, task_obj=init_params)
+
+    process = CrawlerProcess(get_project_settings())
+    f()
     process.start()
     process.stop()
 
@@ -132,3 +156,4 @@ def starter():
 
 if __name__ == '__main__':
     starter_latest_all()
+    # starter_latest_by_record(8278)