feat: 调整拉数据策略

handle/pull_sample_data.py

@@ -48,6 +48,13 @@ def pull_by_es(size: int = 20):
                 "must": [
                     {
                         "term": {
+                            "company_type": {
+                                "value": "1"
+                            }
+                        }
+                    },
+                    {
+                        "term": {
                             "deleted": {
                                 "value": "0"
                             }
@@ -159,11 +166,20 @@ def pull_by_max(size=100000):
     pass
 
 
+def pull(size):
+    if size > 10000:
+        pull_by_max(size)
+    else:
+        pull_by_es(500)
+        pass
+    pass
+
+
 if __name__ == '__main__':
     log.info(f"input args: {sys.argv}")
 
     if len(sys.argv) >= 2:
-        pull_by_max(size=int(sys.argv[1]))
+        pull(size=int(sys.argv[1]))
     else:
-        pull_by_max(size=1000)
+        pull(size=1000)
     pass