modify

libs/es.py

@@ -8,6 +8,7 @@ import os.path
 import re
 import json
 from elasticsearch import Elasticsearch
+import elasticsearch.helpers
 
 
 class ESPerform(object):
@@ -116,3 +117,34 @@ class ESPerform(object):
         except:
             logging.error("catch exception,err_msg:%s" % traceback.format_exc())
             return False
+
+    @classmethod
+    def es_helpers_bulk(cls,es_cli,data_list,sub_index_name,auto_create_index=False,doc_type="_doc"):
+        try:
+            assert (es_cli is not None)
+
+            official_index_name = cls.get_official_index_name(sub_index_name)
+            index_exists = es_cli.indices.exists(official_index_name)
+            if not index_exists:
+                if not auto_create_index:
+                    logging.error("index:%s is not existing,bulk data error!" % official_index_name)
+                    return False
+                else:
+                    cls.create_index(es_cli,sub_index_name)
+                    cls.put_index_mapping(es_cli,sub_index_name)
+
+            bulk_actions = []
+            for data in data_list:
+                bulk_actions.append({
+                    '_op_type': 'index',
+                    '_index': official_index_name,
+                    '_type': doc_type,
+                    '_id': data['id'],
+                    '_source': data,
+                })
+            elasticsearch.helpers.bulk(es_cli,bulk_actions)
+
+            return True
+        except:
+            logging.error("catch exception,err_msg:%s" % traceback.format_exc())
+            return False
\ No newline at end of file

trans2es/type_info.py

@@ -7,6 +7,7 @@ import logging
 import traceback
 import django.db.models
 from django.conf import settings
+from libs.es import ESPerform
 import elasticsearch
 import elasticsearch.helpers
 import sys
@@ -196,10 +197,11 @@ class TypeInfo(object):
 
         stage_2_time = time.time()
 
-        es_result = self.elasticsearch_bulk_insert_data(
-            sub_index_name=sub_index_name,
+        es_result = ESPerform.es_helpers_bulk(
+            es_cli=es,
             data_list=data_list,
-            es=es,
+            sub_index_name=sub_index_name,
+            auto_create_index=True
         )
 
         stage_3_time = time.time()