Add option to pass kwargs to scan and bulk in reindex helper

Changelog.rst

@@ -7,6 +7,8 @@ Changelog
 -----------
 
  * Using insecure SSL configuration (``verify_cert=False``) raises a warning
+ * enable ``reindex`` helper to accept any kwargs for underlying ``bulk`` and
+   ``scan`` calls
 
 1.3.0 (2014-12-31)
 ------------------

elasticsearch/helpers/__init__.py

@@ -236,7 +236,9 @@ def scan(client, query=None, scroll='5m', preserve_order=False, **kwargs):
         if scroll_id is None:
             break
 
-def reindex(client, source_index, target_index, query=None, target_client=None, chunk_size=500, scroll='5m'):
+def reindex(client, source_index, target_index, query=None, target_client=None,
+        chunk_size=500, scroll='5m', scan_kwargs={}, bulk_kwargs={}):
+
     """
     Reindex all documents from one index that satisfy a given query
     to another, potentially (if `target_client` is specified) on a different cluster.
@@ -256,14 +258,18 @@ def reindex(client, source_index, target_index, query=None, target_client=None,
     :arg chunk_size: number of docs in one chunk sent to es (default: 500)
     :arg scroll: Specify how long a consistent view of the index should be
         maintained for scrolled search
+    :arg scan_kwargs: additional kwargs to be passed to
+        :func:`~elasticsearch.helpers.scan`
+    :arg bulk_kwargs: additional kwargs to be passed to
+        :func:`~elasticsearch.helpers.bulk`
     """
     target_client = client if target_client is None else target_client
 
-    docs = scan(client, query=query, index=source_index, scroll=scroll)
+    docs = scan(client, query=query, index=source_index, scroll=scroll, **scan_kwargs)
     def _change_doc_index(hits, index):
         for h in hits:
             h['_index'] = index
             yield h
 
     return bulk(target_client, _change_doc_index(docs, target_index),
-        chunk_size=chunk_size, stats_only=True)
+        chunk_size=chunk_size, stats_only=True, **bulk_kwargs)

test_elasticsearch/test_server/test_helpers.py

@@ -215,6 +215,15 @@ class TestReindex(ElasticsearchTestCase):
             bulk.append({"answer": x, "correct": x == 42})
         self.client.bulk(bulk, refresh=True)
 
+    def test_reindex_passes_kwargs_to_scan_and_bulk(self):
+        helpers.reindex(self.client, "test_index", "prod_index", scan_kwargs={'doc_type': 'answers'}, bulk_kwargs={'refresh': True})
+
+        self.assertTrue(self.client.indices.exists("prod_index"))
+        self.assertFalse(self.client.indices.exists_type(index='prod_index', doc_type='questions'))
+        self.assertEquals(50, self.client.count(index='prod_index', doc_type='answers')['count'])
+
+        self.assertEquals({"answer": 42, "correct": True}, self.client.get(index="prod_index", doc_type="answers", id=42)['_source'])
+
     def test_reindex_accepts_a_query(self):
         helpers.reindex(self.client, "test_index", "prod_index", query={"query": {"filtered": {"filter": {"term": {"_type": "answers"}}}}})
         self.client.indices.refresh()