From 1ee42ca17c54f9758091f35deef9f7f65ec35663 Mon Sep 17 00:00:00 2001
From: Honza Kral <honza.kral@gmail.com>
Date: Thu, 1 Aug 2013 15:05:29 +0200
Subject: [PATCH] Added a simple reindex helper

---
 elasticsearch/helpers.py                       | 10 ++++++++++
 test_elasticsearch/test_server/test_helpers.py | 17 +++++++++++++++++
 2 files changed, 27 insertions(+)

diff --git a/elasticsearch/helpers.py b/elasticsearch/helpers.py
index 505c8cdb..e7dec2bb 100644
--- a/elasticsearch/helpers.py
+++ b/elasticsearch/helpers.py
@@ -41,3 +41,13 @@ def scan(client, query=None, scroll='5m', **kwargs):
             yield hit
         scroll_id = resp['_scroll_id']
 
+def reindex(client, source_index, target_index, target_client=None, chunk_size=500, scroll='5m'):
+    target_client = client if target_client is None else target_index
+
+    docs = scan(client, index=source_index, scroll=scroll)
+    def _change_doc_index(hits, index):
+        for h in hits:
+            h['_index'] = index
+            yield h
+
+    return bulk_index(target_client, _change_doc_index(docs, target_index), chunk_size=chunk_size)
diff --git a/test_elasticsearch/test_server/test_helpers.py b/test_elasticsearch/test_server/test_helpers.py
index 951045e9..57c0ac7f 100644
--- a/test_elasticsearch/test_server/test_helpers.py
+++ b/test_elasticsearch/test_server/test_helpers.py
@@ -24,3 +24,20 @@ class TestScan(ElasticTestCase):
         self.assertEquals(100, len(docs))
         self.assertEquals(set(map(str, range(100))), set(d['_id'] for d in docs))
         self.assertEquals(set(range(100)), set(d['_source']['answer'] for d in docs))
+
+class TestReindex(ElasticTestCase):
+    def test_all_documents_get_moved(self):
+        bulk = []
+        for x in range(100):
+            bulk.append({"index": {"_index": "test_index", "_type": "answers" if x % 2 == 0 else "questions", "_id": x}})
+            bulk.append({"answer": x, "correct": x == 42})
+        self.client.bulk(bulk, refresh=True)
+
+        helpers.reindex(self.client, "test_index", "prod_index")
+        self.client.indices.refresh()
+
+        self.assertTrue(self.client.indices.exists("prod_index"))
+        self.assertTrue(self.client.indices.exists_type("prod_index", "answers"))
+        self.assertTrue(self.client.indices.exists_type("prod_index", "questions"))
+        self.assertEquals(100, self.client.count(index='prod_index')['count'])
+