Respect publish host when present on a node.

Closes #251. Thanks danpilch and bowensong for the patch!

elasticsearch/transport.py

@@ -1,4 +1,3 @@
-import re
 import time
 from itertools import chain
 
@@ -8,9 +7,6 @@ from .serializer import JSONSerializer, Deserializer, DEFAULT_SERIALIZERS
 from .exceptions import ConnectionError, TransportError, SerializationError, \
                         ConnectionTimeout, ImproperlyConfigured
 
-# get ip/port from "127.0.0.1:9200"
-ADDRESS_RE = re.compile(r'^(?P<host>[\.:0-9a-f]*):(?P<port>[0-9]+)?$')
-
 
 def get_host_info(node_info, host):
     """
@@ -218,15 +214,23 @@ class Transport(object):
             raise
 
         hosts = []
-        address = self.connection_class.transport_schema + '_address'
+        address_key = self.connection_class.transport_schema + '_address'
         for n in node_info['nodes'].values():
-            match = ADDRESS_RE.search(n.get(address, ''))
+            host = {}
-            if not match:
+            address = n.get(address_key, '')
+            if '/' in address:
+                host['host'], address = address.split('/', 1)
+
+            # malformed address
+            if ':' not in address:
                 continue
 
-            host = match.groupdict()
+            ip, port = address.rsplit(':', 1)
-            if 'port' in host:
+
-                host['port'] = int(host['port'])
+            # use the ip if not overridden by publish_host
+            host.setdefault('host', ip)
+            host['port'] = int(port)
+
             host = self.host_info_callback(n, host)
             if host is not None:
                 hosts.append(host)

test_elasticsearch/test_transport.py

@@ -38,6 +38,26 @@ CLUSTER_NODES = '''{
     }
 }'''
 
+CLUSTER_NODE_PUBLISH_HOST = '''{
+    "ok" : true,
+    "cluster_name" : "super_cluster",
+    "nodes" : {
+        "wE_6OGBNSjGksbONNncIbg" : {
+            "name": "Thunderbird",
+            "transport_address": "obsidian/192.168.1.60:9300",
+            "host": "192.168.1.60",
+            "ip": "192.168.1.60",
+            "version": "2.1.0",
+            "build": "72cd1f1",
+            "http_address": "obsidian/192.168.1.60:9200",
+            "attributes": {
+                "testattr": "test"
+            }
+        }
+    }
+}'''
+
+
 class TestHostsInfoCallback(TestCase):
     def test_master_only_nodes_are_ignored(self):
         nodes = [
@@ -155,6 +175,14 @@ class TestTransport(TestCase):
         self.assertEquals(1, len(t.connection_pool.connections))
         self.assertEquals('http://1.1.1.1:123', t.get_connection().host)
 
+    def test_sniff_will_pick_up_published_host(self):
+        t = Transport([{'data': CLUSTER_NODE_PUBLISH_HOST}], connection_class=DummyConnection)
+        t.sniff_hosts()
+
+        self.assertEquals(1, len(t.connection_pool.connections))
+        self.assertEquals('http://obsidian:9200', t.get_connection().host)
+
+
     def test_sniff_on_start_fetches_and_uses_nodes_list_for_its_schema(self):
         class DummyThriftConnection(DummyConnection):
             transport_schema = 'thrift'