Fix bug with container name with spaces (#42608)

tvaron3 · web-flow · commit ae93966d330c · 2025-08-22T13:24:32.000-04:00
* fix bug with container name with spaces

* update changelog

* fix test

* fix tests

* remove unused line
diff --git a/sdk/cosmos/azure-cosmos/CHANGELOG.md b/sdk/cosmos/azure-cosmos/CHANGELOG.md
@@ -10,6 +10,7 @@
 #### Bugs Fixed
 * Improved the resilience of Database Account Read metadata operation against short-lived network issues by increasing number of retries. See [PR 42525](https://github.com/Azure/azure-sdk-for-python/pull/42525).
 * Fixed bug where during health checks read regions were marked as unavailable for write operations. See [PR 42525](https://github.com/Azure/azure-sdk-for-python/pull/42525).
+* Fixed bug where containers named with spaces or special characters using session consistency would fall back to eventual consistency. See [PR 42608](https://github.com/Azure/azure-sdk-for-python/pull/42608)
 * Fixed bug where `excluded_locations` was not being honored for some metadata calls. See [PR 42266](https://github.com/Azure/azure-sdk-for-python/pull/42266).
 
 #### Other Changes
diff --git a/sdk/cosmos/azure-cosmos/azure/cosmos/_base.py b/sdk/cosmos/azure-cosmos/azure/cosmos/_base.py
@@ -31,6 +31,7 @@
 from typing import Dict, Any, List, Mapping, Optional, Sequence, Union, Tuple, TYPE_CHECKING
 
 from urllib.parse import quote as urllib_quote
+from urllib.parse import unquote as urllib_unquote
 from urllib.parse import urlsplit
 from azure.core import MatchConditions
 
@@ -358,6 +359,8 @@ def set_session_token_header(
             # then update from session container
             if headers[http_constants.HttpHeaders.ConsistencyLevel] == documents.ConsistencyLevel.Session and \
                     cosmos_client_connection.session:
+                # urllib_unquote is used to decode the path, as it may contain encoded characters
+                path = urllib_unquote(path)
                 # populate session token from the client's session container
                 session_token = (
                     cosmos_client_connection.session.get_session_token(path,
@@ -387,6 +390,8 @@ async def set_session_token_header_async(
             if headers[http_constants.HttpHeaders.ConsistencyLevel] == documents.ConsistencyLevel.Session and \
                     cosmos_client_connection.session:
                 # populate session token from the client's session container
+                # urllib_unquote is used to decode the path, as it may contain encoded characters
+                path = urllib_unquote(path)
                 session_token = \
                     await cosmos_client_connection.session.get_session_token_async(path,
                                                                 options.get('partitionKey'),
@@ -608,19 +613,21 @@ def GetItemContainerInfo(self_link: str, alt_content_path: str, resource_id: str
 
     self_link = TrimBeginningAndEndingSlashes(self_link) + "/"
 
-    index = IndexOfNth(self_link, "/", 4)
+    end_index = IndexOfNth(self_link, "/", 4)
+    start_index = IndexOfNth(self_link, "/", 3)
 
-    if index != -1:
-        collection_id = self_link[0:index]
+    if start_index != -1 and end_index != -1:
+        # parse only the collection rid from the path as it's unique across databases
+        collection_rid = self_link[start_index + 1:end_index]
 
         if "colls" in self_link:
             # this is a collection request
             index_second_slash = IndexOfNth(alt_content_path, "/", 2)
             if index_second_slash == -1:
                 collection_name = alt_content_path + "/colls/" + urllib_quote(resource_id)
-                return collection_id, collection_name
+                return collection_rid, collection_name
             collection_name = alt_content_path
-            return collection_id, collection_name
+            return collection_rid, collection_name
         raise ValueError(
             "Response Not from Server Partition, self_link: {0}, alt_content_path: {1}, id: {2}".format(
                 self_link, alt_content_path, resource_id
diff --git a/sdk/cosmos/azure-cosmos/tests/test_session.py b/sdk/cosmos/azure-cosmos/tests/test_session.py
@@ -86,10 +86,36 @@ def test_session_token_sm_for_ops(self):
         assert self.created_db.client_connection.last_response_headers.get(HttpHeaders.SessionToken) is not None
         assert self.created_db.client_connection.last_response_headers.get(HttpHeaders.SessionToken) != batch_response_token
 
+    def test_session_token_with_space_in_container_name(self):
+
+        # Session token should not be sent for control plane operations
+        test_container = self.created_db.create_container(
+            "Container with space" + str(uuid.uuid4()),
+            PartitionKey(path="/pk"),
+            raw_response_hook=test_config.no_token_response_hook
+        )
+        try:
+            # Session token should be sent for document read/batch requests only - verify it is not sent for write requests
+            created_document = test_container.create_item(body={'id': '1' + str(uuid.uuid4()), 'pk': 'mypk'},
+                                                                   raw_response_hook=test_config.no_token_response_hook)
+            response_session_token = created_document.get_response_headers().get(HttpHeaders.SessionToken)
+            read_item = test_container.read_item(item=created_document['id'], partition_key='mypk',
+                                                          raw_response_hook=test_config.token_response_hook)
+            query_iterable = test_container.query_items(
+                "SELECT * FROM c WHERE c.id = '" + str(created_document['id']) + "'",
+                partition_key='mypk',
+                raw_response_hook=test_config.token_response_hook)
+            for _ in query_iterable:
+                pass
+
+            assert (read_item.get_response_headers().get(HttpHeaders.SessionToken) ==
+                    response_session_token)
+        finally:
+            self.created_db.delete_container(test_container)
+
     def test_session_token_mwr_for_ops(self):
         # For multiple write regions, all document requests should send out session tokens
         # We will use fault injection to simulate the regions the emulator needs
-        first_region_uri: str = test_config.TestConfig.local_host.replace("localhost", "127.0.0.1")
         custom_transport = FaultInjectionTransport()
 
         # Inject topology transformation that would make Emulator look like a multiple write region account
diff --git a/sdk/cosmos/azure-cosmos/tests/test_session_async.py b/sdk/cosmos/azure-cosmos/tests/test_session_async.py
@@ -10,8 +10,8 @@
 from _fault_injection_transport_async import FaultInjectionTransportAsync
 import azure.cosmos.exceptions as exceptions
 import test_config
-from azure.cosmos.aio import CosmosClient, _retry_utility_async
-from azure.cosmos import DatabaseProxy, PartitionKey
+from azure.cosmos.aio import CosmosClient, _retry_utility_async, DatabaseProxy
+from azure.cosmos import PartitionKey
 from azure.cosmos.http_constants import StatusCodes, SubStatusCodes, HttpHeaders
 from azure.core.pipeline.transport._aiohttp import AioHttpTransportResponse
 from azure.core.rest import HttpRequest, AsyncHttpResponse
@@ -87,10 +87,37 @@ async def test_session_token_swr_for_ops_async(self):
         assert self.created_db.client_connection.last_response_headers.get(HttpHeaders.SessionToken) is not None
         assert self.created_db.client_connection.last_response_headers.get(HttpHeaders.SessionToken) != batch_response_token
 
+    async def test_session_token_with_space_in_container_name_async(self):
+
+        # Session token should not be sent for control plane operations
+        test_container = await self.created_db.create_container(
+            "Container with space" + str(uuid.uuid4()),
+            PartitionKey(path="/pk"),
+            raw_response_hook=test_config.no_token_response_hook
+        )
+        try:
+            # Session token should be sent for document read/batch requests only - verify it is not sent for write requests
+            created_document = await test_container.create_item(body={'id': '1' + str(uuid.uuid4()), 'pk': 'mypk'},
+                                                          raw_response_hook=test_config.no_token_response_hook)
+            response_session_token = created_document.get_response_headers().get(HttpHeaders.SessionToken)
+            read_item = await test_container.read_item(item=created_document['id'], partition_key='mypk',
+                                                 raw_response_hook=test_config.token_response_hook)
+            query_iterable = test_container.query_items(
+                "SELECT * FROM c WHERE c.id = '" + str(created_document['id']) + "'",
+                partition_key='mypk',
+                raw_response_hook=test_config.token_response_hook)
+
+            async for _ in query_iterable:
+                pass
+
+            assert (read_item.get_response_headers().get(HttpHeaders.SessionToken) ==
+                    response_session_token)
+        finally:
+            await self.created_db.delete_container(test_container)
+
     async def test_session_token_mwr_for_ops_async(self):
         # For multiple write regions, all document requests should send out session tokens
         # We will use fault injection to simulate the regions the emulator needs
-        first_region_uri: str = test_config.TestConfig.local_host.replace("localhost", "127.0.0.1")
         custom_transport = FaultInjectionTransportAsync()
 
         # Inject topology transformation that would make Emulator look like a multiple write region account
