fixV2(core): Resolve PydanticSerializationError for Google FunctionCall

llama-index-core/llama_index/core/base/llms/types.py

@@ -549,16 +549,12 @@ def from_str(
 
     def _recursive_serialization(self, value: Any) -> Any:
         if isinstance(value, BaseModel):
-            value.model_rebuild()  # ensures all fields are initialized and serializable
-            return value.model_dump()  # type: ignore
+            value.model_rebuild()
+            return value.model_dump()
         if isinstance(value, dict):
-            return {
-                key: self._recursive_serialization(value)
-                for key, value in value.items()
-            }
+            return {k: self._recursive_serialization(v) for k, v in value.items()}
         if isinstance(value, list):
             return [self._recursive_serialization(item) for item in value]
-
         if isinstance(value, bytes):
             return base64.b64encode(value).decode("utf-8")
 

llama-index-integrations/graph_rag/llama-index-graph-rag-cognee/pyproject.toml

@@ -34,7 +34,7 @@ requires-python = ">=3.11,<=3.13"
 readme = "README.md"
 license = "MIT"
 dependencies = [
-    "cognee[neo4j, postgres, qdrant]",
+    "cognee[neo4j, postgres, qdrant]==0.1.26",
     "httpx~=0.27.0",
     "graphistry",
     "llama-index-core>=0.13.0,<0.15",

llama-index-integrations/graph_stores/llama-index-graph-stores-kuzu/llama_index/graph_stores/kuzu/kuzu_property_graph.py

@@ -182,30 +182,29 @@ def _create_vector_index(self, table_name: str) -> None:
         if not self.use_vector_index or table_name != "Chunk":
             return
 
-        # Check if chunk_embedding_index already exists
         existing_indexes_result = self.connection.execute(
             "CALL SHOW_INDEXES() RETURN *"
         )
         for row in existing_indexes_result:
-            if len(row) > 1 and row[1] == "chunk_embedding_index":
+            if "embedding" in str(row):
                 return
 
-        # Check if table has any data - Kuzu requires data before creating vector index
         count_result = self.connection.execute(
             f"MATCH (n:{table_name}) RETURN COUNT(n)"
         )
         if not any(int(row[0]) > 0 for row in count_result):
             return
 
-        # Create vector index for Chunk table
         self.connection.execute("""
-        CALL CREATE_VECTOR_INDEX(
-            'Chunk',
-            'chunk_embedding_index',
-            'embedding',
-            metric := 'cosine'
-        )
-        """)
+            CALL CREATE_VECTOR_INDEX(
+                'Chunk',
+                'embedding',
+                USING {
+                    metric_type: 'COSINE',
+                    index_type: 'HNSW'
+                }
+            )
+            """)
 
     def _ensure_vector_indexes(self) -> None:
         """Ensure vector indexes are created for Chunk table only."""

llama-index-integrations/memory/llama-index-memory-mem0/llama_index/memory/mem0/base.py

@@ -47,15 +47,13 @@ def search(self, query: str, **kwargs) -> Optional[Dict[str, Any]]:
 
 
 class Mem0Context(BaseModel):
-    user_id: Optional[str] = None
-    agent_id: Optional[str] = None
-    run_id: Optional[str] = None
+    user_id: Optional[str] = Field(default=None, description="User ID")
+    agent_id: Optional[str] = Field(default=None, description="Agent ID")
+    run_id: Optional[str] = Field(default=None, description="Run ID")
 
-    @model_validator(mode="after")
+    @model_validator(mode="before")
     def check_at_least_one_assigned(cls, values):
-        if not any(
-            getattr(values, field) for field in ["user_id", "agent_id", "run_id"]
-        ):
+        if not any(values.get(field) for field in ["user_id", "agent_id", "run_id"]):
             raise ValueError(
                 "At least one of 'user_id', 'agent_id', or 'run_id' must be assigned."
             )
@@ -142,10 +140,7 @@ def from_config(
     ):
         primary_memory = LlamaIndexMemory.from_defaults()
 
-        try:
-            context = Mem0Context(**context)
-        except Exception as e:
-            raise ValidationError(f"Context validation error: {e}")
+        context = Mem0Context(**context)
 
         client = Memory.from_config(config_dict=config)
         return cls(

llama-index-integrations/readers/llama-index-readers-web/tests/test_readers_oxylabs.py

@@ -1,5 +1,4 @@
-from unittest.mock import MagicMock, AsyncMock
-
+from unittest.mock import MagicMock, AsyncMock, patch
 import pytest
 import sys
 
@@ -30,12 +29,15 @@
     ("urls", "additional_params", "return_value", "expected_output"),
     [READER_TEST_PARAM],
 )
+@patch("llama_index.readers.web.oxylabs_web.base.version")
 def test_sync_oxylabs_reader(
+    mock_version: MagicMock,
     urls: list[str],
     additional_params: dict,
     return_value: dict,
     expected_output: str,
 ):
+    mock_version.return_value = "0.1.0"
     reader = OxylabsWebReader(
         username="OXYLABS_USERNAME",
         password="OXYLABS_PASSWORD",
@@ -57,12 +59,15 @@ def test_sync_oxylabs_reader(
     [READER_TEST_PARAM],
 )
 @pytest.mark.asyncio
+@patch("llama_index.readers.web.oxylabs_web.base.version")
 async def test_async_oxylabs_reader(
+    mock_version: MagicMock,
     urls: list[str],
     additional_params: dict,
     return_value: dict,
     expected_output: str,
 ):
+    mock_version.return_value = "0.1.0"
     reader = OxylabsWebReader(
         username="OXYLABS_USERNAME",
         password="OXYLABS_PASSWORD",