Merge branch 'master' into openai-embeddings-not-respecting-chunk-size

Author: aubford

.github/workflows/codspeed.yml

@@ -0,0 +1,42 @@
+name: CodSpeed
+
+on:
+  push:
+    branches:
+        - master
+  pull_request:
+    paths:
+      - 'libs/core/**'
+  # `workflow_dispatch` allows CodSpeed to trigger backtest
+  # performance analysis in order to generate initial data.
+  workflow_dispatch:
+
+jobs:
+  codspeed:
+    name: Run benchmarks
+    runs-on: codspeed-macro
+    steps:
+      - uses: actions/checkout@v4
+
+      # We have to use 3.12, 3.13 is not yet supported
+      - name: Install uv
+        uses: astral-sh/setup-uv@v5
+        with:
+          python-version: "3.12"
+
+      # Using this action is still necessary for CodSpeed to work
+      - uses: actions/setup-python@v3
+        with:
+          python-version: "3.12"
+
+      - name: install deps
+        run: uv sync --group test
+        working-directory: ./libs/core
+
+      - name: Run benchmarks
+        uses: CodSpeedHQ/action@v3
+        with:
+          token: ${{ secrets.CODSPEED_TOKEN }}
+          run: |
+            cd libs/core
+            uv run --no-sync pytest ./tests/benchmarks --codspeed

.gitignore

@@ -59,6 +59,7 @@ coverage.xml
 *.py,cover
 .hypothesis/
 .pytest_cache/
+.codspeed/
 
 # Translations
 *.mo

README.md

@@ -17,6 +17,7 @@
 [![Open in Dev Containers](https://img.shields.io/static/v1?label=Dev%20Containers&message=Open&color=blue&logo=visualstudiocode&style=flat-square)](https://vscode.dev/redirect?url=vscode://ms-vscode-remote.remote-containers/cloneInVolume?url=https://github.com/langchain-ai/langchain)
 [<img src="https://github.com/codespaces/badge.svg" title="Open in Github Codespace" width="150" height="20">](https://codespaces.new/langchain-ai/langchain)
 [![Twitter](https://img.shields.io/twitter/url/https/twitter.com/langchainai.svg?style=social&label=Follow%20%40LangChainAI)](https://twitter.com/langchainai)
+[![CodSpeed Badge](https://img.shields.io/endpoint?url=https://codspeed.io/badge.json)](https://codspeed.io/langchain-ai/langchain)
 
 > [!NOTE]
 > Looking for the JS/TS library? Check out [LangChain.js](https://github.com/langchain-ai/langchainjs).

docs/docs/concepts/retrieval.mdx

@@ -92,7 +92,7 @@ structured_model = model.with_structured_output(Questions)
 
 # Define the system prompt
 system = """You are a helpful assistant that generates multiple sub-questions related to an input question. \n
-The goal is to break down the input into a set of sub-problems / sub-questions that can be answers in isolation. \n"""
+The goal is to break down the input into a set of sub-problems / sub-questions that can be answered independently. \n"""
 
 # Pass the question to the model
 question = """What are the main components of an LLM-powered autonomous agent system?"""

docs/docs/how_to/code_splitter.ipynb

@@ -40,7 +40,7 @@
     "\n",
     "To view the list of separators for a given language, pass a value from this enum into\n",
     "```python\n",
-    "RecursiveCharacterTextSplitter.get_separators_for_language`\n",
+    "RecursiveCharacterTextSplitter.get_separators_for_language\n",
     "```\n",
     "\n",
     "To instantiate a splitter that is tailored for a specific language, pass a value from the enum into\n",

docs/docs/versions/migrating_chains/map_rerank_docs_chain.ipynb

@@ -13,7 +13,7 @@
     "- Map a process to the set of documents, where the process includes generating a score;\n",
     "- Rank the results by score and return the maximum.\n",
     "\n",
-    "A common process in this scenario is question-answering using pieces of context from a document. Forcing the model to generate score along with its answer helps to select for answers generated only by relevant context.\n",
+    "A common process in this scenario is question-answering using pieces of context from a document. Forcing the model to generate a score along with its answer helps to select for answers generated only by relevant context.\n",
     "\n",
     "An [LangGraph](https://langchain-ai.github.io/langgraph/) implementation allows for the incorporation of [tool calling](/docs/concepts/tool_calling) and other features for this problem. Below we will go through both `MapRerankDocumentsChain` and a corresponding LangGraph implementation on a simple example for illustrative purposes."
    ]

libs/community/langchain_community/retrievers/google_vertex_ai_search.py

@@ -167,6 +167,8 @@ def _convert_website_search_response(
             doc_metadata = document_dict.get("struct_data", {})
             doc_metadata["id"] = document_dict["id"]
             doc_metadata["source"] = derived_struct_data.get("link", "")
+            if derived_struct_data.get("title") is not None:
+                doc_metadata["title"] = derived_struct_data.get("title")
 
             if chunk_type not in derived_struct_data:
                 continue

libs/community/langchain_community/vectorstores/azure_cosmos_db_no_sql.py

@@ -6,6 +6,7 @@
 from typing import TYPE_CHECKING, Any, Dict, Iterable, List, Optional, Tuple
 
 import numpy as np
+from langchain_core._api import deprecated
 from langchain_core.documents import Document
 from langchain_core.embeddings import Embeddings
 from langchain_core.vectorstores import VectorStore
@@ -40,6 +41,11 @@ class CosmosDBQueryType(str, Enum):
     HYBRID = "hybrid"
 
 
+@deprecated(
+    since="0.3.22",
+    removal="1.0",
+    alternative_import="langchain_azure_ai.vectorstores.AzureCosmosDBNoSqlVectorSearch",
+)
 class AzureCosmosDBNoSqlVectorSearch(VectorStore):
     """`Azure Cosmos DB for NoSQL` vector store.
 

libs/core/Makefile

@@ -64,6 +64,9 @@ spell_check:
 spell_fix:
 	uv run --all-groups codespell --toml pyproject.toml -w
 
+benchmark:
+	uv run pytest tests/benchmarks --codspeed
+
 ######################
 # HELP
 ######################

libs/core/langchain_core/_api/beta_decorator.py

@@ -124,7 +124,7 @@ async def awarning_emitting_wrapper(*args: Any, **kwargs: Any) -> Any:
             _name = _name or obj.__qualname__
             old_doc = obj.__doc__
 
-            def finalize(wrapper: Callable[..., Any], new_doc: str) -> T:
+            def finalize(wrapper: Callable[..., Any], new_doc: str) -> T:  # noqa: ARG001
                 """Finalize the annotation of a class."""
                 # Can't set new_doc on some extension objects.
                 with contextlib.suppress(AttributeError):
@@ -190,7 +190,7 @@ def __set_name__(self, owner: Union[type, None], set_name: str) -> None:
                     if _name == "<lambda>":
                         _name = set_name
 
-            def finalize(wrapper: Callable[..., Any], new_doc: str) -> Any:
+            def finalize(wrapper: Callable[..., Any], new_doc: str) -> Any:  # noqa: ARG001
                 """Finalize the property."""
                 return _BetaProperty(
                     fget=obj.fget, fset=obj.fset, fdel=obj.fdel, doc=new_doc

libs/core/langchain_core/_api/deprecation.py

@@ -204,7 +204,7 @@ async def awarning_emitting_wrapper(*args: Any, **kwargs: Any) -> Any:
             _name = _name or obj.__qualname__
             old_doc = obj.__doc__
 
-            def finalize(wrapper: Callable[..., Any], new_doc: str) -> T:
+            def finalize(wrapper: Callable[..., Any], new_doc: str) -> T:  # noqa: ARG001
                 """Finalize the deprecation of a class."""
                 # Can't set new_doc on some extension objects.
                 with contextlib.suppress(AttributeError):
@@ -234,7 +234,7 @@ def warn_if_direct_instance(
                 raise ValueError(msg)
             old_doc = obj.description
 
-            def finalize(wrapper: Callable[..., Any], new_doc: str) -> T:
+            def finalize(wrapper: Callable[..., Any], new_doc: str) -> T:  # noqa: ARG001
                 return cast(
                     "T",
                     FieldInfoV1(
@@ -255,7 +255,7 @@ def finalize(wrapper: Callable[..., Any], new_doc: str) -> T:
                 raise ValueError(msg)
             old_doc = obj.description
 
-            def finalize(wrapper: Callable[..., Any], new_doc: str) -> T:
+            def finalize(wrapper: Callable[..., Any], new_doc: str) -> T:  # noqa: ARG001
                 return cast(
                     "T",
                     FieldInfoV2(
@@ -315,7 +315,7 @@ def __set_name__(self, owner: Union[type, None], set_name: str) -> None:
                     if _name == "<lambda>":
                         _name = set_name
 
-            def finalize(wrapper: Callable[..., Any], new_doc: str) -> T:
+            def finalize(wrapper: Callable[..., Any], new_doc: str) -> T:  # noqa: ARG001
                 """Finalize the property."""
                 return cast(
                     "T",

libs/core/langchain_core/caches.py

@@ -27,6 +27,8 @@
 from collections.abc import Sequence
 from typing import Any, Optional
 
+from typing_extensions import override
+
 from langchain_core.outputs import Generation
 from langchain_core.runnables import run_in_executor
 
@@ -194,6 +196,7 @@ def update(self, prompt: str, llm_string: str, return_val: RETURN_VAL_TYPE) -> N
             del self._cache[next(iter(self._cache))]
         self._cache[(prompt, llm_string)] = return_val
 
+    @override
     def clear(self, **kwargs: Any) -> None:
         """Clear cache."""
         self._cache = {}
@@ -227,6 +230,7 @@ async def aupdate(
         """
         self.update(prompt, llm_string, return_val)
 
+    @override
     async def aclear(self, **kwargs: Any) -> None:
         """Async clear cache."""
         self.clear()

libs/core/langchain_core/callbacks/file.py

@@ -5,6 +5,8 @@
 from pathlib import Path
 from typing import TYPE_CHECKING, Any, Optional, TextIO, cast
 
+from typing_extensions import override
+
 from langchain_core.callbacks import BaseCallbackHandler
 from langchain_core.utils.input import print_text
 
@@ -38,6 +40,7 @@ def __del__(self) -> None:
         """Destructor to cleanup when done."""
         self.file.close()
 
+    @override
     def on_chain_start(
         self, serialized: dict[str, Any], inputs: dict[str, Any], **kwargs: Any
     ) -> None:
@@ -50,17 +53,17 @@ def on_chain_start(
         """
         if "name" in kwargs:
             name = kwargs["name"]
+        elif serialized:
+            name = serialized.get("name", serialized.get("id", ["<unknown>"])[-1])
         else:
-            if serialized:
-                name = serialized.get("name", serialized.get("id", ["<unknown>"])[-1])
-            else:
-                name = "<unknown>"
+            name = "<unknown>"
         print_text(
             f"\n\n\033[1m> Entering new {name} chain...\033[0m",
             end="\n",
             file=self.file,
         )
 
+    @override
     def on_chain_end(self, outputs: dict[str, Any], **kwargs: Any) -> None:
         """Print out that we finished a chain.
 
@@ -70,6 +73,7 @@ def on_chain_end(self, outputs: dict[str, Any], **kwargs: Any) -> None:
         """
         print_text("\n\033[1m> Finished chain.\033[0m", end="\n", file=self.file)
 
+    @override
     def on_agent_action(
         self, action: AgentAction, color: Optional[str] = None, **kwargs: Any
     ) -> Any:
@@ -83,6 +87,7 @@ def on_agent_action(
         """
         print_text(action.log, color=color or self.color, file=self.file)
 
+    @override
     def on_tool_end(
         self,
         output: str,
@@ -109,6 +114,7 @@ def on_tool_end(
         if llm_prefix is not None:
             print_text(f"\n{llm_prefix}", file=self.file)
 
+    @override
     def on_text(
         self, text: str, color: Optional[str] = None, end: str = "", **kwargs: Any
     ) -> None:
@@ -123,6 +129,7 @@ def on_text(
         """
         print_text(text, color=color or self.color, end=end, file=self.file)
 
+    @override
     def on_agent_finish(
         self, finish: AgentFinish, color: Optional[str] = None, **kwargs: Any
     ) -> None:

libs/core/langchain_core/callbacks/manager.py

@@ -22,7 +22,7 @@
 from uuid import UUID
 
 from langsmith.run_helpers import get_tracing_context
-from typing_extensions import Self
+from typing_extensions import Self, override
 
 from langchain_core.callbacks.base import (
     BaseCallbackHandler,
@@ -364,19 +364,16 @@ async def _ahandle_event_for_handler(
             event = getattr(handler, event_name)
             if asyncio.iscoroutinefunction(event):
                 await event(*args, **kwargs)
+            elif handler.run_inline:
+                event(*args, **kwargs)
             else:
-                if handler.run_inline:
-                    event(*args, **kwargs)
-                else:
-                    await asyncio.get_event_loop().run_in_executor(
-                        None,
-                        cast(
-                            "Callable",
-                            functools.partial(
-                                copy_context().run, event, *args, **kwargs
-                            ),
-                        ),
-                    )
+                await asyncio.get_event_loop().run_in_executor(
+                    None,
+                    cast(
+                        "Callable",
+                        functools.partial(copy_context().run, event, *args, **kwargs),
+                    ),
+                )
     except NotImplementedError as e:
         if event_name == "on_chat_model_start":
             message_strings = [get_buffer_string(m) for m in args[1]]
@@ -1401,6 +1398,7 @@ def on_chain_start(
             inheritable_metadata=self.inheritable_metadata,
         )
 
+    @override
     def on_tool_start(
         self,
         serialized: Optional[dict[str, Any]],
@@ -1456,6 +1454,7 @@ def on_tool_start(
             inheritable_metadata=self.inheritable_metadata,
         )
 
+    @override
     def on_retriever_start(
         self,
         serialized: Optional[dict[str, Any]],
@@ -1927,6 +1926,7 @@ async def on_chain_start(
             inheritable_metadata=self.inheritable_metadata,
         )
 
+    @override
     async def on_tool_start(
         self,
         serialized: Optional[dict[str, Any]],
@@ -2017,6 +2017,7 @@ async def on_custom_event(
             metadata=self.metadata,
         )
 
+    @override
     async def on_retriever_start(
         self,
         serialized: Optional[dict[str, Any]],
@@ -2422,12 +2423,11 @@ def _configure(
                     for handler in callback_manager.handlers
                 ):
                     callback_manager.add_handler(var_handler, inheritable)
-            else:
-                if not any(
-                    isinstance(handler, handler_class)
-                    for handler in callback_manager.handlers
-                ):
-                    callback_manager.add_handler(var_handler, inheritable)
+            elif not any(
+                isinstance(handler, handler_class)
+                for handler in callback_manager.handlers
+            ):
+                callback_manager.add_handler(var_handler, inheritable)
     return callback_manager