How to add standard tests to an integration

When creating either a custom class for yourself or a new tool to publish in a LangChain integration, it is important to add standard tests to ensure it works as expected. This guide will show you how to add standard tests to a tool, and you can Skip to the test templates for implementing tests for each integration.

Setup

First, let's install 2 dependencies:

• langchain-core will define the interfaces we want to import to define our custom tool.
• langchain-tests will provide the standard tests we want to use. Recommended to pin to the latest version:

note

Because added tests in new versions of langchain-tests can break your CI/CD pipelines, we recommend pinning the version of langchain-tests to avoid unexpected changes.

%pip install -U langchain-core langchain-tests pytest pytest-socket

Let's say we're publishing a package, langchain_parrot_link, that exposes a tool called ParrotMultiplyTool:

langchain_parrot_link/tools.py

from langchain_core.tools import BaseTool


class ParrotMultiplyTool(BaseTool):
    name: str = "ParrotMultiplyTool"
    description: str = (
        "Multiply two numbers like a parrot. Parrots always add "
        "eighty for their matey."
    )

    def _run(self, a: int, b: int) -> int:
        return a * b + 80

And we'll assume you've structured your package the same way as the main LangChain packages:

/
├── langchain_parrot_link/
│   └── tools.py
└── tests/
    ├── unit_tests/
    │   └── test_tools.py
    └── integration_tests/
        └── test_tools.py

Add and configure standard tests

There are 2 namespaces in the langchain-tests package:

• unit tests (langchain_tests.unit_tests): designed to be used to test the tool in isolation and without access to external services
• integration tests (langchain_tests.integration_tests): designed to be used to test the tool with access to external services (in particular, the external service that the tool is designed to interact with).

Both types of tests are implemented as pytest class-based test suites.

By subclassing the base classes for each type of standard test (see below), you get all of the standard tests for that type, and you can override the properties that the test suite uses to configure the tests.

Standard tools tests

Here's how you would configure the standard unit tests for the custom tool, e.g. in tests/test_tools.py:

tests/unit_tests/test_tools.py

from typing import Type

from langchain_parrot_link.tools import ParrotMultiplyTool
from langchain_tests.unit_tests import ToolsUnitTests


class TestParrotMultiplyToolUnit(ToolsUnitTests):
    @property
    def tool_constructor(self) -> Type[ParrotMultiplyTool]:
        return ParrotMultiplyTool

    @property
    def tool_constructor_params(self) -> dict:
        # if your tool constructor instead required initialization arguments like
        # `def __init__(self, some_arg: int):`, you would return those here
        # as a dictionary, e.g.: `return {'some_arg': 42}`
        return {}

    @property
    def tool_invoke_params_example(self) -> dict:
        """
        Returns a dictionary representing the "args" of an example tool call.

        This should NOT be a ToolCall dict - i.e. it should not
        have {"name", "id", "args"} keys.
        """
        return {"a": 2, "b": 3}

tests/integration_tests/test_tools.py

from typing import Type

from langchain_parrot_link.tools import ParrotMultiplyTool
from langchain_tests.integration_tests import ToolsIntegrationTests


class TestParrotMultiplyToolIntegration(ToolsIntegrationTests):
    @property
    def tool_constructor(self) -> Type[ParrotMultiplyTool]:
        return ParrotMultiplyTool

    @property
    def tool_constructor_params(self) -> dict:
        # if your tool constructor instead required initialization arguments like
        # `def __init__(self, some_arg: int):`, you would return those here
        # as a dictionary, e.g.: `return {'some_arg': 42}`
        return {}

    @property
    def tool_invoke_params_example(self) -> dict:
        """
        Returns a dictionary representing the "args" of an example tool call.

        This should NOT be a ToolCall dict - i.e. it should not
        have {"name", "id", "args"} keys.
        """
        return {"a": 2, "b": 3}

and you would run these with the following commands from your project root

# run unit tests without network access
pytest --disable-socket --allow-unix-socket tests/unit_tests

# run integration tests
pytest tests/integration_tests

Standard test templates per component:

Above, we implement the unit and integration standard tests for a tool. Below are the templates for implementing the standard tests for each component:

Chat Models

tests/unit_tests/test_chat_models.py

from typing import Tuple, Type

from langchain_parrot_link.chat_models import ChatParrotLink
from langchain_tests.unit_tests import ChatModelUnitTests


class TestChatParrotLinkUnit(ChatModelUnitTests):
    @property
    def chat_model_class(self) -> Type[ChatParrotLink]:
        return ChatParrotLink

    @property
    def chat_model_params(self) -> dict:
        return {"model": "bird-brain-001", "temperature": 0}

tests/integration_tests/test_chat_models.py

from typing import Type

from langchain_parrot_link.chat_models import ChatParrotLink
from langchain_tests.integration_tests import ChatModelIntegrationTests


class TestChatParrotLinkIntegration(ChatModelIntegrationTests):
    @property
    def chat_model_class(self) -> Type[ChatParrotLink]:
        return ChatParrotLink

    @property
    def chat_model_params(self) -> dict:
        return {"model": "bird-brain-001", "temperature": 0}

Embedding Models

tests/unit_tests/test_embeddings.py

from typing import Tuple, Type

from langchain_parrot_link.embeddings import ParrotLinkEmbeddings
from langchain_tests.unit_tests import EmbeddingsUnitTests


class TestParrotLinkEmbeddingsUnit(EmbeddingsUnitTests):
    @property
    def embeddings_class(self) -> Type[ParrotLinkEmbeddings]:
        return ParrotLinkEmbeddings

    @property
    def embedding_model_params(self) -> dict:
        return {"model": "nest-embed-001", "temperature": 0}

tests/integration_tests/test_embeddings.py

from typing import Type

from langchain_parrot_link.embeddings import ParrotLinkEmbeddings
from langchain_tests.integration_tests import EmbeddingsIntegrationTests


class TestParrotLinkEmbeddingsIntegration(EmbeddingsIntegrationTests):
    @property
    def embeddings_class(self) -> Type[ParrotLinkEmbeddings]:
        return ParrotLinkEmbeddings

    @property
    def embedding_model_params(self) -> dict:
        return {"model": "nest-embed-001", "temperature": 0}

Tools/Toolkits

Note: The standard tests for tools/toolkits are implemented in the example in the main body of this guide too.

tests/unit_tests/test_tools.py

from typing import Type

from langchain_parrot_link.tools import ParrotMultiplyTool
from langchain_tests.unit_tests import ToolsUnitTests


class TestParrotMultiplyToolUnit(ToolsUnitTests):
    @property
    def tool_constructor(self) -> Type[ParrotMultiplyTool]:
        return ParrotMultiplyTool

    @property
    def tool_constructor_params(self) -> dict:
        # if your tool constructor instead required initialization arguments like
        # `def __init__(self, some_arg: int):`, you would return those here
        # as a dictionary, e.g.: `return {'some_arg': 42}`
        return {}

    @property
    def tool_invoke_params_example(self) -> dict:
        """
        Returns a dictionary representing the "args" of an example tool call.

        This should NOT be a ToolCall dict - i.e. it should not
        have {"name", "id", "args"} keys.
        """
        return {"a": 2, "b": 3}

tests/integration_tests/test_tools.py

from typing import Type

from langchain_parrot_link.tools import ParrotMultiplyTool
from langchain_tests.integration_tests import ToolsIntegrationTests


class TestParrotMultiplyToolIntegration(ToolsIntegrationTests):
    @property
    def tool_constructor(self) -> Type[ParrotMultiplyTool]:
        return ParrotMultiplyTool

    @property
    def tool_constructor_params(self) -> dict:
        # if your tool constructor instead required initialization arguments like
        # `def __init__(self, some_arg: int):`, you would return those here
        # as a dictionary, e.g.: `return {'some_arg': 42}`
        return {}

    @property
    def tool_invoke_params_example(self) -> dict:
        """
        Returns a dictionary representing the "args" of an example tool call.

        This should NOT be a ToolCall dict - i.e. it should not
        have {"name", "id", "args"} keys.
        """
        return {"a": 2, "b": 3}

Vector Stores

tests/integration_tests/test_vectorstores_sync.py

from typing import AsyncGenerator, Generator

import pytest
from langchain_core.vectorstores import VectorStore
from langchain_parrot_link.vectorstores import ParrotVectorStore
from langchain_standard_tests.integration_tests.vectorstores import (
    AsyncReadWriteTestSuite,
    ReadWriteTestSuite,
)


class TestSync(ReadWriteTestSuite):
    @pytest.fixture()
    def vectorstore(self) -> Generator[VectorStore, None, None]:  # type: ignore
        """Get an empty vectorstore for unit tests."""
        store = ParrotVectorStore()
        # note: store should be EMPTY at this point
        # if you need to delete data, you may do so here
        try:
            yield store
        finally:
            # cleanup operations, or deleting data
            pass


class TestAsync(AsyncReadWriteTestSuite):
    @pytest.fixture()
    async def vectorstore(self) -> AsyncGenerator[VectorStore, None]:  # type: ignore
        """Get an empty vectorstore for unit tests."""
        store = ParrotVectorStore()
        # note: store should be EMPTY at this point
        # if you need to delete data, you may do so here
        try:
            yield store
        finally:
            # cleanup operations, or deleting data
            pass