langchain_community is missing a dependency

[stevegrubb]

Checked other resources

• I added a very descriptive title to this issue.
• I searched the LangChain documentation with the integrated search.
• I used the GitHub search to find a similar question and didn't find it.
• I am sure that this is a bug in LangChain rather than my code.
• The bug is not resolved by updating to the latest stable version of LangChain (or the specific integration package).

Example Code

pip show langchain_community | grep Requires

Error Message and Stack Trace (if applicable)

No response

Description

langchain_community has a dependency on unstructured that is not captured. If you look at the document_loaders/unstructured.py file, you will see several imports of unstructured. It is not captured in the dependencies:

$ pip3 show langchain_community | grep Requires
Requires: SQLAlchemy, langsmith, PyYAML, dataclasses-json, tenacity, langchain-core, langchain, numpy, aiohttp, requests

Even though the system info below is RHEL 8 era, the current code shows poetry.lock does not have unstructured listed anywhere. This causes crashes on downloads.

System Info

System Information

OS: Linux
OS Version: #1 SMP PREEMPT_DYNAMIC Sun Nov 17 18:52:19 UTC 2024
Python Version: 3.9.20 (main, Sep 26 2024, 20:59:47)
[GCC 8.5.0 20210514 (Red Hat 8.5.0-22)]

Package Information

langchain_core: 0.2.43
langchain: 0.2.9
langchain_community: 0.2.9
langsmith: 0.1.147
langchain_text_splitters: 0.2.4

Optional packages not installed

langgraph
langserve

Other Dependencies

aiohttp: 3.11.11
async-timeout: 4.0.3
dataclasses-json: 0.6.7
httpx: 0.28.1
jsonpatch: 1.33
langsmith-pyo3: Installed. No version info available.
numpy: 1.26.4
orjson: 3.10.14
packaging: 24.2
pydantic: 2.10.4
PyYAML: 6.0.2
requests: 2.32.3
requests-toolbelt: 1.0.0
SQLAlchemy: 2.0.36
tenacity: 8.5.0
typing-extensions: 4.12.2

[ccurme]

unstructured and almost all other third party integration dependencies are deliberately omitted from the explicit dependencies of langchain-community. They are often not imported until a class is instantiated and we make an effort to raise an informative error message when missing. Due to the number of integrations in langchain-community, it would not be practical to manage all their dependencies simultaneously.

To mitigate this, we recommend new integrations be implemented as separate packages. There is an existing package for langchain-unstructured that explicitly depends on unstructured. This package is recommended over the abstractions in langchain-community.