fastapi0.87.0 - ModuleNotFoundError: No module named 'httpx' while import TestClient from fastapi.testclient

[anoopks12]

First Check

• I added a very descriptive title to this issue.
• I used the GitHub search to find a similar issue and didn't find it.
• I searched the FastAPI documentation, with the integrated search.
• I already searched in Google "How to X in FastAPI" and didn't find any information.
• I already read and followed all the tutorial in the docs and didn't find an answer.
• I already checked if it is not related to FastAPI but to Pydantic.
• I already checked if it is not related to FastAPI but to Swagger UI.
• I already checked if it is not related to FastAPI but to ReDoc.

Commit to Help

• I commit to help with one of those options 👆

Example Code

pip insall fastapi

>>> print(fastapi.__version__)
0.87.0

>>> from fastapi.testclient import TestClient
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
  File "C:\ProgramData\Anaconda3\lib\site-packages\fastapi\testclient.py", line 1, in <module>
    from starlette.testclient import TestClient as TestClient  # noqa
  File "C:\ProgramData\Anaconda3\lib\site-packages\starlette\testclient.py", line 15, in <module>
    import httpx
ModuleNotFoundError: No module named 'httpx'

Description

Getting error ModuleNotFoundError: No module named 'httpx' while import TestClient from fastapi.testclient.
It looks like an issue specific to the version 0.87.0 (which is released on 13-Nov-2022)
Its working in version 0.86.0

Operating System

Windows

Operating System Details

No response

FastAPI Version

0.87.0

Python Version

3.8

Additional Context

No response

[jgould22]

Yes, starlette has changed the TestClient to use HTTPX instead of requests and 0.87.0 upgrades starlette.

https://github.com/tiangolo/fastapi/releases/tag/0.87.0

The docs have been updated here https://fastapi.tiangolo.com/tutorial/testing/ to reflect the change.

[iudeen]

Just do pip install httpx or pip install fastapi[all] —upgrade

If the solution above works for you, please close the issue. 😊

[tiangolo]

Yep, install HTTPX, that's it 🚀

[mezhaka]

@tiangolo I think there's something wrong with installing httpx--we are pulling a transitive dependency of starlette directly: FastAPI -> Starlette -> httpx. Next time starlette migrates from httpx to httpy, this question will pop again.

Solution:
The httpx is included if one does pip install starlette[full]. Maybe FastAPI can pull starlette[full] if I do pip install fastapi[all]? This would allow starlette to manage its dependencies.

For the conda users: there is a starlette-full conda package that pulls httpx automatically, so just put starlett-full in your environment along with fastapi. It is better to have a first order transitive dependency in you environment FastAPI -> Starlette, rather the second order: FastAPI -> Starlette -> httpx.

@tiangolo I noticed there is no fastapi-all package on Anaconda? Can we create that as well? Then installing fastapi-all can pull starlette-full and this transitive dependency on httpx would be hidden. Who is maintaining current Anaconda package?

[ShantanuKumar]

yeah we use tool like pants which relies on transitive dependencies being explicit and hence it complains that it can't find httpx. So installing manually transitive dependencies isn't great.

[iudeen]

You don't have to manually install if you had used fastapi[all]. Check project.toml for more understanding on optional dependencies.

Also, since TestClient is not needed for "normal" operations, httpx is optional as well.

On a side note, it was clearly mentioned in release notes of both FastAPI and starlette about transitioning from requests to httpx

[pamelafox]

In terms of best practices, does it seem reasonable to put 'fastapi[all]' in requirements-dev.txt but just 'fastapi' in requirements.txt, so that additional libraries needed for testing are only pulled in development mode? That's my new approach.

[pythonweb2]

@pamelafox if you use a package manager like poetry, pdm, or pipenv, you can specify development dependencies so that you can exclude certain dependencies from production instances.