Merge pull request #892 from mlco2/copilot/fix-887

Remove Hatch completely and replace with UV + setuptools

Author: benoit-cty

.conda/recipe.yaml

@@ -25,7 +25,8 @@ requirements:
   host:
     - python
     - pip
-    - hatchling
+    - setuptools
+    - wheel
   run:
     # Will be replaced by the dependencies from the pyproject.toml file
     - dependencies

.github/copilot-instructions.md

@@ -41,7 +41,6 @@ Here's what you need to know to navigate and contribute effectively.
 - **docs/** - Sphinx documentation (build with `uv run task docs`)
 - **examples/** - Usage examples and demos
 - **deploy/** - Deployment configurations and scripts
-- **requirements/** - Dependency specifications for different environments
 
 ## Development Setup
 
@@ -149,7 +148,6 @@ uv run task -l
 - **pyproject.toml**: Main project configuration
 - **.codecarbon.config**: Runtime configuration for API connection
 - **docker-compose.yml**: Local development with PostgreSQL
-- **requirements/**: Pinned dependencies for different environments
 
 ## Documentation and Help
 

.github/workflows/package.yml

@@ -20,8 +20,7 @@ jobs:
           run: uv python install 3.12
         - name: Check versions
           run: |
-              uv pip install --system requests
-              python3 .github/check_version.py -o
+              uv run python .github/check_version.py -o
         - name: Build pip package
           run: |
               uv build

.github/workflows/python-publish.yml

@@ -23,8 +23,7 @@ jobs:
       run: uv python install 3.12
     - name: Check versions
       run: |
-          uv pip install --system requests
-          python3 .github/check_version.py
+          uv run python .github/check_version.py
     - name: Build package
       run: uv build
     - name: Publish package

CONTRIBUTING.md

@@ -144,15 +144,9 @@ To test CodeCarbon, it is useful to stress your computer to make it use its full
 <!-- TOC --><a name="update-all-dependancies"></a>
 ### Update all dependancies
 
-```sh
-uv pip compile requirements.in -o requirements.txt
-```
-
 For multiple requirement files:
-
 ```sh
-uv pip compile requirements/dev.in -o requirements/dev.txt
-uv pip compile requirements/test.in -o requirements/test.txt
+uv sync --upgrade
 ```
 
 <!-- TOC --><a name="branching-and-pull-requests"></a>
@@ -256,7 +250,7 @@ You can also run `pre-commit` with `uv run pre-commit run -v` if you have some c
 Dependencies are defined in different places:
 
 -   In [pyproject.toml](pyproject.toml#L28), those are all the dependencies.
--   In [requirements.txt](requirements.txt) and [requirements/](requirements/), those are locked dependencies managed by [Hatch plugin pip-compile](https://github.com/juftin/hatch-pip-compile), do not edit them.
+-   In [uv.lock](uv.lock), those are the locked dependencies managed by UV, do not edit them.
 -   In [.conda/meta.yaml](.conda/meta.yaml#L21), those are the dependencies for the Conda pacakge targeting Python 3.7 and higher versions.
 
 
@@ -268,8 +262,8 @@ To make generating documentation easier, we use [`sphinx` package](https://www.s
 
 In order to make changes, edit the `.rst` files that are in the `/docs/edit` folder, and then run in root folder:
 
-```
-uv run docs
+```sh
+uv run --only-group doc task docs
 ```
 
 to regenerate the html files.
@@ -281,8 +275,6 @@ to regenerate the html files.
 - Create a PR bumping the version with `uv run bumpver update --patch`. For a release candidate, use `uv run bumpver update --set-version 3.0.0_rc1`.
 - Run `uv run python3 .github/check_version.py` to check version consistancy.
 - Update the dependencies with `uv sync --upgrade`
-- Export a requirements.txt `uv pip freeze > requirements.txt`.
-- Keep pre-commit up-to-date with `uv run task precommit-update`.
 - [Build Documentation](#documentation) with `uv run --only-group doc task docs`.
 - Push the changes.
 - Merge the PR.
@@ -299,13 +291,11 @@ Start a Docker image in the same directory and bind-mount the current directory
 
 Inside the docker container, run:
 
--   `conda install -y conda-build conda-verify conda-forge::hatchling`
+-   `conda install -y conda-build conda-verify`
 -   `cd /data && mkdir -p /conda_dist`
 -   `conda build --python 3.11 .conda/ -c conda-forge --output-folder /conda_dist`
 -   `anaconda upload --user codecarbon /conda_dist/noarch/codecarbon-*.tar.bz2`
 
-
-
 <!-- TOC --><a name="api-and-dashboard"></a>
 ## API and Dashboard
 

carbonserver/tests/TESTING.md

@@ -1,8 +1,6 @@
 # Testing the carbon server
 
-
-## Behaviors to test :
-
+## Behaviors to test
 
 ### Domain logic
 To test the interface exposed by entities, in memory repositories can be used to isolate the domain logic from technical
@@ -29,55 +27,34 @@ To test the interface exposed by entities, in memory repositories can be used to
 Domain logic can be tested at a higher level, with more complex setups, in service/use cases tests.
 Use case tests mocks repositories returns to focus on
 
-
-
 ### Routers
 To test a router, validation parameters are ensured by pydantic (available on the swagger documentation),
 and logic is tested by interfaces.
 A Postman collection of requests is available: ```carbonserver/tests/postman/TestCollection.postman_collection.json```.
 
-
-
 ### Integration
 - Database: in the CI, a prod-like database can be used to test features on real data (TODO)
 - Code Carbon package: Launch a train scenario (TODO)
 
-
 ## Running the tests:
 
-### Install the test setup
-
-In a virtual environment, install and build the api package:
 ```bash
-git clone git@github.com:mlco2/codecarbon.git
-git checkout api
-cd carbonserver
-python -m setup install
-python -m setup build   # Verify the build on local environment
-pip install -r requirements-dev.txt # Install test dependencies
+uv run --extra api task test-api-unit # Unit tests on api
+uv run --extra api task test-api-integ # Integration tests
 ```
 
-
-### Run tests
-```bash
-hatch run api:test-unit # Unit tests on api
-hatch run api:test-integ # Integration tests
-```
-
-
 To test the HTTP layer, you can also deploy a local instance:
 
 ```bash
 cd carbonserver/
-uvicorn main:app --reload
+uv run --extra api uvicorn main:app --reload
 ```
 
 Swagger documentation is available at http://localhost:8000/docs
 
 
 ### Run locally the CI
 
-
 To test the full build process, the Github Actions workflow can be executed locally with act ([install available here](https://raw.githubusercontent.com/nektos/act/master/install.sh)):
 ```bash
 # Build patched dockerfile from project root

carbonserver/tests/api/integration/test_api_black_box.py

@@ -3,7 +3,7 @@
 export CODECARBON_API_URL=http://localhost:8008
 export DATABASE_URL=postgresql://codecarbon-user:supersecret@localhost:5480/codecarbon_db
 Then execute the tests:
-hatch run api:test-integ
+uv run task test-api-integ
 """
 
 import os

codecarbon/data/hardware/cpu_dataset_builder/README.md

@@ -4,10 +4,10 @@ To update the CPU database, you have to run:
 
 ```bash
 cd codecarbon/data/hardware/cpu_dataset_builder
-hatch run pip install playwright beautifulsoup4
-hatch run python intel_cpu_scrapper.py
-hatch run python amd_cpu_scrapper.py
-hatch run python merge_scrapped_cpu_power.py
+uv run pip install playwright beautifulsoup4
+uv run python intel_cpu_scrapper.py
+uv run python amd_cpu_scrapper.py
+uv run python merge_scrapped_cpu_power.py
 ```
 
 Then commit the changes to the CSV files.

codecarbon/data/hardware/cpu_dataset_builder/amd_cpu_scrapper.py

@@ -1,8 +1,8 @@
 """
 This script uses Playwright to scrape AMD CPU data from their website and save it as a CSV file.
 
-hatch run pip install playwright
-hatch run python amd_cpu_scrapper.py
+uv run pip install playwright
+uv run python amd_cpu_scrapper.py
 """
 
 import asyncio

codecarbon/data/hardware/cpu_dataset_builder/intel_cpu_scrapper.py

@@ -5,8 +5,8 @@
 This script uses requests and BeautifulSoup to scrape CPU names and their associated TDP values
 from the Intel ARK website using the advanced filter search URL.
 
-hatch run pip install beautifulsoup4
-hatch run python intel_cpu_scrapper.py
+uv run pip install beautifulsoup4
+uv run python intel_cpu_scrapper.py
 
 """