Conversation
| The models are automatically generated from the JSON schemas in the UCP | ||
| Specification. | ||
|
|
||
| Clone the UCP main repository in the same folder: |
There was a problem hiding this comment.
Consider adding as a submodule instead.
| [project] | ||
| name = "ucp-sdk" | ||
| version = "0.1.0" | ||
| version = "0.2.0" |
There was a problem hiding this comment.
Should this get in sync with the spec version? 2026.01.23 ?
| # pylint: disable=all | ||
| # pyformat: disable | ||
| """ | ||
|
|
There was a problem hiding this comment.
Why move this to a temporarily generated file? These settings take effect across subdirectories too.
There was a problem hiding this comment.
Is this tool specific to the Python repo, or is is needed for Javascript as well? Should it live in the main ucp repo instead?
| # Run generation using uv | ||
| # We use --use-schema-description to use descriptions from JSON schema as docstrings | ||
| # We use --field-constraints to include validation constraints (regex, min/max, etc.) | ||
| # Note: Formatters removed as they can hang on large schemas |
There was a problem hiding this comment.
Might be better to comment that formatting is done as a post-processing step. We still want to reformat the code to minimize accidental whitespace changes.
|
|
||
| echo "Generating Pydantic models from $SCHEMA_DIR..." | ||
| echo "Preprocessing schemas..." | ||
| python3 preprocess_schemas.py |
|
|
||
| echo "Formatting generated models..." | ||
| uv run ruff format "$OUTPUT_DIR" | ||
| uv run ruff check --fix --config "$OUTPUT_DIR/ruff.toml" "$OUTPUT_DIR" 2>&1 | grep -E "^(All checks passed|Fixed|Found)" || echo "Formatting complete" |
There was a problem hiding this comment.
If we remove --config "$OUTPUT_DIR/ruff.toml" , won't it use the settings from pyproject.toml at the top?
|
|
||
| if __name__ == "__main__": | ||
| script_dir = Path(__file__).parent | ||
| input_schemas = script_dir / "ucp" / "source" |
There was a problem hiding this comment.
Consider taking command line arguments instead.
| --allow-extra-fields | ||
|
|
||
| echo "Formatting generated models..." | ||
| uv run ruff format "$OUTPUT_DIR" |
There was a problem hiding this comment.
$OUTPUT_DIR can be omitted; the whole project will be reformatted.
|
|
||
| input_dir_abs = output_path.parent.resolve() | ||
| for ref_file in ["service.json", "capability.json", "payment_handler.json"]: | ||
| ref_path = input_dir_abs / ref_file |
There was a problem hiding this comment.
Suggesting to add the following condition if this is running on windows
if sys.platform == "win32":
ref_path = f'/{ref_path.as_posix()}'
The reason is the ref_path generated causes several issues when running on a windows
- The file path generated includes backslashes that act as special JSON characters, this causes errors when loading the JSON:
json.decoder.JSONDecodeError: Invalid \escape
file://C:\\Users\\samal\\development\\ucp\\sdk\\python\\temp_schemas\\schemas\\service.json
- The file path prefix requires three
/afterfile:
Standard URI format for local files on Windows requires three slashes to properly handle the drive letter:
WIth this addition the ref_path is formatted correctly for windows and the data generation using datamodel_code_generator runs without errors
3 participants
This pull request introduces major improvements to the model generation workflow, schema preprocessing, and project structure for the UCP Python SDK. The most significant changes include the addition of a schema preprocessing script to resolve compatibility issues, updates to the model generation script to use preprocessed schemas, a refactor of capability and discovery profile models for clarity and separation, and enhanced formatting and linting for generated code.
Schema Preprocessing and Model Generation:
preprocess_schemas.pyscript to preprocess JSON schemas, resolving internal references and removing problematic definitions to ensure compatibility withdatamodel-code-generator. Thegenerate_models.shscript is updated to invoke this preprocessing step and to format/lint generated models using Ruff. [1] [2] [3] [4]Model and Schema Refactoring:
Formatting and Linting Improvements:
pyproject.tomland now generate a dedicatedruff.tomlin the models directory for more lenient linting of auto-generated code. The model generation script now applies formatting and lint fixes automatically. [1] [2] [3] [4]Documentation and Setup Updates:
README.mdto clarify repository cloning instructions and document the need to clone the main UCP repository for schema regeneration. [1] [2]Cleanup and Deprecation:
_internal.pymodels file, as its functionality has been replaced by the new schema-based model structure.