Buckets:
AnonymousSpottedToad/candidate_bucket / codebase /python /ThalesGroup_fred /scripts /patch_docling_rapidocr_openvino.py
| #!/usr/bin/env python3 | |
| """Patch Docling's RapidOCR defaults to support OpenVINO artifact lookup. | |
| Docling 2.78.0 accepts `RapidOcrOptions.backend="openvino"` but its | |
| `RapidOcrModel._default_models` mapping only contains `onnxruntime` and | |
| `torch`. When `artifacts_path` is configured, Docling indexes this mapping | |
| with `"openvino"` and crashes with `KeyError`. | |
| This build-time patch keeps the application code clean by extending the | |
| installed Docling module inside the image. OpenVINO uses the same ONNX model | |
| files as the onnxruntime backend, so mirroring those default paths is enough. | |
| """ | |
| from __future__ import annotations | |
| from importlib.util import find_spec | |
| from pathlib import Path | |
| PATCH_SENTINEL = "# FRED Docker patch: RapidOCR OpenVINO artifact defaults" | |
| PATCH_SNIPPET = f""" | |
| {PATCH_SENTINEL} | |
| if "openvino" not in RapidOcrModel._default_models: | |
| RapidOcrModel._default_models["openvino"] = {{ | |
| key: dict(value) | |
| for key, value in RapidOcrModel._default_models["onnxruntime"].items() | |
| }} | |
| """.lstrip() | |
| def main() -> None: | |
| spec = find_spec("docling.models.stages.ocr.rapid_ocr_model") | |
| if spec is None or spec.origin is None: | |
| raise RuntimeError("Could not locate docling.models.stages.ocr.rapid_ocr_model") | |
| module_path = Path(spec.origin) | |
| content = module_path.read_text(encoding="utf-8") | |
| if PATCH_SENTINEL in content: | |
| print(module_path) | |
| print("Docling RapidOCR OpenVINO patch already present.") | |
| return | |
| module_path.write_text(content.rstrip() + "\n\n" + PATCH_SNIPPET, encoding="utf-8") | |
| print(module_path) | |
| print("Patched Docling RapidOCR OpenVINO defaults.") | |
| if __name__ == "__main__": | |
| main() | |
Xet Storage Details
- Size:
- 1.7 kB
- Xet hash:
- 0456e20ca572df5b0a161154007275f796fb5dcc8425d5d17e6dd55018e091d3
·
Xet efficiently stores files, intelligently splitting them into unique chunks and accelerating uploads and downloads. More info.