diff --git a/pkgs/by-name/ow/owocr/package.nix b/pkgs/by-name/ow/owocr/package.nix index b0900bee6478..8071dccc5e44 100644 --- a/pkgs/by-name/ow/owocr/package.nix +++ b/pkgs/by-name/ow/owocr/package.nix @@ -39,7 +39,7 @@ python3Packages.buildPythonApplication { pyjson5 # Google Lens google-cloud-vision manga-ocr - rapidocr-onnxruntime + rapidocr requests # winRT OCR ]; diff --git a/pkgs/development/python-modules/docling-jobkit/default.nix b/pkgs/development/python-modules/docling-jobkit/default.nix index 423c92b124e4..7cfe20ee9478 100644 --- a/pkgs/development/python-modules/docling-jobkit/default.nix +++ b/pkgs/development/python-modules/docling-jobkit/default.nix @@ -19,7 +19,7 @@ # optional dependencies tesserocr, - rapidocr-onnxruntime, + rapidocr, onnxruntime, ray, @@ -68,7 +68,7 @@ buildPythonPackage rec { optional-dependencies = { tesserocr = [ tesserocr ]; rapidocr = [ - rapidocr-onnxruntime + rapidocr onnxruntime ]; ray = [ ray ]; diff --git a/pkgs/development/python-modules/docling-serve/default.nix b/pkgs/development/python-modules/docling-serve/default.nix index b8398621e836..6d1074cfff11 100644 --- a/pkgs/development/python-modules/docling-serve/default.nix +++ b/pkgs/development/python-modules/docling-serve/default.nix @@ -16,7 +16,7 @@ websockets, tesserocr, typer, - rapidocr-onnxruntime, + rapidocr, onnxruntime, torch, torchvision, @@ -81,7 +81,7 @@ buildPythonPackage rec { tesserocr ]; rapidocr = [ - rapidocr-onnxruntime + rapidocr onnxruntime ]; cpu = [ diff --git a/pkgs/development/python-modules/docling/default.nix b/pkgs/development/python-modules/docling/default.nix index aa0d05259557..783bb524f318 100644 --- a/pkgs/development/python-modules/docling/default.nix +++ b/pkgs/development/python-modules/docling/default.nix @@ -30,7 +30,7 @@ pypdfium2, python-docx, python-pptx, - rapidocr-onnxruntime, + rapidocr, requests, rtree, scipy, @@ -90,7 +90,7 @@ buildPythonPackage rec { pypdfium2 python-docx python-pptx - rapidocr-onnxruntime + rapidocr requests rtree scipy @@ -112,7 +112,7 @@ buildPythonPackage rec { ]; rapidocr = [ onnxruntime - rapidocr-onnxruntime + rapidocr ]; tesserocr = [ tesserocr diff --git a/pkgs/development/python-modules/rapidocr-onnxruntime/default.nix b/pkgs/development/python-modules/rapidocr-onnxruntime/default.nix index 97295a4216b2..6e369d6dcc97 100644 --- a/pkgs/development/python-modules/rapidocr-onnxruntime/default.nix +++ b/pkgs/development/python-modules/rapidocr-onnxruntime/default.nix @@ -7,10 +7,8 @@ replaceVars, setuptools, - colorlog, pyclipper, opencv-python, - omegaconf, numpy, six, shapely, @@ -23,13 +21,13 @@ requests, }: let - version = "3.3.1"; + version = "1.4.4"; src = fetchFromGitHub { owner = "RapidAI"; repo = "RapidOCR"; tag = "v${version}"; - hash = "sha256-EgVBMQX+E8ejUd/6FUQ+uJoWjrQSVznpPcc2gA2wAOE="; + hash = "sha256-x0VELDKOffxbV3v0aDFJFuDC4YfsGM548XWgINmRc3M="; }; models = @@ -64,51 +62,62 @@ buildPythonPackage { ]; postPatch = '' - mkdir -p rapidocr/models + mv setup_onnxruntime.py setup.py - ln -s ${models}/* rapidocr/models + ln -s ${models}/* rapidocr_onnxruntime/models - # Magic patch from upstream - what does this even do?? - echo "from .rapidocr.main import RapidOCR, VisRes" > __init__.py + echo "from .rapidocr_onnxruntime.main import RapidOCR, VisRes" > __init__.py ''; - # Upstream expects the source files to be under rapidocr/rapidocr - # instead of rapidocr for the wheel to build correctly. + # Upstream expects the source files to be under rapidocr_onnxruntime/rapidocr_onnxruntime + # instead of rapidocr_onnxruntime for the wheel to build correctly. preBuild = '' - mkdir rapidocr_t - mv rapidocr rapidocr_t - mv rapidocr_t rapidocr + mkdir rapidocr_onnxruntime_t + mv rapidocr_onnxruntime rapidocr_onnxruntime_t + mv rapidocr_onnxruntime_t rapidocr_onnxruntime ''; # Revert the above hack postBuild = '' - mv rapidocr rapidocr_t - mv rapidocr_t/* . + mv rapidocr_onnxruntime rapidocr_onnxruntime_t + mv rapidocr_onnxruntime_t/* . ''; build-system = [ setuptools ]; dependencies = [ - colorlog - numpy - omegaconf - onnxruntime - opencv-python - pillow pyclipper - pyyaml - requests - shapely + opencv-python + numpy six + shapely + pyyaml + pillow + onnxruntime tqdm ]; - pythonImportsCheck = [ "rapidocr" ]; + pythonImportsCheck = [ "rapidocr_onnxruntime" ]; - # As of version 2.1.0, 61 out of 70 tests require internet access. - # It's just not plausible to manually pick out ones that actually work - # in a hermetic build environment anymore :( - doCheck = false; + nativeCheckInputs = [ + pytestCheckHook + requests + ]; + + # These are tests for different backends. + disabledTestPaths = [ + "tests/test_vino.py" + "tests/test_paddle.py" + ]; + + disabledTests = [ + # Needs Internet access + "test_long_img" + ]; + + # rapidocr-onnxruntime has been renamed to rapidocr by upstream since 2.0.0. However, some packages like open-webui still requires rapidocr-onnxruntime 1.4.4. Therefore we set no auto update here. + # nixpkgs-update: no auto update + passthru.skipBulkUpdate = true; meta = { # This seems to be related to https://github.com/microsoft/onnxruntime/issues/10038 @@ -117,8 +126,8 @@ buildPythonPackage { changelog = "https://github.com/RapidAI/RapidOCR/releases/tag/${src.tag}"; description = "Cross platform OCR Library based on OnnxRuntime"; homepage = "https://github.com/RapidAI/RapidOCR"; - license = with lib.licenses; [ asl20 ]; - maintainers = with lib.maintainers; [ pluiedev ]; - mainProgram = "rapidocr"; + license = lib.licenses.asl20; + maintainers = with lib.maintainers; [ wrvsrx ]; + mainProgram = "rapidocr_onnxruntime"; }; } diff --git a/pkgs/development/python-modules/rapidocr-onnxruntime/setup-py-override-version-checking.patch b/pkgs/development/python-modules/rapidocr-onnxruntime/setup-py-override-version-checking.patch index f940c1c29dc1..3800ded38c6a 100644 --- a/pkgs/development/python-modules/rapidocr-onnxruntime/setup-py-override-version-checking.patch +++ b/pkgs/development/python-modules/rapidocr-onnxruntime/setup-py-override-version-checking.patch @@ -1,25 +1,22 @@ -diff --git a/setup.py b/setup.py -index 16938a4..5167972 100644 ---- a/setup.py -+++ b/setup.py +diff --git i/setup_onnxruntime.py w/setup_onnxruntime.py +index e9572e9..f5f3b32 100644 +--- i/setup_onnxruntime.py ++++ w/setup_onnxruntime.py @@ -6,7 +6,6 @@ from pathlib import Path from typing import List, Union - + import setuptools -from get_pypi_latest_version import GetPyPiLatestVersion - - + + def read_txt(txt_path: Union[Path, str]) -> List[str]: -@@ -25,20 +24,7 @@ def get_readme(): - - - MODULE_NAME = "rapidocr" +@@ -25,17 +24,7 @@ def get_readme(): + + + MODULE_NAME = "rapidocr_onnxruntime" - -obtainer = GetPyPiLatestVersion() --try: -- latest_version = obtainer(MODULE_NAME) --except Exception as e: -- latest_version = "0.0.0" +-latest_version = obtainer(MODULE_NAME) -VERSION_NUM = obtainer.version_add_one(latest_version, add_patch=True) - -if len(sys.argv) > 2: @@ -29,6 +26,6 @@ index 16938a4..5167972 100644 - VERSION_NUM = matched_versions -sys.argv = sys.argv[:2] +VERSION_NUM = "@version@" - + project_urls = { "Documentation": "https://rapidai.github.io/RapidOCRDocs", diff --git a/pkgs/development/python-modules/rapidocr/default.nix b/pkgs/development/python-modules/rapidocr/default.nix new file mode 100644 index 000000000000..85bc9dc958a2 --- /dev/null +++ b/pkgs/development/python-modules/rapidocr/default.nix @@ -0,0 +1,123 @@ +{ + lib, + buildPythonPackage, + fetchFromGitHub, + + fetchzip, + replaceVars, + + setuptools, + colorlog, + pyclipper, + opencv-python, + omegaconf, + numpy, + six, + shapely, + pyyaml, + pillow, + onnxruntime, + tqdm, + + pytestCheckHook, + requests, +}: +let + version = "3.3.1"; + + src = fetchFromGitHub { + owner = "RapidAI"; + repo = "RapidOCR"; + tag = "v${version}"; + hash = "sha256-EgVBMQX+E8ejUd/6FUQ+uJoWjrQSVznpPcc2gA2wAOE="; + }; + + models = + fetchzip { + url = "https://github.com/RapidAI/RapidOCR/releases/download/v1.1.0/required_for_whl_v1.3.0.zip"; + hash = "sha256-j/0nzyvu/HfNTt5EZ+2Phe5dkyPOdQw/OZTz0yS63aA="; + stripRoot = false; + } + + "/required_for_whl_v1.3.0/resources/models"; +in +buildPythonPackage { + pname = "rapidocr"; + inherit version src; + pyproject = true; + + sourceRoot = "${src.name}/python"; + + # HACK: + # Upstream uses a very unconventional structure to organize the packages, and we have to coax the + # existing infrastructure to work with it. + # See https://github.com/RapidAI/RapidOCR/blob/02829ef986bc2a5c4f33e9c45c9267bcf2d07a1d/.github/workflows/gen_whl_to_pypi_rapidocr_ort.yml#L80-L92 + # for the "intended" way of building this package. + + # The setup.py supplied by upstream tries to determine the current version by + # fetching the latest version of the package from PyPI, and then bumping the version number. + # This is not allowed in the Nix build environment as we do not have internet access, + # hence we patch that out and get the version from the build environment directly. + patches = [ + (replaceVars ./setup-py-override-version-checking.patch { + inherit version; + }) + ]; + + postPatch = '' + mkdir -p rapidocr/models + + ln -s ${models}/* rapidocr/models + + echo "from .rapidocr.main import RapidOCR, VisRes" > __init__.py + ''; + + # Upstream expects the source files to be under rapidocr/rapidocr + # instead of rapidocr for the wheel to build correctly. + preBuild = '' + mkdir rapidocr_t + mv rapidocr rapidocr_t + mv rapidocr_t rapidocr + ''; + + # Revert the above hack + postBuild = '' + mv rapidocr rapidocr_t + mv rapidocr_t/* . + ''; + + build-system = [ setuptools ]; + + dependencies = [ + colorlog + numpy + omegaconf + onnxruntime + opencv-python + pillow + pyclipper + pyyaml + requests + shapely + six + tqdm + ]; + + pythonImportsCheck = [ "rapidocr" ]; + + # As of version 2.1.0, 61 out of 70 tests require internet access. + # It's just not plausible to manually pick out ones that actually work + # in a hermetic build environment anymore :( + doCheck = false; + + meta = { + # This seems to be related to https://github.com/microsoft/onnxruntime/issues/10038 + # Also some related issue: https://github.com/NixOS/nixpkgs/pull/319053#issuecomment-2167713362 + badPlatforms = [ "aarch64-linux" ]; + changelog = "https://github.com/RapidAI/RapidOCR/releases/tag/${src.tag}"; + description = "Cross platform OCR Library based on OnnxRuntime"; + homepage = "https://github.com/RapidAI/RapidOCR"; + license = with lib.licenses; [ asl20 ]; + maintainers = with lib.maintainers; [ pluiedev ]; + mainProgram = "rapidocr"; + }; +} diff --git a/pkgs/development/python-modules/rapidocr/setup-py-override-version-checking.patch b/pkgs/development/python-modules/rapidocr/setup-py-override-version-checking.patch new file mode 100644 index 000000000000..f940c1c29dc1 --- /dev/null +++ b/pkgs/development/python-modules/rapidocr/setup-py-override-version-checking.patch @@ -0,0 +1,34 @@ +diff --git a/setup.py b/setup.py +index 16938a4..5167972 100644 +--- a/setup.py ++++ b/setup.py +@@ -6,7 +6,6 @@ from pathlib import Path + from typing import List, Union + + import setuptools +-from get_pypi_latest_version import GetPyPiLatestVersion + + + def read_txt(txt_path: Union[Path, str]) -> List[str]: +@@ -25,20 +24,7 @@ def get_readme(): + + + MODULE_NAME = "rapidocr" +- +-obtainer = GetPyPiLatestVersion() +-try: +- latest_version = obtainer(MODULE_NAME) +-except Exception as e: +- latest_version = "0.0.0" +-VERSION_NUM = obtainer.version_add_one(latest_version, add_patch=True) +- +-if len(sys.argv) > 2: +- match_str = " ".join(sys.argv[2:]) +- matched_versions = obtainer.extract_version(match_str) +- if matched_versions: +- VERSION_NUM = matched_versions +-sys.argv = sys.argv[:2] ++VERSION_NUM = "@version@" + + project_urls = { + "Documentation": "https://rapidai.github.io/RapidOCRDocs", diff --git a/pkgs/top-level/python-packages.nix b/pkgs/top-level/python-packages.nix index 23cf263f1d75..dd9d3daaa483 100644 --- a/pkgs/top-level/python-packages.nix +++ b/pkgs/top-level/python-packages.nix @@ -15589,6 +15589,8 @@ self: super: with self; { rapidgzip = callPackage ../development/python-modules/rapidgzip { inherit (pkgs) nasm; }; + rapidocr = callPackage ../development/python-modules/rapidocr { }; + rapidocr-onnxruntime = callPackage ../development/python-modules/rapidocr-onnxruntime { }; rapt-ble = callPackage ../development/python-modules/rapt-ble { };