From e7fd8f8ca1b63d7c368397ba36d76c8415c0dc5c Mon Sep 17 00:00:00 2001 From: wrvsrx Date: Tue, 26 Aug 2025 18:15:15 +0800 Subject: [PATCH 1/3] python3Packages.rapidocr: rename `rapidocr-onnxruntime` to `rapidocr` --- pkgs/by-name/op/open-webui/package.nix | 2 +- pkgs/by-name/ow/owocr/package.nix | 2 +- pkgs/development/python-modules/docling-jobkit/default.nix | 4 ++-- pkgs/development/python-modules/docling-serve/default.nix | 4 ++-- pkgs/development/python-modules/docling/default.nix | 6 +++--- .../{rapidocr-onnxruntime => rapidocr}/default.nix | 3 +-- .../setup-py-override-version-checking.patch | 0 pkgs/top-level/python-packages.nix | 2 +- 8 files changed, 11 insertions(+), 12 deletions(-) rename pkgs/development/python-modules/{rapidocr-onnxruntime => rapidocr}/default.nix (97%) rename pkgs/development/python-modules/{rapidocr-onnxruntime => rapidocr}/setup-py-override-version-checking.patch (100%) diff --git a/pkgs/by-name/op/open-webui/package.nix b/pkgs/by-name/op/open-webui/package.nix index 0ac600275fd2..5b7407472592 100644 --- a/pkgs/by-name/op/open-webui/package.nix +++ b/pkgs/by-name/op/open-webui/package.nix @@ -192,7 +192,7 @@ python3Packages.buildPythonApplication rec { pyxlsb qdrant-client rank-bm25 - rapidocr-onnxruntime + rapidocr redis requests restrictedpython 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/default.nix similarity index 97% rename from pkgs/development/python-modules/rapidocr-onnxruntime/default.nix rename to pkgs/development/python-modules/rapidocr/default.nix index 97295a4216b2..85bc9dc958a2 100644 --- a/pkgs/development/python-modules/rapidocr-onnxruntime/default.nix +++ b/pkgs/development/python-modules/rapidocr/default.nix @@ -41,7 +41,7 @@ let + "/required_for_whl_v1.3.0/resources/models"; in buildPythonPackage { - pname = "rapidocr-onnxruntime"; + pname = "rapidocr"; inherit version src; pyproject = true; @@ -68,7 +68,6 @@ buildPythonPackage { ln -s ${models}/* rapidocr/models - # Magic patch from upstream - what does this even do?? echo "from .rapidocr.main import RapidOCR, VisRes" > __init__.py ''; diff --git a/pkgs/development/python-modules/rapidocr-onnxruntime/setup-py-override-version-checking.patch b/pkgs/development/python-modules/rapidocr/setup-py-override-version-checking.patch similarity index 100% rename from pkgs/development/python-modules/rapidocr-onnxruntime/setup-py-override-version-checking.patch rename to pkgs/development/python-modules/rapidocr/setup-py-override-version-checking.patch diff --git a/pkgs/top-level/python-packages.nix b/pkgs/top-level/python-packages.nix index e49a820de7b6..73ca850bbe65 100644 --- a/pkgs/top-level/python-packages.nix +++ b/pkgs/top-level/python-packages.nix @@ -15553,7 +15553,7 @@ self: super: with self; { rapidgzip = callPackage ../development/python-modules/rapidgzip { inherit (pkgs) nasm; }; - rapidocr-onnxruntime = callPackage ../development/python-modules/rapidocr-onnxruntime { }; + rapidocr = callPackage ../development/python-modules/rapidocr { }; rapt-ble = callPackage ../development/python-modules/rapt-ble { }; From 3f31d64ddf68deedd17ea1d6698f958aa2a61020 Mon Sep 17 00:00:00 2001 From: wrvsrx Date: Wed, 3 Sep 2025 00:02:21 +0800 Subject: [PATCH 2/3] python3Packages.rapidocr-onnxruntime: init at 1.4.4 --- .../rapidocr-onnxruntime/default.nix | 133 ++++++++++++++++++ .../setup-py-override-version-checking.patch | 31 ++++ pkgs/top-level/python-packages.nix | 2 + 3 files changed, 166 insertions(+) create mode 100644 pkgs/development/python-modules/rapidocr-onnxruntime/default.nix create mode 100644 pkgs/development/python-modules/rapidocr-onnxruntime/setup-py-override-version-checking.patch diff --git a/pkgs/development/python-modules/rapidocr-onnxruntime/default.nix b/pkgs/development/python-modules/rapidocr-onnxruntime/default.nix new file mode 100644 index 000000000000..6e369d6dcc97 --- /dev/null +++ b/pkgs/development/python-modules/rapidocr-onnxruntime/default.nix @@ -0,0 +1,133 @@ +{ + lib, + buildPythonPackage, + fetchFromGitHub, + + fetchzip, + replaceVars, + + setuptools, + pyclipper, + opencv-python, + numpy, + six, + shapely, + pyyaml, + pillow, + onnxruntime, + tqdm, + + pytestCheckHook, + requests, +}: +let + version = "1.4.4"; + + src = fetchFromGitHub { + owner = "RapidAI"; + repo = "RapidOCR"; + tag = "v${version}"; + hash = "sha256-x0VELDKOffxbV3v0aDFJFuDC4YfsGM548XWgINmRc3M="; + }; + + 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-onnxruntime"; + 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 = '' + mv setup_onnxruntime.py setup.py + + ln -s ${models}/* rapidocr_onnxruntime/models + + echo "from .rapidocr_onnxruntime.main import RapidOCR, VisRes" > __init__.py + ''; + + # Upstream expects the source files to be under rapidocr_onnxruntime/rapidocr_onnxruntime + # instead of rapidocr_onnxruntime for the wheel to build correctly. + preBuild = '' + mkdir rapidocr_onnxruntime_t + mv rapidocr_onnxruntime rapidocr_onnxruntime_t + mv rapidocr_onnxruntime_t rapidocr_onnxruntime + ''; + + # Revert the above hack + postBuild = '' + mv rapidocr_onnxruntime rapidocr_onnxruntime_t + mv rapidocr_onnxruntime_t/* . + ''; + + build-system = [ setuptools ]; + + dependencies = [ + pyclipper + opencv-python + numpy + six + shapely + pyyaml + pillow + onnxruntime + tqdm + ]; + + pythonImportsCheck = [ "rapidocr_onnxruntime" ]; + + nativeCheckInputs = [ + pytestCheckHook + requests + ]; + + # These are tests for different backends. + disabledTestPaths = [ + "tests/test_vino.py" + "tests/test_paddle.py" + ]; + + disabledTests = [ + # Needs Internet access + "test_long_img" + ]; + + # rapidocr-onnxruntime has been renamed to rapidocr by upstream since 2.0.0. However, some packages like open-webui still requires rapidocr-onnxruntime 1.4.4. Therefore we set no auto update here. + # nixpkgs-update: no auto update + passthru.skipBulkUpdate = true; + + meta = { + # This seems to be related to https://github.com/microsoft/onnxruntime/issues/10038 + # Also some related issue: https://github.com/NixOS/nixpkgs/pull/319053#issuecomment-2167713362 + badPlatforms = [ "aarch64-linux" ]; + changelog = "https://github.com/RapidAI/RapidOCR/releases/tag/${src.tag}"; + description = "Cross platform OCR Library based on OnnxRuntime"; + homepage = "https://github.com/RapidAI/RapidOCR"; + license = 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 new file mode 100644 index 000000000000..3800ded38c6a --- /dev/null +++ b/pkgs/development/python-modules/rapidocr-onnxruntime/setup-py-override-version-checking.patch @@ -0,0 +1,31 @@ +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]: +@@ -25,17 +24,7 @@ def get_readme(): + + + MODULE_NAME = "rapidocr_onnxruntime" +- +-obtainer = GetPyPiLatestVersion() +-latest_version = obtainer(MODULE_NAME) +-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 73ca850bbe65..7a128d6be7a0 100644 --- a/pkgs/top-level/python-packages.nix +++ b/pkgs/top-level/python-packages.nix @@ -15555,6 +15555,8 @@ self: super: with self; { rapidocr = callPackage ../development/python-modules/rapidocr { }; + rapidocr-onnxruntime = callPackage ../development/python-modules/rapidocr-onnxruntime { }; + rapt-ble = callPackage ../development/python-modules/rapt-ble { }; rarfile = callPackage ../development/python-modules/rarfile { inherit (pkgs) libarchive; }; From 898a1df732414f7d293a207459c8e47679b33364 Mon Sep 17 00:00:00 2001 From: wrvsrx Date: Tue, 26 Aug 2025 18:16:29 +0800 Subject: [PATCH 3/3] open-webui: fix build --- pkgs/by-name/op/open-webui/package.nix | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/pkgs/by-name/op/open-webui/package.nix b/pkgs/by-name/op/open-webui/package.nix index 5b7407472592..0ac600275fd2 100644 --- a/pkgs/by-name/op/open-webui/package.nix +++ b/pkgs/by-name/op/open-webui/package.nix @@ -192,7 +192,7 @@ python3Packages.buildPythonApplication rec { pyxlsb qdrant-client rank-bm25 - rapidocr + rapidocr-onnxruntime redis requests restrictedpython