rapidocrlite 5.0.0a6__tar.gz

rapidocrlite-5.0.0a6/.github/check_manylinux_tag.sh

@@ -0,0 +1,25 @@
+#!/bin/bash
+# Fail unless every wheel in the directory is a manylinux2014 / manylinux_2_17 wheel.
+set -euo pipefail
+
+dir=${1:-wheelhouse}
+python3 -m pip install -q auditwheel
+mapfile -t wheels < <(find "$dir" -type f -name '*.whl' | sort)
+if [[ ${#wheels[@]} -eq 0 ]]; then
+  echo "no wheels under $dir" >&2
+  exit 1
+fi
+for wheel in "${wheels[@]}"; do
+  base=$(basename "$wheel")
+  if ! echo "$base" | grep -Eq 'manylinux2014|manylinux_2_17'; then
+    echo "platform tag missing from filename: $base" >&2
+    exit 1
+  fi
+  show=$(python3 -m auditwheel show "$wheel")
+  printf '%s\n' "$show"
+  if ! printf '%s\n' "$show" | grep -Eq 'manylinux2014|manylinux_2_17'; then
+    echo "auditwheel show did not report manylinux2014/manylinux_2_17: $base" >&2
+    exit 1
+  fi
+  echo "tag_ok $base"
+done

rapidocrlite-5.0.0a6/.github/workflows/build-wheels.yml

@@ -0,0 +1,398 @@
+name: build wheels
+
+on:
+  push:
+    branches:
+      - main
+    tags:
+      - "v5*"
+  pull_request:
+    branches:
+      - main
+  workflow_dispatch:
+
+concurrency:
+  group: wheels-${{ github.workflow }}-${{ github.ref }}
+  cancel-in-progress: true
+
+permissions:
+  contents: read
+
+jobs:
+  sdist:
+    name: sdist
+    runs-on: ubuntu-latest
+    steps:
+      - uses: actions/checkout@v4
+        with:
+          submodules: recursive
+
+      - uses: actions/setup-python@v5
+        with:
+          python-version: "3.12"
+
+      - name: Build sdist
+        run: |
+          python -m pip install -U build
+          python -m build --sdist
+          (cd dist && sha256sum ./*.tar.gz > SHA256SUMS)
+
+      - uses: actions/upload-artifact@v4
+        with:
+          name: rapidocrlite-sdist
+          path: |
+            dist/*.tar.gz
+            dist/SHA256SUMS
+          if-no-files-found: error
+
+  wheels-x86_64:
+    name: wheels x86_64
+    runs-on: ubuntu-latest
+    steps:
+      - uses: actions/checkout@v4
+        with:
+          submodules: recursive
+
+      - uses: actions/setup-python@v5
+        with:
+          python-version: "3.12"
+
+      - name: Cache tiny models
+        uses: actions/cache@v4
+        with:
+          path: ci-models
+          key: rapidocr-ci-models-tiny-v1
+
+      - name: Download tiny models when the cache is empty
+        run: python python/tests/fetch_ci_models.py
+
+      - name: Build manylinux2014 wheels
+        uses: pypa/cibuildwheel@v2.23.3
+        env:
+          CIBW_ARCHS: x86_64
+
+      - name: Check manylinux2014 / manylinux_2_17 tags
+        run: bash .github/check_manylinux_tag.sh wheelhouse
+
+      - uses: actions/upload-artifact@v4
+        with:
+          name: rapidocrlite-wheels-x86_64
+          path: wheelhouse/*.whl
+          if-no-files-found: error
+
+  wheels-aarch64:
+    name: wheels aarch64
+    runs-on: ubuntu-24.04-arm
+    outputs:
+      started: ${{ steps.mark.outputs.started }}
+    steps:
+      - name: Mark the native ARM runner as available
+        id: mark
+        run: echo "started=true" >> "$GITHUB_OUTPUT"
+
+      - uses: actions/checkout@v4
+        with:
+          submodules: recursive
+
+      - uses: actions/setup-python@v5
+        with:
+          python-version: "3.12"
+
+      - name: Cache tiny models
+        uses: actions/cache@v4
+        with:
+          path: ci-models
+          key: rapidocr-ci-models-tiny-v1
+
+      - name: Download tiny models when the cache is empty
+        run: python python/tests/fetch_ci_models.py
+
+      - name: Build manylinux2014 wheels
+        uses: pypa/cibuildwheel@v2.23.3
+        env:
+          CIBW_ARCHS: aarch64
+
+      - name: Check manylinux2014 / manylinux_2_17 tags
+        run: bash .github/check_manylinux_tag.sh wheelhouse
+
+      - uses: actions/upload-artifact@v4
+        with:
+          name: rapidocrlite-wheels-aarch64
+          path: wheelhouse/*.whl
+          if-no-files-found: error
+
+  wheels-aarch64-qemu:
+    name: wheels aarch64 qemu
+    needs: wheels-aarch64
+    if: >-
+      always() &&
+      needs.wheels-aarch64.result == 'failure' &&
+      needs.wheels-aarch64.outputs.started != 'true'
+    runs-on: ubuntu-latest
+    steps:
+      - uses: actions/checkout@v4
+        with:
+          submodules: recursive
+
+      - uses: actions/setup-python@v5
+        with:
+          python-version: "3.12"
+
+      - name: QEMU for aarch64
+        uses: docker/setup-qemu-action@v3
+        with:
+          platforms: arm64
+
+      - name: Cache tiny models
+        uses: actions/cache@v4
+        with:
+          path: ci-models
+          key: rapidocr-ci-models-tiny-v1
+
+      - name: Download tiny models when the cache is empty
+        run: python python/tests/fetch_ci_models.py
+
+      - name: Build manylinux2014 wheels
+        uses: pypa/cibuildwheel@v2.23.3
+        env:
+          CIBW_ARCHS: aarch64
+
+      - name: Check manylinux2014 / manylinux_2_17 tags
+        run: bash .github/check_manylinux_tag.sh wheelhouse
+
+      - uses: actions/upload-artifact@v4
+        with:
+          name: rapidocrlite-wheels-aarch64
+          path: wheelhouse/*.whl
+          if-no-files-found: error
+
+  wheels-macos-arm64:
+    name: wheels macOS arm64 (Metal)
+    runs-on: macos-14
+    steps:
+      - uses: actions/checkout@v4
+        with:
+          submodules: recursive
+
+      - uses: actions/setup-python@v5
+        with:
+          python-version: "3.12"
+
+      - name: Cache tiny models
+        uses: actions/cache@v4
+        with:
+          path: ci-models
+          key: rapidocr-ci-models-tiny-v1
+
+      - name: Download tiny models when the cache is empty
+        run: python python/tests/fetch_ci_models.py
+
+      - name: Build macOS 14 arm64 wheels with Metal
+        uses: pypa/cibuildwheel@v2.23.3
+        env:
+          CIBW_PLATFORM: macos
+          CIBW_BUILD_VERBOSITY: 1
+          CIBW_ARCHS: arm64
+          CIBW_SKIP: "pp* *-macosx_x86_64 *-macosx_universal2"
+
+      - name: Check macOS 14 arm64 tags
+        run: |
+          ls wheelhouse
+          n=$(ls wheelhouse/*.whl | wc -l)
+          test "$n" -eq 5
+          for w in wheelhouse/*.whl; do
+            case "$(basename "$w")" in
+              *-macosx_14_0_arm64.whl) ;;
+              *) echo "unexpected tag: $w" >&2; exit 1 ;;
+            esac
+          done
+
+      - uses: actions/upload-artifact@v4
+        with:
+          name: rapidocrlite-wheels-macos-arm64
+          path: wheelhouse/*.whl
+          if-no-files-found: error
+
+  wheels-windows-x64:
+    name: wheels Windows x64
+    runs-on: windows-latest
+    steps:
+      - uses: actions/checkout@v4
+        with:
+          submodules: recursive
+
+      - uses: actions/setup-python@v5
+        with:
+          python-version: "3.12"
+
+      - name: Cache tiny models
+        uses: actions/cache@v4
+        with:
+          path: ci-models
+          key: rapidocr-ci-models-tiny-v1
+
+      - name: Download tiny models when the cache is empty
+        run: python python/tests/fetch_ci_models.py
+
+      - name: Build Windows x64 wheels
+        uses: pypa/cibuildwheel@v2.23.3
+        env:
+          CIBW_PLATFORM: windows
+          CIBW_ARCHS: AMD64
+          CIBW_SKIP: "pp* *-win32 *-win_arm64"
+          CIBW_BUILD_VERBOSITY: 1
+
+      - name: Check Windows tags
+        shell: bash
+        run: |
+          ls wheelhouse
+          test "$(ls wheelhouse/*-win_amd64.whl | wc -l)" -eq 5
+
+      # GitHub's Windows runners are usually AVX2-only, so run the AVX-512
+      # kernels under Intel SDE (Sapphire Rapids) against the reference texts.
+      - name: AVX-512 check under Intel SDE
+        shell: bash
+        run: |
+          curl -sSL -o sde.tar.xz https://downloadmirror.intel.com/843185/sde-external-9.48.0-2024-11-25-win.tar.xz
+          mkdir sde && tar -xf sde.tar.xz -C sde --strip-components=1
+          python -m pip install wheelhouse/*cp312-cp312-win_amd64.whl
+          PPOCR_CHECK_ISAS=avx512 PPOCR_CHECK_TEXTS_ONLY=1 PYTHONIOENCODING=utf-8 \
+            ./sde/sde.exe -spr -- python python/tests/ci_wheel_check.py
+
+      - uses: actions/upload-artifact@v4
+        with:
+          name: rapidocrlite-wheels-windows-x64
+          path: wheelhouse/*.whl
+          if-no-files-found: error
+
+  wheels-windows-x86:
+    name: wheels Windows x86 (win32)
+    runs-on: windows-latest
+    steps:
+      - uses: actions/checkout@v4
+        with:
+          submodules: recursive
+
+      - uses: actions/setup-python@v5
+        with:
+          python-version: "3.12"
+
+      - name: Cache tiny models
+        uses: actions/cache@v4
+        with:
+          path: ci-models
+          key: rapidocr-ci-models-tiny-v1
+
+      - name: Download tiny models when the cache is empty
+        run: python python/tests/fetch_ci_models.py
+
+      - name: Build Windows x86 (win32) wheels
+        uses: pypa/cibuildwheel@v2.23.3
+        env:
+          CIBW_PLATFORM: windows
+          CIBW_ARCHS: x86
+          CIBW_SKIP: "pp* *-win_amd64 *-win_arm64"
+          CIBW_BUILD_VERBOSITY: 1
+
+      - name: Check Windows tags
+        shell: bash
+        run: |
+          ls wheelhouse
+          test "$(ls wheelhouse/*-win32.whl | wc -l)" -eq 5
+
+      - uses: actions/upload-artifact@v4
+        with:
+          name: rapidocrlite-wheels-windows-x86
+          path: wheelhouse/*.whl
+          if-no-files-found: error
+
+  dist:
+    name: SHA256SUMS
+    needs: [sdist, wheels-x86_64, wheels-aarch64, wheels-aarch64-qemu, wheels-macos-arm64, wheels-windows-x64, wheels-windows-x86]
+    if: >-
+      always() &&
+      needs.sdist.result == 'success' &&
+      needs.wheels-x86_64.result == 'success' &&
+      needs.wheels-macos-arm64.result == 'success' &&
+      needs.wheels-windows-x64.result == 'success' &&
+      needs.wheels-windows-x86.result == 'success' &&
+      (needs.wheels-aarch64.result == 'success' || needs.wheels-aarch64-qemu.result == 'success')
+    runs-on: ubuntu-latest
+    steps:
+      - uses: actions/download-artifact@v4
+        with:
+          pattern: rapidocrlite-*
+          merge-multiple: true
+          path: dist
+
+      - name: Write SHA256SUMS
+        run: |
+          rm -f dist/SHA256SUMS
+          (cd dist && sha256sum ./*.whl ./*.tar.gz | tee SHA256SUMS)
+
+      - uses: actions/upload-artifact@v4
+        with:
+          name: rapidocrlite-dist
+          path: |
+            dist/*.whl
+            dist/*.tar.gz
+            dist/SHA256SUMS
+          if-no-files-found: error
+
+  publish:
+    name: PyPI
+    needs: dist
+    # always(): dist itself runs after a skipped job (aarch64 qemu), and without
+    # it the skipped ancestor would skip publishing as well.
+    if: >-
+      always() &&
+      needs.dist.result == 'success' &&
+      github.event_name == 'push' &&
+      startsWith(github.ref, 'refs/tags/v5')
+    runs-on: ubuntu-latest
+    environment: pypi
+    env:
+      PYPI_API_TOKEN_V5_NG: ${{ secrets.PYPI_API_TOKEN_V5_NG }}
+    permissions:
+      id-token: write
+      contents: read
+    steps:
+      - uses: actions/download-artifact@v4
+        with:
+          name: rapidocrlite-dist
+          path: dist
+
+      - name: Check tag matches package version
+        env:
+          TAG: ${{ github.ref_name }}
+        run: |
+          v="${TAG#v}"
+          ls dist
+          ls dist/rapidocrlite-"$v".tar.gz
+          bad=$(ls dist/*.whl | grep -v "/rapidocrlite-$v-" || true)
+          if [ -n "$bad" ]; then echo "version mismatch: $bad" >&2; exit 1; fi
+
+      - name: Keep only distributions in the upload directory
+        run: |
+          mkdir -p pypi
+          cp dist/*.whl dist/*.tar.gz pypi/
+          ls pypi
+
+      - name: Publish to PyPI (API token secret)
+        if: env.PYPI_API_TOKEN_V5_NG != ''
+        uses: pypa/gh-action-pypi-publish@release/v1
+        with:
+          packages-dir: pypi
+          password: ${{ secrets.PYPI_API_TOKEN_V5_NG }}
+          skip-existing: true
+          attestations: false
+          verbose: true
+
+      - name: Publish to PyPI (Trusted Publishing)
+        if: env.PYPI_API_TOKEN_V5_NG == ''
+        uses: pypa/gh-action-pypi-publish@release/v1
+        with:
+          packages-dir: pypi
+          skip-existing: true
+          attestations: false
+          verbose: true

rapidocrlite-5.0.0a6/.github/workflows/metal-macos.yml

@@ -0,0 +1,65 @@
+name: metal macos
+
+on:
+  push:
+    branches:
+      - main
+  pull_request:
+    branches:
+      - main
+  workflow_dispatch:
+
+concurrency:
+  group: metal-${{ github.workflow }}-${{ github.ref }}
+  cancel-in-progress: true
+
+permissions:
+  contents: read
+
+jobs:
+  metal:
+    name: macos-14 Metal
+    runs-on: macos-14
+    steps:
+      - uses: actions/checkout@v4
+        with:
+          submodules: recursive
+
+      - uses: actions/setup-python@v5
+        with:
+          python-version: "3.12"
+
+      - name: Install shader tools
+        run: brew install glslang spirv-cross ninja
+
+      - name: Cache tiny models
+        uses: actions/cache@v4
+        with:
+          path: ci-models
+          key: rapidocr-ci-models-tiny-v1
+
+      - name: Download tiny models when the cache is empty
+        run: python python/tests/fetch_ci_models.py
+
+      - name: Build with the Metal backend (no Vulkan loader)
+        run: |
+          cmake -S . -B build-metal -G Ninja \
+            -DCMAKE_BUILD_TYPE=Release \
+            -DPPOCR_BUILD_STATIC=OFF \
+            -DPPOCR_BUILD_TESTS=ON \
+            -DPPOCR_BUILD_EXAMPLES=ON \
+            -DPPOCR_ENABLE_VULKAN=OFF \
+            -DPPOCR_ENABLE_METAL=ON | tee configure.log
+          grep -q "Metal backend enabled" configure.log
+          cmake --build build-metal -j"$(sysctl -n hw.ncpu)"
+
+      - name: C API smoke
+        run: ctest --test-dir build-metal --output-on-failure
+
+      - name: Install Python test dependencies
+        run: |
+          python -m pip install -r python/tests/ci-constraints.txt \
+            pytest PyYAML requests tqdm
+
+      - name: Metal checks
+        run: bash tests/run_metal_macos.sh build-metal

rapidocrlite-5.0.0a6/.github/workflows/vulkan-lavapipe.yml

@@ -0,0 +1,76 @@
+name: vulkan lavapipe
+
+on:
+  push:
+    branches:
+      - main
+  pull_request:
+    branches:
+      - main
+  workflow_dispatch:
+
+concurrency:
+  group: vulkan-${{ github.workflow }}-${{ github.ref }}
+  cancel-in-progress: true
+
+permissions:
+  contents: read
+
+jobs:
+  lavapipe:
+    name: lavapipe
+    runs-on: ubuntu-latest
+    steps:
+      - uses: actions/checkout@v4
+        with:
+          submodules: recursive
+
+      - uses: actions/setup-python@v5
+        with:
+          python-version: "3.12"
+
+      - name: Install the software Vulkan stack
+        run: |
+          sudo apt-get update
+          sudo apt-get install -y --no-install-recommends \
+            mesa-vulkan-drivers vulkan-tools vulkan-validationlayers \
+            glslang-tools ninja-build cmake g++
+
+      - name: Cache tiny models
+        uses: actions/cache@v4
+        with:
+          path: ci-models
+          key: rapidocr-ci-models-tiny-v1
+
+      - name: Download tiny models when the cache is empty
+        run: python python/tests/fetch_ci_models.py
+
+      - name: Build with embedded Vulkan shaders
+        run: |
+          cmake -S . -B build-vk -G Ninja \
+            -DCMAKE_BUILD_TYPE=Release \
+            -DCMAKE_C_COMPILER=gcc -DCMAKE_CXX_COMPILER=g++ \
+            -DPPOCR_BUILD_STATIC=OFF \
+            -DPPOCR_BUILD_TESTS=ON \
+            -DPPOCR_BUILD_EXAMPLES=ON \
+            -DPPOCR_ENABLE_VULKAN=ON
+          cmake --build build-vk -j"$(nproc)"
+
+      - name: Confirm the library does not link libvulkan
+        run: |
+          if objdump -p build-vk/libppocr.so | grep -E 'NEEDED[[:space:]]+libvulkan'; then
+            echo "libppocr.so must dlopen libvulkan, not DT_NEEDED it" >&2
+            exit 1
+          fi
+
+      - name: Install Python test dependencies
+        run: |
+          python -m pip install -r python/tests/ci-constraints.txt \
+            pytest PyYAML requests tqdm
+
+      - name: Lavapipe hello, kernels, and fallback
+        env:
+          VK_ICD_FILENAMES: /usr/share/vulkan/icd.d/lvp_icd.json
+          PPOCR_VULKAN_DEVICE_INDEX: "0"
+          PPOCR_VULKAN_VALIDATION: "1"
+        run: bash tests/run_vulkan_lavapipe.sh build-vk

rapidocrlite-5.0.0a6/.gitignore

@@ -0,0 +1,27 @@
+# Local build products and generated analysis output
+/build*/
+/bench/
+__pycache__/
+*.pyc
+/python/dist/
+/python/*.egg-info/
+/python/src/*.egg-info/
+/python/.pytest_cache/
+/ci-models/
+/wheelhouse/
+/dist/
+/python/src/rapidocr/models/*
+!/python/src/rapidocr/models/.gitkeep
+/**/graphify-out/
+/build-msvc-env.txt
+
+# Nested vendor repository metadata; vendor source is committed as normal files.
+/third_party/Vulkan-Headers/.git/
+
+# Editor and OS files
+.vs/
+.idea/
+*.user
+*.suo
+.DS_Store
+Thumbs.db

rapidocrlite-5.0.0a6/.gitmodules

@@ -0,0 +1,3 @@
+[submodule "third_party/Vulkan-Headers"]
+	path = third_party/Vulkan-Headers
+	url = https://github.com/KhronosGroup/Vulkan-Headers.git