view mupdf-source/thirdparty/tesseract/.github/workflows/cmake.yml @ 46:7ee69f120f19 default tip

>>>>> tag v1.26.5+1 for changeset b74429b0f5c4
author Franz Glasner <fzglas.hg@dom66.de>
date Sat, 11 Oct 2025 17:17:30 +0200
parents b50eed0cc0ef
children
line wrap: on
line source

name: cmake
# cmake build of tesseract and training tools on ubuntu and macOS homebrew using Ninja.
# test command line version of tesseract. run basicapitest.
on:
  #push:
  schedule:
    - cron: 0 21 * * *

jobs:
  basictests:
    name: ${{ matrix.config.name }}
    runs-on: ${{ matrix.config.os }}
    strategy:
      fail-fast: false
      matrix:
        config:
          - { name: macos-14-clang-15-cmake, os: macos-14, cxx: clang++ } # default
          - { name: macos-14-gcc-14-cmake, os: macos-14, cxx: g++-14 } #installed
          - { name: macos-15-clang-cmake, os: macos-15, cxx: clang++ } # default

          - { name: ubuntu-22.04-clang-15-cmake, os: ubuntu-22.04, cxx: clang++-15 } #installed

          - { name: ubuntu-22.04-gcc-12-cmake, os: ubuntu-22.04, cxx: g++-12 } #installed
          - { name: ubuntu-22.04-gcc-11-cmake, os: ubuntu-22.04, cxx: g++-11 } #installed
          - { name: ubuntu-20.04-gcc-10-cmake, os: ubuntu-20.04, cxx: g++-10 } #installed
          - { name: ubuntu-20.04-gcc-9-cmake, os: ubuntu-20.04, cxx: g++-9 } #installed

    steps:
      - name: Install compilers on Linux
        run: |
             sudo apt-get update
             sudo apt-get install ${{ matrix.config.cxx }} -y
        if: runner.os == 'Linux'

      - name: Install dependencies on Linux
        run: |
           sudo apt-get install autoconf libleptonica-dev -y
           sudo apt-get install libarchive-dev libcurl4-openssl-dev -y
           sudo apt-get install libpango1.0-dev -y
           sudo apt-get install cabextract -y
           sudo apt-get install ninja-build -y
           cmake --version
        if: runner.os == 'Linux'

      - name: Install dependencies on macOS
        run: |
           brew install autoconf automake
           brew install leptonica
           # brew install libarchive
           brew install pango
           brew install icu4c && brew link icu4c
           brew install cabextract
           brew install ninja
           ninja --version
           cmake --version
           clang++ --version
           g++ --version
        if: runner.os == 'macOS'

      - name: Checkout Source
        uses: actions/checkout@v4
        with:
             submodules: recursive

      - name: Configure Tesseract (Linux)
        run: |
             mkdir build
             mkdir inst
             cmake \
               -S . \
               -B build \
               -G Ninja \
               -DCMAKE_BUILD_TYPE=Release \
               -DOPENMP_BUILD=OFF \
               -DCMAKE_CXX_COMPILER=${{ matrix.config.cxx }} \
               -DCMAKE_INSTALL_PREFIX:PATH=inst
        if: runner.os == 'Linux'

      - name: Configure Tesseract (macOS)
        shell: bash
        run: |
             set -e
             export PKG_CONFIG_PATH=$(brew --prefix)/opt/icu4c/lib/pkgconfig:$(brew --prefix)/opt/libarchive/lib/pkgconfig:/$(brew --prefix)/opt/libffi/lib/pkgconfig:$PKG_CONFIG_PATH
             export LDFLAGS="-L/usr/local/opt/icu4c/lib"
             export CPPFLAGS="-I/usr/local/opt/icu4c/include"
             mkdir build
             mkdir inst
             cmake \
               -S . \
               -B build \
               -G Ninja \
               -DCMAKE_BUILD_TYPE=Release \
               -DOPENMP_BUILD=OFF \
               -DCMAKE_CXX_COMPILER=${{ matrix.config.cxx }} \
               -DCMAKE_INSTALL_PREFIX:PATH=inst
        if: runner.os == 'macOS'

      - name: Build Tesseract
        run: |
             cmake --build build --config Release --target install

      - name: Display Tesseract Version
        run: |
             build/inst/bin/tesseract -v

      - name: Display Training Tools Version
        run: |
             build/inst/bin/lstmtraining -v
             build/inst/bin/text2image -v

      - name: Download fonts, tessdata and langdata required for tests
        run: |
             git clone https://github.com/egorpugin/tessdata tessdata_unittest
             cp tessdata_unittest/fonts/* test/testing/
             mv tessdata_unittest/* ../

      - name: List languages in different tessdata-dir
        run: |
             build/inst/bin/tesseract  --list-langs --tessdata-dir ../tessdata
             build/inst/bin/tesseract  --list-langs --tessdata-dir ../tessdata_best
             build/inst/bin/tesseract  --list-langs --tessdata-dir ../tessdata_fast

      - name: Run Tesseract on test images in different languages
        run: |
             build/inst/bin/tesseract test/testing/phototest.tif - --oem 1  --tessdata-dir ../tessdata
             build/inst/bin/tesseract test/testing/raaj.tif - -l hin --oem 1   --tessdata-dir ../tessdata
             build/inst/bin/tesseract test/testing/viet.tif - -l vie --oem 1   --tessdata-dir ../tessdata
             build/inst/bin/tesseract test/testing/hebrew.png - -l heb --oem 1   --tessdata-dir ../tessdata
             build/inst/bin/tesseract test/testing/eurotext.tif - -l fra --oem 1 --tessdata-dir ../tessdata_best
             build/inst/bin/tesseract test/testing/arabic.tif - -l ara --oem 1 --psm 6  --tessdata-dir ../tessdata

      - name: Build and run basicapitest (Linux)
        run: |
             export "PKG_CONFIG_PATH=$GITHUB_WORKSPACE/build/inst/lib/pkgconfig/:$PKG_CONFIG_PATH"
             cd test
             ${{ matrix.config.cxx }} -o basicapitest testing/basicapitest.cpp "-I$GITHUB_WORKSPACE/build/inst/include" "-L$GITHUB_WORKSPACE/build/inst/lib" $(pkg-config --cflags --libs tesseract lept libarchive libcurl) -pthread -std=c++17
             ./basicapitest
        if: runner.os == 'Linux'

      - name: Build and run basicapitest (macOS)
        run: |
             export "PKG_CONFIG_PATH=$GITHUB_WORKSPACE/build/inst/lib/pkgconfig/:$(brew --prefix)/opt/libarchive/lib/pkgconfig:$(brew --prefix)/Library/Homebrew/os/mac/pkgconfig/11:$PKG_CONFIG_PATH"
             cd test
             ${{ matrix.config.cxx }} -o basicapitest testing/basicapitest.cpp "-I$GITHUB_WORKSPACE/build/inst/include" "-L$GITHUB_WORKSPACE/build/inst/lib" $(pkg-config --cflags --libs tesseract lept libcurl) -pthread -std=c++17
             ./basicapitest
        if: runner.os == 'macOS'

      - name: Display Compiler Version
        run: |
             ${{ matrix.config.cxx }} --version
             pwd
             ls -la
             # git log -3 --pretty=format:'%h %ad %s | %an'
        if: always()