Merge pull request #12263 from dlorenc/tess

Add tesseract and leptonica packages.

leptonica.yaml

@@ -0,0 +1,42 @@
+package:
+  name: leptonica
+  version: 1.84.1
+  epoch: 0
+  description: Leptonica is an open source library containing software that is broadly useful for image processing and image analysis applications.
+  copyright:
+    - license: BSD-2-Clause
+
+environment:
+  contents:
+    packages:
+      - autoconf
+      - automake
+      - build-base
+      - busybox
+      - ca-certificates-bundle
+      - cmake
+
+pipeline:
+  - uses: git-checkout
+    with:
+      repository: https://github.com/danbloomberg/leptonica
+      tag: ${{package.version}}
+      expected-commit: 7e803e73511fbd320f01314c141d35d2b8491dde
+
+  - uses: cmake/configure
+
+  - uses: cmake/build
+
+  - uses: cmake/install
+
+  - uses: strip
+
+subpackages:
+  - name: leptonica-dev
+    pipeline:
+      - uses: split/dev
+
+update:
+  enabled: true
+  github:
+    identifier: danbloomberg/leptonica

tesseract.yaml

@@ -0,0 +1,154 @@
+package:
+  name: tesseract
+  version: 5.3.4
+  epoch: 0
+  description: Tesseract Open Source OCR Engine
+  copyright:
+    - license: Apache-2.0
+
+environment:
+  contents:
+    packages:
+      - autoconf
+      - automake
+      - build-base
+      - busybox
+      - ca-certificates-bundle
+      - cairo-dev
+      - cmake
+      - expat-dev
+      - fontconfig-config
+      - fontconfig-dev
+      - fribidi-dev
+      - glib-dev
+      - harfbuzz-dev
+      - icu-dev
+      - leptonica-dev
+      - libfontconfig1
+      - libjpeg-turbo-dev
+      - libxft-dev
+      - opencl-dev
+      - pango
+      - pango-dev
+      - pkgconf
+      - pkgconf-dev
+      - zlib-dev
+
+data:
+  - name: langs
+    items:
+      afr:
+      ara:
+      aze:
+      bel:
+      ben:
+      bul:
+      cat:
+      ces:
+      chi_sim:
+      chi_tra:
+      chr:
+      dan:
+      deu:
+      eng:
+      enm:
+      epo:
+      equ:
+      est:
+      eus:
+      fin:
+      fra:
+      frk:
+      frm:
+      glg:
+      grc:
+      heb:
+      hin:
+      hrv:
+      hun:
+      ind:
+      isl:
+      ita:
+      ita_old:
+      jpn:
+      kan:
+      kat:
+      khm:
+      kor:
+      lav:
+      lit:
+      mal:
+      mkd:
+      mlt:
+      msa:
+      nld:
+      nor:
+      osd:
+      pol:
+      por:
+      ron:
+      rus:
+      slk:
+      slv:
+      spa:
+      spa_old:
+      sqi:
+      srp:
+      swa:
+      swe:
+      tam:
+      tel:
+      tgl:
+      tha:
+      tur:
+      ukr:
+      vie:
+
+vars:
+  tessdata-version: 4.1.0
+
+pipeline:
+  - uses: git-checkout
+    with:
+      repository: https://github.com/tesseract-ocr/tesseract
+      tag: ${{package.version}}
+      expected-commit: 8ee020e14cf5be4e3f0e9beb09b6b050a1871854
+
+  # Training data is stored in a separate repository
+  - uses: git-checkout
+    with:
+      repository: https://github.com/tesseract-ocr/tessdata
+      destination: tessdata-${{vars.tessdata-version}}
+      tag: ${{vars.tessdata-version}}
+      expected-commit: 4767ea922bcc460e70b87b1d303ebdfed0897da8
+
+  - runs: |
+      # They have some hardcoded include paths
+      ln -s /usr/include/pango-1.0 /usr/include/pango
+
+  - uses: cmake/configure
+    with:
+      opts: \ --enable-opencl
+
+  - uses: cmake/build
+
+  - uses: cmake/install
+
+  - uses: strip
+
+subpackages:
+  - name: tesseract-dev
+    pipeline:
+      - uses: split/dev
+
+  - range: langs
+    name: tesseract-${{range.key}}
+    pipeline:
+      - runs: |
+          mkdir -p ${{targets.subpkgdir}}/usr/share/tessdata
+          mv tessdata-${{vars.tessdata-version}}/${{range.key}}.traineddata ${{targets.subpkgdir}}/usr/share/tessdata/
+
+update:
+  enabled: true
+  github:
+    identifier: tesseract-ocr/tesseract