pantry/projects/google.com/sentencepiece/package.yml

distributable:
  url: https://github.com/google/sentencepiece/archive/v{{version}}.tar.gz
  strip-components: 1

versions:
  github: google/sentencepiece/releases/tags
  ignore: '1.0.0' # whoops, who tagged this then?

build:
  dependencies:
    cmake.org: ^3
    python.org: ~3.11
    pip.pypa.io: '*'
    freedesktop.org/pkg-config: ~0.29
    protobuf.dev: 25
  working-directory: build
  script:
    - cmake .. $CMAKE_ARGS
    - make --jobs {{ hw.concurrency }} install

    # FIXME: this fails for linux in v0.2.0
    - run: pip install ../python --verbose  --prefix="{{prefix}}"
      if: darwin
  env:
    # pip needs this to find our new package
    PKG_CONFIG_PATH: '$PKG_CONFIG_PATH:{{prefix}}/lib/pkgconfig'
    CMAKE_ARGS:
      - -DCMAKE_INSTALL_PREFIX="{{prefix}}"
      - -DCMAKE_BUILD_TYPE=Release

provides:
  - bin/spm_decode
  - bin/spm_encode
  - bin/spm_export_vocab
  - bin/spm_normalize
  - bin/spm_train

runtime:
  env:
    PYTHONPATH: '{{prefix}}/lib/python3.11/site-packages'

test:
  dependencies:
    gnu.org/wget: '*'
    python.org: ~3.11
  script:
    - wget https://raw.githubusercontent.com/google/sentencepiece/master/data/botchan.txt
    - spm_train --input=botchan.txt --model_prefix=m --vocab_size=1000

    # FIXME: linux build
    - run: python -c 'import sentencepiece as spm; spm.SentencePieceProcessor()'
      if: darwin