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<3.12'
    pip.pypa.io: '*'
    freedesktop.org/pkg-config: ~0.29
  working-directory: build
  script: |
    cmake .. -DCMAKE_INSTALL_PREFIX="{{prefix}}" -DCMAKE_BUILD_TYPE=Release
    make --jobs {{ hw.concurrency }} install

    pip install ../python --verbose  --prefix="{{prefix}}"

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<3.12'
  script: |
    wget https://raw.githubusercontent.com/google/sentencepiece/master/data/botchan.txt
    spm_train --input=botchan.txt --model_prefix=m --vocab_size=1000

    python -c 'import sentencepiece as spm; spm.SentencePieceProcessor()'