nltk

#!/usr/bin/env bash

# This script serves as the NLTK build step of the
# [**Python Buildpack**](https://github.com/heroku/heroku-buildpack-python)
# compiler.
#
# A [buildpack](https://devcenter.heroku.com/articles/buildpacks) is an
# adapter between a Python application and Heroku's runtime.
#
# This script is invoked by [`bin/compile`](/).

# Syntax sugar.
# shellcheck source=bin/utils
source "$BIN_DIR/utils"

# Check that nltk was installed by pip, otherwise obviously not needed
if sp-grep -s nltk; then
    puts-step "Downloading NLTK corpora…"

    nltk_packages_definition="$BUILD_DIR/nltk.txt"

    if [ -f "$nltk_packages_definition" ]; then

        readarray -t nltk_packages < "$nltk_packages_definition"
        puts-step "Downloading NLTK packages: ${nltk_packages[*]}"

        python -m nltk.downloader -d "$BUILD_DIR/.heroku/python/nltk_data" "${nltk_packages[@]}" | indent
        set_env NLTK_DATA "/app/.heroku/python/nltk_data"

    else
        puts-warn "'nltk.txt' not found, not downloading any corpora"
        puts-warn "Learn more: https://devcenter.heroku.com/articles/python-nltk"
    fi
fi