install.sh

#!/usr/bin/env bash

# bash install.sh --stage 0 --stop_stage 2
# bash install.sh --stage 1 --stop_stage 1

verbose=true;
stage=-1
stop_stage=0


# parse options
while true; do
  [ -z "${1:-}" ] && break;  # break if there are no arguments
  case "$1" in
    --*) name=$(echo "$1" | sed s/^--// | sed s/-/_/g);
      eval '[ -z "${'"$name"'+xxx}" ]' && echo "$0: invalid option $1" 1>&2 && exit 1;
      old_value="(eval echo \\$$name)";
      if [ "${old_value}" == "true" ] || [ "${old_value}" == "false" ]; then
        was_bool=true;
      else
        was_bool=false;
      fi

      # Set the variable to the right value-- the escaped quotes make it work if
      # the option had spaces, like --cmd "queue.pl -sync y"
      eval "${name}=\"$2\"";

      # Check that Boolean-valued arguments are really Boolean.
      if $was_bool && [[ "$2" != "true" && "$2" != "false" ]]; then
        echo "$0: expected \"true\" or \"false\": $1 $2" 1>&2
        exit 1;
      fi
      shift 2;
      ;;

    *) break;
  esac
done

work_dir="$(pwd)"
data_dir="$(pwd)/data"
nltk_data_dir="${data_dir}/nltk_data"
nltk_data_tokenizers_dir="${nltk_data_dir}/tokenizers"
pyltp_models_dir="${data_dir}/pyltp_models"

mkdir -p "${data_dir}"
mkdir -p "${nltk_data_dir}"
mkdir -p "${nltk_data_tokenizers_dir}"
mkdir -p "${pyltp_models_dir}"

if [ ${stage} -le 0 ] && [ ${stop_stage} -ge 0 ]; then
  $verbose && echo "stage 0: download nltk data punkt"
  cd "${nltk_data_tokenizers_dir}" || exit 1;

  # https://www.nltk.org/nltk_data/
  if [ ! -d "punkt" ]; then
  # nltk==3.8.1
    wget -c https://raw.githubusercontent.com/nltk/nltk_data/gh-pages/packages/tokenizers/punkt.zip
    unzip punkt.zip
    rm punkt.zip
  fi

  if [ ! -d "punkt_tab" ]; then
    # nltk==3.8.2
    wget -c https://raw.githubusercontent.com/nltk/nltk_data/gh-pages/packages/tokenizers/punkt_tab.zip
    unzip punkt_tab.zip
    rm punkt_tab.zip
  fi
fi


if [ ${stage} -le 1 ] && [ ${stop_stage} -ge 1 ]; then
  $verbose && echo "stage 0: download pyltp models"
  cd "${pyltp_models_dir}" || exit 1;

  # pyltp
  # https://ltp.ai/download.html
  if [ ! -e "resources.json" ]; then
    wget -c http://model.scir.yunfutech.com/model/ltp_data_v3.4.0.zip
    unzip ltp_data_v3.4.0.zip
    rm ltp_data_v3.4.0.zip
  fi

fi