#!/bin/bash

base_dir=$(readlink -f "$(dirname "$0")")

if [ $# -ne 3 ]; then
  >&2 printf 'usage:  %s wiki-depth markov-depth word-count\n' "${0##*/}"
  exit 1
fi

depth=$1
shift

if [ ! -s "${base_dir}/.words.${depth}" ]; then

  urls=()

  for i in $(seq "${depth}"); do
    urls=(
      $(
        "${base_dir}/dive-into-wikipedia.sh" "${urls[@]}" \
          | sort -u
      )
    )
    echo $i >&2
  done

  printf '%s\n' "${urls[@]}" \
    | parallel -j0 -n1 curl -x 'socks5://127.0.0.1:9050' -s \
    | "${base_dir}/remove-tags.sh" \
    | tr -C 'a-zA-ZäöüÄÖÜß' '\n' \
    | grep -vxF '' \
    | sort -u \
    > "${base_dir}/.words.${depth}"

fi

"${base_dir}/markov" "${base_dir}/.words.${depth}" "$@"