archive_goo_blog.sh

#!/usr/bin/env bash


__check() {
  if [[ $# == 0 ]]; then
    echo "usage: $0 <ENV> [spn opt]..."
    exit 0
  fi

  if ! command -v curl git &>/dev/null; then
    echo "install: curl, git" >&1
    exit 1
  fi
}

__spn() {
  local script_path
  script_path='wayback-machine-spn-scripts/spn.sh'
  if ! [[ -x "$script_path" ]]; then
    git clone --single-branch --depth 1 https://github.com/overcast07/wayback-machine-spn-scripts
    chmod +x "$script_path"
  fi
  . "$script_path" "$@"
}

__main() {
  local blog_id page_source i
  blog_id="$1"
  shift
  page_source=$(mktemp)
  i=1
  while :; do
    curl -s "https://blog.goo.ne.jp/${blog_id}/arcv/?page=${i}" | tee "$page_source"
    if grep -qF '該当する記事はありません。' "$page_source"; then
      break
    fi
    ((i++))
  done > urls.txt
  cat urls.txt | grep -oE "https://blog\.goo\.ne\.jp/${blog_id}/e/[^"]+" | sort -V | uniq 1<>urls.txt
  __spn -p 1 -w 60 -o "https://blog\.goo\.ne\.jp/${blog_id}/e/[^\"]+" $@ urls.txt
}

__check $@
__main $@