check-links

#!/bin/bash
set -euo pipefail

UA="tldr-pages-bad-url-checker (+https://github.com/tldr-pages/tldr; implemented by @sbrl)"

pushd "$HOME/.tldr/cache/pages" >/dev/null 2>&1

records=$(rg --sort-files -o '<(https?:[^>]+)>' -r '$1' | cat)
tempdir=$(mktemp -d)

echo "tempdir: $tempdir" >&2

for record in $records; do
  file=$(echo "$record" | cut -d':' -f1)
  url=$(echo "$record" | cut -d':' -f2-)
  md5=$(echo "$url" | md5sum | cut -d' ' -f1)

  status_code=
  if [[ -f $tempdir/$md5 ]]; then
    status_code=$(cat "$tempdir/$md5")
  else
    status_code=$(curl --user-agent "$UA" -so /dev/null -w "%{http_code}" -I "$url" || echo 0)
    if [[ $status_code == 405 ]]; then
      status_code=$(curl --user-agent "$UA" -so /dev/null -w "%{http_code}" "$url" || echo 0)
    fi
    echo "$status_code" >"$tempdir/$md5"
  fi

  echo "$status_code $url" >&2

  if [[ $status_code -lt 200 ]] || [[ $status_code -ge 400 ]]; then 
    echo "$file $url $status_code"; 
  fi
done

popd >/dev/null 2>&1