find-https.bash

#                                                                                                                                                                                                                  
# Example usage:
#   $ git clone --depth=1 git://git.archlinux.org/svntogit/packages.git
#   $ ( cd packages >/dev/null && find-https ) >unused-https.txt
#

has_https_available() {
  local https_url url_eff

  https_url=https${1#http}

  url_eff=$(curl -sLfI -o/dev/null -w '%{url_effective}' "$https_url") &&
    [[ $url_eff = https:* ]]
}

print_srcinfo() {
  local bin=$HOME/code/pkgbuild-introspection/introspect

  # introspect from p-i is substantially faster than makepkg.
  if type -p "$bin"; then
    "$bin" "$1"
  else
    makepkg --printsrcinfo "$1"
  fi
}

find_https() {
  local v

  mapfile -t sources < <(print_srcinfo "$d" | awk '$1 == "source" && $3 ~ /http:/ { print $3 }')

  for v in "${sources[@]}"; do
    # strip source renaming
    [[ $v = *::* ]] && v=${v#*::}

    if has_https_available "$v"; then
      d=${d%/trunk/} d=${d##*/}
      printf '%s\t%s\n' "$d" "https${v#http}"
      continue 2
    fi
  done
}

if (( $# )); then
  search=("$@")
else
  search=(*/trunk/PKGBUILD)
fi

MAX_JOBS=${MAX_JOBS:-20}

for d in "${search[@]}"; do
  while (( $(jobs -p | wc -l) >= MAX_JOBS )); do
    # we could use 'wait -n' here, but it doesn't seem to play nicely with
    # interrupting the script and just hangs forever.
    sleep 1s
  done

  find_https "$d" &
done

wait

find-unused-gpg.bash

#!/bin/bash

#
# Attempts to find packages where gpg signatures are available, but not
# referenced by the PKGBUILD. This script expects to run on a git clone of
# Arch's SVN repos, i.e.
#
#   git://git.archlinux.org/svntogit/packages.git/
#   git://git.archlinux.org/svntogit/community.git/
#
# Example usage:
#   $ git clone --depth=1 git://git.archlinux.org/svntogit/packages.git
#   $ ( cd packages >/dev/null && find-unused-gpg ) >unused-gpg.txt
#

curl_silent() {
  curl -gsLfI -o/dev/null "$@"
}

has_gpg() {
  local url=$1 ext=$2

  if [[ $url = http:* ]]; then
    if curl_silent "https${url#http}.$ext"; then
      echo "https${url#http}.$ext"
      return 0
    fi
  fi

  curl_silent "$url.$ext" && echo "$url.$ext"
}

print_srcinfo() {
  local bin=$HOME/code/pkgbuild-introspection/introspect

  # introspect from p-i is substantially faster than makepkg.
  if type -p "$bin"; then
    "$bin" "$1"
  else
    makepkg --printsrcinfo "$1"
  fi
}

find_unused_gpg() {
  local d=$1 sigurl
  declare -A sources

  while read -r src; do
    sources["$src"]=1
  done < <(print_srcinfo "$d" | awk '$1 ~ /^source/ && $3 ~ /http:/ { print $3 }')

  for v in "${!sources[@]}"; do
    # skip non-remote files
    [[ $v = *://* ]] || continue

    # skip things which look like signatures already
    [[ $v = *.@(sig|asc) ]] && continue

    # strip source renaming
    [[ $v = *::* ]] && v=${v#*::}

    for ext in sig asc; do
      (( 'sources[$v.$ext]' )) && break 2
    done

    for ext in sig asc; do
      if sigurl=$(has_gpg "$v" "$ext"); then
        printf '%s\t%s\n' "$1" "$sigurl"
        break
      fi
    done
  done
}

if (( $# )); then
  search=("$@")
else
  search=(*/trunk/PKGBUILD)
fi

MAX_JOBS=${MAX_JOBS:-20}

for d in "${search[@]}"; do
  while (( $(jobs -p | wc -l) >= MAX_JOBS )); do
    # we could use 'wait -n' here, but it doesn't seem to play nicely with
    # interrupting the script and just hangs forever.
    sleep
  done

  find_unused_gpg "$d" &
done

wait