path: root/bin/common-functions

#!/bin/sh

# contains functions used by more than one script

# shellcheck disable=SC2039

# TODO: include link depenendencies in run-depends metadata

# TODO: have full information (currently in files) in database

# TODO: remove state files / metadata files

if [ -z "${base_dir}" ]; then
  # just to make shellcheck happy
  . 'conf/default.conf'
fi

# find_pkgbuilds package repository git_repository git_revision mod_git_revision
# find the PKGBUILD and modification of $package from $repository
# sets $PKGBUILD and $PKGBUILD_mod

find_pkgbuilds() {

  local package="$1"
  local repository="$2"
  local git_repository="$3"
  local git_revision="$4"
  local mod_git_revision="$5"

  local repo_path
  eval 'repo_path="${repo_paths__'"${git_repository}"'}"'

  PKGBUILD=$(
    git -C "${repo_path}" archive "${git_revision}" -- "${package}/repos/" 2> /dev/null | \
      tar -t 2> /dev/null | \
      grep "$(printf '^%s-.*/PKGBUILD' "$(str_to_regex "${package}/repos/${repository}")")" | \
      grep -v -- '-i686/PKGBUILD$' | \
      grep -v -- '[-/]\(staging\|testing\|unstable\)-[^/]\+/PKGBUILD$' | \
      sort | \
      tail -n1
  )

  PKGBUILD_mod=$(
    git -C "${repo_paths__archlinux32}" archive "${mod_git_revision}" 2> /dev/null | \
      tar -t "${repository}/${package}/PKGBUILD" 2> /dev/null
  ) || true

  if [ -z "${PKGBUILD}" ] && \
    [ -z "${PKGBUILD_mod}" ]; then
    >&2 printf 'Neither PKGBUILD nor modification of PKGBUILD found for package "%s" from %s (%s), revisions %s and %s.\n' \
      "${package}" \
      "${repository}" \
      "${git_repository}" \
      "${git_revision}" \
      "${mod_git_revision}"
    return 1
  fi

}

# find_repository_with_commit commit
# find the repository which has $commit

find_repository_with_commit() {

  local repository

  for repository in ${repo_names}; do
    # shellcheck disable=SC2016
    if [ "$(eval git -C "$(printf '"${repo_paths__%s}"' "${repository}")" cat-file -t '"$1"' 2> /dev/null)" = "commit" ]; then
      echo "${repository}"
      return 0
    fi
  done
  >&2 printf 'find_repository_with_commit: Cannot find repository with commit "%s"\n' "$1"
  exit 1

}

# find_git_repository_to_package_repository repository
# find the git repository which tracks the package repository $repository

find_git_repository_to_package_repository() {

  local repository
  local package_repository
  local repo_path

  package_repository="$1"

  if [ "$1" = 'build-support' ]; then
    echo 'packages'
    return 0
  fi

  for repository in ${repo_names}; do
    if [ "${repository}" = "archlinux32" ]; then
      continue
    fi
    eval 'repo_path="${repo_paths__'"${repository}"'}"'
    if git -C "${repo_path}" archive "$(cat "${work_dir}/${repository}.revision")" -- | \
      tar -t --wildcards '*/repos' | \
      grep '^\([^/]\+/\)\{3\}PKGBUILD$' | \
      cut -d/ -f3 | \
      sed 's|-[^-]\+$||' | \
      sort -u | \
      grep -qxF "${package_repository}"; then
      echo "${repository}"
      return 0
    fi
  done
  >&2 echo "can't find git repository with package repository '$1'"
  exit 1

}

# generate_package_metadata $package $git_revision $mod_git_revision $repository
# or
# generate_package_metadata $package.$git_revision.$mod_git_revision.$repository
# generate the meta data files of a package (dependencies, built packages, ...)

generate_package_metadata() {

  local package="$1"
  local git_revision="$2"
  local mod_git_revision="$3"
  local repository="$4"
  local file_prefix
  local file
  local PKGBUILD

  if [ $# -eq 1 ]; then
    # second form
    repository="${package##*.}"
    package="${package%.*}"
    mod_git_revision="${package##*.}"
    package="${package%.*}"
    git_revision="${package##*.}"
    package="${package%.*}"
  fi

  file_prefix="${work_dir}/package-infos/${package}.${git_revision}.${mod_git_revision}.${repository}"

  if [ -e "${file_prefix}.builds" ] && \
    [ -e "${file_prefix}.build-depends" ] && \
    [ -e "${file_prefix}.run-depends" ] && \
    [ -e "${file_prefix}.groups" ] && \
    [ -e "${file_prefix}.packages" ]; then
    return 0
  fi

  if ! make_source_info "${package}" "${repository}" "${git_revision}" "${mod_git_revision}" "${file_prefix}.SRCINFO"; then
    printf '"make_source_info %s %s %s %s %s" failed.\n' "${package}" "${repository}" "${git_revision}" "${mod_git_revision}" "${file_prefix}.SRCINFO"
    exit 1
  fi
  if [ ! -s "${file_prefix}.SRCINFO" ]; then
    >&2 printf '"%s" not created by "make_source_info" - eh, what?' "${file_prefix}.SRCINFO"
    exit 1
  fi

  # otherwise this just calls for trouble
  sed -i '
    /^[^=]*=\s*$/d
    s/_i686\(\s*=\)/\1/
  ' "${file_prefix}.SRCINFO"

  # extract "groups" = groups \cup provides
  grep "$(printf '^\t\\(groups\\|provides\\) = ')" "${file_prefix}.SRCINFO" | \
    cut -d= -f2 | \
    sed 's|^\s\+||; s|[<>]$||' | \
    sort -u > \
    "${file_prefix}.groups"

  # extract "packages" = pkgname
  grep '^pkgname = ' "${file_prefix}.SRCINFO" | \
    cut -d= -f2 | \
    sed 's|^\s\+||; s|[<>]$||' | \
    sort -u > \
    "${file_prefix}.packages"

  # extract "builds" = provides \cup pkgname \cup groups
  cat "${file_prefix}.groups" "${file_prefix}.packages" | \
    sort -u > \
    "${file_prefix}.builds"

  # extract "build-depends" = makedepends \cup depends \cup \{ base, base-devel \} \setminus "builds"
  {
    {
      printf 'all_depend = %s\n' 'base' 'base-devel'
      sed -n "$(
        printf '/^\t%s = /p\n' \
          'depends' \
          'makedepends'
      )" "${file_prefix}.SRCINFO"
    } | \
      cut -d= -f2 | \
      sed 's|^\s\+||; s|[<>]$||' | \
      sort -u
    sed 'p' "${file_prefix}.builds"
  } | \
    sort | \
    uniq -u > \
    "${file_prefix}.build-depends"

  # extract "run-depends" = depends \cup \{ base \} \setminus "builds"
  {
    {
      printf 'all_depend = %s\n' 'base'
      sed -n "$(printf '/^\tdepends = /p')" "${file_prefix}.SRCINFO"
    } | \
      cut -d= -f2 | \
      sed 's|^\s\+||; s|[<>]$||' | \
      sort -u
    sed 'p' "${file_prefix}.builds"
  } | \
    sort | \
    uniq -u > \
    "${file_prefix}.run-depends"

  rm "${file_prefix}.SRCINFO"

}

# delete_old_metadata
# delete old (=unneeded) meta data of packages

delete_old_metadata() {

  local current_metadata

  current_metadata=$(
    find "${work_dir}/package-infos" -mindepth 1 -maxdepth 1 -printf '%f\n' | \
      sed '
        s|\.[^.]\+$||
        s|\.\([^.]\+\)\.\([^.]\+\)\.\([^.]\+\)$| \1 \2 \3|
      ' | \
      sort -u
  )

  ( # the new shell is intentional
    # what we have
    echo "${current_metadata}"

    # package-states should stay
    find "${work_dir}/package-states" -mindepth 1 -maxdepth 1 -printf '%f\n' | \
      sed '
        s|\.\([^.]\+\)\.\([^.]\+\)\.\([^.]\+\)\.[^.]\+$| \1 \2 \3|
      ' | \
      sort -u | \
      sed 'p'

    # build-list items should stay
    sed 'p' "${work_dir}/build-list"

    tmp_dir=$(mktemp -d 'tmp.common-functions.delete_old_metadata.XXXXXXXXXX' --tmpdir)
    trap 'rm -rf --one-file-system "${tmp_dir}"' EXIT

    echo "${current_metadata}" | \
      sort -k1,1 > \
      "${tmp_dir}/current-metadata"

    # the newest of the following should stay:
    {
      # deletion-list items
      cat "${work_dir}/deletion-list"
      # all packages in the repos
      for repo in ${repo_names}; do
        eval 'git -C "${repo_paths__'"${repo}"'}" archive '"$(cat "${work_dir}/${repo}.revision")" | \
          tar -t | \
          sed '
            s|/$||
            /\//d
          '
      done
    } | \
      sort -u | \
      join -j 1 -o 2.2,2.3,2.4,2.1 - "${tmp_dir}/current-metadata" | \
      sort -k4,4 > \
      "${tmp_dir}/find-newest-revisions"

    uniq -uf3 < \
      "${tmp_dir}/find-newest-revisions" | \
      awk '{print $4 " " $1 " " $2 " " $3}' | \
      sed 'p'

    uniq -Df3 < \
      "${tmp_dir}/find-newest-revisions" | \
      uniq --group=append -f3 | \
      {
        revs=''
        mod_revs=''
        opkg=''
        orepo=''
        while read -r rev mod_rev repo pkg; do

          if [ -z "${rev}" ] && \
            [ -z "${mod_rev}" ] && \
            [ -z "${repo}" ] && \
            [ -z "${pkg}" ]; then

            printf '%s %s %s %s\n' \
              "$(
                printf '%s\n' ${revs} | \
                  find_newest_of_git_revisions
              )" \
              "$(
                printf '%s\n' ${mod_revs} | \
                  find_newest_of_git_revisions
              )" \
              "${orepo}" \
              "${opkg}"

            revs=''
            mod_revs=''
            orepo=''
            opkg=''
            continue
          fi
          revs=$(
            # shellcheck disable=SC2086
            printf '%s\n' ${revs} ${rev} | \
              sort -u
          )
          mod_revs=$(
            # shellcheck disable=SC2086
            printf '%s\n' ${mod_revs} ${mod_rev} | \
              sort -u
          )
          orepo="${repo}"
          opkg="${pkg}"
        done
      } | \
      awk '{print $4 " " $1 " " $2 " " $3}' | \
      sed 'p'
  ) | \
    sort | \
    uniq -u | \
    while read -r pkg rev mod_rev repo; do
      rm -f "${work_dir}/package-infos/${pkg}.${rev}.${mod_rev}.${repo}."*
    done
}

# repository_of_package $package.$repo_revision.$mod_repo_revision.$repository
# print which (stable) repository a package belongs to

repository_of_package() {
  local package="$1"
  local repository="${package##*.}"
  package="${package%.*}"
  local a32_rev="${package##*.}"
  package="${package%.*.*}"

  case "${repository}" in
    'multilib')
      if git -C "${repo_paths__archlinux32}" archive --format=tar "${a32_rev}" -- 'extra-from-multilib' | \
        tar -Ox | \
        grep -qFx "${package%.*.*.*}"; then
        echo 'extra'
      else
        echo 'community'
      fi
      ;;
    *)
      echo "${repository}"
  esac
}

# official_or_community $package.$repo_revision.$mod_repo_revision.$repository $ending
# print wether the specified package is an official package (print
# $ending) or a community package (print 'community-$ending') or a
# build-suppor package (print 'build-support')

official_or_community() {
  local prepo
  prepo=$(repository_of_package "$1")

  if [ "${prepo}" = 'community' ]; then
    echo 'community-'"$2"
  elif [ "${prepo}" = 'build-support' ]; then
    echo 'build-support'
  else
    echo "$2"
  fi
}

# ls_master_mirror $path
# list content of $path on the master mirror (via rsync)

ls_master_mirror() {

  local path="$1"

  ${master_mirror_rsync_command} \
    "${master_mirror_rsync_directory}/${path}/" | \
    grep -v '\s\.$' | \
    awk '{print $5}'

}

# TODO: the actions of remove_old_package_versions should be done
# on basis of the information in the database

# remove_old_package_versions $arch $repository $package_file

# removes all older (not-newer) versions of $package_file
# in all repositories not-older (newer) than $repository

# TODO: should remove all other version (also newer) from
# some repositories :-/

# A package is considered not newer if
#  a) its version is not newer
# A package is considered older if
#  b) its version is older or
#  c) if it's "not newer" and its architecture is 'any' and different or
#  d) if it's "not newer" and the other architecture is 'any' and different

# this ensures an any package may replace arch-specific packages of the same version and vice versa

remove_old_package_versions() {

  local arch="$1"
  local repository="$2"
  local package="$3"

  local pkgname
  local epoch
  local pkgver
  local pkgrel
  local sub_pkgrel
  pkgname="${package%-*}"
  pkgrel="${pkgname##*-}"
  sub_pkgrel="${pkgrel##*.}"
  if [ "${sub_pkgrel}" = "${pkgrel}" ]; then
    sub_pkgrel='0'
  else
    pkgrel="${pkgrel%.*}"
  fi
  pkgname="${pkgname%-*}"
  pkgver="${pkgname##*-}"
  epoch="${pkgver%%:*}"
  if [ "${epoch}" = "${pkgver}" ]; then
    epoch='0'
  else
    pkgver="${pkgver#*:}"
  fi
  pkgname="${pkgname%-*}"

  # shellcheck disable=SC2016
  {
    printf 'SELECT "bogus",CONCAT(from_base64("%s"),"/",from_base64("%s")),1,from_base64("%s");\n' \
      "$(
        printf '%s' "${repository}" | \
          base64 -w0
      )" \
      "$(
        printf '%s' "${package}" | \
          base64 -w0
      )" \
      "$(
        printf '%s' "${package}" | \
          sed '
            s/^.*-\([^-]\+-[^-]\+\)-[^-]\+$/\1/
          ' | \
          base64 -w0
      )"
    printf 'SELECT '
    printf '`binary_packages`.`id`,'
    printf 'CONCAT(`repositories`.`name`,"/",'
    mysql_package_name_query
    printf '),'
    # should we delete packages of identical version?
    printf 'IF((`more_stable_repos`.`id`!=`repositories`.`id`) AND (`more_stable_repos`.`stability`=`repositories`.`stability`),2,0),'
    printf 'CONCAT('
      printf 'IF(`binary_packages`.`epoch`=0,"",CONCAT(`binary_packages`.`epoch`,":")),'
      printf '`binary_packages`.`pkgver`,"-",'
      printf '`binary_packages`.`pkgrel`,".",'
      printf '`binary_packages`.`sub_pkgrel`'
    printf ')'
    printf ' FROM `binary_packages`'
    printf ' JOIN `%s` ON `%s`.`id`=`binary_packages`.`%s`' \
      'repositories'  'repositories'  'repository' \
      'architectures' 'architectures' 'architecture'
    printf ' JOIN `repository_stability_relations` ON `repository_stability_relations`.`less_stable`=`repositories`.`stability`'
    printf ' JOIN `repositories` AS `more_stable_repos` ON `repository_stability_relations`.`more_stable`=`more_stable_repos`.`stability`'
    # name must match
    printf ' WHERE `binary_packages`.`pkgname`=from_base64("%s")' \
      "$(printf '%s' "${package%-*-*-*}" | base64 -w0)"
    # repository, where package should be deleted, should be less stable
    printf ' AND `more_stable_repos`.`name`=from_base64("%s")' \
      "$(printf '%s' "${repository}" | base64 -w0)"
    printf ';\n'
  } | \
    ${mysql_command} --raw --batch | \
    sed '
      /^\S\+\sCONCAT(/d
      y/\t/ /
    ' | \
    expand_version 4 | \
    sort -k4V,4 -k3r,3 | \
    shrink_version 4 | \
    sed -n '
      /^bogus /q
      p
    ' | \
    cut -d' ' -f1,2 >&2

  # repositories in which older packages should be deleted
  local delete_older_repositories
  # repositories in which not-newer packages should be deleted
  local delete_not_newer_repositories

  if echo "${standalone_package_repositories}" | \
    grep -qxF "${repository}"; then

    delete_older_repositories="${repository}"
    delete_not_newer_repositories=''

  elif echo "${staging_package_repositories}" | \
    grep -qxF "${repository}"; then

    delete_older_repositories="${repository}"
    delete_not_newer_repositories=$(
      echo "${staging_package_repositories}" | \
        grep -vxF "${repository}"
    ) || true

  elif echo "${testing_package_repositories}" | \
    grep -qxF "${repository}"; then

    delete_older_repositories=$(
      printf '%s\n' "${staging_package_repositories}" "${repository}"
    )
    delete_not_newer_repositories=$(
      echo "${testing_package_repositories}" | \
        grep -vxF "${repository}"
    ) || true

  elif echo "${stable_package_repositories}" | \
    grep -qxF "${repository}"; then

    delete_older_repositories=$(
      printf '%s\n' "${staging_package_repositories}" "${testing_package_repositories}" "${repository}"
    )
    delete_not_newer_repositories=$(
      echo "${stable_package_repositories}" | \
        grep -vxF "${repository}"
    ) || true

  else

    >&2 printf 'remove_old_package_versions: Unknown repository "%s".\n' "${repository}"
    return 1

  fi

  ( # the new shell is intentional
    tmp_dir=$(mktemp -d 'tmp.common-functions.remove_old_package_versions.XXXXXXXXXX' --tmpdir)
    trap 'rm -rf --one-file-system "${tmp_dir}"' EXIT

    {
      # the architecture of the package (any vs. i686)
      package_arch="${package##*-}"
      package_arch="${package_arch%%.*}"
      if [ "${package_arch}" = 'any' ]; then
        package_arch_regex_inverter='!'
      else
        unset package_arch_regex_inverter
      fi

      for repo in ${delete_older_repositories}; do
        ls_master_mirror "${arch}/${repo}" | \
          sed -n '
            /\.pkg\.tar\.xz$/!d
            s|-\([^-]\+-[^-]\+\)-\([^-]\+\)$| \1 \2|
            /^'"$(str_to_regex "${package%-*-*-*}")"' / {
              s|^|2 '"${arch} ${repo}"' |
              / any\.pkg\.tar\.xz$/'"${package_arch_regex_inverter}"'{
                s|^2|0|
              }
              p
            }
          '
      done
      for repo in ${delete_not_newer_repositories}; do
        ls_master_mirror "${arch}/${repo}" | \
          sed -n '
            /\.pkg\.tar\.xz$/!d
            s|-\([^-]\+-[^-]\+\)-\([^-]\+\)$| \1 \2|
            /^'"$(str_to_regex "${package%-*-*-*}")"' / {
              s|^|0 '"${arch} ${repo}"' |
              p
            }
          '
      done
      echo "${package%-*}" | \
        sed 's|^.*-\([^-]\+-[^-]\+\)$|1 %cut% %it% %here% \1|'

      # the generated list contains the following columns:
      # $delete-if-newer-vs-not-older $arch-directory $repo-directory $pkgname $pkgver-$pkgrel $pkg-arch.pkg.tar.xz
    } | \
      expand_version 5 | \
      sort -k5V,5 -k1n,1 | \
      shrink_version 5 | \
      sed -n '
        /^1 %cut% %it% %here% /q
        s/^[02] //
        s/ \(\S\+\)$/-\1/
        p
      ' | \
      sort -u > \
      "${tmp_dir}/packages-to-delete"
    # this file contains a list of packages to be deleted, one on each line:
    # $architecture-directory $repository-directory $package-name $pkgver-$pkgrel-$package-architecture.pkg.tar.xz

    cut -d' ' -f1,2 < \
      "${tmp_dir}/packages-to-delete" | \
      grep -vxF "${arch} ${repository}" | \
      sort -u > \
      "${tmp_dir}/repositories-to-modify"

    # fetch all databases being modified
    while read -r del_arch del_repo; do
      mkdir -p "${tmp_dir}/${del_arch}/${del_repo}"
      ${master_mirror_rsync_command} \
        "${master_mirror_rsync_directory}/${del_arch}/${del_repo}/${del_repo}.db."* \
        "${master_mirror_rsync_directory}/${del_arch}/${del_repo}/${del_repo}.files."* \
        "${tmp_dir}/${del_arch}/${del_repo}/"
    done < \
      "${tmp_dir}/repositories-to-modify"

    while read -r del_arch del_repo del_package _; do
      if [ "${del_arch}/${del_repo}" = "${arch}/${repository}" ]; then
        # we do not repo-remove the package in the target repository
        continue
      fi
      repo-remove -q "${tmp_dir}/${del_arch}/${del_repo}/${del_repo}.db.tar.gz" \
        "${del_package}"
    done < \
      "${tmp_dir}/packages-to-delete"

    # upload modified databases
    while read -r del_arch del_repo; do
      ${master_mirror_rsync_command} \
        "${tmp_dir}/${del_arch}/${del_repo}/${del_repo}.db."* \
        "${tmp_dir}/${del_arch}/${del_repo}/${del_repo}.files."* \
        "${master_mirror_rsync_directory}/${del_arch}/${del_repo}/"
    done < \
      "${tmp_dir}/repositories-to-modify"

    # shellcheck disable=SC2016
    sed '
      s/\.pkg\.tar\.xz$//
      s/^\S\+ //
      s/-\([^-. ]\+\)\(-[^- ]\+\)$/-\1.0\2/
      s/ \([^-: ]\+\(-[^- ]\+\)\{2\}\)$/ 0:\1/
      s/ \([^-.]\+\):\([^-:]\+\)-\([^-.]\+\)\.\([^-.]\+\)-\([^-]\+\)$/ \1 \2 \3 \4 \5/
    ' "${tmp_dir}/packages-to-delete" | \
      while read -r repo pkgname epoch pkgver pkgrel sub_pkgrel arch; do
        printf 'DELETE FROM `binary_packages` WHERE'
        printf ' `binary_packages`.`%s`=(SELECT `%s`.`id` FROM `%s` WHERE `%s`.`name`=from_base64("%s")) AND' \
          'architecture' 'architectures' 'architectures' 'architectures' "$(printf '%s' "${arch}" | base64 -w0)" \
          'repository'   'repositories'  'repositories'  'repositories'  "$(printf '%s' "${repo}" | base64 -w0)"
        printf ' `binary_packages`.`%s`=from_base64("%s") AND' \
          'pkgname' "$(printf '%s' "${pkgname}" | base64 -w0)" \
          'epoch' "$(printf '%s' "${epoch}" | base64 -w0)" \
          'pkgver' "$(printf '%s' "${pkgver}" | base64 -w0)" \
          'pkgrel' "$(printf '%s' "${pkgrel}" | base64 -w0)" \
          'sub_pkgrel' "$(printf '%s' "${sub_pkgrel}" | base64 -w0)" | \
          sed 's/ AND$//'
        printf ';\n'
      done | \
      ${mysql_command}

    sed '
      s| \(\S\+\)$|-\1|
      y| |/|
      s|^|rm "|
      s|$|"|
      p
      s|"$|.sig"|
    ' "${tmp_dir}/packages-to-delete" | \
      ${master_mirror_sftp_command}
  )

}

# wait_some_time $minimum $diff
# wait between minimum and minimum+diff seconds (diff defaults to 30)

wait_some_time() {
  local minimum=$1
  local diff=$2
  local random

  if [ -z "${diff}" ]; then
    diff=30
  fi

  random=$(
    dd if='/dev/urandom' count=1 2> /dev/null | \
      cksum | \
      cut -d' ' -f1
  )

  sleep $((minimum + random % diff))
}

# str_to_regex $string
# escape dots for use in regex

str_to_regex() {
  echo "$1" | \
    sed '
      s|[.[]|\\\0|g
    '
}

# make_source_info $package $repository $git_revision $mod_git_revision $output
# create .SRCINFO from PKGBUILD within git repositories, output to $output

make_source_info() {

  local package="$1"
  local repository="$2"
  local git_revision="$3"
  local mod_git_revision="$4"
  local output="$5"

  local git_repo
  local PKGBUILD
  local PKGBUILD_mod

  git_repo=$(find_repository_with_commit "${git_revision}")

  if [ -z "${git_repo}" ]; then
    return 1
  fi

  find_pkgbuilds "${package}" "${repository}" "${git_repo}" "${git_revision}" "${mod_git_revision}"

  ( # the new shell is intentional

    tmp_dir=$(mktemp -d "${work_dir}/tmp.make_source_info.XXXXXX")
    trap 'rm -rf --one-file-system "${tmp_dir}"' EXIT

    extract_source_directory "${git_repo}" "${git_revision}" "${mod_git_revision}" "${tmp_dir}" '0'

    {
      cd "${tmp_dir}"
      makepkg --printsrcinfo
      cd ..
    } |
    if [ "${package%-i18n}-i18n" = "${package}" ]; then
      sed '
        1 a \\tdepends = '"${package%-i18n}"'
      '
    else
      cat
    fi > \
      "${output}"

  )

}

# recursively_umount_and_rm $dir
# umount all mountpoints in $dir which are also in $dir's
#  filesystem, possibly also $dir itself and then
#  rm -rf --one-file-system $dir

recursively_umount_and_rm() {
  local dir="$1"

  if [ -z "${dir}" ]; then
    >&2 echo 'ERROR: recursively_umount_and_rm requires an argument'
    exit 42
  fi

  find "${dir}" \
    -xdev -depth -type d \
    -exec 'mountpoint' '-q' '{}' ';' \
    -exec 'sudo' 'umount' '-l' '{}' ';'
  rm -rf --one-file-system "${dir}"
}

# mangle_pkgbuild $PKGBUILD [$sub_pkgrel]
#  mangle $arch in PKBUILDs to contain i486, i586, i686
#  append $sub_pkgrel to the pkgrel

mangle_pkgbuild() {
  local PKGBUILD="$1"
  local sub_pkgrel="$2"

  if [ -n "${sub_pkgrel}" ]; then
    sub_pkgrel=".${sub_pkgrel}"
  fi

  sed -i '
    /^arch=[^#]*any/!{
      /^arch=(/s/(/(i686 /
    }
    s/^\(\s*pkgrel=\)['"'"'"]\?\([0-9.]\+\)['"'"'"]\?\s*\(#.*\)\?$/\1"\2'"${sub_pkgrel}"'"/
  ' "${PKGBUILD}"
}

# find_newest_of_git_revisions
# find newest git revision of the ones provided at stdin
# (assuming linear history)

find_newest_of_git_revisions() {
  local revisions
  local repo
  revisions=$(cat)

  if [ "$(
    echo "${revisions}" | \
      wc -l
    )" -eq 1 ]; then

    echo "${revisions}"
    return

  fi

  repo=$(
    find_repository_with_commit \
      "$(
        echo "${revisions}" | \
          grep -xm1 '[0-9a-f]\{40\}'
      )"
  )

  eval 'repo="${repo_paths__'"${repo}"'}"'

  echo "${revisions}" | \
    xargs -rn1 git -C "${repo}" rev-parse | \
    {
      newest=''
      while read -r current; do
        if [ -z "${newest}" ] || \
          git -C "${repo}" merge-base --is-ancestor "${newest}" "${current}"; then
            newest="${current}"
          fi
      done
      echo "${newest}"
    }
}

# find_package_repository_to_package $package $git_repository $git_commit
# find the package repository a package from a given git repository
# belongs to

find_package_repository_to_package() {

  local package="$1"
  local git_repository="$2"
  local git_commit="$3"
  local repo_path
  local repo

  eval 'repo_path="${repo_paths__'"${git_repository}"'}"'

  if [ "${git_repository}" = 'archlinux32' ]; then
    repo=$(
      git -C "${repo_path}" archive "${git_commit}" -- | \
        tar -t --wildcards "*/${package}/" | \
        cut -d/ -f1 | \
        sort -u
    )
  else
    repo=$(
      git -C "${repo_path}" archive "${git_commit}" -- "${package}/repos" 2> /dev/null | \
        tar -t | \
        cut -d/ -f3 | \
        grep -vxF '' | \
        grep -v 'staging\|testing\|-unstable' | \
        grep -v -- '-i686$' | \
        sed 's|-[^-]\+$||' | \
        sort -u
    )
  fi

  if [ -z "${repo}" ]; then
    return 1
  fi

  if [ "$(
    echo "${repo}" | \
      wc -l
    )" -ne 1 ]; then
    return 1
  fi

  echo "${repo}"

}

# extract_source_directory $git_repo $rev $mod_rev $output $sub_pkgrel
# extract files found in the svn/git source directories
# $PKGBUILD and $PKGBUILD_mod are expected to be set correctly

extract_source_directory() {

  local git_repo="$1"
  local rev="$2"
  local mod_rev="$3"
  local output="$4"
  local sub_pkgrel="$5"

  if [ -n "${PKGBUILD}" ]; then
    eval 'git -C "${repo_paths__'"${git_repo}"'}" archive "${rev}" -- "${PKGBUILD%/*}"' | \
      tar -x --strip-components=3 -C "${output}"
  fi

  if [ -n "${PKGBUILD_mod}" ]; then
    git -C "${repo_paths__archlinux32}" archive "${mod_rev}" -- "${PKGBUILD_mod%/*}" | \
      tar -x --overwrite --exclude 'PKGBUILD' --strip-components=2 -C "${output}" 2> /dev/null || \
      true
    git -C "${repo_paths__archlinux32}" archive "${mod_rev}" -- "${PKGBUILD_mod}" | \
      tar -Ox "${PKGBUILD_mod}" >> \
      "${output}/PKGBUILD"
  fi

  mangle_pkgbuild "${output}/PKGBUILD" "${sub_pkgrel}"

  # shellcheck disable=SC2016
  sed -i '/^\$Id\$$/d' "${output}/PKGBUILD"

}

# find_dependencies_on_build_list $package $git_revision $mod_git_revision $repository
# return a list of packages on the build list which are (run- / build- / check-time)
# dependencies of the given package

find_dependencies_on_build_list() {

  local package="$1"
  local git_revision="$2"
  local mod_git_revision="$3"
  local repository="$4"

  generate_package_metadata "${package}" "${git_revision}" "${mod_git_revision}" "${repository}"

  {
    cat "${work_dir}/package-infos/${package}.${git_revision}.${mod_git_revision}.${repository}.build-depends"
    awk '{print $1 "." $2 "." $3 "." $4}' < \
      "${work_dir}/build-list" | \
      sed '
        s|^|'"${work_dir}"'/package-infos/|
        s|$|\.builds|
      ' | \
      xargs -r cat | \
      sort -u
  } | \
    sort | \
    uniq -d

}

# download_sources_by_hash $package $repository $git_revision $git_mod_revision
# try to download all sources by their hash into the current directory
# returns 0 if any source was downloaded and 1 otherwise

download_sources_by_hash() {

  local package="$1"
  local repository="$2"
  local git_revision="$3"
  local git_mod_revision="$4"

  local return_value=1
  local tmp_dir
  local sum_type
  local arch_suffix

  tmp_dir=$(mktemp -d 'tmp.common-functions.download_sources_by_hash.XXXXXXXXXX' --tmpdir)

  if ! make_source_info "${package}" "${repository}" "${git_revision}" "${git_mod_revision}" "${tmp_dir}/.SRCINFO"; then
    >&2 echo 'download_sources_by_hash: make_source_info failed.'
    rm -rf --one-file-system "${tmp_dir}"
    return 1
  fi

  if ! [ -s "${tmp_dir}/.SRCINFO" ]; then
    >&2 echo 'download_sources_by_hash: ".SRCINFO" has not been created by make_source_info.'
    rm -rf --one-file-system "${tmp_dir}"
    return 1
  fi

  for arch_suffix in '' '_i686'; do
    for sum_type in 'sha256sum' 'sha512sum'; do
      grep '^\s*'"${sum_type}s${arch_suffix}"' = ' "${tmp_dir}/.SRCINFO" | \
        sed 's|^.* = ||' | \
        cat -n > \
        "${tmp_dir}/sums"
      grep '^\s*source'"${arch_suffix}"' = ' "${tmp_dir}/.SRCINFO" | \
        sed '
          s|^.* = ||
          s|::.*$||
          s|.*/||
        ' | \
        cat -n > \
        "${tmp_dir}/urls"
      if [ "$(wc -l < "${tmp_dir}/sums")" -eq "$(wc -l < "${tmp_dir}/urls")" ]; then
        join -1 1 -2 1 -o 1.2,2.2 "${tmp_dir}/sums" "${tmp_dir}/urls" > \
          "${tmp_dir}/joined"
        while read -r sum file; do
          if [ "${sum}" = 'SKIP' ]; then
            continue
          fi
          if echo "${sum}  ${file}" | \
            ${sum_type} -c > /dev/null 2>&1; then
            # the correct source is already there
            continue
          fi
          if wget -O "${tmp_dir}/transfer" "${source_by_hash_mirror}${sum}"; then
            mv "${tmp_dir}/transfer" "${file}"
            return_value=0
          fi
        done < \
          "${tmp_dir}/joined"
      fi
    done
  done

  rm -rf --one-file-system "${tmp_dir}"
  return ${return_value}

}

# expand_version $column_num
# add "0:" to version in $colum_num-th column if no ":" is there (epoch)
# add "+0" to version in $colum_num-th column if no "+" is there (git count/hash)

expand_version() {
  local column_num
  column_num="$1"

  sed '
    /^\(\S\+\s\+\)\{'"$((column_num-1))"'\}\S*+/! s/^\(\(\S\+\s\+\)\{'"$((column_num-1))"'\}\S*\)-/\1+0-/
    /^\(\S\+\s\+\)\{'"$((column_num-1))"'\}\S*:/! s/^\(\(\S\+\s\+\)\{'"$((column_num-1))"'\}\)/\10:/
  '
}

# shrink_version $column_num
# remove "0:" from version in $colum_num-th column (epoch)
# remove "+0" from version in $colum_num-th column (git count/hash)

shrink_version() {
  local column_num
  column_num="$1"

  sed '
    s/^\(\(\S\+\s\+\)\{'"$((column_num-1))"'\}\S*\)+0-/\1-/
    s/^\(\(\S\+\s\+\)\{'"$((column_num-1))"'\}\)0:/\1/
  '
}

# find_biggest_subset_of_packages $omega $keep $all_builds $all_depends [ $force ]

# Return (to stdout) the biggest subset A of the packages in $omega whose
# runtime dependencies in $omega \cup $keep are also in A

# $all_builds and $all_depends either point to an empty file - then they will get
# filled with cached data for subsequent calls - or to the same files of a previous
# call

# If non-empty, $force contains packages which are assumed to match the above
# condition without checking.

# The arguments are names of files with one $package.$revision.$mod_revision.$repository
# per line.

find_biggest_subset_of_packages() {

  ( # the new shell is intentional
    omega="$1"
    keep="$2"
    all_builds="$3"
    all_depends="$4"
    if [ $# -eq 4 ]; then
      force='/dev/null'
    elif [ $# -eq 5 ]; then
      force="$5"
    else
      >&2 printf 'find_biggest_subset_of_packages: Wrong number of arguments: %s given, 4 or 5 expected.' "$#"
      return 2
    fi

    if [ ! -s "${all_builds}" ]; then
      find "${work_dir}/package-infos/" -maxdepth 1 -name '*.builds' \
        -exec sed '
          s|^|{} |
          s|^\S\+/||
          s|\.builds | |
        ' {} \; | \
        sort -k2,2 > \
        "${all_builds}"
    fi

    if [ ! -s "${all_depends}" ]; then
      find "${work_dir}/package-infos/" -maxdepth 1 -name '*.run-depends' \
        -exec sed '
          s|^|{} |
          s|^\S\+/||
          s|\.run-depends | |
        ' {} \; | \
        grep -v ' base$' | \
        sort -k2,2 > \
        "${all_depends}"
    fi

    sort -u "${omega}" | \
      sponge "${omega}"

    temp_dir=$(mktemp -d 'tmp.common-functions.find_biggest_subset_of_packages.XXXXXXXXXX' --tmpdir)
    trap 'rm -rf --one-file-system "${temp_dir}"' EXIT

    {
      sort -u "${keep}"
      cat "${force}" "${force}"
    } | \
      sort | \
      uniq -u > \
      "${temp_dir}/keep.new"
    touch "${temp_dir}/keep"

    while [ -s "${temp_dir}/keep.new" ]; do
      cat "${temp_dir}/keep.new" "${temp_dir}/keep" | \
        sort -u | \
        sponge "${temp_dir}/keep"

      {
        # we append all packages which are run-dependencies of keep-packages
        # to the keep-list
        sed '
          s|^|'"${work_dir}"'/package-infos/|
          s|$|.run-depends|
        ' "${temp_dir}/keep" | \
          xargs -r grep -HF '' | \
          sed '
            s|^.*/||
            s|\.run-depends:| |
          ' | \
          sort -u | \
          sort -k2,2 | \
          uniq -f1 | \
          join -1 2 -2 2 -o 2.1 - "${all_builds}"

        # we append all packages with run-dependencies on the keep-list
        # to the keep-list
        sed '
          s|^|'"${work_dir}"'/package-infos/|
          s|$|.builds|
        ' "${temp_dir}/keep" | \
          xargs -r grep -HF '' | \
          sed '
            s|^.*/||
            s|\.builds:| |
          ' | \
          sort -u | \
          sort -k2,2 | \
          uniq -f1 | \
          join -1 2 -2 2 -o 2.1 - "${all_depends}"
      } | \
        sort -u | \
        join -1 1 -2 1 -o 2.1 - "${omega}" | \
        sort -u > \
        "${temp_dir}/keep.new"

      # "new" is only what has not been there before and what is not forced
      cat "${temp_dir}/keep" "${temp_dir}/keep" "${force}" "${force}" "${temp_dir}/keep.new" | \
        sort | \
        uniq -u | \
        sponge "${temp_dir}/keep.new"
    done

    cat "${omega}" "${temp_dir}/keep" "${temp_dir}/keep" | \
      sort | \
      uniq -u

  )

}

# sort_square_bracket_content $file
# sort the content of [] in $file, print to stdout

sort_square_bracket_content() {
  local file
  local line
  local token
  local token_list
  local rest
  file="$1"

  while read -r line; do
    printf '%s ' "${line}" | \
      tr ' ' '\n' | \
      while read -r token; do
        if echo "${token}" | \
          grep -qF '['; then
          printf '%s[' "${token%[*}"
          token="${token##*[}"
          token_list="${token%,}"
          while ! echo "${token_list}" | \
            grep -qF ']'; do
            read -r token
            token_list=$(
              printf '%s\n' \
                "${token_list}" \
                "${token%,}"
            )
          done
          rest="]${token_list#*]}"
          token_list="${token_list%%]*}"
          token=$(
            printf '%s' "${token_list}" | \
              sort | \
              sed 's|$|,|'
            printf '%s' "${rest}"
          )
        fi
        printf '%s\n' "${token}"
      done | \
      tr '\n' ' ' | \
      sed '
        s|, ]|]|g
        s| $||
      '
    printf '\n'
  done < \
    "${file}"
}

# smoothen_namcap_log $file
# remove unneccesary differences from namcap-logs:
#   - remove architecture specific information
#   - sort lines
#   - sort content of square brackets

smoothen_namcap_log() {
  local file
  file="$1"
  # shellcheck disable=SC2016
  sort_square_bracket_content "${file}" | \
    sed '
      # normalize architecture specific information
      s|i[34567]86|$ARCH|g
      s|x86\([-_]64\)\?|$ARCH|g
      # remove haskell hashes
      s|\('"'"'[^'"'"']*-[0-9.]\+\)-[a-zA-Z0-9]\{1,22\}\(-ghc[^'"'"']*'"'"'\)|\1\2|g
    ' | \
    sort | \
    sponge "${file}"
}

# print_list_of_archaic_packages $source1 $source2 ...
# print a list of packages which have not been touched for a while,
# but which are still in the pipeline, e.g. in $source1, $source2 or ...

print_list_of_archaic_packages() {
  for source in "$@"; do
    case "${source}" in
      'testing')
        # packages remaining longer than $max_package_age_testing days in testing will be marked tested if no bug for them exists on FS32
        find "${work_dir}/package-states" -mindepth 1 -maxdepth 1 -name '*.testing' -mtime "+${max_package_age_testing}" \
          -exec head -n1 {} \; | \
          "${base_dir}/bin/modify-package-state" -n --tested /dev/stdin
        # packages remaining longer than $max_package_age_broken_testing days in testing (w/o being tested!) will be considered outdated
        # and no longer block other packages from being moved
        find "${work_dir}/package-states" -mindepth 1 -maxdepth 1 -name '*.testing' -mtime "+${max_package_age_broken_testing}" -printf '%f\n' | \
          sed '
            s|\.testing$||
          '
      ;;
      'build-list')
        while read -r pkg rev mod_rev repo; do
          git_repo=$(
            find_repository_with_commit "${rev}"
          )
          eval repo_path='"${repo_paths__'"${git_repo}"'}"'
          commit_date=$(
            git -C "${repo_path}" show -s --format=%ct "${rev}"
          )
          mod_commit_date=$(
            git -C "${repo_paths__archlinux32}" show -s --format=%ct "${mod_rev}"
          )
          if [ "${mod_commit_date}" -gt "${commit_date}" ]; then
            commit_date="${mod_commit_date}"
          fi
          # packages remaining longer than $max_package_age_build_list days on the build list
          if [ "$((commit_date + 24*60*60*max_package_age_build_list))" -lt "$(date '+%s')" ]; then
            printf '%s %s %s %s\n' \
              "${pkg}" \
              "${rev}" \
              "${mod_rev}" \
              "${repo}"
          fi
        done < \
          "${work_dir}/build-list"
      ;;
      'staging')
        # packages remaining longer than $max_package_age_staging days in staging
        find "${work_dir}/package-states" -mindepth 1 -maxdepth 1 -name '*.done' -mtime "+${max_package_age_staging}" -printf '%f\n' | \
          sed '
            s|\.done$||
          '
      ;;
      *)
        >&2 printf 'unknown archaic-source "%s" - skipped.\n' "${source}"
      ;;
    esac
  done | \
    sort -u
}

# modification_revision_link "${mod_rev}" "${repo}" "${pkg}"
# print the given modification revision possibly with a html link to github

modification_revision_link() {
  local mod_rev="$1"
  local repo="$2"
  local pkg="$3"

  if git -C "${repo_paths__archlinux32}" archive "${mod_rev}" -- "${repo}/${pkg}/PKGBUILD" > /dev/null 2>&1; then
    printf '<a href="https://github.com/archlinux32/packages/tree/%s/%s/%s">%s</a>\n' \
      "${mod_rev}" \
      "${repo}" \
      "${pkg}" \
      "${mod_rev}"
  else
    printf '%s\n' \
      "${mod_rev}"
  fi
}

# trigger_mirror_refreshs
# trigger a refresh of capable tier 1 mirrors (as backup for master mirror)

trigger_mirror_refreshs() {
  local tmp_file

  tmp_file=$(mktemp "tmp.common-functions.trigger_mirror_refreshs.XXXXXXXXXX" --tmpdir)
  date '+%s' > \
    "${tmp_file}"
  ${master_mirror_rsync_command} \
    "${tmp_file}" \
    "${master_mirror_rsync_directory}/lastupdate"
  rm "${tmp_file}"
  for trigger_url in ${mirror_refresh_trigger_urls}; do
    screen -S trigger-mirror-update -d -m curl -L "${trigger_url}"
  done
}

# extract_pkgname_epoch_pkgver_pkgrel_sub_pkgrel_arch_from_package_name
extract_pkgname_epoch_pkgver_pkgrel_sub_pkgrel_arch_from_package_name() {
  pkgname="$1"
  pkgname="${pkgname%.pkg.tar.xz}"
  arch="${pkgname##*-}"
  pkgname="${pkgname%-*}"
  sub_pkgrel="${pkgname##*-}"
  pkgname="${pkgname%-*}"
  pkgrel="${sub_pkgrel%.*}"
  if [ "${pkgrel}" = "${sub_pkgrel}" ]; then
    sub_pkgrel='0'
  else
    sub_pkgrel="${sub_pkgrel##*.}"
  fi
  epoch="${pkgname##*-}"
  pkgname="${pkgname%-*}"
  pkgver="${epoch#*:}"
  if [ "${pkgver}" = "${epoch}" ]; then
    epoch='0'
  else
    epoch="${epoch%%:*}"
  fi
}