path: root/bin/common-functions

#!/bin/sh

# contains functions used by more than one script

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}"'}"'

  # TODO: remove, when FS#54696 closed
  local _perl_list_moreutils_repair_flag
  if [ "${package}" = 'perl-list-moreutils' ]; then
    _perl_list_moreutils_repair_flag='-r'
  fi

  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 ${_perl_list_moreutils_repair_flag} | \
      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 ${repo_names}; do
    if [ "$(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

}

# package_locked_or_blocked package git_revision mod_git_revision repository
# return if package - of given repository and revisions - is [locked or blocked]

package_locked_or_blocked() {
  [ -f "${work_dir}/package-states/$1.$2.$3.$4.locked" ] || \
    [ -f "${work_dir}/package-states/$1.$2.$3.$4.blocked" ]
}

# 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}.depends" ] && \
    [ -e "${file_prefix}.needs" ] && \
    [ -e "${file_prefix}.groups" ] && \
    [ -e "${file_prefix}.packages" ]; then
    return 0
  fi

  if ! make_source_info "${package}" "${repository}" "${git_revision}" "${mod_git_revision}" "${file_prefix}.SRCINFO"; then
    echo "make_source_info failed."
    exit 1
  fi
  if [ ! -s "${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 "depends" = makedepends \cup checkdepends \cup depends \cup \{ base, base-devel \}
  (
    printf 'all_depend = %s\n' 'base' 'base-devel'
    sed -n "$(printf '/^pkgname = /q;/^\tdepends = /p')" "${file_prefix}.SRCINFO"
    grep "$(printf '^\t\\(makedepends\\|checkdepends\\) = ')" "${file_prefix}.SRCINFO"
  ) | \
    cut -d= -f2 | \
    sed 's|^\s\+||; s|[<>]$||' | \
    sort -u > \
    "${file_prefix}.depends"

  # extract "needs" = "depends" \setminus ( "builds" \cup \{ base-devel \} )
  (
    cat "${file_prefix}.depends"
    sed 'p' "${file_prefix}.builds"
    printf '%s\n' 'base-devel' 'base-devel'
  ) | \
    sort | \
    uniq -u > \
    "${file_prefix}.needs"

  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" -maxdepth 1 -printf '%f\n' | \
      sed '
        s|\.[^.]\+$||
        s|\.\([^.]\+\)\.\([^.]\+\)\.\([^.]\+\)$| \1 \2 \3|
      ' | \
      sort -u
  )

  (
    # what we have
    echo "${current_metadata}"

    # package-states should stay
    find "${work_dir}/package-states" -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=$(mktemp -d)
    trap 'rm -rf --one-file-system "${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}'

}

# remove_old_package_versions $directory $package_file
# removes all other versions of $package_file in $directory on the master mirror

remove_old_package_versions() {

  local directory="$1"
  local package="$2"

  local pkgname="${package%-*-*-*.pkg.tar.xz}"

  ${master_mirror_rsync_command} \
    --recursive \
    --delete \
    $( \
      ls_master_mirror "${directory}" | \
        grep "^$(str_to_regex "${pkgname}")\(-[^-]\+\)\{3\}\.pkg\.tar\.xz\(\.sig\)\?\$" | \
        grep -v "^$(str_to_regex "${package}")\(\.sig\)\?\$" | \
        sed 's|^|--include=|'
    ) \
    '--exclude=*' \
    ./ \
    "${master_mirror_rsync_directory}/${directory}/"

}

# 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|\.|\\.|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}"

  (

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

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

    (
      cd "${tmp_dir}"
      makepkg --printsrcinfo
    ) > \
      "${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}"
}

# insert_break_loops_orders $file
# inserts the correct number of 'break_loops' order into $file

insert_break_loops_orders() {
  local file="$1"
  local looped_packages
  local line

  looped_packages=$(
    find "${work_dir}/build-list.loops" -maxdepth 1 | \
      grep '/loop_[0-9]\+$' | \
      xargs -r cat | \
      sort -u | \
      wc -l
  )

  (
    while read -r line; do
      if [ "${line}" = 'break_loops' ]; then
        if [ "${looped_packages}" -eq 0 ]; then
          continue
        fi
        looped_packages=$((looped_packages-1))
      fi
      printf '%s\n' "${line}"
    done < \
      "${file}"
    while [ ${looped_packages} -gt 0 ]; do
      looped_packages=$((looped_packages-1))
      echo 'break_loops'
    done
  ) | \
    sponge "${file}"
}

# append_architectures $PKGBUILD
#  mangle $arch in PKBUILDs to contain i486, i586, i686

append_architectures() {
  local PKGBUILD="$1"
  sed -i '/^arch=(/s/(/(i686 /' "${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
# 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 repo_path
  local repo

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

  repo=$(
    git -C "${repo_path}" archive "$(cat "${work_dir}/${git_repository}.revision")" -- "${package}/repos" 2> /dev/null | \
      tar -t | \
      cut -d/ -f3 | \
      grep -vxF '' | \
      grep -v 'staging\|testing\|-unstable' | \
      grep -v -- '-i686$' | \
      sed 's|-[^-]\+$||' | \
      sort -u
  )

  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
# 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"

  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

  append_architectures "${output}/PKGBUILD"

  # 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 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}.${repository}.needs"
    grep -vxF 'break_loops' "${work_dir}/build-list" | \
      awk '{print $1 "." $2 "." $3 "." $4}' | \
      sed "
        s|^|${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)

  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" > \
          "${tmp_dir}/joined"
        while read -r sum file; do
          if echo "${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}

}