path: root/lib/common-functions

#!/bin/sh

# contains functions used by more than one script

# shellcheck disable=SC2039,SC2119,SC2120

if [ -z "${base_dir}" ]; then
  # just to make shellcheck happy
  . '../lib/load-configuration'
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/${repository}-*/PKGBUILD" 2> /dev/null | \
      tar -t 2> /dev/null | \
      grep -- '/PKGBUILD$' | \
      grep -v -- '-i686/PKGBUILD$' | \
      grep -v -- '[-/]\(staging\|testing\|unstable\)-[^/]\+/PKGBUILD$' | \
      sort | \
      tail -n1
  )

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

  repository=$(
    # shellcheck disable=SC2016
    {
      printf 'SELECT `git_repositories`.`name` FROM `git_repositories`'
      mysql_join_git_repositories_upstream_repositories
      printf ' WHERE `upstream_repositories`.`name`=from_base64("%s");\n' \
        "$(printf '%s' "$1" | base64 -w0)"
    } | \
      mysql_run_query
  )
  if [ -z "${repository}" ]; then
    >&2 echo "can't find git repository with package repository '$1'"
    exit 1
  else
    echo "${repository}"
    return 0
  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: once binary_packages_in_repositories is funcitonal, we should
# use the id of that

# remove_old_package_versions $binary_package_id

# removes all older versions of the given package from all repositories
# less[1] stable than the current repository, as well as any different
# version of the same package from equally[2] stable repositories

# 1] determined by `repository_stability_relations`
# 2] identical `repository`.`stability`

remove_old_package_versions() {

# TODO: needs to remove binary_packages_in_repositories, too

  ( # 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

    binary_package_id="$1"

    # shellcheck disable=SC2016
    {
      printf 'SELECT '
      printf '`d_bp`.`id`,'
      printf 'IF(`d_r`.`stability`=`o_r`.`stability` AND `d_bp`.`id`!=`o_bp`.`id`,1,0),'
      printf 'CONCAT('
        printf 'IF(`d_bp`.`epoch`=0,"",CONCAT(`d_bp`.`epoch`,":")),'
        printf '`d_bp`.`pkgver`,"-",'
        printf '`d_bp`.`pkgrel`,".",'
        printf '`d_bp`.`sub_pkgrel`'
      printf '),'
      printf 'IF(`d_r`.`id`=`o_r`.`id`,1,0),'
      printf '`d_ra`.`name`,'
      printf '`d_r`.`name`,'
      printf '`d_bp`.`pkgname`,'
      printf 'CONCAT(`d_ra`.`name`,"/",'
      printf '`d_r`.`name`,"/",'
      mysql_package_name_query 'd_bp' 'd_bpa'
      printf ')'
      printf ' FROM `binary_packages` AS `d_bp`'
      mysql_join_binary_packages_binary_packages_in_repositories 'd_bp' 'd_bpir'
      mysql_join_binary_packages_in_repositories_repositories 'd_bpir' 'd_r'
      printf ' AND `d_r`.`is_on_master_mirror`'
      mysql_join_repositories_architectures 'd_r' 'd_ra'
      mysql_join_binary_packages_architectures 'd_bp' 'd_bpa'
      printf ' JOIN `binary_packages` AS `o_bp` ON `d_bp`.`pkgname`=`o_bp`.`pkgname`'
      mysql_join_binary_packages_binary_packages_in_repositories 'o_bp' 'o_bpir'
      mysql_join_binary_packages_in_repositories_repositories 'o_bpir' 'o_r'
      printf ' AND `o_r`.`is_on_master_mirror`'
      printf ' AND `o_r`.`architecture`=`d_r`.`architecture`'
      printf ' JOIN `repository_stability_relations`'
      printf ' ON `repository_stability_relations`.`less_stable`=`d_r`.`stability`'
      printf ' AND `repository_stability_relations`.`more_stable`=`o_r`.`stability`'
      printf ' WHERE `o_bp`.`id`=from_base64("%s")' \
        "$(
          printf '%s' "${binary_package_id}" | \
            base64 -w0
        )"
      printf ';\n'
    } | \
      mysql_run_query | \
      tr '\t' ' ' | \
      expand_version 3 | \
      sort -k3V,3 -k2r,2 | \
      shrink_version 3 | \
      sed -n '
        /^'"${binary_package_id}"' /q
        s/^\([0-9]\+ \)\(\S\+ \)\{2\}/\1/
        h
        /^[0-9]\+ 0 / {
          s/^\(\S\+ \)\{2\}//
          s/ \S\+$//
          w'"${tmp_dir}"'/repo-removes
          g
        }
        s/^\(\S\+ \)\{5\}//
        w'"${tmp_dir}"'/sftp-removes
        s/$/.sig/
        w'"${tmp_dir}"'/sftp-removes
        g
        s/ .*$//
        w'"${tmp_dir}"'/db-removes
      '

    # repo-remove packages
    while read -r arch repo pkgname; do
      mkdir "${tmp_dir}/transit"
      ${master_mirror_rsync_command} \
        "${master_mirror_rsync_directory}/${arch}/${repo}/${repo}.db."* \
        "${master_mirror_rsync_directory}/${arch}/${repo}/${repo}.files."* \
        "${tmp_dir}/transit/"
      repo-remove "${tmp_dir}/transit/${repo}.db.tar.gz" "${pkgname}"
      ${master_mirror_rsync_command} \
        "${tmp_dir}/repos/${arch}/${repo}/${repo}.db."* \
        "${tmp_dir}/repos/${arch}/${repo}/${repo}.files."* \
        "${master_mirror_rsync_directory}/${arch}/${repo}/"
      rm -rf --one-file-system "${tmp_dir}/transit"
    done < \
      "${tmp_dir}/repo-removes"

    # sftp-remove packages
    if [ -s "${tmp_dir}/sftp-removes" ]; then
      sed '
        s|^|rm "|
        s|$|"|
      ' "${tmp_dir}/sftp-removes" | \
        ${master_mirror_sftp_command}
    fi

    # db-remove packages
    if [ -s "${tmp_dir}/db-removes" ]; then
      # shellcheck disable=SC2016
      {
        printf 'CREATE TEMPORARY TABLE `del` (`id` BIGINT NOT NULL AUTO_INCREMENT, PRIMARY KEY (`id`));\n'
        printf 'LOAD DATA LOCAL INFILE "%s" INTO TABLE `del` (`id`);\n' "${tmp_dir}/db-removes"
        printf 'DELETE `binary_packages` FROM `binary_packages` JOIN `del` ON `binary_packages`.`id`=`del`.`id`;\n'
      } | \
        mysql_run_query
    fi

  )

}

# 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 "${tmp_dir}"
      # some additional info
      printf 'upstream_git_repository = %s\n' "${git_repo}"
      printf 'PKGBUILD = %s\n' "${PKGBUILD}"
      printf 'PKGBUILD_mod = %s\n' "${PKGBUILD_mod}"
      if [ "${package%-i18n}-i18n" = "${package}" ]; then
        # *-i18n packages should have this dependency
        printf '\tdepends = %s\n' "${package%-i18n}"
      elif [ "${package%-doc}-doc" = "${package}" ]; then
        # *-doc packages should have this dependency
        printf '\tdepends = %s\n' "${package%-doc}"
      elif [ "${package%-docs}-docs" = "${package}" ]; then
        # *-doc packages should have this dependency
        printf '\tdepends = %s\n' "${package%-docs}"
      fi
      makepkg --printsrcinfo
      cd ..
    } > \
      "${output}" || \
      rm -f "${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
#  remove "lib32-" from {make,check,opt,}depends

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

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

  if grep -q '^\s*pkgname=["'"'"']\?lib32-' "${PKGBUILD}"; then
    sed -i '
      s/^\(\s*pkgrel=\)['"'"'"]\?\([0-9]\+\)\.[0-9]*['"'"'"]\?\s*\(#.*\)\?$/\1"\2"/
    ' "${PKGBUILD}"
  fi

  sed -i '
    /^arch=[^#]*any/!{
      /^arch=(/s/(/(i686 i486 pentium3 /
    }
    s/^\(\s*pkgrel=\)['"'"'"]\?\([0-9.]\+\)['"'"'"]\?\s*\(#.*\)\?$/\1"\2'"${sub_pkgrel}"'"/
  ' "${PKGBUILD}"

  sed -i '
    /^\s*\(make\|check\|opt\|\)depends\(_[^=[:space:]]\+\)\?=(/ {
      :a
        /^\s*\(\S[^=]*\)=(\([^()]*\(#[^\n]*\n\)\?\)*)/! {
          $! {
            N
            ba
          }
        }
        s/\(["'"'"'([:space:]]\)lib32-/\1/g
    }
  ' "${PKGBUILD}"
}

# 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"
  # shellcheck disable=SC2034
  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

  # we do not want to update pkgver, so we just undefine it
  printf 'unset -f pkgver\n' >> \
    "${output}/PKGBUILD"

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

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

  # we don't want write permissions on the PKGBUILD - otherwise pkgver()
  # will change the version! (**HACK**)
  chmod oga-w "${output}/PKGBUILD"

}

# 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 'sha1sum' '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/
  '
}

# 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}"
}

# 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
}

# irc_say $channel [copy]
# say content of stdin in irc channel $channel (default: #archlinux32)
# and print copy to stdout if 'copy' is given

# shellcheck disable=SC2120
irc_say() {
  local channel
  if [ -z "$1" ]; then
    channel='#archlinux32'
  else
    channel="$1"
  fi
  if [ -p "${irc_dir}/${channel}/in" ]; then
    if [ "$2" = 'copy' ]; then
      pee cat 'sponge '"${irc_dir}/${channel}/in"
    else
      sponge "${irc_dir}/${channel}/in"
    fi
  fi
}

# calculate_script_checksum
# calculate and print a checksum of the main script and all scripts in lib/
calculate_script_checksum() {
  {
    sha512sum "$0"
    find "${base_dir}/lib" -type f \
      -exec sha512sum '{}' \;
  } | \
    sort | \
    awk '{print $1}' | \
    sha512sum | \
    awk '{print $1}'
}

# verbose_flock
# flock wrapper with some informational output on error
verbose_flock() {
  local err=0
  flock "$@" || {
    err=$?
    lsof +c0 "/proc/$$/fd/$(
      printf '%s\n' "$@" | \
        grep -vm1 '^-'
    )" || true
    printf 'FYI: I am %s.\n' "$$"
    return ${err}
  }
}