path: root/brep/handler/upload/upload-bindist-clean.in

#!/usr/bin/env bash

# file      : brep/handler/upload/upload-bindist-clean.in
# license   : MIT; see accompanying LICENSE file

# Remove expired package configuration directories created by the
# upload-bindist handler.
#
# Specifically, perform the following steps:
#
# - Recursively scan the specified root directory and collect the package
#   configuration directories with age older than the specified timeout (in
#   minutes). Recognize the package configuration directories by matching the
#   *-????-??-??T??:??:??Z* pattern and calculate their age based on the
#   modification time of the packages.sha256 file they may contain. If
#   packages.sha256 doesn't exist in the configuration directory, then
#   consider it as still being prepared and skip.
#
# - Iterate over the expired package configuration directories and for each of
#   them:
#
#   - Lock the root directory.
#
#   - Re-check the expiration criteria.
#
#   - Remove the package configuration symlink if it refers to this directory.
#
#   - Remove this directory.
#
#   - Remove all the the parent directories of this directory which become
#     empty, up to (but excluding) the root directory.
#
#   - Unlock the root directory.
#
usage="usage: $0 <root> <timeout>"

# Diagnostics.
#
verbose= #true

# The root directory lock timeout (in seconds).
#
lock_timeout=60

trap "{ exit 1; }" ERR
set -o errtrace     # Trap in functions and subshells.
set -o pipefail     # Fail if any pipeline command fails.
shopt -s lastpipe   # Execute last pipeline command in the current shell.
shopt -s nullglob   # Expand no-match globs to nothing rather than themselves.

function info () { echo "$*" 1>&2; }
function error () { info "$*"; exit 1; }
function trace () { if [ "$verbose" ]; then info "$*"; fi }

# Trace a command line, quoting empty arguments as well as those that contain
# spaces.
#
function trace_cmd () # <cmd> <arg>...
{
  if [[ "$verbose" ]]; then
    local s="+"
    while [ $# -gt 0 ]; do
      if [ -z "$1" -o -z "${1##* *}" ]; then
        s="$s '$1'"
      else
        s="$s $1"
      fi

      shift
    done

    info "$s"
  fi
}

# Trace and run a command.
#
function run () # <cmd> <arg>...
{
  trace_cmd "$@"
  "$@"
}

if [[ "$#" -ne 2 ]]; then
  error "$usage"
fi

# Package configurations root directory.
#
root_dir="${1%/}"
shift

if [[ -z "$root_dir" ]]; then
  error "$usage"
fi

if [[ ! -d "$root_dir" ]]; then
  error "'$root_dir' does not exist or is not a directory"
fi

# Package configuration directories timeout.
#
timeout="$1"
shift

if [[ ! "$timeout" =~ ^[0-9]+$ ]]; then
  error "$usage"
fi

# Note that while the '%s' date format is not POSIX, it is supported on both
# Linux and FreeBSD.
#
expiration=$(($(date -u +"%s") - $timeout * 60))

# Collect the list of expired package configuration directories.
#
expired_dirs=()

run find "$root_dir" -type d -name "*-????-??-??T??:??:??Z*" | while read d; do
  f="$d/packages.sha256"

  # Note that while the -r date option is not POSIX, it is supported on both
  # Linux and FreeBSD.
  #
  trace_cmd date -u -r "$f" +"%s"
  if t="$(date -u -r "$f" +"%s" 2>/dev/null)" && (($t <= $expiration)); then
    expired_dirs+=("$d")
  fi
done

if [[ "${#expired_dirs[@]}" -eq 0 ]]; then
  exit 0 # Nothing to do.
fi

# Make sure the root directory lock file exists.
#
lock="$root_dir/upload.lock"
run touch "$lock"

# Remove the expired package configuration directories, symlinks which refer
# to them, and the parent directories which become empty.
#
for d in "${expired_dirs[@]}"; do
  # Deduce the path of the potential package configuration symlink that may
  # refer to this package configuration directory by stripping the
  # -<timestamp>[-<number>] suffix.
  #
  l="$(sed -n -re 's/^(.+)-[0-9]{4}-[0-9]{2}-[0-9]{2}T[0-9]{2}:[0-9]{2}:[0-9]{2}Z(-[0-9]+)?$/\1/p' <<<"$d")"
  if [[ -z "$l" ]]; then
    error "invalid name '$d' for package configuration directory"
  fi

  f="$d/packages.sha256"

  # Open the reading file descriptor and lock the root directory. Fail if
  # unable to lock before timeout.
  #
  trace "+ exec {lfd}<$lock"
  exec {lfd}<"$lock"

  if ! run flock -w "$lock_timeout" "$lfd"; then
    error "unable to lock root directory"
  fi

  # Now, as the lock is acquired, recheck the package configuration directory
  # expiration criteria (see above) and, if it still holds, remove this
  # directory, the package configuration symlink if it refers to it, and all
  # the parent directories which become empty up to (but excluding) the root
  # directory.
  #
  trace_cmd date -u -r "$f" +"%s"
  if t="$(date -u -r "$f" +"%s" 2>/dev/null)" && (($t <= $expiration)); then
    # Remove the package configuration symlink.
    #
    # Do this first to avoid dangling symlinks which may potentially be
    # exposed by brep.
    #
    # Note that while the realpath utility is not POSIX, it is present on
    # both Linux and FreeBSD.
    #
    if [[ -L "$l" ]]; then
      p="$(realpath "$l")"
      if [[ "$p" == "$d" ]]; then
        run rm "$l"
      fi
    fi

    # Remove the package configuration directory.
    #
    # Note that this directory contains files copied from a subdirectory of
    # upload-data. These files are normally owned by the Apache2 user/group
    # and have rw-r--r-- permissions. This script is normally executed as the
    # brep user/group and thus the uploads root directory and all its
    # subdirectories must have read, write, and execute permissions granted to
    # the brep user, for example, by using ACL (see INSTALL file for
    # details). Since cp preserves the file permissions by default, these
    # files effective permissions will normally be r-- (read-only) for this
    # script. In this case rm pops up the 'remove write-protected regular
    # file' prompt by default prior to removing these files. To suppress the
    # prompt we will pass the -f option to rm.
    #
    run rm -rf "$d"

    # Remove the empty parent directories.
    #
    # Note that we iterate until the rmdir command fails, presumably because a
    # directory is not empty.
    #
    d="$(dirname "$d")"
    while [[ "$d" != "$root_dir" ]]; do
      trace_cmd rmdir "$d"
      if rmdir "$d" 2>/dev/null; then
        d="$(dirname "$d")"
      else
        break
      fi
    done
  fi

  # Close the file descriptor and unlock the root directory.
  #
  trace "+ exec {lfd}<&-"
  exec {lfd}<&-
done