Ticket #1952: untar.txt

#! /bin/sh
#
# by dw 2013-11-12
#
# relies on tar to autodetect the used compression method (gz, bz2, xz, ...)
# supported formats (posix, ...) depend on the found version of tar
# due to proprietory options works only with GNU tar (1.26-1.29)
# list and copyout working
# rm, mkdir, rmdir partly working (only uncompressed tar archives)
# copyin works partly (only uncompressed tar archives, archive rootdir)
# besides modifying the tar file can cause problems (duplicate files, ...)
# and should be avoided
#
# by dw 2017-04-09
# experimental support for bsdtar added, problems
#         - timestamps require LANG=C for english month names
#         - with LANG=C can't handle filenames containing utf-8
# experimental support for star added
#
# by dw 2018-06-03
# star 1.5.3 from openSUSE Leap 15.0 supports auto detection of xz
# date format conversion added, list and copyout working with star
# error messages with GNU tar created archives containing 
# special character filenames
# => GNU tar is the default, star is first fallback

TAR=$( which tar 2>/dev/null ) #  || exit 1
if [ -z "$TAR" ]; then
  TAR=$( which star 2>/dev/null ) # || exit 1
fi
if [ -z "$TAR" ]; then
  TAR=$( which bsdtar 2>/dev/null ) || exit 1
fi

mctarfs_list ()
{
  # link count is hardcoded to 1
  # can't use $6 for the filename in case of blanks or other separators
  # mc expects: AAAAAAA NNN OOOOOOOO GGGGGGGG SSSSSSSS DATETIME [PATH/]FILENAME [-> [PATH/]FILENAME[/]]]
  # gawk strips blanks => trailing blanks of filenames are lost
  # filenames may contain blanks - field splitting does not work for them

  case "$TAR" in
    */tar)
      # GNU tar
      # output of GNU tar 1.26 for tar tvf:
      # drwxr-xr-x user/group      0 2013-08-22 20:27 directory/
      # GNU tar timestamps always contain ":" and can be used as anchor

      $TAR tvf "$1" | gawk '
      BEGIN { FS="" }
      {
        split($0,b," ")
        sub(/\//, " ", b[2])
        split(b[4], DATE, "-")
        name_offs = index($0, b[5])+6
        printf "%s 1 %s %s %s-%s-%s %s ./%s\n", b[1], b[2], b[3], DATE[2], DATE[3], DATE[1], b[5], substr($0, name_offs)
      }'
      ;;
    */bsdtar)
      # BSD tar
      # bsdtar uses LANG for the date format
      # setting LANG to en_US.UTF-8 to get a specific date format breaks handling of non-utf8 archives
      # output of bsdtar-3.1.2 tar tvf:
      # drwxr-xr-x  0 user   group       0 Oct  5  2016 directory/
      # -rw-r--r--  0 user   group   68981 Oct  1  2016 directory/file
      # drwxr-xr-x  0 user   group       0 Dec 13 20:59 directory/
      # -rw-r--r--  0 user   group    2013 Sep 28  2013 dateformat.tst
      # the date/timestamp format is supported by mc BUT do not always contain ":", "YYYY" or "hh:mm"
      # if b[8] =~ ":" this should always be the first match on the line, use as anchor
      # if b[8] =~ "YYYY" this could match the filesize => name_offs has to be bigger than parts
      LANG_BAK="$LANG"
      export LANG=en_US.UTF-8

      $TAR tvf "$1" | gawk '
      BEGIN { FS="" }
      {
        split($0,b," ")
        parts = length(b[1] b[2] b[3] b[4] b[5] b[6] b[7]) + 7
        name_offs = index($0, " "b[8]" ") + 6
        if (index(b[8], ":") != 0)
          name_offs += 1
        else if (name_offs < parts)
        {
          new_part = substr($0, parts)
          name_offs = index(new_part, " "b[8]" ") + 5 + parts
        }
        printf "%s 1 %s %s %s %s %s %s ./%s\n", b[1], b[3], b[4], b[5], b[6], b[7], b[8], substr($0, name_offs)
      }'
      LANG="$LANG_BAK"
      ;;
    */star)
      # output of star 1.5.3 tar tvf:
      #      5 -rw-r--r--  user/group Sep 28 13:16 2013 dateformat.tst
      #      0 drwxr-xr-x  user/group Aug 22 20:36 2013 tartest_dir_level_1/
      # date conversion to MM-DD-YYYY hh:mm[:ss]
      # b[1] size
      # b[2] permissions
      # b[3] user group
      # b[4] MMM
      # b[5] DD
      # b[6] hh:mm
      # b[7] YYYY

      $TAR tvf "$1" 2> /dev/null | gawk '
      BEGIN {
         FS=""
         # Copied from uarc/uzoo
         split("Jan:Feb:Mar:Apr:May:Jun:Jul:Aug:Sep:Oct:Nov:Dec", month_list, ":")
         for (i=1; i<=12; i++) {
            month[month_list[i]] = i
         }
      }
      {
        split($0,b," ")
        sub(/\//, " ", b[3])
        split(b[4], DATE, "-")
        name_offs = index($0, b[6]) + 11
        printf "%s 1 %s %s %02d-%02d-%04d %s ./%s\n", b[2], b[3], b[1], month[b[4]], b[5], b[7], b[6], substr($0, name_offs)
      }'
      ;;
    *)
      echo "unsupported tar implementation"
      exit 1
    ;;
  esac
}

mctarfs_copyin ()
{
  $TAR rf "$1" "$2" >/dev/null
}

mctarfs_copyout ()
{
  case "$TAR" in
    */star)
      $TAR x -to-stdout -f "$1" "$2" > "$3" 2> /dev/null || $TAR x -to-stdout -f "$1" "./$2" > "$3"
      ;;
    *)
      # GNU tar, bsdtar
      $TAR xOf "$1" "$2" > "$3" 2> /dev/null || $TAR xOf "$1" "./$2" > "$3"
      # file in archive begins with ./ , e.g. from find generated file list
      # $TAR xOf "$1" "./$2" > "$3"
    ;;
  esac
}

mctarfs_mkdir ()
{
  pwd="$( pwd )"
  dir=$( mktemp -d "${MC_TMPDIR:-/tmp}/mctmpdir-utar.XXXXXX" ) || exit 1
  cd "$dir" || exit 1
  mkdir -p "$2"
  $TAR rf "$1" "$2" >/dev/null
  cd "$pwd" || exit 1
  rm -rf "$dir"
}

mctarfs_rm ()
{
  $TAR --delete -f "$1" "$2" >/dev/null
}

# override any locale for dates
LC_DATE=C
export LC_DATE

umask 077

if [ -z "$TAR" ]; then
  echo "Error: could not find tar." >&2
  exit 1
fi

cmd="$1"
shift

case "$cmd" in
  # Workaround for a bug in mc - directories must precede files to
  # avoid duplicate entries, so we sort output by "permission string" drwxr-xr-x
  list)    mctarfs_list    "$@" | sort -k 8 ;;
  rm)      mctarfs_rm      "$@" ;;
  rmdir)   mctarfs_rm      "$@" ;;
  mkdir)   mctarfs_mkdir   "$@" ;;
  copyin)  mctarfs_copyin  "$@" ;;
  copyout) mctarfs_copyout "$@" ;;
  *) exit 1 ;;
esac
exit 0