path: root/tests/tar.test

#!/bin/bash

[ -f testing.sh ] && . testing.sh

#testing "name" "command" "result" "infile" "stdin"

# For reproducibility: TZ=UTC, umask 0002, override ownership and timestamp
export TZ=utc
umask 0002
TAR='tar c --owner root --group sys --mtime @1234567890'

# 255 bytes, longest VFS name
LONG=0123456789abcdef0123456789abcdef
LONG=$LONG$LONG$LONG$LONG$LONG$LONG$LONG$LONG
LONG=${LONG:1:255}

# We check both sha1sum (to ensure binary identical output) and list contents.

# Check hash of first N 512 byte frames to ensure result is binary identical.
function SUM()
{
  # Different tars add variable trailing NUL padding (1024 bytes is just
  # minimum) so look at first N 512-byte frames when analyzing header content.
  tee save.dat | head -c $(($1*512)) | sha1sum | sed "s/ .*//"
}

# List tarball contents, converting variable tabs into one space
function LST()
{
  tar tv "$@" | sed 's/[ \t][ \t]*/ /g'
}

# Check that stored empty file is binary identical and decodes as expected.
touch file
testing "store file" "$TAR file | SUM 3" \
  "2735f3a18d770dd0d7145d76108532f72bef9927\n" "" ""
testing "pass file" "$TAR file | LST" \
  "-rw-rw-r-- root/sys 0 2009-02-13 23:31 file\n" "" ""

# Two files from -T list
touch file1 file2
testing "-T newline" "$TAR -T input | LST" \
  "-rw-rw-r-- root/sys 0 2009-02-13 23:31 file1\n-rw-rw-r-- root/sys 0 2009-02-13 23:31 file2\n" "file1\nfile2\n" ""
testing "-T null" "$TAR --null -T input | LST" \
  "-rw-rw-r-- root/sys 0 2009-02-13 23:31 file1\n-rw-rw-r-- root/sys 0 2009-02-13 23:31 file2\n" "file1\0file2\0" ""

# User "root" is UID 0 and group "sys" is GID 3 (on Linux, BSD, and Mac),
# inherited from Bell Labs Unix v7

# Note: testing both "tar c" and "tar -c" here.
testing "specify UID, fetch GID" "tar -c --owner nobody:65534 --group sys --mtime @0 file | LST" \
  "-rw-rw-r-- nobody/sys 0 1970-01-01 00:00 file\n" "" ""
testing "fetch UID, specify GID" "tar c --owner root --group nobody:65534 --mtime @0 file | LST" \
  "-rw-rw-r-- root/nobody 0 1970-01-01 00:00 file\n" "" ""

# Large values switch from ascii numbers to a binary format.
testing "huge values" "tar c --owner 9999999 --group 8888888 --mtime @0 file | SUM 3" \
  "396b07fd2f80eeb312462e3bfb7dc1325dc6bcfb\n" "" ""

testcmd "longname" "tf $FILES/tar/long_path.tar" \
  "$(printf 'long file name%86cTRAILING' ' ' | tr ' ' _)\n" "" ""

touch -t 198701231234.56 file
testing "pass mtime" \
  "tar c --owner root --group sys file | LST --full-time" \
  "-rw-rw-r-- root/sys 0 1987-01-23 12:34:56 file\n" "" ""

testing "adjust mode" \
  "tar c --owner root --group sys --mode a+x file | LST --full-time" \
  "-rwxrwxr-x root/sys 0 1987-01-23 12:34:56 file\n" "" ""

mkdir dir
testing "store dir" "$TAR dir | SUM 3" \
  "85add1060cfe831ca0cdc945158efe6db485b81e\n" "" ""
testing "pass dir" "$TAR dir | LST" \
  "drwxrwxr-x root/sys 0 2009-02-13 23:31 dir/\n" "" ""

# note: does _not_ include dir entry in archive, just file
touch dir/file
testing "store file in dir" "$TAR dir/file | SUM 3" \
  "d9e7fb3884430d29e7eed0dc04a2593dd260df14\n" "" ""

# Test recursion with one file so filesystem sort order can't change result
testing "store dir and dir/file" "$TAR dir | SUM 3" \
  "a4e35f87e28c4565b60ba01dbe79e431914f8788\n" "" ""

testing "pass dir/file" "$TAR dir | LST" \
  "drwxrwxr-x root/sys 0 2009-02-13 23:31 dir/\n-rw-rw-r-- root/sys 0 2009-02-13 23:31 dir/file\n" "" ""

echo boing > dir/that
testing "tar C" "$TAR -C dir that | SUM 3" \
  "d469d4bc06def2d8808400ba30025ca295d05e4f\n" "" ""

ln dir/file dir/hardlink
testing "store hardlink" "$TAR dir/file dir/hardlink | SUM 3" \
  "519de8abd1b32debd495a0fc1d96082184abbdcc\n" "" ""

skipnot mkfifo dir/fifo 2>/dev/null
testing "create dir/fifo" "$TAR dir/fifo | SUM 3" \
  "cad477bd0fc5173d0a43f4774f514035456960e6\n" "" ""

# test L and K records

# 4+96=100 (biggest short name), 4+97=101 (shortest long name)
touch dir/${LONG:1:96} dir/${LONG:1:97}
testing "create long fname" "$TAR dir/${LONG:1:97} dir/${LONG:1:96} | SUM 3" \
  "d70018505fa5df19ae73498cfc74d0281601e42e\n" "" ""

# MacOS X has different symlink permissions, skip these tests there
[ "$(uname)" == Darwin ] && SKIP=999

  # / and .. only stripped from name, not symlink target.
  ln -s ../name.././.. dir/link
  testing "create symlink" "$TAR dir/link | SUM 3" \
    "f841bf9d757c655c5d37f30be62acb7ae24f433c\n" "" ""

  ln dir/link dir/hlink
  testing "create hardlink to symlink" "$TAR dir/link dir/hlink | SUM 3" \
    "de571a6dbf09e1485e513ad13a178b1729267452\n" "" ""

  ln -s dir/${LONG:1:96} dir/lshort
  ln -s dir/${LONG:1:97} dir/llong
  testing "create long symlink" "$TAR dir/lshort dir/llong | SUM 3" \
    "07eaf397634b5443dbf2d3ec38a4302150fcfe82\n" "" ""

  skipnot ln -s $LONG dir/${LONG:5} 2>/dev/null
  testing "create long->long" "$TAR dir/${LONG:5} | SUM 7" \
    "b9e24f53e27496c5125445230d201b4a36ff7398\n" "" ""

  # absolute and relative link names, broken and not
  ln -s file dir/linkok
  testing "create symlink" "$TAR dir/linkok | SUM 3" \
    "f5669cfd179ddcdd5ca9f8a1561a99e11e0a08b1\n" "" ""

SKIP=0 # End of tests that don't match MacOS symlink permissions

symlink_perms=lrwxrwxrwx
[ "$(uname)" == "Darwin" ] && symlink_perms=lrwxrwxr-x

ln -s /dev/null dir/linknull
testing "pass absolute symlink" "$TAR dir/linknull | LST" \
  "$symlink_perms root/sys 0 2009-02-13 23:31 dir/linknull -> /dev/null\n" "" ""

ln -s rel/broken dir/relbrok
testing "pass broken symlink" "$TAR dir/relbrok | LST" \
  "$symlink_perms root/sys 0 2009-02-13 23:31 dir/relbrok -> rel/broken\n" "" ""

ln -s /does/not/exist dir/linkabsbrok
testing "pass broken absolute symlink" "$TAR dir/linkabsbrok | LST" \
  "$symlink_perms root/sys 0 2009-02-13 23:31 dir/linkabsbrok -> /does/not/exist\n" \
  "" ""

nulldev=1,3 # devtmpfs values
[ "$(uname)" == "Darwin" ] && nulldev=3,2

testing "pass /dev/null" \
  "tar c --mtime @0 --group sys /dev/null 2>/dev/null | LST" \
  "crw-rw-rw- root/sys $nulldev 1970-01-01 00:00 dev/null\n" "" ""
testing "--absolute-names" \
  "tar c --mtime @0 --group sys --absolute-names /dev/null 2>/dev/null | LST" \
  "crw-rw-rw- root/sys $nulldev 1970-01-01 00:00 /dev/null\n" "" ""

# compression types
testing "autodetect gzip" 'LST -f "$FILES"/tar/tar.tgz' \
  "drwxr-x--- enh/eng 0 2017-05-13 01:05 dir/\n-rw-r----- enh/eng 12 2017-05-13 01:05 dir/file\n" \
  "" ""

testing "manually specify bz2" 'LST -jf "$FILES"/tar/tar.tbz2' \
  "drwxr-x--- enh/eng 0 2017-05-13 01:05 dir/\n-rw-r----- enh/eng 12 2017-05-13 01:05 dir/file\n" \
  "" ""

# -I
testing "-I gzip c" "$TAR -Igzip file | file - | grep -o 'gzip compressed'" \
  "gzip compressed\n" "" ""
testing "-I gzip t" 'LST -Igzip -f "$FILES"/tar/tar.tgz' \
  "drwxr-x--- enh/eng 0 2017-05-13 01:05 dir/\n-rw-r----- enh/eng 12 2017-05-13 01:05 dir/file\n" \
  "" ""

skipnot mknod -m 660 dir/char c 12 34 2>/dev/null && chgrp sys dir/char
NOSPACE=1 testing "character special" "tar --mtime @0 -cf test.tar dir/char && rm -f dir/char && tar xf test.tar && ls -l --full-time dir/char" \
  "crw-rw---- 1 root sys 12, 34 1970-01-01 00:00:00.000000000 +0000 dir/char\n"\
  "" ""

skipnot mknod -m 660 dir/block b 23 45 2>/dev/null && chgrp sys dir/block
NOSPACE=1 testing "block special" "tar --mtime @0 -cf test.tar dir/block && rm -f dir/block && tar xf test.tar && ls -l --full-time dir/block" \
  "brw-rw---- 1 root sys 23, 45 1970-01-01 00:00:00.000000000 +0000 dir/block\n"\
  "" ""

skipnot chown nobody:nogroup dir/file 2>/dev/null
testing "ownership" "$TAR dir/file | SUM 3" \
  "2d7b96c7025987215f5a41f10eaa84311160afdb\n" "" ""

mkdir -p dd/sub/blah &&
tar cf test.tar dd/sub/blah &&
rm -rf dd/sub &&
skipnot ln -s ../.. dd/sub
toyonly testing "symlink out of cwd" \
  "tar xf test.tar 2> /dev/null || echo yes ; [ ! -e dd/sub/blah ] && echo yes" \
  "yes\nyes\n" "" ""

# If not root can't preserve ownership, so don't try yet.

testing "extract dir/file from tar" \
  "tar xvCf dd $FILES/tar/tar.tar && stat -c '%A %Y %n' dd/dir dd/dir/file" \
  "dir/\ndir/file\ndrwxr-x--- 1494637555 dd/dir\n-rw-r----- 1494637555 dd/dir/file\n" \
  "" ""

testing "extract dir/file from tgz (autodetect)" \
  "tar xvCf dd $FILES/tar/tar.tgz && stat -c '%A %Y %n' dd/dir dd/dir/file" \
  "dir/\ndir/file\ndrwxr-x--- 1494637555 dd/dir\n-rw-r----- 1494637555 dd/dir/file\n" \
  "" ""

toyonly testing "cat tgz | extract dir/file (autodetect)" \
  "cat $FILES/tar/tar.tgz | tar xvC dd && stat -c '%A %Y %n' dd/dir dd/dir/file" \
  "dir/\ndir/file\ndrwxr-x--- 1494637555 dd/dir\n-rw-r----- 1494637555 dd/dir/file\n" \
  "" ""

testing "extract dir/file from tbz2 (autodetect)" \
  "tar xvCf dd $FILES/tar/tar.tbz2 && stat -c '%A %Y %n' dd/dir dd/dir/file" \
  "dir/\ndir/file\ndrwxr-x--- 1494637555 dd/dir\n-rw-r----- 1494637555 dd/dir/file\n" \
  "" ""

toyonly testing "cat tbz | extract dir/file (autodetect)" \
  "cat $FILES/tar/tar.tbz2 | tar xvC dd && stat -c '%A %Y %n' dd/dir dd/dir/file" \
  "dir/\ndir/file\ndrwxr-x--- 1494637555 dd/dir\n-rw-r----- 1494637555 dd/dir/file\n" \
  "" ""

mkdir path && ln -s "$(which gzip)" "$(which tar)" path/ && [ -x path/gzip ] ||
  ((++SKIP))
toyonly testing "autodetect falls back to gzip -d when no zcat" \
  "PATH=path; tar tf $FILES/tar/tar.tgz" "dir/\ndir/file\n" "" ""
rm -rf path

# Only run sparse tests if filesystem can handle sparse files @4k granularity
dd if=/dev/zero bs=4k count=1 seek=1 of=blah.img 2>/dev/null
[ $(du blah.img | sed 's/[ \t].*//') -ne 4 ] && SKIP=999
rm -f blah.img
[ "$(uname)" == "Darwin" ] && SKIP=999

  yes | head -n $((1<<18)) > bang
  {
    dd bs=$((1<<16)) count=1 status=none
    dd bs=8192 seek=14 count=1 status=none
    dd bs=4096 seek=64 count=5 status=none
  } < bang > fweep
  testing "sparse without overflow" "$TAR --sparse fweep | SUM 3" \
    "50dc56c3c7eed163f0f37c0cfc2562852a612ad0\n" "" ""
  rm bang fweep

  for i in 1 3 5 7 9 14 27 36 128 256 300 304
  do
    dd if=/dev/zero of=fweep bs=65536 seek=$i count=1 2>/dev/null
  done
  testing "sparse single overflow" "$TAR --sparse fweep | SUM 6" \
    "81d59c3a7470201f92d60e63a43318ddde893f6d\n" "" ""
  rm fweep

  for i in $(seq 8 3 200)
  do
    dd if=/dev/zero of=fweep bs=65536 seek=$i count=1 2>/dev/null
    dd if=/dev/zero of=fweep2 bs=65536 seek=$i count=1 2>/dev/null
  done
  truncate -s 20m fweep2
  testing "sparse double overflow" "$TAR --sparse fweep | SUM 7" \
    "024aacd955e45f89bafedb3f37c8d39b4d556471\n" "" ""

  tar c --sparse fweep > fweep.tar
  rm fweep
  testing "sparse extract" "tar xf fweep.tar && $TAR --sparse fweep | SUM 4" \
    "b949d3a3b4c6457c873f1ea9918fd9029c5ed4b3\n" "" ""
  testing "sparse tvf" \
    "tar tvf fweep.tar | grep -wq 13172736 && echo right size" "right size\n" \
    "" ""
  rm fweep fweep.tar

  tar c --sparse fweep2 > fweep2.tar
  rm fweep2
  testing "sparse extract hole at end" \
    "tar xf fweep2.tar && $TAR --sparse fweep2 | SUM 4" \
    "807664bcad0e827793318ff742991d6f006b2127\n" "" ""
  rm fweep2 fweep2.tar

SKIP=0 # End of sparse tests

mkdir -p links
touch links/orig
ln links/{orig,link1}
ln links/{orig,link2}
testcmd 'links' '-cf test.tar links' '' '' ''
rm -rf links

mkdir links
for i in {0..12}; do touch links/orig$i; ln links/{orig,link}$i; done
testcmd 'links2' '-cf test.tar links' '' '' ''
rm -rf links

install -m 000 -d folder/skip/oof &&
testcmd 'exclude' '--exclude skip -cvf tar.tar folder && echo yes' \
  'folder/\nyes\n' '' ''
rm -rf folder tar.tar

mkdir -p one/two; echo hello > one/two/three; tar czf test.tar one/two/three
rm one/two/three; mkdir one/two/three
testcmd 'replace dir with file' '-xf test.tar && cat one/two/three' \
  'hello\n' '' ''
rm -rf one test.tar

mkdir ..dotsdir
testing "create ..dotsdir" "$TAR ..dotsdir | SUM 3" \
  "62ff23c9b427020331992b9bc71f082099c1411f\n" "" ""

testing "pass ..dotsdir" "$TAR ..dotsdir | LST" \
  "drwxrwxr-x root/sys 0 2009-02-13 23:31 ..dotsdir/\n" "" ""
rmdir ..dotsdir

mkdir -p one/two/three/four/five
touch one/two/three/four/five/six
testing "--strip" "$TAR one | tar t --strip=2 --show-transformed | grep six" \
  "three/four/five/six\n" "" ""

# toybox tar --xform depends on toybox sed
[ -z "$TEST_HOST" ] && ! sed --tarxform '' </dev/null 2>/dev/null && SKIP=99

mkdir uno
ln -s tres uno/dos
touch uno/tres
ln uno/tres uno/quatro
LL() { LST --show-transformed-names $XX | sed 's/^.* 23:31 //'; }
TT() { $TAR --no-recursion uno uno/{dos,tres,quatro} "$@" | LL; }
testing 'xform S' \
  "TT --xform 's/uno/one/S;s/dos/two/S;s/tres/three/S;s/quatro/four/S'" \
  "one/\none/two -> tres\none/three\none/four link to one/three\n" "" ""

testing 'xform flags=rh starts with all disabled' \
  "TT --xform 's/uno/one/;flags=rh;s/dos/two/;s/tres/three/;s/quatro/four/'" \
  "one/\none/two -> tres\none/three\none/four link to one/three\n" "" ""

testing 'xform flags=rHhsS toggles' \
  "TT --xform 's/uno/one/;flags=rHhsS;s/dos/two/;s/tres/three/;s/quatro/four/'"\
  "one/\none/two -> tres\none/three\none/four link to one/three\n" "" ""

testing 'xform flags= is not a delta from previous' \
  "TT --xform 'flags=s;flags=rh;s/uno/one/;s/dos/two/;s/tres/three/;s/quatro/four/'" \
  "one/\none/two -> tres\none/three\none/four link to one/three\n" "" ""

testing 'xform H' \
  "TT --xform 'flags=rsH;s/uno/one/;s/dos/two/;s/tres/three/;s/quatro/four/'" \
  "one/\none/two -> three\none/three\none/four link to uno/tres\n" "" ""

testing 'xform R' \
  "TT --xform 'flags=rshR;s/uno/one/;s/dos/two/;s/tres/three/;s/quatro/four/'" \
  "uno/\nuno/dos -> three\nuno/tres\nuno/quatro link to one/three\n" "" ""

testing "xform path" "$TAR one --xform=s@three/four/@zero@ | tar t | grep six" \
  "one/two/zerofive/six\n" "" ""

testing "xform trailing slash special case" \
  "$TAR --xform 's#^.+/##x' one/two/three/four/five | tar t" 'five/\nsix\n' '' ''

# The quoting works because default IFS splits on whitepace not ;
testing "xform extract all" \
  "XX='--xform s/uno/one/;s/dos/two/;s/tres/three/;s/quatro/four/' TT" \
  'one/\none/two -> three\none/three\none/four link to one/three\n' '' ''

testing 'xform extract S' \
  "XX='--xform s/uno/one/S;s/dos/two/S;s/tres/three/S;s/quatro/four/S' TT" \
  "one/\none/two -> tres\none/three\none/four link to one/three\n" "" ""

testing 'xform extract H' \
  "XX='--xform flags=rs;s/uno/one/;s/dos/two/;s/tres/three/;s/quatro/four/' TT"\
  "one/\none/two -> three\none/three\none/four link to uno/tres\n" "" ""

testing 'xform extract R' \
  "XX='--xform flags=sh;s/uno/one/;s/dos/two/;s/tres/three/;s/quatro/four/' TT"\
  "uno/\nuno/dos -> three\nuno/tres\nuno/quatro link to one/three\n" "" ""

rm -rf uno
SKIP=0
rm -rf one

testing '-P' "$TAR -P --no-recursion -C / /// .. | SUM 3" \
  "a3e94211582da121845d823374d8f41ead62d7bd\n" "" ""

testing 'without -P' "$TAR --no-recursion -C / /// .. 2>/dev/null | SUM 3" \
  "077d03243e247b074806904885e6da272fd5857a\n" "" ""

# Wildcards: --exclude, include (create/extract * cmdline/recursive)
# --anchored, --wildcards, --wildcards-match-slash
# --no-* versions of each. Span coverage, switching on/off...

#pattern a.c
#  abcd dabc a/c da/c
#  top/*

mkdir sub && cd sub && mkdir -p a da top/a top/da &&
touch abcd dabc a/c da/c top/abcd top/dabc top/a/c top/da/c &&
$TAR -f ../sub.tar abcd dabc a da top && cd .. || exit 1

# TODO I have not made wildcard state changes positional.

testing 'wildcards do not affect creation cmdline args' \
  '$TAR -C sub --wildcards a.cd abcd dabc a da top 2>/dev/null | cmp - sub.tar' \
  '' '' ''

testing 'creation --exclude --no-wildcards'\
  '$TAR -C sub --no-wildcards --exclude=d?bc abcd dabc | LL' \
  'abcd\ndabc\n' '' ''


testing 'creation --wildcards --exclude'\
  '$TAR -C sub --wildcards --exclude=d?bc abcd dabc | LL' \
  'abcd\n' '' ''

# TODO: do we need to set DIRTREE_BREADTH at top level? Come up with test if so.
mkdir sub2
touch sub2/{ephebe,klatch,djelibeybi}
testing 'tsort' '$TAR -c sub2 --sort=name | tar t' \
  'sub2/\nsub2/djelibeybi\nsub2/ephebe\nsub2/klatch\n' '' ''

touch file
testing './file bug' 'tar c ./file > tar.tar && tar t ./file < tar.tar' \
  './file\n' '' ''

if false
then
# Sequencing issues that leak implementation details out the interface
testing "what order are --xform, --strip, and --exclude processed in?"
testing "--xform vs ../ removal and adding / to dirs"

chmod 700 dir
tar cpf tar.tgz dir/file
#chmod 700 dir
#tar xpf file
#ls -ld dir/file

# restore ownership of file, dir, and symlink

# merge add_to_tar and write_longname,
# filter, incl or excl and anchored/wildcards

# extract file not under cwd
# exclusion defaults to --no-anchored and --wildcards-match-slash
#  both incl and excl

# catch symlink overwrite
# add dir with no trailing slash
# don't allow hardlink target outside cwd
# extract dir/file without dir in tarball
# create with and without each flag
# --owner --group --numeric-owner
# extract with and without each flag
# --owner 0 --group 0
# set symlink owner
# >256 hardlink inodes
#   // remove leading / and any .. entries from saved name
#  // exclusion defaults to --no-anchored and --wildcards-match-slash
# //bork blah../thing blah/../thing blah/../and/../that blah/.. ../blah
# tar tv --owner --group --mtime
# extract file within dir date correct
# name ending in /.. or just ".." as a name


fi

rm -f save.dat