#!/bin/sh
# Copyright 2009 by Denys Vlasenko
# Licensed under GPLv2, see file LICENSE in this source tree.

. ./testing.sh

unset LANG
unset LANGUAGE
unset LC_COLLATE
unset LC_ALL
umask 022

rm -rf tar.tempdir 2>/dev/null
mkdir tar.tempdir && cd tar.tempdir || exit 1

# testing "test name" "script" "expected result" "file input" "stdin"

testing "Empty file is not a tarball" '\
tar xvf - 2>&1; echo $?
' "\
tar: short read
1
" \
"" ""
SKIP=

optional FEATURE_SEAMLESS_GZ
# In NOMMU case, "invalid magic" message comes from gunzip child process.
# Otherwise, it comes from tar.
# Need to fix output up to avoid false positive.
testing "Empty file is not a tarball.tar.gz" '\
{ tar xvzf - 2>&1; echo $?; } | grep -Fv "invalid magic"
' "\
tar: short read
1
" \
"" ""
SKIP=

testing "Two zeroed blocks is a ('truncated') empty tarball" '\
dd if=/dev/zero bs=512 count=2 2>/dev/null | tar xvf - 2>&1; echo $?
' "\
0
" \
"" ""
SKIP=

testing "Twenty zeroed blocks is an empty tarball" '\
dd if=/dev/zero bs=512 count=20 2>/dev/null | tar xvf - 2>&1; echo $?
' "\
0
" \
"" ""
SKIP=

# "tar cf test.tar input input_dir/ input_hard1 input_hard2 input_hard1 input_dir/ input":
# GNU tar 1.26 records as hardlinks:
#  input_hard2 -> input_hard1
#  input_hard1 -> input_hard1 (!!!)
#  input_dir/file -> input_dir/file
#  input -> input
# As of 1.24.0, we don't record last two: for them, nlink==1
# and we check for "hardlink"ness only files with nlink!=1
# We also don't use "hrw-r--r--" notation for hardlinks in "tar tv" listing.
optional FEATURE_TAR_CREATE FEATURE_LS_SORTFILES
testing "tar hardlinks and repeated files" '\
rm -rf input_* test.tar 2>/dev/null
>input_hard1
ln input_hard1 input_hard2
mkdir input_dir
>input_dir/file
chmod -R 644 *
chmod    755 input_dir
tar cf test.tar input input_dir/ input_hard1 input_hard2 input_hard1 input_dir/ input
tar tvf test.tar | sed "s/.*[0-9] input/input/"
rm -rf input_dir
tar xf test.tar 2>&1
echo Ok: $?
ls -l . input_dir/* | grep input_ | sed "s/\\(^[^ ]*\\) .* input/\\1 input/"
' "\
input
input_dir/
input_dir/file
input_hard1
input_hard2 -> input_hard1
input_hard1 -> input_hard1
input_dir/
input_dir/file
input
Ok: 0
-rw-r--r-- input_dir/file
drwxr-xr-x input_dir
-rw-r--r-- input_hard1
-rw-r--r-- input_hard2
" \
"" ""
SKIP=

optional FEATURE_TAR_CREATE FEATURE_LS_SORTFILES
testing "tar hardlinks mode" '\
rm -rf input_* test.tar 2>/dev/null
>input_hard1
chmod 741 input_hard1
ln input_hard1 input_hard2
mkdir input_dir
ln input_hard1 input_dir
ln input_hard2 input_dir
chmod 550 input_dir
# On some filesystems, input_dir/input_hard2 is returned by readdir
# BEFORE input_dir/input_hard1! Thats why we cant just "tar cf ... input_*":
tar cf test.tar input_dir/input_hard* input_hard*
tar tvf test.tar | sed "s/.*[0-9] input/input/"
chmod 770 input_dir
rm -rf input_*
tar xf test.tar 2>&1
echo Ok: $?
ls -l . input_dir/* | grep "input.*hard" | sed "s/\\(^[^ ]*\\) .* input/\\1 input/"
' "\
input_dir/input_hard1
input_dir/input_hard2 -> input_dir/input_hard1
input_hard1 -> input_dir/input_hard1
input_hard2 -> input_dir/input_hard1
Ok: 0
-rwxr----x input_dir/input_hard1
-rwxr----x input_dir/input_hard2
-rwxr----x input_hard1
-rwxr----x input_hard2
" \
"" ""
SKIP=

optional FEATURE_TAR_CREATE FEATURE_LS_SORTFILES
testing "tar symlinks mode" '\
rm -rf input_* test.tar 2>/dev/null
>input_file
chmod 741 input_file
ln -s input_file input_soft
mkdir input_dir
ln input_file input_dir
ln input_soft input_dir
chmod 550 input_dir
tar cf test.tar input_dir/* input_[fs]*
tar tvf test.tar | sed "s/.*[0-9] input/input/" | sort
chmod 770 input_dir
rm -rf input_*
tar xf test.tar 2>&1
echo Ok: $?
ls -l . input_dir/* | grep "input_[fs]" | sed "s/\\(^[^ ]*\\) .* input/\\1 input/"
' "\
input_dir/input_file
input_dir/input_soft -> input_file
input_file -> input_dir/input_file
input_soft -> input_dir/input_soft
Ok: 0
-rwxr----x input_dir/input_file
lrwxrwxrwx input_file
-rwxr----x input_file
lrwxrwxrwx input_file
" \
"" ""
SKIP=

optional FEATURE_TAR_CREATE FEATURE_TAR_LONG_OPTIONS
testing "tar --overwrite" "\
rm -rf input_* test.tar 2>/dev/null
ln input input_hard
tar cf test.tar input_hard
echo WRONG >input
# --overwrite opens 'input_hard' without unlinking,
# thus 'input_hard' still linked to 'input' and we write 'Ok' into it
tar xf test.tar --overwrite 2>&1 && cat input
" "\
Ok
" \
"Ok\n" ""
SKIP=

test x"$SKIP_KNOWN_BUGS" = x"" && {
# Needs to be run under non-root for meaningful test
optional FEATURE_TAR_CREATE
testing "tar writing into read-only dir" '\
rm -rf input_* test.tar 2>/dev/null
mkdir input_dir
>input_dir/input_file
chmod 550 input_dir
tar cf test.tar input_dir
tar tvf test.tar | sed "s/.*[0-9] input/input/"
chmod 770 input_dir
rm -rf input_*
tar xf test.tar 2>&1
echo Ok: $?
ls -l input_dir/* . | grep input_ | sed "s/\\(^[^ ]*\\) .* input/\\1 input/"
chmod 770 input_dir
' "\
input_dir/
input_dir/input_file
Ok: 0
-rw-r--r-- input_dir/input_file
dr-xr-x--- input_dir
" \
"" ""
SKIP=
}

# Had a bug where on extract autodetect first "switched off" -z
# and then failed to recognize .tgz extension
optional FEATURE_TAR_CREATE FEATURE_SEAMLESS_GZ
testing "tar extract tgz" "\
dd count=1 bs=1M if=/dev/zero of=F0 2>/dev/null
tar -czf F0.tgz F0
rm F0
tar -xzvf F0.tgz && echo Ok
rm F0 || echo BAD
" "\
F0
Ok
" \
"" ""
SKIP=

# Do we detect XZ-compressed data (even w/o .tar.xz or txz extension)?
# (the uuencoded hello_world.txz contains one empty file named "hello_world")
optional UUDECODE FEATURE_TAR_AUTODETECT FEATURE_SEAMLESS_XZ
testing "tar extract txz" "\
uudecode -o input && tar tf input && echo Ok
" "\
hello_world
Ok
" \
"" "\
begin-base64 644 hello_world.txz
/Td6WFoAAATm1rRGAgAhARYAAAB0L+Wj4AX/AEldADQZSe6ODIZQ3rSQ8kAJ
SnMPTX+XWGKW3Yu/Rwqg4Ik5wqgQKgVH97J8yA8IvZ4ahaCQogUNHRkXibr2
Q615wcb2G7fJU49AhWAAAAAAUA8gu9DyXfAAAWWADAAAAB5FXGCxxGf7AgAA
AAAEWVo=
====
"
SKIP=

# On extract, everything up to and including last ".." component is stripped
optional FEATURE_TAR_CREATE
testing "tar strips /../ on extract" "\
rm -rf input_* test.tar 2>/dev/null
mkdir input_dir
echo Ok >input_dir/file
tar cf test.tar ./../tar.tempdir/input_dir/../input_dir 2>&1
rm -rf input_* 2>/dev/null
tar -vxf test.tar 2>&1
cat input_dir/file 2>&1
" "\
tar: removing leading './../tar.tempdir/input_dir/../' from member names
input_dir/
input_dir/file
Ok
" \
"" ""
SKIP=

# attack.tar.bz2 has symlink pointing to a system file
# followed by a regular file with the same name
# containing "root::0:0::/root:/bin/sh":
#  lrwxrwxrwx root/root passwd -> /tmp/passwd
#  -rw-r--r-- root/root passwd
# naive tar implementation may end up creating the symlink
# and then writing into it.
# The correct implementation unlinks target before
# creating the second file.
# We test that /tmp/passwd remains empty:
optional UUDECODE FEATURE_SEAMLESS_BZ2
testing "tar does not extract into symlinks" "\
>>/tmp/passwd && uudecode -o input && tar xf input 2>&1 && rm passwd; cat /tmp/passwd; echo \$?
" "\
0
" \
"" "\
begin-base64 644 attack.tar.bz2
QlpoOTFBWSZTWRVn/bIAAKt7hMqwAEBAAP2QAhB0Y96AAACACCAAlISgpqe0
po0DIaDynqAkpDRP1ANAhiYNSPR8VchKhAz0AK59+DA6FcMKBggOARIJdVHL
DGllrjs20ATUgR1HmccBX3EhoMnpMJaNyggmxgLDMz54lBnBTJO/1L1lbMS4
l4/V8LDoe90yiWJhOJvIypgEfxdyRThQkBVn/bI=
====
"
SKIP=
# And same with -k
optional UUDECODE FEATURE_SEAMLESS_BZ2
testing "tar -k does not extract into symlinks" "\
>>/tmp/passwd && uudecode -o input && tar xf input -k 2>&1 && rm passwd; cat /tmp/passwd; echo \$?
" "\
tar: can't open 'passwd': File exists
0
" \
"" "\
begin-base64 644 attack.tar.bz2
QlpoOTFBWSZTWRVn/bIAAKt7hMqwAEBAAP2QAhB0Y96AAACACCAAlISgpqe0
po0DIaDynqAkpDRP1ANAhiYNSPR8VchKhAz0AK59+DA6FcMKBggOARIJdVHL
DGllrjs20ATUgR1HmccBX3EhoMnpMJaNyggmxgLDMz54lBnBTJO/1L1lbMS4
l4/V8LDoe90yiWJhOJvIypgEfxdyRThQkBVn/bI=
====
"
SKIP=


cd .. && rm -rf tar.tempdir || exit 1

exit $FAILCOUNT