tests/file.test

#!/bin/bash

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

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

test_line() {
  ONE="$1"; shift; TWO="\"$FILES\"/$1 | sed 's/^.*: //'"; shift
  testcmd "$ONE" "$TWO" "$@"
}

ln -s "$FILES"/java.class symlink
LINK=$(readlink symlink)
ln -s "$FILES"/java.klass dangler
BROKEN=$(readlink dangler)

mkdir directory
testing "directory" "file ." ".: directory\n" "" ""
rmdir directory
touch empty
testing "empty" "file empty" "empty: empty\n" "" ""
rm -f empty

testing "script" "file input | grep -o ' script'" " script\n" "#!/bin/bash\n" ""
testing "script with spaces" "file input | grep -o ' script'" " script\n" \
  "#!  /bin/bash\n" ""
testing "env script" "file input | egrep -o '(python|script)' | sort" \
  "python\nscript\n" "#! /usr/bin/env python\n" ""
testing "ascii" "file input" "input: ASCII text\n" "Hello, world!\n" ""
testing "utf-8" \
  "file \"$FILES\"/utf8/japan.txt | egrep -o '(UTF-8|text)' | LANG=c sort" \
  "UTF-8\ntext\n" "" ""

# TODO each of these has multiple options we could test
testing "java class" \
  "file \"$FILES\"/java.class | egrep -o '(Java class|version 53.0)'"\
  "Java class\nversion 53.0\n" "" ""

echo "cafebabe000000020100000700000003000040000000d9300000000e0100000c8000000200014000000098500000000e" | xxd -r -p > universal
testcmd "mach-o universal" "universal" \
  "universal: Mach-O universal binary with 2 architectures: [x86_64] [arm64]\n" "" ""
rm universal

test_line "tar file" "tar/tar.tar" "POSIX tar archive (GNU)\n" "" ""
testing "gzip data" "file \"$FILES\"/tar/tar.tgz | grep -o 'gzip compressed data'" \
    "gzip compressed data\n" "" ""
test_line "bzip2 data" "tar/tar.tbz2" \
  "bzip2 compressed data, block size = 900k\n" "" ""
test_line "7z file" "tar/tar.7z" "7-zip archive data, version 0.4\n" "" ""
testing "zip file" \
  "file $FILES/zip/example.zip | egrep -o '(Zip archive data|at least v1.0 to extract)'" \
    "Zip archive data\nat least v1.0 to extract\n" "" ""

echo R0lGODlhIAAgAMZHAAAAABYWFiYmJioqKi4uLjIy | base64 -d > gif
testing "gif file" "file gif" "gif: GIF image data, version 89a, 32 x 32\n" "" ""
rm -f gif

# TODO: check in a genuine minimal .dex

echo "6465780a3033350038ca8f6ce910f94e" | xxd -r -p > android.dex
testing "Android .dex" "file android.dex | egrep -o '(dex file|version 035)'" \
  "dex file\nversion 035\n" "" ""
rm -f android.dex

# These actually test a lot of the ELF code: 32-/64-bit, arm/arm64, PT_INTERP,
# the two kinds of NDK ELF note, BuildID, and stripped/not stripped.
toyonly test_line "Android NDK full ELF note" "elf/ndk-elf-note-full" \
    "ELF shared object, 64-bit LSB arm64, dynamic (/system/bin/linker64), for Android 24, built by NDK r19b (5304403), BuildID=0c712b8af424d57041b85326f0000fadad38ee0a, not stripped\n" "" ""
toyonly test_line "Android NDK short ELF note" "elf/ndk-elf-note-short" \
    "ELF shared object, 32-bit LSB arm, EABI5, soft float, dynamic (/system/bin/linker), for Android 28, BuildID=da6a5f4ca8da163b9339326e626d8a3c, stripped\n" "" ""
toyonly test_line "ELF static fdpic" "elf/fdstatic" \
    "ELF executable (fdpic), 32-bit MSB sh, static, stripped\n" "" ""

testing "broken symlink" "file dangler" "dangler: broken symbolic link to $BROKEN\n" "" ""
testing "symlink" "file symlink" "symlink: symbolic link to $LINK\n" "" ""
testing "symlink -h" "file -h symlink" "symlink: symbolic link to $LINK\n" "" ""
testing "symlink -L" \
  "file -L symlink | egrep -o '(symlink:|Java class|version 53.0)'" \
  "symlink:\nJava class\nversion 53.0\n" "" ""

# Some host versions say "-" some "/dev/stdin"...
testing "- pipe" "cat $FILES/java.class | file - | egrep -o '(Java class|version 53.0)'" \
  "Java class\nversion 53.0\n" "" ""
testing "- redirect" \
  "file - <$FILES/java.class | egrep -o '(Java class|version 53.0)'" \
  "Java class\nversion 53.0\n" "" ""

zero_dev="1/5"
[ "$(uname)" == "Darwin" ] && zero_dev="3/3"
testing "/dev/zero" "file /dev/zero" "/dev/zero: character special ($zero_dev)\n" "" ""
testing "- </dev/zero" "file - </dev/zero | grep -ow data" "data\n" "" ""

testcmd 'ar' 'input | grep -o " ar archive"' ' ar archive\n' \
  "$(echo -e '!<arch>\nhello/          0           0     0     644     6         `\nworld')\n" ""
testcmd 'cpio' 'input' 'input: ASCII cpio archive (SVR4 with no CRC)\n' \
  '07070103344745000081A4000003E800' ''

#TODO block fifo socket
#can't stat (unopenable)
#file
#  readerror
#  elf
#  png (grayscale, color RGB, indexed color, grayscale with alpha, color RGBA
#    X x X x-bit/{non-}interlaced
#  gif87
#  jpeg
#  xz
#  Ogg (buncha types)
#  wav audio (buncha types)
#  truetype font/collection
#  Opentype font
#  LLVM IR bitcode
#  PEM certificate
#  pe32
#  BMP
#  Linux perf
#  Android sparse image file
#  Android boot image
#  Android DTB/DTBO
#  Android Binary XML
#  #! shell script