Jingwen Chen | 475b3cc | 2021-01-05 21:45:16 -0500 | [diff] [blame] | 1 | """Recognize image file formats based on their first few bytes.""" |
| 2 | |
| 3 | from os import PathLike |
| 4 | |
| 5 | __all__ = ["what"] |
| 6 | |
| 7 | #-------------------------# |
| 8 | # Recognize image headers # |
| 9 | #-------------------------# |
| 10 | |
| 11 | def what(file, h=None): |
| 12 | f = None |
| 13 | try: |
| 14 | if h is None: |
| 15 | if isinstance(file, (str, PathLike)): |
| 16 | f = open(file, 'rb') |
| 17 | h = f.read(32) |
| 18 | else: |
| 19 | location = file.tell() |
| 20 | h = file.read(32) |
| 21 | file.seek(location) |
| 22 | for tf in tests: |
| 23 | res = tf(h, f) |
| 24 | if res: |
| 25 | return res |
| 26 | finally: |
| 27 | if f: f.close() |
| 28 | return None |
| 29 | |
| 30 | |
| 31 | #---------------------------------# |
| 32 | # Subroutines per image file type # |
| 33 | #---------------------------------# |
| 34 | |
| 35 | tests = [] |
| 36 | |
| 37 | def test_jpeg(h, f): |
| 38 | """JPEG data in JFIF or Exif format""" |
| 39 | if h[6:10] in (b'JFIF', b'Exif'): |
| 40 | return 'jpeg' |
| 41 | |
| 42 | tests.append(test_jpeg) |
| 43 | |
| 44 | def test_png(h, f): |
| 45 | if h.startswith(b'\211PNG\r\n\032\n'): |
| 46 | return 'png' |
| 47 | |
| 48 | tests.append(test_png) |
| 49 | |
| 50 | def test_gif(h, f): |
| 51 | """GIF ('87 and '89 variants)""" |
| 52 | if h[:6] in (b'GIF87a', b'GIF89a'): |
| 53 | return 'gif' |
| 54 | |
| 55 | tests.append(test_gif) |
| 56 | |
| 57 | def test_tiff(h, f): |
| 58 | """TIFF (can be in Motorola or Intel byte order)""" |
| 59 | if h[:2] in (b'MM', b'II'): |
| 60 | return 'tiff' |
| 61 | |
| 62 | tests.append(test_tiff) |
| 63 | |
| 64 | def test_rgb(h, f): |
| 65 | """SGI image library""" |
| 66 | if h.startswith(b'\001\332'): |
| 67 | return 'rgb' |
| 68 | |
| 69 | tests.append(test_rgb) |
| 70 | |
| 71 | def test_pbm(h, f): |
| 72 | """PBM (portable bitmap)""" |
| 73 | if len(h) >= 3 and \ |
| 74 | h[0] == ord(b'P') and h[1] in b'14' and h[2] in b' \t\n\r': |
| 75 | return 'pbm' |
| 76 | |
| 77 | tests.append(test_pbm) |
| 78 | |
| 79 | def test_pgm(h, f): |
| 80 | """PGM (portable graymap)""" |
| 81 | if len(h) >= 3 and \ |
| 82 | h[0] == ord(b'P') and h[1] in b'25' and h[2] in b' \t\n\r': |
| 83 | return 'pgm' |
| 84 | |
| 85 | tests.append(test_pgm) |
| 86 | |
| 87 | def test_ppm(h, f): |
| 88 | """PPM (portable pixmap)""" |
| 89 | if len(h) >= 3 and \ |
| 90 | h[0] == ord(b'P') and h[1] in b'36' and h[2] in b' \t\n\r': |
| 91 | return 'ppm' |
| 92 | |
| 93 | tests.append(test_ppm) |
| 94 | |
| 95 | def test_rast(h, f): |
| 96 | """Sun raster file""" |
| 97 | if h.startswith(b'\x59\xA6\x6A\x95'): |
| 98 | return 'rast' |
| 99 | |
| 100 | tests.append(test_rast) |
| 101 | |
| 102 | def test_xbm(h, f): |
| 103 | """X bitmap (X10 or X11)""" |
| 104 | if h.startswith(b'#define '): |
| 105 | return 'xbm' |
| 106 | |
| 107 | tests.append(test_xbm) |
| 108 | |
| 109 | def test_bmp(h, f): |
| 110 | if h.startswith(b'BM'): |
| 111 | return 'bmp' |
| 112 | |
| 113 | tests.append(test_bmp) |
| 114 | |
| 115 | def test_webp(h, f): |
| 116 | if h.startswith(b'RIFF') and h[8:12] == b'WEBP': |
| 117 | return 'webp' |
| 118 | |
| 119 | tests.append(test_webp) |
| 120 | |
| 121 | def test_exr(h, f): |
| 122 | if h.startswith(b'\x76\x2f\x31\x01'): |
| 123 | return 'exr' |
| 124 | |
| 125 | tests.append(test_exr) |
| 126 | |
| 127 | #--------------------# |
| 128 | # Small test program # |
| 129 | #--------------------# |
| 130 | |
| 131 | def test(): |
| 132 | import sys |
| 133 | recursive = 0 |
| 134 | if sys.argv[1:] and sys.argv[1] == '-r': |
| 135 | del sys.argv[1:2] |
| 136 | recursive = 1 |
| 137 | try: |
| 138 | if sys.argv[1:]: |
| 139 | testall(sys.argv[1:], recursive, 1) |
| 140 | else: |
| 141 | testall(['.'], recursive, 1) |
| 142 | except KeyboardInterrupt: |
| 143 | sys.stderr.write('\n[Interrupted]\n') |
| 144 | sys.exit(1) |
| 145 | |
| 146 | def testall(list, recursive, toplevel): |
| 147 | import sys |
| 148 | import os |
| 149 | for filename in list: |
| 150 | if os.path.isdir(filename): |
| 151 | print(filename + '/:', end=' ') |
| 152 | if recursive or toplevel: |
| 153 | print('recursing down:') |
| 154 | import glob |
| 155 | names = glob.glob(os.path.join(glob.escape(filename), '*')) |
| 156 | testall(names, recursive, 0) |
| 157 | else: |
| 158 | print('*** directory (use -r) ***') |
| 159 | else: |
| 160 | print(filename + ':', end=' ') |
| 161 | sys.stdout.flush() |
| 162 | try: |
| 163 | print(what(filename)) |
| 164 | except OSError: |
| 165 | print('*** not found ***') |
| 166 | |
| 167 | if __name__ == '__main__': |
| 168 | test() |