| 1 | n/a | """Routines to help recognizing sound files. |
|---|
| 2 | n/a | |
|---|
| 3 | n/a | Function whathdr() recognizes various types of sound file headers. |
|---|
| 4 | n/a | It understands almost all headers that SOX can decode. |
|---|
| 5 | n/a | |
|---|
| 6 | n/a | The return tuple contains the following items, in this order: |
|---|
| 7 | n/a | - file type (as SOX understands it) |
|---|
| 8 | n/a | - sampling rate (0 if unknown or hard to decode) |
|---|
| 9 | n/a | - number of channels (0 if unknown or hard to decode) |
|---|
| 10 | n/a | - number of frames in the file (-1 if unknown or hard to decode) |
|---|
| 11 | n/a | - number of bits/sample, or 'U' for U-LAW, or 'A' for A-LAW |
|---|
| 12 | n/a | |
|---|
| 13 | n/a | If the file doesn't have a recognizable type, it returns None. |
|---|
| 14 | n/a | If the file can't be opened, OSError is raised. |
|---|
| 15 | n/a | |
|---|
| 16 | n/a | To compute the total time, divide the number of frames by the |
|---|
| 17 | n/a | sampling rate (a frame contains a sample for each channel). |
|---|
| 18 | n/a | |
|---|
| 19 | n/a | Function what() calls whathdr(). (It used to also use some |
|---|
| 20 | n/a | heuristics for raw data, but this doesn't work very well.) |
|---|
| 21 | n/a | |
|---|
| 22 | n/a | Finally, the function test() is a simple main program that calls |
|---|
| 23 | n/a | what() for all files mentioned on the argument list. For directory |
|---|
| 24 | n/a | arguments it calls what() for all files in that directory. Default |
|---|
| 25 | n/a | argument is "." (testing all files in the current directory). The |
|---|
| 26 | n/a | option -r tells it to recurse down directories found inside |
|---|
| 27 | n/a | explicitly given directories. |
|---|
| 28 | n/a | """ |
|---|
| 29 | n/a | |
|---|
| 30 | n/a | # The file structure is top-down except that the test program and its |
|---|
| 31 | n/a | # subroutine come last. |
|---|
| 32 | n/a | |
|---|
| 33 | n/a | __all__ = ['what', 'whathdr'] |
|---|
| 34 | n/a | |
|---|
| 35 | n/a | from collections import namedtuple |
|---|
| 36 | n/a | |
|---|
| 37 | n/a | SndHeaders = namedtuple('SndHeaders', |
|---|
| 38 | n/a | 'filetype framerate nchannels nframes sampwidth') |
|---|
| 39 | n/a | |
|---|
| 40 | n/a | SndHeaders.filetype.__doc__ = ("""The value for type indicates the data type |
|---|
| 41 | n/a | and will be one of the strings 'aifc', 'aiff', 'au','hcom', |
|---|
| 42 | n/a | 'sndr', 'sndt', 'voc', 'wav', '8svx', 'sb', 'ub', or 'ul'.""") |
|---|
| 43 | n/a | SndHeaders.framerate.__doc__ = ("""The sampling_rate will be either the actual |
|---|
| 44 | n/a | value or 0 if unknown or difficult to decode.""") |
|---|
| 45 | n/a | SndHeaders.nchannels.__doc__ = ("""The number of channels or 0 if it cannot be |
|---|
| 46 | n/a | determined or if the value is difficult to decode.""") |
|---|
| 47 | n/a | SndHeaders.nframes.__doc__ = ("""The value for frames will be either the number |
|---|
| 48 | n/a | of frames or -1.""") |
|---|
| 49 | n/a | SndHeaders.sampwidth.__doc__ = ("""Either the sample size in bits or |
|---|
| 50 | n/a | 'A' for A-LAW or 'U' for u-LAW.""") |
|---|
| 51 | n/a | |
|---|
| 52 | n/a | def what(filename): |
|---|
| 53 | n/a | """Guess the type of a sound file.""" |
|---|
| 54 | n/a | res = whathdr(filename) |
|---|
| 55 | n/a | return res |
|---|
| 56 | n/a | |
|---|
| 57 | n/a | |
|---|
| 58 | n/a | def whathdr(filename): |
|---|
| 59 | n/a | """Recognize sound headers.""" |
|---|
| 60 | n/a | with open(filename, 'rb') as f: |
|---|
| 61 | n/a | h = f.read(512) |
|---|
| 62 | n/a | for tf in tests: |
|---|
| 63 | n/a | res = tf(h, f) |
|---|
| 64 | n/a | if res: |
|---|
| 65 | n/a | return SndHeaders(*res) |
|---|
| 66 | n/a | return None |
|---|
| 67 | n/a | |
|---|
| 68 | n/a | |
|---|
| 69 | n/a | #-----------------------------------# |
|---|
| 70 | n/a | # Subroutines per sound header type # |
|---|
| 71 | n/a | #-----------------------------------# |
|---|
| 72 | n/a | |
|---|
| 73 | n/a | tests = [] |
|---|
| 74 | n/a | |
|---|
| 75 | n/a | def test_aifc(h, f): |
|---|
| 76 | n/a | import aifc |
|---|
| 77 | n/a | if not h.startswith(b'FORM'): |
|---|
| 78 | n/a | return None |
|---|
| 79 | n/a | if h[8:12] == b'AIFC': |
|---|
| 80 | n/a | fmt = 'aifc' |
|---|
| 81 | n/a | elif h[8:12] == b'AIFF': |
|---|
| 82 | n/a | fmt = 'aiff' |
|---|
| 83 | n/a | else: |
|---|
| 84 | n/a | return None |
|---|
| 85 | n/a | f.seek(0) |
|---|
| 86 | n/a | try: |
|---|
| 87 | n/a | a = aifc.open(f, 'r') |
|---|
| 88 | n/a | except (EOFError, aifc.Error): |
|---|
| 89 | n/a | return None |
|---|
| 90 | n/a | return (fmt, a.getframerate(), a.getnchannels(), |
|---|
| 91 | n/a | a.getnframes(), 8 * a.getsampwidth()) |
|---|
| 92 | n/a | |
|---|
| 93 | n/a | tests.append(test_aifc) |
|---|
| 94 | n/a | |
|---|
| 95 | n/a | |
|---|
| 96 | n/a | def test_au(h, f): |
|---|
| 97 | n/a | if h.startswith(b'.snd'): |
|---|
| 98 | n/a | func = get_long_be |
|---|
| 99 | n/a | elif h[:4] in (b'\0ds.', b'dns.'): |
|---|
| 100 | n/a | func = get_long_le |
|---|
| 101 | n/a | else: |
|---|
| 102 | n/a | return None |
|---|
| 103 | n/a | filetype = 'au' |
|---|
| 104 | n/a | hdr_size = func(h[4:8]) |
|---|
| 105 | n/a | data_size = func(h[8:12]) |
|---|
| 106 | n/a | encoding = func(h[12:16]) |
|---|
| 107 | n/a | rate = func(h[16:20]) |
|---|
| 108 | n/a | nchannels = func(h[20:24]) |
|---|
| 109 | n/a | sample_size = 1 # default |
|---|
| 110 | n/a | if encoding == 1: |
|---|
| 111 | n/a | sample_bits = 'U' |
|---|
| 112 | n/a | elif encoding == 2: |
|---|
| 113 | n/a | sample_bits = 8 |
|---|
| 114 | n/a | elif encoding == 3: |
|---|
| 115 | n/a | sample_bits = 16 |
|---|
| 116 | n/a | sample_size = 2 |
|---|
| 117 | n/a | else: |
|---|
| 118 | n/a | sample_bits = '?' |
|---|
| 119 | n/a | frame_size = sample_size * nchannels |
|---|
| 120 | n/a | if frame_size: |
|---|
| 121 | n/a | nframe = data_size / frame_size |
|---|
| 122 | n/a | else: |
|---|
| 123 | n/a | nframe = -1 |
|---|
| 124 | n/a | return filetype, rate, nchannels, nframe, sample_bits |
|---|
| 125 | n/a | |
|---|
| 126 | n/a | tests.append(test_au) |
|---|
| 127 | n/a | |
|---|
| 128 | n/a | |
|---|
| 129 | n/a | def test_hcom(h, f): |
|---|
| 130 | n/a | if h[65:69] != b'FSSD' or h[128:132] != b'HCOM': |
|---|
| 131 | n/a | return None |
|---|
| 132 | n/a | divisor = get_long_be(h[144:148]) |
|---|
| 133 | n/a | if divisor: |
|---|
| 134 | n/a | rate = 22050 / divisor |
|---|
| 135 | n/a | else: |
|---|
| 136 | n/a | rate = 0 |
|---|
| 137 | n/a | return 'hcom', rate, 1, -1, 8 |
|---|
| 138 | n/a | |
|---|
| 139 | n/a | tests.append(test_hcom) |
|---|
| 140 | n/a | |
|---|
| 141 | n/a | |
|---|
| 142 | n/a | def test_voc(h, f): |
|---|
| 143 | n/a | if not h.startswith(b'Creative Voice File\032'): |
|---|
| 144 | n/a | return None |
|---|
| 145 | n/a | sbseek = get_short_le(h[20:22]) |
|---|
| 146 | n/a | rate = 0 |
|---|
| 147 | n/a | if 0 <= sbseek < 500 and h[sbseek] == 1: |
|---|
| 148 | n/a | ratecode = 256 - h[sbseek+4] |
|---|
| 149 | n/a | if ratecode: |
|---|
| 150 | n/a | rate = int(1000000.0 / ratecode) |
|---|
| 151 | n/a | return 'voc', rate, 1, -1, 8 |
|---|
| 152 | n/a | |
|---|
| 153 | n/a | tests.append(test_voc) |
|---|
| 154 | n/a | |
|---|
| 155 | n/a | |
|---|
| 156 | n/a | def test_wav(h, f): |
|---|
| 157 | n/a | import wave |
|---|
| 158 | n/a | # 'RIFF' <len> 'WAVE' 'fmt ' <len> |
|---|
| 159 | n/a | if not h.startswith(b'RIFF') or h[8:12] != b'WAVE' or h[12:16] != b'fmt ': |
|---|
| 160 | n/a | return None |
|---|
| 161 | n/a | f.seek(0) |
|---|
| 162 | n/a | try: |
|---|
| 163 | n/a | w = wave.openfp(f, 'r') |
|---|
| 164 | n/a | except (EOFError, wave.Error): |
|---|
| 165 | n/a | return None |
|---|
| 166 | n/a | return ('wav', w.getframerate(), w.getnchannels(), |
|---|
| 167 | n/a | w.getnframes(), 8*w.getsampwidth()) |
|---|
| 168 | n/a | |
|---|
| 169 | n/a | tests.append(test_wav) |
|---|
| 170 | n/a | |
|---|
| 171 | n/a | |
|---|
| 172 | n/a | def test_8svx(h, f): |
|---|
| 173 | n/a | if not h.startswith(b'FORM') or h[8:12] != b'8SVX': |
|---|
| 174 | n/a | return None |
|---|
| 175 | n/a | # Should decode it to get #channels -- assume always 1 |
|---|
| 176 | n/a | return '8svx', 0, 1, 0, 8 |
|---|
| 177 | n/a | |
|---|
| 178 | n/a | tests.append(test_8svx) |
|---|
| 179 | n/a | |
|---|
| 180 | n/a | |
|---|
| 181 | n/a | def test_sndt(h, f): |
|---|
| 182 | n/a | if h.startswith(b'SOUND'): |
|---|
| 183 | n/a | nsamples = get_long_le(h[8:12]) |
|---|
| 184 | n/a | rate = get_short_le(h[20:22]) |
|---|
| 185 | n/a | return 'sndt', rate, 1, nsamples, 8 |
|---|
| 186 | n/a | |
|---|
| 187 | n/a | tests.append(test_sndt) |
|---|
| 188 | n/a | |
|---|
| 189 | n/a | |
|---|
| 190 | n/a | def test_sndr(h, f): |
|---|
| 191 | n/a | if h.startswith(b'\0\0'): |
|---|
| 192 | n/a | rate = get_short_le(h[2:4]) |
|---|
| 193 | n/a | if 4000 <= rate <= 25000: |
|---|
| 194 | n/a | return 'sndr', rate, 1, -1, 8 |
|---|
| 195 | n/a | |
|---|
| 196 | n/a | tests.append(test_sndr) |
|---|
| 197 | n/a | |
|---|
| 198 | n/a | |
|---|
| 199 | n/a | #-------------------------------------------# |
|---|
| 200 | n/a | # Subroutines to extract numbers from bytes # |
|---|
| 201 | n/a | #-------------------------------------------# |
|---|
| 202 | n/a | |
|---|
| 203 | n/a | def get_long_be(b): |
|---|
| 204 | n/a | return (b[0] << 24) | (b[1] << 16) | (b[2] << 8) | b[3] |
|---|
| 205 | n/a | |
|---|
| 206 | n/a | def get_long_le(b): |
|---|
| 207 | n/a | return (b[3] << 24) | (b[2] << 16) | (b[1] << 8) | b[0] |
|---|
| 208 | n/a | |
|---|
| 209 | n/a | def get_short_be(b): |
|---|
| 210 | n/a | return (b[0] << 8) | b[1] |
|---|
| 211 | n/a | |
|---|
| 212 | n/a | def get_short_le(b): |
|---|
| 213 | n/a | return (b[1] << 8) | b[0] |
|---|
| 214 | n/a | |
|---|
| 215 | n/a | |
|---|
| 216 | n/a | #--------------------# |
|---|
| 217 | n/a | # Small test program # |
|---|
| 218 | n/a | #--------------------# |
|---|
| 219 | n/a | |
|---|
| 220 | n/a | def test(): |
|---|
| 221 | n/a | import sys |
|---|
| 222 | n/a | recursive = 0 |
|---|
| 223 | n/a | if sys.argv[1:] and sys.argv[1] == '-r': |
|---|
| 224 | n/a | del sys.argv[1:2] |
|---|
| 225 | n/a | recursive = 1 |
|---|
| 226 | n/a | try: |
|---|
| 227 | n/a | if sys.argv[1:]: |
|---|
| 228 | n/a | testall(sys.argv[1:], recursive, 1) |
|---|
| 229 | n/a | else: |
|---|
| 230 | n/a | testall(['.'], recursive, 1) |
|---|
| 231 | n/a | except KeyboardInterrupt: |
|---|
| 232 | n/a | sys.stderr.write('\n[Interrupted]\n') |
|---|
| 233 | n/a | sys.exit(1) |
|---|
| 234 | n/a | |
|---|
| 235 | n/a | def testall(list, recursive, toplevel): |
|---|
| 236 | n/a | import sys |
|---|
| 237 | n/a | import os |
|---|
| 238 | n/a | for filename in list: |
|---|
| 239 | n/a | if os.path.isdir(filename): |
|---|
| 240 | n/a | print(filename + '/:', end=' ') |
|---|
| 241 | n/a | if recursive or toplevel: |
|---|
| 242 | n/a | print('recursing down:') |
|---|
| 243 | n/a | import glob |
|---|
| 244 | n/a | names = glob.glob(os.path.join(filename, '*')) |
|---|
| 245 | n/a | testall(names, recursive, 0) |
|---|
| 246 | n/a | else: |
|---|
| 247 | n/a | print('*** directory (use -r) ***') |
|---|
| 248 | n/a | else: |
|---|
| 249 | n/a | print(filename + ':', end=' ') |
|---|
| 250 | n/a | sys.stdout.flush() |
|---|
| 251 | n/a | try: |
|---|
| 252 | n/a | print(what(filename)) |
|---|
| 253 | n/a | except OSError: |
|---|
| 254 | n/a | print('*** not found ***') |
|---|
| 255 | n/a | |
|---|
| 256 | n/a | if __name__ == '__main__': |
|---|
| 257 | n/a | test() |
|---|