Guido van Rossum | 17ed1ae | 1993-06-01 13:21:04 +0000 | [diff] [blame] | 1 | # This module contains several routines that help recognizing sound |
| 2 | # files. |
| 3 | # |
| 4 | # Function whathdr() recognizes various types of sound file headers. |
| 5 | # It understands almost all headers that SOX can decode. |
| 6 | # |
| 7 | # The return tuple contains the following items, in this order: |
| 8 | # - file type (as SOX understands it) |
| 9 | # - sampling rate (0 if unknown or hard to decode) |
| 10 | # - number of channels (0 if unknown or hard to decode) |
| 11 | # - number of frames in the file (-1 if unknown or hard to decode) |
| 12 | # - number of bits/sample, or 'U' for U-LAW, or 'A' for A-LAW |
| 13 | # |
| 14 | # If the file doesn't have a recognizable type, it returns None. |
| 15 | # If the file can't be opened, IOError is raised. |
| 16 | # |
| 17 | # To compute the total time, divide the number of frames by the |
| 18 | # sampling rate (a frame contains a sample for each channel). |
| 19 | # |
| 20 | # Function whatraw() calls the "whatsound" program and interprets its |
| 21 | # output. You'll have to guess the sampling rate by listening though! |
| 22 | # |
| 23 | # Function what() calls whathdr() and if it doesn't recognize the file |
| 24 | # then calls whatraw(). |
| 25 | # |
| 26 | # Finally, the function test() is a simple main program that calls |
| 27 | # what() for all files mentioned on the argument list. For directory |
| 28 | # arguments it calls what() for all files in that directory. Default |
| 29 | # argument is "." (testing all files in the current directory). The |
| 30 | # option -r tells it to recurse down directories found inside |
| 31 | # explicitly given directories. |
| 32 | # |
| 33 | # The file structure is top-down except that the test program and its |
| 34 | # subroutine come last. |
| 35 | |
| 36 | |
| 37 | #------------------------------------------------------# |
| 38 | # Guess the type of any sound file, raw or with header # |
| 39 | #------------------------------------------------------# |
| 40 | |
| 41 | def what(filename): |
| 42 | res = whathdr(filename) |
| 43 | if not res: |
| 44 | res = whatraw(filename) |
| 45 | return res |
| 46 | |
| 47 | |
| 48 | #-----------------------------# |
| 49 | # Guess the type of raw sound # |
| 50 | #-----------------------------# |
| 51 | |
| 52 | def whatraw(filename): |
| 53 | # Assume it's always 1 channel, byte-sized samples |
| 54 | # Don't assume anything about the rate |
| 55 | import os |
| 56 | from stat import ST_SIZE |
| 57 | # XXX "whatsound" should be part of the distribution somehow... |
| 58 | cmd = 'whatsound ' + filename + ' 2>/dev/null' |
Guido van Rossum | 5cfa5df | 1993-06-23 09:30:50 +0000 | [diff] [blame] | 59 | cmd = 'PATH=$PATH:/ufs/guido/bin/sgi\n' + cmd |
Guido van Rossum | 17ed1ae | 1993-06-01 13:21:04 +0000 | [diff] [blame] | 60 | pipe = os.popen(cmd, 'r') |
| 61 | data = pipe.read() |
| 62 | sts = pipe.close() |
| 63 | if sts: |
| 64 | return None |
| 65 | if data[:13] == '-t raw -b -s ': |
| 66 | type = 'sb' |
| 67 | sample_size = 8 |
| 68 | elif data[:13] == '-t raw -b -u ': |
| 69 | type = 'ub' |
| 70 | sample_size = 8 |
| 71 | elif data[:13] == '-t raw -b -U ': |
| 72 | type = 'ul' |
| 73 | sample_size = 'U' |
| 74 | else: |
| 75 | return None |
| 76 | try: |
| 77 | frame_count = os.stat(filename)[ST_SIZE] |
| 78 | except IOError: |
| 79 | frame_count = -1 |
| 80 | return type, 0, 1, frame_count, sample_size |
| 81 | |
| 82 | |
| 83 | #-------------------------# |
| 84 | # Recognize sound headers # |
| 85 | #-------------------------# |
| 86 | |
| 87 | def whathdr(filename): |
| 88 | f = open(filename, 'r') |
| 89 | h = f.read(512) |
| 90 | for tf in tests: |
| 91 | res = tf(h, f) |
| 92 | if res: |
| 93 | return res |
| 94 | return None |
| 95 | |
| 96 | |
| 97 | #-----------------------------------# |
| 98 | # Subroutines per sound header type # |
| 99 | #-----------------------------------# |
| 100 | |
| 101 | tests = [] |
| 102 | |
| 103 | def test_aifc(h, f): |
| 104 | import aifc |
| 105 | if h[:4] <> 'FORM': |
| 106 | return None |
| 107 | if h[8:12] == 'AIFC': |
| 108 | fmt = 'aifc' |
| 109 | elif h[8:12] == 'AIFF': |
| 110 | fmt = 'aiff' |
| 111 | else: |
| 112 | return None |
| 113 | f.seek(0) |
| 114 | try: |
| 115 | a = aifc.openfp(f, 'r') |
| 116 | except (EOFError, aifc.Error): |
| 117 | return None |
| 118 | return (fmt, a.getframerate(), a.getnchannels(), \ |
| 119 | a.getnframes(), 8*a.getsampwidth()) |
| 120 | |
| 121 | tests.append(test_aifc) |
| 122 | |
| 123 | |
| 124 | def test_au(h, f): |
| 125 | if h[:4] == '.snd': |
| 126 | f = get_long_be |
| 127 | elif h[:4] in ('\0ds.', 'dns.'): |
| 128 | f = get_long_le |
| 129 | else: |
| 130 | return None |
| 131 | type = 'au' |
| 132 | hdr_size = f(h[4:8]) |
| 133 | data_size = f(h[8:12]) |
| 134 | encoding = f(h[12:16]) |
| 135 | rate = f(h[16:20]) |
| 136 | nchannels = f(h[20:24]) |
| 137 | sample_size = 1 # default |
| 138 | if encoding == 1: |
| 139 | sample_bits = 'U' |
| 140 | elif encoding == 2: |
| 141 | sample_bits = 8 |
| 142 | elif encoding == 3: |
| 143 | sample_bits = 16 |
| 144 | sample_size = 2 |
| 145 | else: |
| 146 | sample_bits = '?' |
| 147 | frame_size = sample_size * nchannels |
| 148 | return type, rate, nchannels, data_size/frame_size, sample_bits |
| 149 | |
| 150 | tests.append(test_au) |
| 151 | |
| 152 | |
| 153 | def test_hcom(h, f): |
| 154 | if h[65:69] <> 'FSSD' or h[128:132] <> 'HCOM': |
| 155 | return None |
| 156 | divisor = get_long_be(h[128+16:128+20]) |
| 157 | return 'hcom', 22050/divisor, 1, -1, 8 |
| 158 | |
| 159 | tests.append(test_hcom) |
| 160 | |
| 161 | |
| 162 | def test_voc(h, f): |
| 163 | if h[:20] <> 'Creative Voice File\032': |
| 164 | return None |
| 165 | sbseek = get_short_le(h[20:22]) |
| 166 | rate = 0 |
| 167 | if 0 <= sbseek < 500 and h[sbseek] == '\1': |
| 168 | ratecode = ord(h[sbseek+4]) |
| 169 | rate = int(1000000.0 / (256 - ratecode)) |
| 170 | return 'voc', rate, 1, -1, 8 |
| 171 | |
| 172 | tests.append(test_voc) |
| 173 | |
| 174 | |
| 175 | def test_wav(h, f): |
| 176 | # 'RIFF' <len> 'WAVE' 'fmt ' <len> |
| 177 | if h[:4] <> 'RIFF' or h[8:12] <> 'WAVE' or h[12:16] <> 'fmt ': |
| 178 | return None |
| 179 | style = get_short_le(h[20:22]) |
| 180 | nchannels = get_short_le(h[22:24]) |
| 181 | rate = get_long_le(h[24:28]) |
| 182 | sample_bits = get_short_le(h[34:36]) |
| 183 | return 'wav', rate, nchannels, -1, sample_bits |
| 184 | |
| 185 | tests.append(test_wav) |
| 186 | |
| 187 | |
| 188 | def test_8svx(h, f): |
| 189 | if h[:4] <> 'FORM' or h[8:12] <> '8SVX': |
| 190 | return None |
| 191 | # Should decode it to get #channels -- assume always 1 |
| 192 | return '8svx', 0, 1, 0, 8 |
| 193 | |
| 194 | tests.append(test_8svx) |
| 195 | |
| 196 | |
| 197 | def test_sndt(h, f): |
| 198 | if h[:5] == 'SOUND': |
| 199 | nsamples = get_long_le(h[8:12]) |
| 200 | rate = get_short_le(h[20:22]) |
| 201 | return 'sndt', rate, 1, nsamples, 8 |
| 202 | |
| 203 | tests.append(test_sndt) |
| 204 | |
| 205 | |
| 206 | def test_sndr(h, f): |
| 207 | if h[:2] == '\0\0': |
| 208 | rate = get_short_le(h[2:4]) |
| 209 | if 4000 <= rate <= 25000: |
| 210 | return 'sndr', rate, 1, -1, 8 |
| 211 | |
| 212 | tests.append(test_sndr) |
| 213 | |
| 214 | |
| 215 | #---------------------------------------------# |
| 216 | # Subroutines to extract numbers from strings # |
| 217 | #---------------------------------------------# |
| 218 | |
| 219 | def get_long_be(s): |
| 220 | return (ord(s[0])<<24) | (ord(s[1])<<16) | (ord(s[2])<<8) | ord(s[3]) |
| 221 | |
| 222 | def get_long_le(s): |
| 223 | return (ord(s[3])<<24) | (ord(s[2])<<16) | (ord(s[1])<<8) | ord(s[0]) |
| 224 | |
| 225 | def get_short_be(s): |
| 226 | return (ord(s[0])<<8) | ord(s[1]) |
| 227 | |
| 228 | def get_short_le(s): |
| 229 | return (ord(s[1])<<8) | ord(s[0]) |
| 230 | |
| 231 | |
| 232 | #--------------------# |
| 233 | # Small test program # |
| 234 | #--------------------# |
| 235 | |
| 236 | def test(): |
| 237 | import sys |
| 238 | recursive = 0 |
| 239 | if sys.argv[1:] and sys.argv[1] == '-r': |
| 240 | del sys.argv[1:2] |
| 241 | recursive = 1 |
| 242 | try: |
| 243 | if sys.argv[1:]: |
| 244 | testall(sys.argv[1:], recursive, 1) |
| 245 | else: |
| 246 | testall(['.'], recursive, 1) |
| 247 | except KeyboardInterrupt: |
| 248 | sys.stderr.write('\n[Interrupted]\n') |
| 249 | sys.exit(1) |
| 250 | |
| 251 | def testall(list, recursive, toplevel): |
| 252 | import sys |
| 253 | import os |
| 254 | for filename in list: |
| 255 | if os.path.isdir(filename): |
| 256 | print filename + '/:', |
| 257 | if recursive or toplevel: |
| 258 | print 'recursing down:' |
| 259 | import glob |
| 260 | names = glob.glob(os.path.join(filename, '*')) |
| 261 | testall(names, recursive, 0) |
| 262 | else: |
| 263 | print '*** directory (use -r) ***' |
| 264 | else: |
| 265 | print filename + ':', |
| 266 | sys.stdout.flush() |
| 267 | try: |
| 268 | print what(filename) |
| 269 | except IOError: |
| 270 | print '*** not found ***' |