src/hcom.c

   1 /* libSoX Macintosh HCOM format.
   2  * These are really FSSD type files with Huffman compression,
   3  * in MacBinary format.
   4  * TODO: make the MacBinary format optional (so that .data files
   5  * are also acceptable).  (How to do this on output?)
   6  *
   7  * September 25, 1991
   8  * Copyright 1991 Guido van Rossum And Sundry Contributors
   9  * This source code is freely redistributable and may be used for
  10  * any purpose.  This copyright notice must be maintained.
  11  * Guido van Rossum And Sundry Contributors are not responsible for
  12  * the consequences of using this software.
  13  *
  14  * April 28, 1998 - Chris Bagwell (cbagwell@sprynet.com)
  15  *
  16  *  Rearranged some functions so that they are declared before they are
  17  *  used, clearing up some compiler warnings.  Because these functions
  18  *  passed floats, it helped some dumb compilers pass stuff on the
  19  *  stack correctly.
  20  *
  21  */
  22
  23 #include "sox_i.h"
  24 #include <assert.h>
  25 #include <string.h>
  26 #include <stdlib.h>
  27 #include <errno.h>
  28
  29 /* FIXME: eliminate these 2 functions */
  30
  31 static void put32_be(unsigned char **p, int32_t val)
  32 {
  33   *(*p)++ = (val >> 24) & 0xff;
  34   *(*p)++ = (val >> 16) & 0xff;
  35   *(*p)++ = (val >> 8) & 0xff;
  36   *(*p)++ = val & 0xff;
  37 }
  38
  39 static void put16_be(unsigned char **p, int val)
  40 {
  41   *(*p)++ = (val >> 8) & 0xff;
  42   *(*p)++ = val & 0xff;
  43 }
  44
  45 /* Dictionary entry for Huffman (de)compression */
  46 typedef struct {
  47         long frequ;
  48         short dict_leftson;
  49         short dict_rightson;
  50 } dictent;
  51
  52 typedef struct {
  53   /* Static data from the header */
  54   dictent *dictionary;
  55   int32_t checksum;
  56   int deltacompression;
  57   /* Engine state */
  58   long huffcount;
  59   long cksum;
  60   int dictentry;
  61   int nrbits;
  62   uint32_t current;
  63   short sample;
  64   /* Dictionary */
  65   dictent *de;
  66   int32_t new_checksum;
  67   int nbits;
  68   int32_t curword;
  69
  70   /* Private data used by writer */
  71   unsigned char *data;          /* Buffer allocated with lsx_malloc */
  72   size_t size;               /* Size of allocated buffer */
  73   size_t pos;                /* Where next byte goes */
  74 } priv_t;
  75
  76 static int startread(sox_format_t * ft)
  77 {
  78         priv_t *p = (priv_t *) ft->priv;
  79         int i;
  80         char buf[5];
  81         uint32_t datasize, rsrcsize;
  82         uint32_t huffcount, checksum, compresstype, divisor;
  83         unsigned short dictsize;
  84         int rc;
  85
  86
  87         /* Skip first 65 bytes of header */
  88         rc = lsx_skipbytes(ft, (size_t) 65);
  89         if (rc)
  90             return rc;
  91
  92         /* Check the file type (bytes 65-68) */
  93         if (lsx_reads(ft, buf, (size_t)4) == SOX_EOF || strncmp(buf, "FSSD", (size_t)4) != 0)
  94         {
  95                 lsx_fail_errno(ft,SOX_EHDR,"Mac header type is not FSSD");
  96                 return (SOX_EOF);
  97         }
  98
  99         /* Skip to byte 83 */
 100         rc = lsx_skipbytes(ft, (size_t) 83-69);
 101         if (rc)
 102             return rc;
 103
 104         /* Get essential numbers from the header */
 105         lsx_readdw(ft, &datasize); /* bytes 83-86 */
 106         lsx_readdw(ft, &rsrcsize); /* bytes 87-90 */
 107
 108         /* Skip the rest of the header (total 128 bytes) */
 109         rc = lsx_skipbytes(ft, (size_t) 128-91);
 110         if (rc != 0)
 111             return rc;
 112
 113         /* The data fork must contain a "HCOM" header */
 114         if (lsx_reads(ft, buf, (size_t)4) == SOX_EOF || strncmp(buf, "HCOM", (size_t)4) != 0)
 115         {
 116                 lsx_fail_errno(ft,SOX_EHDR,"Mac data fork is not HCOM");
 117                 return (SOX_EOF);
 118         }
 119
 120         /* Then follow various parameters */
 121         lsx_readdw(ft, &huffcount);
 122         lsx_readdw(ft, &checksum);
 123         lsx_readdw(ft, &compresstype);
 124         if (compresstype > 1)
 125         {
 126                 lsx_fail_errno(ft,SOX_EHDR,"Bad compression type in HCOM header");
 127                 return (SOX_EOF);
 128         }
 129         lsx_readdw(ft, &divisor);
 130         if (divisor == 0 || divisor > 4)
 131         {
 132                 lsx_fail_errno(ft,SOX_EHDR,"Bad sampling rate divisor in HCOM header");
 133                 return (SOX_EOF);
 134         }
 135         lsx_readw(ft, &dictsize);
 136
 137         /* Translate to sox parameters */
 138         ft->encoding.encoding = SOX_ENCODING_HCOM;
 139         ft->encoding.bits_per_sample = 8;
 140         ft->signal.rate = 22050 / divisor;
 141         ft->signal.channels = 1;
 142
 143         /* Allocate memory for the dictionary */
 144         p->dictionary = lsx_malloc(511 * sizeof(dictent));
 145
 146         /* Read dictionary */
 147         for(i = 0; i < dictsize; i++) {
 148                 lsx_readsw(ft, &(p->dictionary[i].dict_leftson));
 149                 lsx_readsw(ft, &(p->dictionary[i].dict_rightson));
 150                 lsx_debug("%d %d",
 151                        p->dictionary[i].dict_leftson,
 152                        p->dictionary[i].dict_rightson);
 153         }
 154         rc = lsx_skipbytes(ft, (size_t) 1); /* skip pad byte */
 155         if (rc)
 156             return rc;
 157
 158         /* Initialized the decompression engine */
 159         p->checksum = checksum;
 160         p->deltacompression = compresstype;
 161         if (!p->deltacompression)
 162                 lsx_debug("HCOM data using value compression");
 163         p->huffcount = huffcount;
 164         p->cksum = 0;
 165         p->dictentry = 0;
 166         p->nrbits = -1; /* Special case to get first byte */
 167
 168         return (SOX_SUCCESS);
 169 }
 170
 171 static size_t read_samples(sox_format_t * ft, sox_sample_t *buf, size_t len)
 172 {
 173         register priv_t *p = (priv_t *) ft->priv;
 174         int done = 0;
 175         unsigned char sample_rate;
 176
 177         if (p->nrbits < 0) {
 178                 /* The first byte is special */
 179                 if (p->huffcount == 0)
 180                         return 0; /* Don't know if this can happen... */
 181                 if (lsx_readb(ft, &sample_rate) == SOX_EOF)
 182                 {
 183                         return (0);
 184                 }
 185                 p->sample = sample_rate;
 186                 *buf++ = SOX_UNSIGNED_8BIT_TO_SAMPLE(p->sample,);
 187                 p->huffcount--;
 188                 p->nrbits = 0;
 189                 done++;
 190                 len--;
 191                 if (len == 0)
 192                         return done;
 193         }
 194
 195         while (p->huffcount > 0) {
 196                 if(p->nrbits == 0) {
 197                         lsx_readdw(ft, &(p->current));
 198                         if (lsx_eof(ft))
 199                         {
 200                                 lsx_fail_errno(ft,SOX_EOF,"unexpected EOF in HCOM data");
 201                                 return (0);
 202                         }
 203                         p->cksum += p->current;
 204                         p->nrbits = 32;
 205                 }
 206                 if(p->current & 0x80000000) {
 207                         p->dictentry =
 208                                 p->dictionary[p->dictentry].dict_rightson;
 209                 } else {
 210                         p->dictentry =
 211                                 p->dictionary[p->dictentry].dict_leftson;
 212                 }
 213                 p->current = p->current << 1;
 214                 p->nrbits--;
 215                 if(p->dictionary[p->dictentry].dict_leftson < 0) {
 216                         short datum;
 217                         datum = p->dictionary[p->dictentry].dict_rightson;
 218                         if (!p->deltacompression)
 219                                 p->sample = 0;
 220                         p->sample = (p->sample + datum) & 0xff;
 221                         p->huffcount--;
 222                         *buf++ = SOX_UNSIGNED_8BIT_TO_SAMPLE(p->sample,);
 223                         p->dictentry = 0;
 224                         done++;
 225                         len--;
 226                         if (len == 0)
 227                                 break;
 228                 }
 229         }
 230
 231         return done;
 232 }
 233
 234 static int stopread(sox_format_t * ft)
 235 {
 236         register priv_t *p = (priv_t *) ft->priv;
 237
 238         if (p->huffcount != 0)
 239         {
 240                 lsx_fail_errno(ft,SOX_EFMT,"not all HCOM data read");
 241                 return (SOX_EOF);
 242         }
 243         if(p->cksum != p->checksum)
 244         {
 245                 lsx_fail_errno(ft,SOX_EFMT,"checksum error in HCOM data");
 246                 return (SOX_EOF);
 247         }
 248         free(p->dictionary);
 249         p->dictionary = NULL;
 250         return (SOX_SUCCESS);
 251 }
 252
 253 #define BUFINCR (10*BUFSIZ)
 254
 255 static int startwrite(sox_format_t * ft)
 256 {
 257   priv_t * p = (priv_t *) ft->priv;
 258
 259   p->size = BUFINCR;
 260   p->pos = 0;
 261   p->data = lsx_malloc(p->size);
 262   return SOX_SUCCESS;
 263 }
 264
 265 static size_t write_samples(sox_format_t * ft, const sox_sample_t *buf, size_t len)
 266 {
 267   priv_t *p = (priv_t *) ft->priv;
 268   sox_sample_t datum;
 269   size_t i;
 270
 271   if (len == 0)
 272     return 0;
 273
 274   if (p->pos + len > p->size) {
 275     p->size = ((p->pos + len) / BUFINCR + 1) * BUFINCR;
 276     p->data = lsx_realloc(p->data, p->size);
 277   }
 278
 279   for (i = 0; i < len; i++) {
 280     SOX_SAMPLE_LOCALS;
 281     datum = *buf++;
 282     p->data[p->pos++] = SOX_SAMPLE_TO_UNSIGNED_8BIT(datum, ft->clips);
 283   }
 284
 285   return len;
 286 }
 287
 288 static void makecodes(int e, int c, int s, int b, dictent newdict[511], long codes[256], long codesize[256])
 289 {
 290   assert(b);                    /* Prevent stack overflow */
 291   if (newdict[e].dict_leftson < 0) {
 292     codes[newdict[e].dict_rightson] = c;
 293     codesize[newdict[e].dict_rightson] = s;
 294   } else {
 295     makecodes(newdict[e].dict_leftson, c, s + 1, b << 1, newdict, codes, codesize);
 296     makecodes(newdict[e].dict_rightson, c + b, s + 1, b << 1, newdict, codes, codesize);
 297   }
 298 }
 299
 300 static void putcode(sox_format_t * ft, long codes[256], long codesize[256], unsigned c, unsigned char **df)
 301 {
 302   priv_t *p = (priv_t *) ft->priv;
 303   long code, size;
 304   int i;
 305
 306   code = codes[c];
 307   size = codesize[c];
 308   for(i = 0; i < size; i++) {
 309     p->curword <<= 1;
 310     if (code & 1)
 311       p->curword += 1;
 312     p->nbits++;
 313     if (p->nbits == 32) {
 314       put32_be(df, p->curword);
 315       p->new_checksum += p->curword;
 316       p->nbits = 0;
 317       p->curword = 0;
 318     }
 319     code >>= 1;
 320   }
 321 }
 322
 323 static void compress(sox_format_t * ft, unsigned char **df, int32_t *dl)
 324 {
 325   priv_t *p = (priv_t *) ft->priv;
 326   int samplerate;
 327   unsigned char *datafork = *df;
 328   unsigned char *ddf, *dfp;
 329   short dictsize;
 330   int frequtable[256];
 331   long codes[256], codesize[256];
 332   dictent newdict[511];
 333   int i, sample, j, k, d, l, frequcount;
 334
 335   sample = *datafork;
 336   memset(frequtable, 0, sizeof(frequtable));
 337   memset(codes, 0, sizeof(codes));
 338   memset(codesize, 0, sizeof(codesize));
 339   memset(newdict, 0, sizeof(newdict));
 340
 341   for (i = 1; i < *dl; i++) {
 342     d = (datafork[i] - (sample & 0xff)) & 0xff; /* creates absolute entries LMS */
 343     sample = datafork[i];
 344     datafork[i] = d;
 345     assert(d >= 0 && d <= 255); /* check our table is accessed correctly */
 346     frequtable[d]++;
 347   }
 348   p->de = newdict;
 349   for (i = 0; i < 256; i++)
 350     if (frequtable[i] != 0) {
 351       p->de->frequ = -frequtable[i];
 352       p->de->dict_leftson = -1;
 353       p->de->dict_rightson = i;
 354       p->de++;
 355     }
 356   frequcount = p->de - newdict;
 357   for (i = 0; i < frequcount; i++) {
 358     for (j = i + 1; j < frequcount; j++) {
 359       if (newdict[i].frequ > newdict[j].frequ) {
 360         k = newdict[i].frequ;
 361         newdict[i].frequ = newdict[j].frequ;
 362         newdict[j].frequ = k;
 363         k = newdict[i].dict_leftson;
 364         newdict[i].dict_leftson = newdict[j].dict_leftson;
 365         newdict[j].dict_leftson = k;
 366         k = newdict[i].dict_rightson;
 367         newdict[i].dict_rightson = newdict[j].dict_rightson;
 368         newdict[j].dict_rightson = k;
 369       }
 370     }
 371   }
 372   while (frequcount > 1) {
 373     j = frequcount - 1;
 374     p->de->frequ = newdict[j - 1].frequ;
 375     p->de->dict_leftson = newdict[j - 1].dict_leftson;
 376     p->de->dict_rightson = newdict[j - 1].dict_rightson;
 377     l = newdict[j - 1].frequ + newdict[j].frequ;
 378     for (i = j - 2; i >= 0 && l < newdict[i].frequ; i--)
 379       newdict[i + 1] = newdict[i];
 380     i = i + 1;
 381     newdict[i].frequ = l;
 382     newdict[i].dict_leftson = j;
 383     newdict[i].dict_rightson = p->de - newdict;
 384     p->de++;
 385     frequcount--;
 386   }
 387   dictsize = p->de - newdict;
 388   makecodes(0, 0, 0, 1, newdict, codes, codesize);
 389   l = 0;
 390   for (i = 0; i < 256; i++)
 391     l += frequtable[i] * codesize[i];
 392   l = (((l + 31) >> 5) << 2) + 24 + dictsize * 4;
 393   lsx_debug("  Original size: %6d bytes", *dl);
 394   lsx_debug("Compressed size: %6d bytes", l);
 395   datafork = lsx_malloc((size_t)l);
 396   ddf = datafork + 22;
 397   for(i = 0; i < dictsize; i++) {
 398     put16_be(&ddf, newdict[i].dict_leftson);
 399     put16_be(&ddf, newdict[i].dict_rightson);
 400   }
 401   *ddf++ = 0;
 402   *ddf++ = *(*df)++;
 403   p->new_checksum = 0;
 404   p->nbits = 0;
 405   p->curword = 0;
 406   for (i = 1; i < *dl; i++)
 407     putcode(ft, codes, codesize, *(*df)++, &ddf);
 408   if (p->nbits != 0) {
 409     codes[0] = 0;
 410     codesize[0] = 32 - p->nbits;
 411     putcode(ft, codes, codesize, 0, &ddf);
 412   }
 413   memcpy(datafork, "HCOM", (size_t)4);
 414   dfp = datafork + 4;
 415   put32_be(&dfp, *dl);
 416   put32_be(&dfp, p->new_checksum);
 417   put32_be(&dfp, 1);
 418   samplerate = 22050 / ft->signal.rate + .5;
 419   put32_be(&dfp, samplerate);
 420   put16_be(&dfp, dictsize);
 421   *df = datafork;               /* reassign passed pointer to new datafork */
 422   *dl = l;                      /* and its compressed length */
 423 }
 424
 425 /* End of hcom utility routines */
 426
 427 static int stopwrite(sox_format_t * ft)
 428 {
 429   priv_t *p = (priv_t *) ft->priv;
 430   unsigned char *compressed_data = p->data;
 431   size_t compressed_len = p->pos;
 432   int rc = SOX_SUCCESS;
 433
 434   /* Compress it all at once */
 435   if (compressed_len)
 436     compress(ft, &compressed_data, (int32_t *)&compressed_len);
 437   free(p->data);
 438
 439   /* Write the header */
 440   lsx_writebuf(ft, "\000\001A", (size_t) 3); /* Dummy file name "A" */
 441   lsx_padbytes(ft, (size_t) 65-3);
 442   lsx_writes(ft, "FSSD");
 443   lsx_padbytes(ft, (size_t) 83-69);
 444   lsx_writedw(ft, (unsigned) compressed_len); /* compressed_data size */
 445   lsx_writedw(ft, 0); /* rsrc size */
 446   lsx_padbytes(ft, (size_t) 128 - 91);
 447   if (lsx_error(ft)) {
 448     lsx_fail_errno(ft, errno, "write error in HCOM header");
 449     rc = SOX_EOF;
 450   } else if (lsx_writebuf(ft, compressed_data, compressed_len) != compressed_len) {
 451     /* Write the compressed_data fork */
 452     lsx_fail_errno(ft, errno, "can't write compressed HCOM data");
 453     rc = SOX_EOF;
 454   }
 455   free(compressed_data);
 456
 457   if (rc == SOX_SUCCESS)
 458     /* Pad the compressed_data fork to a multiple of 128 bytes */
 459     lsx_padbytes(ft, 128u - (compressed_len % 128));
 460
 461   return rc;
 462 }
 463
 464 LSX_FORMAT_HANDLER(hcom)
 465 {
 466   static char const * const names[]       = {"hcom", NULL};
 467   static sox_rate_t   const write_rates[] = {22050,22050/2,22050/3,22050/4.,0};
 468   static unsigned     const write_encodings[] = {
 469     SOX_ENCODING_HCOM, 8, 0, 0};
 470   static sox_format_handler_t handler = {SOX_LIB_VERSION_CODE,
 471     "Mac FSSD files with Huffman compression",
 472     names, SOX_FILE_BIG_END|SOX_FILE_MONO,
 473     startread, read_samples, stopread,
 474     startwrite, write_samples, stopwrite,
 475     NULL, write_encodings, write_rates, sizeof(priv_t)
 476   };
 477   return &handler;
 478 }