modules/codec/opus_header.c

   1 /* Copyright (C)2012 Xiph.Org Foundation
   2    File: opus_header.c
   3
   4    Redistribution and use in source and binary forms, with or without
   5    modification, are permitted provided that the following conditions
   6    are met:
   7
   8    - Redistributions of source code must retain the above copyright
   9    notice, this list of conditions and the following disclaimer.
  10
  11    - Redistributions in binary form must reproduce the above copyright
  12    notice, this list of conditions and the following disclaimer in the
  13    documentation and/or other materials provided with the distribution.
  14
  15    THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
  16    ``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
  17    LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
  18    A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO EVENT SHALL THE FOUNDATION OR
  19    CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL,
  20    EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO,
  21    PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
  22    PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
  23    LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING
  24    NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS
  25    SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
  26 */
  27
  28 #ifdef HAVE_CONFIG_H
  29 # include "config.h"
  30 #endif
  31
  32 #include "opus_header.h"
  33 #include <opus.h>
  34 #include <string.h>
  35 #include <stdio.h>
  36 #include <stdlib.h>
  37
  38 #include <vlc_common.h>
  39 #include "../demux/xiph.h"
  40
  41 /* Header contents:
  42   - "OpusHead" (64 bits)
  43   - version number (8 bits)
  44   - Channels C (8 bits)
  45   - Pre-skip (16 bits)
  46   - Sampling rate (32 bits)
  47   - Gain in dB (16 bits, S7.8)
  48   - Mapping (8 bits, 0=single stream (mono/stereo) 1=Vorbis mapping,
  49              2..254: reserved, 255: multistream with no mapping)
  50
  51   - if (mapping != 0)
  52      - N = totel number of streams (8 bits)
  53      - M = number of paired streams (8 bits)
  54      - C times channel origin
  55           - if (C<2*M)
  56              - stream = byte/2
  57              - if (byte&0x1 == 0)
  58                  - left
  59                else
  60                  - right
  61           - else
  62              - stream = byte-M
  63 */
  64
  65 typedef struct {
  66     unsigned char *data;
  67     int maxlen;
  68     int pos;
  69 } Packet;
  70
  71 typedef struct {
  72     const unsigned char *data;
  73     int maxlen;
  74     int pos;
  75 } ROPacket;
  76
  77 static int write_uint32(Packet *p, ogg_uint32_t val)
  78 {
  79     if (p->pos>p->maxlen-4)
  80         return 0;
  81     p->data[p->pos  ] = (val    ) & 0xFF;
  82     p->data[p->pos+1] = (val>> 8) & 0xFF;
  83     p->data[p->pos+2] = (val>>16) & 0xFF;
  84     p->data[p->pos+3] = (val>>24) & 0xFF;
  85     p->pos += 4;
  86     return 1;
  87 }
  88
  89 static int write_uint16(Packet *p, ogg_uint16_t val)
  90 {
  91     if (p->pos>p->maxlen-2)
  92         return 0;
  93     p->data[p->pos  ] = (val    ) & 0xFF;
  94     p->data[p->pos+1] = (val>> 8) & 0xFF;
  95     p->pos += 2;
  96     return 1;
  97 }
  98
  99 static int write_chars(Packet *p, const unsigned char *str, int nb_chars)
 100 {
 101     if (p->pos>p->maxlen-nb_chars)
 102         return 0;
 103     for (int i=0;i<nb_chars;i++)
 104         p->data[p->pos++] = str[i];
 105     return 1;
 106 }
 107
 108 static int read_uint32(ROPacket *p, ogg_uint32_t *val)
 109 {
 110     if (p->pos>p->maxlen-4)
 111         return 0;
 112     *val =  (ogg_uint32_t)p->data[p->pos  ];
 113     *val |= (ogg_uint32_t)p->data[p->pos+1]<< 8;
 114     *val |= (ogg_uint32_t)p->data[p->pos+2]<<16;
 115     *val |= (ogg_uint32_t)p->data[p->pos+3]<<24;
 116     p->pos += 4;
 117     return 1;
 118 }
 119
 120 static int read_uint16(ROPacket *p, ogg_uint16_t *val)
 121 {
 122     if (p->pos>p->maxlen-2)
 123         return 0;
 124     *val =  (ogg_uint16_t)p->data[p->pos  ];
 125     *val |= (ogg_uint16_t)p->data[p->pos+1]<<8;
 126     p->pos += 2;
 127     return 1;
 128 }
 129
 130 static int read_chars(ROPacket *p, unsigned char *str, int nb_chars)
 131 {
 132     if (p->pos>p->maxlen-nb_chars)
 133         return 0;
 134     for (int i=0;i<nb_chars;i++)
 135         str[i] = p->data[p->pos++];
 136     return 1;
 137 }
 138
 139 int opus_header_parse(const unsigned char *packet, int len, OpusHeader *h)
 140 {
 141     char str[9];
 142     ROPacket p;
 143     unsigned char ch;
 144     ogg_uint16_t shortval;
 145
 146     p.data = packet;
 147     p.maxlen = len;
 148     p.pos = 0;
 149     str[8] = 0;
 150     if (len<19)return 0;
 151     read_chars(&p, (unsigned char*)str, 8);
 152     if (memcmp(str, "OpusHead", 8)!=0)
 153         return 0;
 154
 155     if (!read_chars(&p, &ch, 1))
 156         return 0;
 157     h->version = ch;
 158     if((h->version&240) != 0) /* Only major version 0 supported. */
 159         return 0;
 160
 161     if (!read_chars(&p, &ch, 1))
 162         return 0;
 163     h->channels = ch;
 164     if (h->channels == 0)
 165         return 0;
 166
 167     if (!read_uint16(&p, &shortval))
 168         return 0;
 169     h->preskip = shortval;
 170
 171     if (!read_uint32(&p, &h->input_sample_rate))
 172         return 0;
 173
 174     if (!read_uint16(&p, &shortval))
 175         return 0;
 176     h->gain = (short)shortval;
 177
 178     if (!read_chars(&p, &ch, 1))
 179         return 0;
 180     h->channel_mapping = ch;
 181
 182     if (h->channel_mapping != 0)
 183     {
 184         if (!read_chars(&p, &ch, 1))
 185             return 0;
 186
 187         if (ch<1)
 188             return 0;
 189         h->nb_streams = ch;
 190
 191         if (!read_chars(&p, &ch, 1))
 192             return 0;
 193
 194         if (ch>h->nb_streams || (ch+h->nb_streams)>255)
 195             return 0;
 196         h->nb_coupled = ch;
 197
 198         /* Multi-stream support */
 199         for (int i=0;i<h->channels;i++)
 200         {
 201             if (!read_chars(&p, &h->stream_map[i], 1))
 202                 return 0;
 203             if (h->stream_map[i]>(h->nb_streams+h->nb_coupled) && h->stream_map[i]!=255)
 204                 return 0;
 205         }
 206     } else {
 207         if(h->channels>2)
 208             return 0;
 209         h->nb_streams = 1;
 210         h->nb_coupled = h->channels>1;
 211         h->stream_map[0]=0;
 212         h->stream_map[1]=1;
 213     }
 214     /*For version 0/1 we know there won't be any more data
 215       so reject any that have data past the end.*/
 216     if ((h->version==0 || h->version==1) && p.pos != len)
 217         return 0;
 218     return 1;
 219 }
 220
 221 /*
 222  Comments will be stored in the Vorbis style.
 223  It is described in the "Structure" section of
 224     http://www.xiph.org/ogg/vorbis/doc/v-comment.html
 225
 226  However, Opus and other non-vorbis formats omit the "framing_bit".
 227
 228 The comment header is decoded as follows:
 229   1) [vendor_length] = unsigned little endian 32 bits integer
 230   2) [vendor_string] = UTF-8 vector as [vendor_length] octets
 231   3) [user_comment_list_length] = unsigned little endian 32 bits integer
 232   4) iterate [user_comment_list_length] times {
 233      5) [length] = unsigned little endian 32 bits integer
 234      6) this iteration's user comment = UTF-8 vector as [length] octets
 235   }
 236   7) done.
 237 */
 238
 239 static char *comment_init(size_t *length)
 240 {
 241     /*The 'vendor' field should be the actual encoding library used.*/
 242     const char *vendor_string = opus_get_version_string();
 243     int vendor_length = strlen(vendor_string);
 244
 245     int user_comment_list_length = 0;
 246     int len = 8 + 4 + vendor_length + 4;
 247     char *p = malloc(len);
 248     if (p == NULL)
 249         return NULL;
 250
 251     memcpy(p, "OpusTags", 8);
 252     SetDWLE(p + 8, vendor_length);
 253     memcpy(p + 12, vendor_string, vendor_length);
 254     SetDWLE(p + 12 + vendor_length, user_comment_list_length);
 255
 256     *length = len;
 257     return p;
 258 }
 259
 260 static int comment_add(char **comments, size_t *length, const char *tag,
 261                        const char *val)
 262 {
 263     char *p = *comments;
 264     int vendor_length = GetDWLE(p + 8);
 265     size_t user_comment_list_length = GetDWLE(p + 8 + 4 + vendor_length);
 266     size_t tag_len = (tag ? strlen(tag) : 0);
 267     size_t val_len = strlen(val);
 268     size_t len = (*length) + 4 + tag_len + val_len;
 269
 270     p = realloc(p, len);
 271     if (p == NULL)
 272         return 1;
 273
 274     SetDWLE(p + *length, tag_len + val_len);          /* length of comment */
 275     if (tag) memcpy(p + *length + 4, tag, tag_len);         /* comment */
 276     memcpy(p + *length + 4 + tag_len, val, val_len);        /* comment */
 277     SetDWLE(p + 8 + 4 + vendor_length, user_comment_list_length + 1);
 278     *comments = p;
 279     *length = len;
 280     return 0;
 281 }
 282
 283 /* adds padding so that metadata can be updated without rewriting the whole file */
 284 static int comment_pad(char **comments, size_t *length)
 285 {
 286     const unsigned padding = 512; /* default from opus-tools */
 287
 288     char *p = *comments;
 289     /* Make sure there is at least "padding" worth of padding free, and
 290        round up to the maximum that fits in the current ogg segments. */
 291     size_t newlen = ((*length + padding) / 255 + 1) * 255 - 1;
 292     p = realloc(p, newlen);
 293     if (p == NULL)
 294         return 1;
 295
 296     memset(p + *length, 0, newlen - *length);
 297     *comments = p;
 298     *length = newlen;
 299     return 0;
 300 }
 301
 302 int opus_prepare_header(unsigned channels, unsigned rate, OpusHeader *header)
 303 {
 304     header->version = 1;
 305     header->channels = channels;
 306     header->nb_streams = header->channels;
 307     header->nb_coupled = 0;
 308     header->input_sample_rate = rate;
 309     header->gain = 0; // 0dB
 310     header->channel_mapping = header->channels > 8 ? 255 :
 311                               header->channels > 2;
 312
 313     return 0;
 314 }
 315
 316 int opus_write_header(uint8_t **p_extra, int *i_extra, OpusHeader *header)
 317 {
 318     unsigned char header_data[100];
 319     const int packet_size = opus_header_to_packet(header, header_data,
 320                                                   sizeof(header_data));
 321     ogg_packet headers[2];
 322     headers[0].packet = header_data;
 323     headers[0].bytes = packet_size;
 324     headers[0].b_o_s = 1;
 325     headers[0].e_o_s = 0;
 326     headers[0].granulepos = 0;
 327     headers[0].packetno = 0;
 328
 329     size_t comments_length;
 330     char *comments = comment_init(&comments_length);
 331     if (!comments)
 332         return 1;
 333     if (comment_add(&comments, &comments_length, "ENCODER=",
 334                     "VLC media player"))
 335     {
 336         free(comments);
 337         return 1;
 338     }
 339
 340     if (comment_pad(&comments, &comments_length))
 341     {
 342         free(comments);
 343         return 1;
 344     }
 345
 346     headers[1].packet = (unsigned char *) comments;
 347     headers[1].bytes = comments_length;
 348     headers[1].b_o_s = 0;
 349     headers[1].e_o_s = 0;
 350     headers[1].granulepos = 0;
 351     headers[1].packetno = 1;
 352
 353     for (unsigned i = 0; i < ARRAY_SIZE(headers); ++i)
 354     {
 355         if (xiph_AppendHeaders(i_extra, (void **) p_extra,
 356                                headers[i].bytes, headers[i].packet))
 357         {
 358             *i_extra = 0;
 359             *p_extra = NULL;
 360         }
 361     }
 362
 363     return 0;
 364 }
 365
 366 int opus_header_to_packet(const OpusHeader *h, unsigned char *packet, int len)
 367 {
 368     Packet p;
 369     unsigned char ch;
 370
 371     p.data = packet;
 372     p.maxlen = len;
 373     p.pos = 0;
 374     if (len<19)return 0;
 375     if (!write_chars(&p, (const unsigned char*)"OpusHead", 8))
 376         return 0;
 377     /* Version is 1 */
 378     ch = 1;
 379     if (!write_chars(&p, &ch, 1))
 380         return 0;
 381
 382     ch = h->channels;
 383     if (!write_chars(&p, &ch, 1))
 384         return 0;
 385
 386     if (!write_uint16(&p, h->preskip))
 387         return 0;
 388
 389     if (!write_uint32(&p, h->input_sample_rate))
 390         return 0;
 391
 392     if (!write_uint16(&p, h->gain))
 393         return 0;
 394
 395     ch = h->channel_mapping;
 396     if (!write_chars(&p, &ch, 1))
 397         return 0;
 398
 399     if (h->channel_mapping != 0)
 400     {
 401         ch = h->nb_streams;
 402         if (!write_chars(&p, &ch, 1))
 403             return 0;
 404
 405         ch = h->nb_coupled;
 406         if (!write_chars(&p, &ch, 1))
 407             return 0;
 408
 409         /* Multi-stream support */
 410         for (int i=0;i<h->channels;i++)
 411         {
 412             if (!write_chars(&p, &h->stream_map[i], 1))
 413                 return 0;
 414         }
 415     }
 416
 417     return p.pos;
 418 }