You can not select more than 25 topics Topics must start with a letter or number, can include dashes ('-') and can be up to 35 characters long.

384 lines
12KB

  1. /*
  2. * copyright (c) 2002 Mark Hills <mark@pogo.org.uk>
  3. *
  4. * This file is part of FFmpeg.
  5. *
  6. * FFmpeg is free software; you can redistribute it and/or
  7. * modify it under the terms of the GNU Lesser General Public
  8. * License as published by the Free Software Foundation; either
  9. * version 2.1 of the License, or (at your option) any later version.
  10. *
  11. * FFmpeg is distributed in the hope that it will be useful,
  12. * but WITHOUT ANY WARRANTY; without even the implied warranty of
  13. * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
  14. * Lesser General Public License for more details.
  15. *
  16. * You should have received a copy of the GNU Lesser General Public
  17. * License along with FFmpeg; if not, write to the Free Software
  18. * Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA
  19. */
  20. /**
  21. * @file oggvorbis.c
  22. * Ogg Vorbis codec support via libvorbisenc.
  23. * @author Mark Hills <mark@pogo.org.uk>
  24. */
  25. #include <vorbis/vorbisenc.h>
  26. #include "avcodec.h"
  27. #undef NDEBUG
  28. #include <assert.h>
  29. #define OGGVORBIS_FRAME_SIZE 64
  30. #define BUFFER_SIZE (1024*64)
  31. typedef struct OggVorbisContext {
  32. vorbis_info vi ;
  33. vorbis_dsp_state vd ;
  34. vorbis_block vb ;
  35. uint8_t buffer[BUFFER_SIZE];
  36. int buffer_index;
  37. /* decoder */
  38. vorbis_comment vc ;
  39. ogg_packet op;
  40. } OggVorbisContext ;
  41. static int oggvorbis_init_encoder(vorbis_info *vi, AVCodecContext *avccontext) {
  42. double cfreq;
  43. if(avccontext->flags & CODEC_FLAG_QSCALE) {
  44. /* variable bitrate */
  45. if(vorbis_encode_setup_vbr(vi, avccontext->channels,
  46. avccontext->sample_rate,
  47. avccontext->global_quality / (float)FF_QP2LAMBDA))
  48. return -1;
  49. } else {
  50. /* constant bitrate */
  51. if(vorbis_encode_setup_managed(vi, avccontext->channels,
  52. avccontext->sample_rate, -1, avccontext->bit_rate, -1))
  53. return -1;
  54. #ifdef OGGVORBIS_VBR_BY_ESTIMATE
  55. /* variable bitrate by estimate */
  56. if(vorbis_encode_ctl(vi, OV_ECTL_RATEMANAGE_AVG, NULL))
  57. return -1;
  58. #endif
  59. }
  60. /* cutoff frequency */
  61. if(avccontext->cutoff > 0) {
  62. cfreq = avccontext->cutoff / 1000.0;
  63. if(vorbis_encode_ctl(vi, OV_ECTL_LOWPASS_SET, &cfreq))
  64. return -1;
  65. }
  66. return vorbis_encode_setup_init(vi);
  67. }
  68. static int oggvorbis_encode_init(AVCodecContext *avccontext) {
  69. OggVorbisContext *context = avccontext->priv_data ;
  70. ogg_packet header, header_comm, header_code;
  71. uint8_t *p;
  72. unsigned int offset, len;
  73. vorbis_info_init(&context->vi) ;
  74. if(oggvorbis_init_encoder(&context->vi, avccontext) < 0) {
  75. av_log(avccontext, AV_LOG_ERROR, "oggvorbis_encode_init: init_encoder failed") ;
  76. return -1 ;
  77. }
  78. vorbis_analysis_init(&context->vd, &context->vi) ;
  79. vorbis_block_init(&context->vd, &context->vb) ;
  80. vorbis_comment_init(&context->vc);
  81. vorbis_comment_add_tag(&context->vc, "encoder", LIBAVCODEC_IDENT) ;
  82. vorbis_analysis_headerout(&context->vd, &context->vc, &header,
  83. &header_comm, &header_code);
  84. len = header.bytes + header_comm.bytes + header_code.bytes;
  85. avccontext->extradata_size= 64 + len + len/255;
  86. p = avccontext->extradata= av_mallocz(avccontext->extradata_size);
  87. p[0] = 2;
  88. offset = 1;
  89. offset += av_xiphlacing(&p[offset], header.bytes);
  90. offset += av_xiphlacing(&p[offset], header_comm.bytes);
  91. memcpy(&p[offset], header.packet, header.bytes);
  92. offset += header.bytes;
  93. memcpy(&p[offset], header_comm.packet, header_comm.bytes);
  94. offset += header_comm.bytes;
  95. memcpy(&p[offset], header_code.packet, header_code.bytes);
  96. offset += header_code.bytes;
  97. avccontext->extradata_size = offset;
  98. avccontext->extradata= av_realloc(avccontext->extradata, avccontext->extradata_size);
  99. /* vorbis_block_clear(&context->vb);
  100. vorbis_dsp_clear(&context->vd);
  101. vorbis_info_clear(&context->vi);*/
  102. vorbis_comment_clear(&context->vc);
  103. avccontext->frame_size = OGGVORBIS_FRAME_SIZE ;
  104. avccontext->coded_frame= avcodec_alloc_frame();
  105. avccontext->coded_frame->key_frame= 1;
  106. return 0 ;
  107. }
  108. static int oggvorbis_encode_frame(AVCodecContext *avccontext,
  109. unsigned char *packets,
  110. int buf_size, void *data)
  111. {
  112. OggVorbisContext *context = avccontext->priv_data ;
  113. float **buffer ;
  114. ogg_packet op ;
  115. signed short *audio = data ;
  116. int l, samples = data ? OGGVORBIS_FRAME_SIZE : 0;
  117. buffer = vorbis_analysis_buffer(&context->vd, samples) ;
  118. if(context->vi.channels == 1) {
  119. for(l = 0 ; l < samples ; l++)
  120. buffer[0][l]=audio[l]/32768.f;
  121. } else {
  122. for(l = 0 ; l < samples ; l++){
  123. buffer[0][l]=audio[l*2]/32768.f;
  124. buffer[1][l]=audio[l*2+1]/32768.f;
  125. }
  126. }
  127. vorbis_analysis_wrote(&context->vd, samples) ;
  128. while(vorbis_analysis_blockout(&context->vd, &context->vb) == 1) {
  129. vorbis_analysis(&context->vb, NULL);
  130. vorbis_bitrate_addblock(&context->vb) ;
  131. while(vorbis_bitrate_flushpacket(&context->vd, &op)) {
  132. /* i'd love to say the following line is a hack, but sadly it's
  133. * not, apparently the end of stream decision is in libogg. */
  134. if(op.bytes==1)
  135. continue;
  136. memcpy(context->buffer + context->buffer_index, &op, sizeof(ogg_packet));
  137. context->buffer_index += sizeof(ogg_packet);
  138. memcpy(context->buffer + context->buffer_index, op.packet, op.bytes);
  139. context->buffer_index += op.bytes;
  140. // av_log(avccontext, AV_LOG_DEBUG, "e%d / %d\n", context->buffer_index, op.bytes);
  141. }
  142. }
  143. l=0;
  144. if(context->buffer_index){
  145. ogg_packet *op2= (ogg_packet*)context->buffer;
  146. op2->packet = context->buffer + sizeof(ogg_packet);
  147. l= op2->bytes;
  148. avccontext->coded_frame->pts= av_rescale_q(op2->granulepos, (AVRational){1, avccontext->sample_rate}, avccontext->time_base);
  149. //FIXME we should reorder the user supplied pts and not assume that they are spaced by 1/sample_rate
  150. memcpy(packets, op2->packet, l);
  151. context->buffer_index -= l + sizeof(ogg_packet);
  152. memcpy(context->buffer, context->buffer + l + sizeof(ogg_packet), context->buffer_index);
  153. // av_log(avccontext, AV_LOG_DEBUG, "E%d\n", l);
  154. }
  155. return l;
  156. }
  157. static int oggvorbis_encode_close(AVCodecContext *avccontext) {
  158. OggVorbisContext *context = avccontext->priv_data ;
  159. /* ogg_packet op ; */
  160. vorbis_analysis_wrote(&context->vd, 0) ; /* notify vorbisenc this is EOF */
  161. vorbis_block_clear(&context->vb);
  162. vorbis_dsp_clear(&context->vd);
  163. vorbis_info_clear(&context->vi);
  164. av_freep(&avccontext->coded_frame);
  165. av_freep(&avccontext->extradata);
  166. return 0 ;
  167. }
  168. AVCodec oggvorbis_encoder = {
  169. "vorbis",
  170. CODEC_TYPE_AUDIO,
  171. CODEC_ID_VORBIS,
  172. sizeof(OggVorbisContext),
  173. oggvorbis_encode_init,
  174. oggvorbis_encode_frame,
  175. oggvorbis_encode_close,
  176. .capabilities= CODEC_CAP_DELAY,
  177. } ;
  178. static int oggvorbis_decode_init(AVCodecContext *avccontext) {
  179. OggVorbisContext *context = avccontext->priv_data ;
  180. uint8_t *p= avccontext->extradata;
  181. int i, hsizes[3];
  182. unsigned char *headers[3], *extradata = avccontext->extradata;
  183. vorbis_info_init(&context->vi) ;
  184. vorbis_comment_init(&context->vc) ;
  185. if(! avccontext->extradata_size || ! p) {
  186. av_log(avccontext, AV_LOG_ERROR, "vorbis extradata absent\n");
  187. return -1;
  188. }
  189. if(p[0] == 0 && p[1] == 30) {
  190. for(i = 0; i < 3; i++){
  191. hsizes[i] = *p++ << 8;
  192. hsizes[i] += *p++;
  193. headers[i] = p;
  194. p += hsizes[i];
  195. }
  196. } else if(*p == 2) {
  197. unsigned int offset = 1;
  198. p++;
  199. for(i=0; i<2; i++) {
  200. hsizes[i] = 0;
  201. while((*p == 0xFF) && (offset < avccontext->extradata_size)) {
  202. hsizes[i] += 0xFF;
  203. offset++;
  204. p++;
  205. }
  206. if(offset >= avccontext->extradata_size - 1) {
  207. av_log(avccontext, AV_LOG_ERROR,
  208. "vorbis header sizes damaged\n");
  209. return -1;
  210. }
  211. hsizes[i] += *p;
  212. offset++;
  213. p++;
  214. }
  215. hsizes[2] = avccontext->extradata_size - hsizes[0]-hsizes[1]-offset;
  216. #if 0
  217. av_log(avccontext, AV_LOG_DEBUG,
  218. "vorbis header sizes: %d, %d, %d, / extradata_len is %d \n",
  219. hsizes[0], hsizes[1], hsizes[2], avccontext->extradata_size);
  220. #endif
  221. headers[0] = extradata + offset;
  222. headers[1] = extradata + offset + hsizes[0];
  223. headers[2] = extradata + offset + hsizes[0] + hsizes[1];
  224. } else {
  225. av_log(avccontext, AV_LOG_ERROR,
  226. "vorbis initial header len is wrong: %d\n", *p);
  227. return -1;
  228. }
  229. for(i=0; i<3; i++){
  230. context->op.b_o_s= i==0;
  231. context->op.bytes = hsizes[i];
  232. context->op.packet = headers[i];
  233. if(vorbis_synthesis_headerin(&context->vi, &context->vc, &context->op)<0){
  234. av_log(avccontext, AV_LOG_ERROR, "%d. vorbis header damaged\n", i+1);
  235. return -1;
  236. }
  237. }
  238. avccontext->channels = context->vi.channels;
  239. avccontext->sample_rate = context->vi.rate;
  240. avccontext->time_base= (AVRational){1, avccontext->sample_rate};
  241. vorbis_synthesis_init(&context->vd, &context->vi);
  242. vorbis_block_init(&context->vd, &context->vb);
  243. return 0 ;
  244. }
  245. static inline int conv(int samples, float **pcm, char *buf, int channels) {
  246. int i, j, val ;
  247. ogg_int16_t *ptr, *data = (ogg_int16_t*)buf ;
  248. float *mono ;
  249. for(i = 0 ; i < channels ; i++){
  250. ptr = &data[i];
  251. mono = pcm[i] ;
  252. for(j = 0 ; j < samples ; j++) {
  253. val = mono[j] * 32767.f;
  254. if(val > 32767) val = 32767 ;
  255. if(val < -32768) val = -32768 ;
  256. *ptr = val ;
  257. ptr += channels;
  258. }
  259. }
  260. return 0 ;
  261. }
  262. static int oggvorbis_decode_frame(AVCodecContext *avccontext,
  263. void *data, int *data_size,
  264. uint8_t *buf, int buf_size)
  265. {
  266. OggVorbisContext *context = avccontext->priv_data ;
  267. float **pcm ;
  268. ogg_packet *op= &context->op;
  269. int samples, total_samples, total_bytes;
  270. if(!buf_size){
  271. //FIXME flush
  272. return 0;
  273. }
  274. op->packet = buf;
  275. op->bytes = buf_size;
  276. // av_log(avccontext, AV_LOG_DEBUG, "%d %d %d %"PRId64" %"PRId64" %d %d\n", op->bytes, op->b_o_s, op->e_o_s, op->granulepos, op->packetno, buf_size, context->vi.rate);
  277. /* for(i=0; i<op->bytes; i++)
  278. av_log(avccontext, AV_LOG_DEBUG, "%02X ", op->packet[i]);
  279. av_log(avccontext, AV_LOG_DEBUG, "\n");*/
  280. if(vorbis_synthesis(&context->vb, op) == 0)
  281. vorbis_synthesis_blockin(&context->vd, &context->vb) ;
  282. total_samples = 0 ;
  283. total_bytes = 0 ;
  284. while((samples = vorbis_synthesis_pcmout(&context->vd, &pcm)) > 0) {
  285. conv(samples, pcm, (char*)data + total_bytes, context->vi.channels) ;
  286. total_bytes += samples * 2 * context->vi.channels ;
  287. total_samples += samples ;
  288. vorbis_synthesis_read(&context->vd, samples) ;
  289. }
  290. *data_size = total_bytes ;
  291. return buf_size ;
  292. }
  293. static int oggvorbis_decode_close(AVCodecContext *avccontext) {
  294. OggVorbisContext *context = avccontext->priv_data ;
  295. vorbis_info_clear(&context->vi) ;
  296. vorbis_comment_clear(&context->vc) ;
  297. return 0 ;
  298. }
  299. AVCodec oggvorbis_decoder = {
  300. "vorbis",
  301. CODEC_TYPE_AUDIO,
  302. CODEC_ID_VORBIS,
  303. sizeof(OggVorbisContext),
  304. oggvorbis_decode_init,
  305. NULL,
  306. oggvorbis_decode_close,
  307. oggvorbis_decode_frame,
  308. .capabilities= CODEC_CAP_DELAY,
  309. } ;