| 1 | /* |
| 2 | * Copyright (c) 2011 Stefano Sabatini |
| 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 Street, Fifth Floor, Boston, MA 02110-1301 USA |
| 19 | */ |
| 20 | |
| 21 | /** |
| 22 | * @file |
| 23 | * eval audio source |
| 24 | */ |
| 25 | |
| 26 | #include "libavutil/avassert.h" |
| 27 | #include "libavutil/avstring.h" |
| 28 | #include "libavutil/channel_layout.h" |
| 29 | #include "libavutil/eval.h" |
| 30 | #include "libavutil/opt.h" |
| 31 | #include "libavutil/parseutils.h" |
| 32 | #include "avfilter.h" |
| 33 | #include "audio.h" |
| 34 | #include "internal.h" |
| 35 | |
| 36 | static const char * const var_names[] = { |
| 37 | "ch", ///< the value of the current channel |
| 38 | "n", ///< number of frame |
| 39 | "nb_in_channels", |
| 40 | "nb_out_channels", |
| 41 | "t", ///< timestamp expressed in seconds |
| 42 | "s", ///< sample rate |
| 43 | NULL |
| 44 | }; |
| 45 | |
| 46 | enum var_name { |
| 47 | VAR_CH, |
| 48 | VAR_N, |
| 49 | VAR_NB_IN_CHANNELS, |
| 50 | VAR_NB_OUT_CHANNELS, |
| 51 | VAR_T, |
| 52 | VAR_S, |
| 53 | VAR_VARS_NB |
| 54 | }; |
| 55 | |
| 56 | typedef struct { |
| 57 | const AVClass *class; |
| 58 | char *sample_rate_str; |
| 59 | int sample_rate; |
| 60 | int64_t chlayout; |
| 61 | char *chlayout_str; |
| 62 | int nb_channels; ///< number of output channels |
| 63 | int nb_in_channels; ///< number of input channels |
| 64 | int same_chlayout; ///< set output as input channel layout |
| 65 | int64_t pts; |
| 66 | AVExpr **expr; |
| 67 | char *exprs; |
| 68 | int nb_samples; ///< number of samples per requested frame |
| 69 | int64_t duration; |
| 70 | uint64_t n; |
| 71 | double var_values[VAR_VARS_NB]; |
| 72 | double *channel_values; |
| 73 | int64_t out_channel_layout; |
| 74 | } EvalContext; |
| 75 | |
| 76 | static double val(void *priv, double ch) |
| 77 | { |
| 78 | EvalContext *eval = priv; |
| 79 | return eval->channel_values[FFMIN((int)ch, eval->nb_in_channels-1)]; |
| 80 | } |
| 81 | |
| 82 | static double (* const aeval_func1[])(void *, double) = { val, NULL }; |
| 83 | static const char * const aeval_func1_names[] = { "val", NULL }; |
| 84 | |
| 85 | #define OFFSET(x) offsetof(EvalContext, x) |
| 86 | #define FLAGS AV_OPT_FLAG_AUDIO_PARAM|AV_OPT_FLAG_FILTERING_PARAM |
| 87 | |
| 88 | static const AVOption aevalsrc_options[]= { |
| 89 | { "exprs", "set the '|'-separated list of channels expressions", OFFSET(exprs), AV_OPT_TYPE_STRING, {.str = NULL}, .flags = FLAGS }, |
| 90 | { "nb_samples", "set the number of samples per requested frame", OFFSET(nb_samples), AV_OPT_TYPE_INT, {.i64 = 1024}, 0, INT_MAX, FLAGS }, |
| 91 | { "n", "set the number of samples per requested frame", OFFSET(nb_samples), AV_OPT_TYPE_INT, {.i64 = 1024}, 0, INT_MAX, FLAGS }, |
| 92 | { "sample_rate", "set the sample rate", OFFSET(sample_rate_str), AV_OPT_TYPE_STRING, {.str = "44100"}, CHAR_MIN, CHAR_MAX, FLAGS }, |
| 93 | { "s", "set the sample rate", OFFSET(sample_rate_str), AV_OPT_TYPE_STRING, {.str = "44100"}, CHAR_MIN, CHAR_MAX, FLAGS }, |
| 94 | { "duration", "set audio duration", OFFSET(duration), AV_OPT_TYPE_DURATION, {.i64 = -1}, -1, INT64_MAX, FLAGS }, |
| 95 | { "d", "set audio duration", OFFSET(duration), AV_OPT_TYPE_DURATION, {.i64 = -1}, -1, INT64_MAX, FLAGS }, |
| 96 | { "channel_layout", "set channel layout", OFFSET(chlayout_str), AV_OPT_TYPE_STRING, {.str = NULL}, 0, 0, FLAGS }, |
| 97 | { "c", "set channel layout", OFFSET(chlayout_str), AV_OPT_TYPE_STRING, {.str = NULL}, 0, 0, FLAGS }, |
| 98 | { NULL } |
| 99 | }; |
| 100 | |
| 101 | AVFILTER_DEFINE_CLASS(aevalsrc); |
| 102 | |
| 103 | static int parse_channel_expressions(AVFilterContext *ctx, |
| 104 | int expected_nb_channels) |
| 105 | { |
| 106 | EvalContext *eval = ctx->priv; |
| 107 | char *args1 = av_strdup(eval->exprs); |
| 108 | char *expr, *last_expr = NULL, *buf; |
| 109 | double (* const *func1)(void *, double) = NULL; |
| 110 | const char * const *func1_names = NULL; |
| 111 | int i, ret = 0; |
| 112 | |
| 113 | if (!args1) |
| 114 | return AVERROR(ENOMEM); |
| 115 | |
| 116 | if (!eval->exprs) { |
| 117 | av_log(ctx, AV_LOG_ERROR, "Channels expressions list is empty\n"); |
| 118 | return AVERROR(EINVAL); |
| 119 | } |
| 120 | |
| 121 | if (!strcmp(ctx->filter->name, "aeval")) { |
| 122 | func1 = aeval_func1; |
| 123 | func1_names = aeval_func1_names; |
| 124 | } |
| 125 | |
| 126 | #define ADD_EXPRESSION(expr_) do { \ |
| 127 | if (!av_dynarray2_add((void **)&eval->expr, &eval->nb_channels, \ |
| 128 | sizeof(*eval->expr), NULL)) { \ |
| 129 | ret = AVERROR(ENOMEM); \ |
| 130 | goto end; \ |
| 131 | } \ |
| 132 | eval->expr[eval->nb_channels-1] = NULL; \ |
| 133 | ret = av_expr_parse(&eval->expr[eval->nb_channels - 1], expr_, \ |
| 134 | var_names, func1_names, func1, \ |
| 135 | NULL, NULL, 0, ctx); \ |
| 136 | if (ret < 0) \ |
| 137 | goto end; \ |
| 138 | } while (0) |
| 139 | |
| 140 | /* reset expressions */ |
| 141 | for (i = 0; i < eval->nb_channels; i++) { |
| 142 | av_expr_free(eval->expr[i]); |
| 143 | eval->expr[i] = NULL; |
| 144 | } |
| 145 | av_freep(&eval->expr); |
| 146 | eval->nb_channels = 0; |
| 147 | |
| 148 | buf = args1; |
| 149 | while (expr = av_strtok(buf, "|", &buf)) { |
| 150 | ADD_EXPRESSION(expr); |
| 151 | last_expr = expr; |
| 152 | } |
| 153 | |
| 154 | if (expected_nb_channels > eval->nb_channels) |
| 155 | for (i = eval->nb_channels; i < expected_nb_channels; i++) |
| 156 | ADD_EXPRESSION(last_expr); |
| 157 | |
| 158 | if (expected_nb_channels > 0 && eval->nb_channels != expected_nb_channels) { |
| 159 | av_log(ctx, AV_LOG_ERROR, |
| 160 | "Mismatch between the specified number of channel expressions '%d' " |
| 161 | "and the number of expected output channels '%d' for the specified channel layout\n", |
| 162 | eval->nb_channels, expected_nb_channels); |
| 163 | ret = AVERROR(EINVAL); |
| 164 | goto end; |
| 165 | } |
| 166 | |
| 167 | end: |
| 168 | av_free(args1); |
| 169 | return ret; |
| 170 | } |
| 171 | |
| 172 | static av_cold int init(AVFilterContext *ctx) |
| 173 | { |
| 174 | EvalContext *eval = ctx->priv; |
| 175 | int ret = 0; |
| 176 | |
| 177 | if (eval->chlayout_str) { |
| 178 | if (!strcmp(eval->chlayout_str, "same") && !strcmp(ctx->filter->name, "aeval")) { |
| 179 | eval->same_chlayout = 1; |
| 180 | } else { |
| 181 | ret = ff_parse_channel_layout(&eval->chlayout, NULL, eval->chlayout_str, ctx); |
| 182 | if (ret < 0) |
| 183 | return ret; |
| 184 | |
| 185 | ret = parse_channel_expressions(ctx, av_get_channel_layout_nb_channels(eval->chlayout)); |
| 186 | if (ret < 0) |
| 187 | return ret; |
| 188 | } |
| 189 | } else { |
| 190 | /* guess channel layout from nb expressions/channels */ |
| 191 | if ((ret = parse_channel_expressions(ctx, -1)) < 0) |
| 192 | return ret; |
| 193 | |
| 194 | eval->chlayout = av_get_default_channel_layout(eval->nb_channels); |
| 195 | if (!eval->chlayout && eval->nb_channels <= 0) { |
| 196 | av_log(ctx, AV_LOG_ERROR, "Invalid number of channels '%d' provided\n", |
| 197 | eval->nb_channels); |
| 198 | return AVERROR(EINVAL); |
| 199 | } |
| 200 | } |
| 201 | |
| 202 | if (eval->sample_rate_str) |
| 203 | if ((ret = ff_parse_sample_rate(&eval->sample_rate, eval->sample_rate_str, ctx))) |
| 204 | return ret; |
| 205 | eval->n = 0; |
| 206 | |
| 207 | return ret; |
| 208 | } |
| 209 | |
| 210 | static av_cold void uninit(AVFilterContext *ctx) |
| 211 | { |
| 212 | EvalContext *eval = ctx->priv; |
| 213 | int i; |
| 214 | |
| 215 | for (i = 0; i < eval->nb_channels; i++) { |
| 216 | av_expr_free(eval->expr[i]); |
| 217 | eval->expr[i] = NULL; |
| 218 | } |
| 219 | av_freep(&eval->expr); |
| 220 | } |
| 221 | |
| 222 | static int config_props(AVFilterLink *outlink) |
| 223 | { |
| 224 | EvalContext *eval = outlink->src->priv; |
| 225 | char buf[128]; |
| 226 | |
| 227 | outlink->time_base = (AVRational){1, eval->sample_rate}; |
| 228 | outlink->sample_rate = eval->sample_rate; |
| 229 | |
| 230 | eval->var_values[VAR_S] = eval->sample_rate; |
| 231 | eval->var_values[VAR_NB_IN_CHANNELS] = NAN; |
| 232 | eval->var_values[VAR_NB_OUT_CHANNELS] = outlink->channels; |
| 233 | |
| 234 | av_get_channel_layout_string(buf, sizeof(buf), 0, eval->chlayout); |
| 235 | |
| 236 | av_log(outlink->src, AV_LOG_VERBOSE, |
| 237 | "sample_rate:%d chlayout:%s duration:%"PRId64"\n", |
| 238 | eval->sample_rate, buf, eval->duration); |
| 239 | |
| 240 | return 0; |
| 241 | } |
| 242 | |
| 243 | static int query_formats(AVFilterContext *ctx) |
| 244 | { |
| 245 | EvalContext *eval = ctx->priv; |
| 246 | static const enum AVSampleFormat sample_fmts[] = { AV_SAMPLE_FMT_DBLP, AV_SAMPLE_FMT_NONE }; |
| 247 | int64_t chlayouts[] = { eval->chlayout ? eval->chlayout : FF_COUNT2LAYOUT(eval->nb_channels) , -1 }; |
| 248 | int sample_rates[] = { eval->sample_rate, -1 }; |
| 249 | |
| 250 | ff_set_common_formats (ctx, ff_make_format_list(sample_fmts)); |
| 251 | ff_set_common_channel_layouts(ctx, avfilter_make_format64_list(chlayouts)); |
| 252 | ff_set_common_samplerates(ctx, ff_make_format_list(sample_rates)); |
| 253 | |
| 254 | return 0; |
| 255 | } |
| 256 | |
| 257 | static int request_frame(AVFilterLink *outlink) |
| 258 | { |
| 259 | EvalContext *eval = outlink->src->priv; |
| 260 | AVFrame *samplesref; |
| 261 | int i, j; |
| 262 | int64_t t = av_rescale(eval->n, AV_TIME_BASE, eval->sample_rate); |
| 263 | |
| 264 | if (eval->duration >= 0 && t >= eval->duration) |
| 265 | return AVERROR_EOF; |
| 266 | |
| 267 | samplesref = ff_get_audio_buffer(outlink, eval->nb_samples); |
| 268 | if (!samplesref) |
| 269 | return AVERROR(ENOMEM); |
| 270 | |
| 271 | /* evaluate expression for each single sample and for each channel */ |
| 272 | for (i = 0; i < eval->nb_samples; i++, eval->n++) { |
| 273 | eval->var_values[VAR_N] = eval->n; |
| 274 | eval->var_values[VAR_T] = eval->var_values[VAR_N] * (double)1/eval->sample_rate; |
| 275 | |
| 276 | for (j = 0; j < eval->nb_channels; j++) { |
| 277 | *((double *) samplesref->extended_data[j] + i) = |
| 278 | av_expr_eval(eval->expr[j], eval->var_values, NULL); |
| 279 | } |
| 280 | } |
| 281 | |
| 282 | samplesref->pts = eval->pts; |
| 283 | samplesref->sample_rate = eval->sample_rate; |
| 284 | eval->pts += eval->nb_samples; |
| 285 | |
| 286 | return ff_filter_frame(outlink, samplesref); |
| 287 | } |
| 288 | |
| 289 | #if CONFIG_AEVALSRC_FILTER |
| 290 | static const AVFilterPad aevalsrc_outputs[] = { |
| 291 | { |
| 292 | .name = "default", |
| 293 | .type = AVMEDIA_TYPE_AUDIO, |
| 294 | .config_props = config_props, |
| 295 | .request_frame = request_frame, |
| 296 | }, |
| 297 | { NULL } |
| 298 | }; |
| 299 | |
| 300 | AVFilter ff_asrc_aevalsrc = { |
| 301 | .name = "aevalsrc", |
| 302 | .description = NULL_IF_CONFIG_SMALL("Generate an audio signal generated by an expression."), |
| 303 | .query_formats = query_formats, |
| 304 | .init = init, |
| 305 | .uninit = uninit, |
| 306 | .priv_size = sizeof(EvalContext), |
| 307 | .inputs = NULL, |
| 308 | .outputs = aevalsrc_outputs, |
| 309 | .priv_class = &aevalsrc_class, |
| 310 | }; |
| 311 | |
| 312 | #endif /* CONFIG_AEVALSRC_FILTER */ |
| 313 | |
| 314 | #define OFFSET(x) offsetof(EvalContext, x) |
| 315 | #define FLAGS AV_OPT_FLAG_AUDIO_PARAM|AV_OPT_FLAG_FILTERING_PARAM |
| 316 | |
| 317 | static const AVOption aeval_options[]= { |
| 318 | { "exprs", "set the '|'-separated list of channels expressions", OFFSET(exprs), AV_OPT_TYPE_STRING, {.str = NULL}, .flags = FLAGS }, |
| 319 | { "channel_layout", "set channel layout", OFFSET(chlayout_str), AV_OPT_TYPE_STRING, {.str = NULL}, 0, 0, FLAGS }, |
| 320 | { "c", "set channel layout", OFFSET(chlayout_str), AV_OPT_TYPE_STRING, {.str = NULL}, 0, 0, FLAGS }, |
| 321 | { NULL } |
| 322 | }; |
| 323 | |
| 324 | AVFILTER_DEFINE_CLASS(aeval); |
| 325 | |
| 326 | static int aeval_query_formats(AVFilterContext *ctx) |
| 327 | { |
| 328 | AVFilterFormats *formats = NULL; |
| 329 | AVFilterChannelLayouts *layouts; |
| 330 | AVFilterLink *inlink = ctx->inputs[0]; |
| 331 | AVFilterLink *outlink = ctx->outputs[0]; |
| 332 | EvalContext *eval = ctx->priv; |
| 333 | static const enum AVSampleFormat sample_fmts[] = { |
| 334 | AV_SAMPLE_FMT_DBLP, AV_SAMPLE_FMT_NONE |
| 335 | }; |
| 336 | |
| 337 | // inlink supports any channel layout |
| 338 | layouts = ff_all_channel_counts(); |
| 339 | ff_channel_layouts_ref(layouts, &inlink->out_channel_layouts); |
| 340 | |
| 341 | if (eval->same_chlayout) { |
| 342 | layouts = ff_all_channel_counts(); |
| 343 | if (!layouts) |
| 344 | return AVERROR(ENOMEM); |
| 345 | ff_set_common_channel_layouts(ctx, layouts); |
| 346 | } else { |
| 347 | // outlink supports only requested output channel layout |
| 348 | layouts = NULL; |
| 349 | ff_add_channel_layout(&layouts, |
| 350 | eval->out_channel_layout ? eval->out_channel_layout : |
| 351 | FF_COUNT2LAYOUT(eval->nb_channels)); |
| 352 | ff_channel_layouts_ref(layouts, &outlink->in_channel_layouts); |
| 353 | } |
| 354 | |
| 355 | formats = ff_make_format_list(sample_fmts); |
| 356 | if (!formats) |
| 357 | return AVERROR(ENOMEM); |
| 358 | ff_set_common_formats(ctx, formats); |
| 359 | |
| 360 | formats = ff_all_samplerates(); |
| 361 | if (!formats) |
| 362 | return AVERROR(ENOMEM); |
| 363 | ff_set_common_samplerates(ctx, formats); |
| 364 | |
| 365 | return 0; |
| 366 | } |
| 367 | |
| 368 | static int aeval_config_output(AVFilterLink *outlink) |
| 369 | { |
| 370 | AVFilterContext *ctx = outlink->src; |
| 371 | EvalContext *eval = ctx->priv; |
| 372 | AVFilterLink *inlink = ctx->inputs[0]; |
| 373 | int ret; |
| 374 | |
| 375 | if (eval->same_chlayout) { |
| 376 | eval->chlayout = inlink->channel_layout; |
| 377 | |
| 378 | if ((ret = parse_channel_expressions(ctx, inlink->channels)) < 0) |
| 379 | return ret; |
| 380 | } |
| 381 | |
| 382 | eval->n = 0; |
| 383 | eval->nb_in_channels = eval->var_values[VAR_NB_IN_CHANNELS] = inlink->channels; |
| 384 | eval->var_values[VAR_NB_OUT_CHANNELS] = outlink->channels; |
| 385 | eval->var_values[VAR_S] = inlink->sample_rate; |
| 386 | eval->var_values[VAR_T] = NAN; |
| 387 | |
| 388 | eval->channel_values = av_realloc_f(eval->channel_values, |
| 389 | inlink->channels, sizeof(*eval->channel_values)); |
| 390 | if (!eval->channel_values) |
| 391 | return AVERROR(ENOMEM); |
| 392 | |
| 393 | return 0; |
| 394 | } |
| 395 | |
| 396 | #define TS2T(ts, tb) ((ts) == AV_NOPTS_VALUE ? NAN : (double)(ts)*av_q2d(tb)) |
| 397 | |
| 398 | static int filter_frame(AVFilterLink *inlink, AVFrame *in) |
| 399 | { |
| 400 | EvalContext *eval = inlink->dst->priv; |
| 401 | AVFilterLink *outlink = inlink->dst->outputs[0]; |
| 402 | int nb_samples = in->nb_samples; |
| 403 | AVFrame *out; |
| 404 | double t0; |
| 405 | int i, j; |
| 406 | |
| 407 | /* do volume scaling in-place if input buffer is writable */ |
| 408 | out = ff_get_audio_buffer(outlink, nb_samples); |
| 409 | if (!out) |
| 410 | return AVERROR(ENOMEM); |
| 411 | av_frame_copy_props(out, in); |
| 412 | |
| 413 | t0 = TS2T(in->pts, inlink->time_base); |
| 414 | |
| 415 | /* evaluate expression for each single sample and for each channel */ |
| 416 | for (i = 0; i < nb_samples; i++, eval->n++) { |
| 417 | eval->var_values[VAR_N] = eval->n; |
| 418 | eval->var_values[VAR_T] = t0 + i * (double)1/inlink->sample_rate; |
| 419 | |
| 420 | for (j = 0; j < inlink->channels; j++) |
| 421 | eval->channel_values[j] = *((double *) in->extended_data[j] + i); |
| 422 | |
| 423 | for (j = 0; j < outlink->channels; j++) { |
| 424 | eval->var_values[VAR_CH] = j; |
| 425 | *((double *) out->extended_data[j] + i) = |
| 426 | av_expr_eval(eval->expr[j], eval->var_values, eval); |
| 427 | } |
| 428 | } |
| 429 | |
| 430 | av_frame_free(&in); |
| 431 | return ff_filter_frame(outlink, out); |
| 432 | } |
| 433 | |
| 434 | #if CONFIG_AEVAL_FILTER |
| 435 | |
| 436 | static const AVFilterPad aeval_inputs[] = { |
| 437 | { |
| 438 | .name = "default", |
| 439 | .type = AVMEDIA_TYPE_AUDIO, |
| 440 | .filter_frame = filter_frame, |
| 441 | }, |
| 442 | { NULL } |
| 443 | }; |
| 444 | |
| 445 | static const AVFilterPad aeval_outputs[] = { |
| 446 | { |
| 447 | .name = "default", |
| 448 | .type = AVMEDIA_TYPE_AUDIO, |
| 449 | .config_props = aeval_config_output, |
| 450 | }, |
| 451 | { NULL } |
| 452 | }; |
| 453 | |
| 454 | AVFilter ff_af_aeval = { |
| 455 | .name = "aeval", |
| 456 | .description = NULL_IF_CONFIG_SMALL("Filter audio signal according to a specified expression."), |
| 457 | .query_formats = aeval_query_formats, |
| 458 | .init = init, |
| 459 | .uninit = uninit, |
| 460 | .priv_size = sizeof(EvalContext), |
| 461 | .inputs = aeval_inputs, |
| 462 | .outputs = aeval_outputs, |
| 463 | .priv_class = &aeval_class, |
| 464 | }; |
| 465 | |
| 466 | #endif /* CONFIG_AEVAL_FILTER */ |