| /* |
| * Copyright (c) 2021 Paul B Mahol |
| * |
| * This file is part of FFmpeg. |
| * |
| * FFmpeg is free software; you can redistribute it and/or |
| * modify it under the terms of the GNU Lesser General Public |
| * License as published by the Free Software Foundation; either |
| * version 2.1 of the License, or (at your option) any later version. |
| * |
| * FFmpeg is distributed in the hope that it will be useful, |
| * but WITHOUT ANY WARRANTY; without even the implied warranty of |
| * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU |
| * Lesser General Public License for more details. |
| * |
| * You should have received a copy of the GNU Lesser General Public |
| * License along with FFmpeg; if not, write to the Free Software |
| * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA |
| */ |
| |
| /** |
| * @file |
| * Change the PTS/DTS timestamps. |
| */ |
| |
| #include "libavutil/opt.h" |
| #include "libavutil/eval.h" |
| |
| #include "bsf.h" |
| #include "bsf_internal.h" |
| |
| static const char *const var_names[] = { |
| "N", ///< frame number (starting at zero) |
| "TS", |
| "POS", ///< original position in the file of the frame |
| "PREV_INPTS", ///< previous input PTS |
| "PREV_INDTS", ///< previous input DTS |
| "PREV_INDURATION", ///< previous input duration |
| "PREV_OUTPTS", ///< previous output PTS |
| "PREV_OUTDTS", ///< previous output DTS |
| "PREV_OUTDURATION", ///< previous output duration |
| "NEXT_PTS", ///< next input PTS |
| "NEXT_DTS", ///< next input DTS |
| "NEXT_DURATION", ///< next input duration |
| "PTS", ///< original PTS in the file of the frame |
| "DTS", ///< original DTS in the file of the frame |
| "DURATION", ///< original duration in the file of the frame |
| "STARTPTS", ///< PTS at start of movie |
| "STARTDTS", ///< DTS at start of movie |
| "TB", ///< input timebase of the stream |
| "TB_OUT", ///< output timebase of the stream |
| "SR", ///< sample rate of the stream |
| "NOPTS", ///< The AV_NOPTS_VALUE constant |
| NULL |
| }; |
| |
| enum var_name { |
| VAR_N, |
| VAR_TS, |
| VAR_POS, |
| VAR_PREV_INPTS, |
| VAR_PREV_INDTS, |
| VAR_PREV_INDUR, |
| VAR_PREV_OUTPTS, |
| VAR_PREV_OUTDTS, |
| VAR_PREV_OUTDUR, |
| VAR_NEXT_PTS, |
| VAR_NEXT_DTS, |
| VAR_NEXT_DUR, |
| VAR_PTS, |
| VAR_DTS, |
| VAR_DURATION, |
| VAR_STARTPTS, |
| VAR_STARTDTS, |
| VAR_TB, |
| VAR_TB_OUT, |
| VAR_SR, |
| VAR_NOPTS, |
| VAR_VARS_NB |
| }; |
| |
| typedef struct SetTSContext { |
| const AVClass *class; |
| |
| char *ts_str; |
| char *pts_str; |
| char *dts_str; |
| char *duration_str; |
| |
| AVRational time_base; |
| |
| int64_t frame_number; |
| |
| double var_values[VAR_VARS_NB]; |
| |
| AVExpr *ts_expr; |
| AVExpr *pts_expr; |
| AVExpr *dts_expr; |
| AVExpr *duration_expr; |
| |
| AVPacket *prev_inpkt; |
| AVPacket *prev_outpkt; |
| AVPacket *cur_pkt; |
| } SetTSContext; |
| |
| static int setts_init(AVBSFContext *ctx) |
| { |
| SetTSContext *s = ctx->priv_data; |
| int ret; |
| |
| s->prev_inpkt = av_packet_alloc(); |
| s->prev_outpkt = av_packet_alloc(); |
| s->cur_pkt = av_packet_alloc(); |
| if (!s->prev_inpkt || !s->prev_outpkt || !s->cur_pkt) |
| return AVERROR(ENOMEM); |
| |
| if ((ret = av_expr_parse(&s->ts_expr, s->ts_str, |
| var_names, NULL, NULL, NULL, NULL, 0, ctx)) < 0) { |
| av_log(ctx, AV_LOG_ERROR, "Error while parsing ts expression '%s'\n", s->ts_str); |
| return ret; |
| } |
| |
| if ((ret = av_expr_parse(&s->duration_expr, s->duration_str, |
| var_names, NULL, NULL, NULL, NULL, 0, ctx)) < 0) { |
| av_log(ctx, AV_LOG_ERROR, "Error while parsing duration expression '%s'\n", s->duration_str); |
| return ret; |
| } |
| |
| if (s->pts_str) { |
| if ((ret = av_expr_parse(&s->pts_expr, s->pts_str, |
| var_names, NULL, NULL, NULL, NULL, 0, ctx)) < 0) { |
| av_log(ctx, AV_LOG_ERROR, "Error while parsing pts expression '%s'\n", s->pts_str); |
| return ret; |
| } |
| } |
| |
| if (s->dts_str) { |
| if ((ret = av_expr_parse(&s->dts_expr, s->dts_str, |
| var_names, NULL, NULL, NULL, NULL, 0, ctx)) < 0) { |
| av_log(ctx, AV_LOG_ERROR, "Error while parsing dts expression '%s'\n", s->dts_str); |
| return ret; |
| } |
| } |
| |
| if (s->time_base.num > 0 && s->time_base.den > 0) |
| ctx->time_base_out = s->time_base; |
| |
| s->frame_number= 0; |
| s->var_values[VAR_STARTPTS] = AV_NOPTS_VALUE; |
| s->var_values[VAR_STARTDTS] = AV_NOPTS_VALUE; |
| s->var_values[VAR_NOPTS] = AV_NOPTS_VALUE; |
| s->var_values[VAR_TB] = ctx->time_base_in.den ? av_q2d(ctx->time_base_in) : 0; |
| s->var_values[VAR_TB_OUT]= ctx->time_base_out.den ? av_q2d(ctx->time_base_out) : 0; |
| s->var_values[VAR_SR] = ctx->par_in->sample_rate; |
| |
| return 0; |
| } |
| |
| static int setts_filter(AVBSFContext *ctx, AVPacket *pkt) |
| { |
| SetTSContext *s = ctx->priv_data; |
| int64_t new_ts, new_pts, new_dts, new_duration; |
| int ret; |
| |
| ret = ff_bsf_get_packet_ref(ctx, pkt); |
| if (ret < 0 && (ret != AVERROR_EOF || !s->cur_pkt->data)) |
| return ret; |
| |
| if (!s->cur_pkt->data) { |
| av_packet_move_ref(s->cur_pkt, pkt); |
| return AVERROR(EAGAIN); |
| } |
| |
| if (s->var_values[VAR_STARTPTS] == AV_NOPTS_VALUE) |
| s->var_values[VAR_STARTPTS] = s->cur_pkt->pts; |
| |
| if (s->var_values[VAR_STARTDTS] == AV_NOPTS_VALUE) |
| s->var_values[VAR_STARTDTS] = s->cur_pkt->dts; |
| |
| s->var_values[VAR_N] = s->frame_number++; |
| s->var_values[VAR_TS] = s->cur_pkt->dts; |
| s->var_values[VAR_POS] = s->cur_pkt->pos; |
| s->var_values[VAR_PTS] = s->cur_pkt->pts; |
| s->var_values[VAR_DTS] = s->cur_pkt->dts; |
| s->var_values[VAR_DURATION] = s->cur_pkt->duration; |
| s->var_values[VAR_PREV_INPTS] = s->prev_inpkt->pts; |
| s->var_values[VAR_PREV_INDTS] = s->prev_inpkt->dts; |
| s->var_values[VAR_PREV_INDUR] = s->prev_inpkt->duration; |
| s->var_values[VAR_PREV_OUTPTS] = s->prev_outpkt->pts; |
| s->var_values[VAR_PREV_OUTDTS] = s->prev_outpkt->dts; |
| s->var_values[VAR_PREV_OUTDUR] = s->prev_outpkt->duration; |
| s->var_values[VAR_NEXT_PTS] = pkt->pts; |
| s->var_values[VAR_NEXT_DTS] = pkt->dts; |
| s->var_values[VAR_NEXT_DUR] = pkt->duration; |
| |
| new_ts = llrint(av_expr_eval(s->ts_expr, s->var_values, NULL)); |
| new_duration = llrint(av_expr_eval(s->duration_expr, s->var_values, NULL)); |
| |
| if (s->pts_str) { |
| s->var_values[VAR_TS] = s->cur_pkt->pts; |
| new_pts = llrint(av_expr_eval(s->pts_expr, s->var_values, NULL)); |
| } else { |
| new_pts = new_ts; |
| } |
| |
| if (s->dts_str) { |
| s->var_values[VAR_TS] = s->cur_pkt->dts; |
| new_dts = llrint(av_expr_eval(s->dts_expr, s->var_values, NULL)); |
| } else { |
| new_dts = new_ts; |
| } |
| |
| av_packet_unref(s->prev_inpkt); |
| av_packet_unref(s->prev_outpkt); |
| av_packet_move_ref(s->prev_inpkt, s->cur_pkt); |
| av_packet_move_ref(s->cur_pkt, pkt); |
| |
| ret = av_packet_ref(pkt, s->prev_inpkt); |
| if (ret < 0) |
| return ret; |
| |
| pkt->pts = new_pts; |
| pkt->dts = new_dts; |
| pkt->duration = new_duration; |
| |
| ret = av_packet_ref(s->prev_outpkt, pkt); |
| if (ret < 0) |
| av_packet_unref(pkt); |
| |
| return ret; |
| } |
| |
| static void setts_close(AVBSFContext *bsf) |
| { |
| SetTSContext *s = bsf->priv_data; |
| |
| av_packet_free(&s->prev_inpkt); |
| av_packet_free(&s->prev_outpkt); |
| av_packet_free(&s->cur_pkt); |
| |
| av_expr_free(s->ts_expr); |
| s->ts_expr = NULL; |
| av_expr_free(s->pts_expr); |
| s->pts_expr = NULL; |
| av_expr_free(s->dts_expr); |
| s->dts_expr = NULL; |
| av_expr_free(s->duration_expr); |
| s->duration_expr = NULL; |
| } |
| |
| #define OFFSET(x) offsetof(SetTSContext, x) |
| #define FLAGS (AV_OPT_FLAG_VIDEO_PARAM|AV_OPT_FLAG_AUDIO_PARAM|AV_OPT_FLAG_SUBTITLE_PARAM|AV_OPT_FLAG_BSF_PARAM) |
| |
| static const AVOption options[] = { |
| { "ts", "set expression for packet PTS and DTS", OFFSET(ts_str), AV_OPT_TYPE_STRING, {.str="TS"}, 0, 0, FLAGS }, |
| { "pts", "set expression for packet PTS", OFFSET(pts_str), AV_OPT_TYPE_STRING, {.str=NULL}, 0, 0, FLAGS }, |
| { "dts", "set expression for packet DTS", OFFSET(dts_str), AV_OPT_TYPE_STRING, {.str=NULL}, 0, 0, FLAGS }, |
| { "duration", "set expression for packet duration", OFFSET(duration_str), AV_OPT_TYPE_STRING, {.str="DURATION"}, 0, 0, FLAGS }, |
| { "time_base", "set output timebase", OFFSET(time_base), AV_OPT_TYPE_RATIONAL, {.dbl=0}, 0, INT_MAX, FLAGS }, |
| { NULL }, |
| }; |
| |
| static const AVClass setts_class = { |
| .class_name = "setts_bsf", |
| .item_name = av_default_item_name, |
| .option = options, |
| .version = LIBAVUTIL_VERSION_INT, |
| }; |
| |
| const FFBitStreamFilter ff_setts_bsf = { |
| .p.name = "setts", |
| .p.priv_class = &setts_class, |
| .priv_data_size = sizeof(SetTSContext), |
| .init = setts_init, |
| .close = setts_close, |
| .filter = setts_filter, |
| }; |