view rl2.c @ 5482:68c900a2d203 libavformat

Deprecate and mark for deletion the function guess_stream_format(), and clone its code to ffserver_guess_format() in ffserver.c. guess_stream_format() is hackish since it relies on some undocumented properties of the name of the muxers (wich is currently only relevant for the ASF muxer), and has no use outside ffserver.c.
author stefano
date Thu, 31 Dec 2009 14:12:58 +0000
parents 6b79937e78ae
children 536e5527c1e0
line wrap: on
line source

/*
 * RL2 Format Demuxer
 * Copyright (c) 2008 Sascha Sommer (saschasommer@freenet.de)
 *
 * 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
 */

/**
 * RL2 file demuxer
 * @file libavformat/rl2.c
 * @author Sascha Sommer (saschasommer@freenet.de)
 * For more information regarding the RL2 file format, visit:
 *   http://wiki.multimedia.cx/index.php?title=RL2
 *
 * extradata:
 * 2 byte le initial drawing offset within 320x200 viewport
 * 4 byte le number of used colors
 * 256 * 3 bytes rgb palette
 * optional background_frame
 */

#include "libavutil/intreadwrite.h"
#include "avformat.h"

#define EXTRADATA1_SIZE (6 + 256 * 3) ///< video base, clr, palette

#define FORM_TAG MKBETAG('F', 'O', 'R', 'M')
#define RLV2_TAG MKBETAG('R', 'L', 'V', '2')
#define RLV3_TAG MKBETAG('R', 'L', 'V', '3')

typedef struct Rl2DemuxContext {
    unsigned int index_pos[2];   ///< indexes in the sample tables
} Rl2DemuxContext;


/**
 * check if the file is in rl2 format
 * @param p probe buffer
 * @return 0 when the probe buffer does not contain rl2 data, > 0 otherwise
 */
static int rl2_probe(AVProbeData *p)
{

    if(AV_RB32(&p->buf[0]) != FORM_TAG)
        return 0;

    if(AV_RB32(&p->buf[8]) != RLV2_TAG &&
        AV_RB32(&p->buf[8]) != RLV3_TAG)
        return 0;

    return AVPROBE_SCORE_MAX;
}

/**
 * read rl2 header data and setup the avstreams
 * @param s demuxer context
 * @param ap format parameters
 * @return 0 on success, AVERROR otherwise
 */
static av_cold int rl2_read_header(AVFormatContext *s,
                            AVFormatParameters *ap)
{
    ByteIOContext *pb = s->pb;
    AVStream *st;
    unsigned int frame_count;
    unsigned int audio_frame_counter = 0;
    unsigned int video_frame_counter = 0;
    unsigned int back_size;
    int data_size;
    unsigned short encoding_method;
    unsigned short sound_rate;
    unsigned short rate;
    unsigned short channels;
    unsigned short def_sound_size;
    unsigned int signature;
    unsigned int pts_den = 11025; /* video only case */
    unsigned int pts_num = 1103;
    unsigned int* chunk_offset = NULL;
    int* chunk_size = NULL;
    int* audio_size = NULL;
    int i;
    int ret = 0;

    url_fskip(pb,4);          /* skip FORM tag */
    back_size = get_le32(pb); /** get size of the background frame */
    signature = get_be32(pb);
    data_size = get_be32(pb);
    frame_count = get_le32(pb);

    /* disallow back_sizes and frame_counts that may lead to overflows later */
    if(back_size > INT_MAX/2  || frame_count > INT_MAX / sizeof(uint32_t))
        return AVERROR_INVALIDDATA;

    encoding_method = get_le16(pb);
    sound_rate = get_le16(pb);
    rate = get_le16(pb);
    channels = get_le16(pb);
    def_sound_size = get_le16(pb);

    /** setup video stream */
    st = av_new_stream(s, 0);
    if(!st)
         return AVERROR(ENOMEM);

    st->codec->codec_type = CODEC_TYPE_VIDEO;
    st->codec->codec_id = CODEC_ID_RL2;
    st->codec->codec_tag = 0;  /* no fourcc */
    st->codec->width = 320;
    st->codec->height = 200;

    /** allocate and fill extradata */
    st->codec->extradata_size = EXTRADATA1_SIZE;

    if(signature == RLV3_TAG && back_size > 0)
        st->codec->extradata_size += back_size;

    st->codec->extradata = av_mallocz(st->codec->extradata_size +
                                          FF_INPUT_BUFFER_PADDING_SIZE);
    if(!st->codec->extradata)
        return AVERROR(ENOMEM);

    if(get_buffer(pb,st->codec->extradata,st->codec->extradata_size) !=
                      st->codec->extradata_size)
        return AVERROR(EIO);

    /** setup audio stream if present */
    if(sound_rate){
        pts_num = def_sound_size;
        pts_den = rate;

        st = av_new_stream(s, 0);
        if (!st)
            return AVERROR(ENOMEM);
        st->codec->codec_type = CODEC_TYPE_AUDIO;
        st->codec->codec_id = CODEC_ID_PCM_U8;
        st->codec->codec_tag = 1;
        st->codec->channels = channels;
        st->codec->bits_per_coded_sample = 8;
        st->codec->sample_rate = rate;
        st->codec->bit_rate = st->codec->channels * st->codec->sample_rate *
            st->codec->bits_per_coded_sample;
        st->codec->block_align = st->codec->channels *
            st->codec->bits_per_coded_sample / 8;
        av_set_pts_info(st,32,1,rate);
    }

    av_set_pts_info(s->streams[0], 32, pts_num, pts_den);

    chunk_size =   av_malloc(frame_count * sizeof(uint32_t));
    audio_size =   av_malloc(frame_count * sizeof(uint32_t));
    chunk_offset = av_malloc(frame_count * sizeof(uint32_t));

    if(!chunk_size || !audio_size || !chunk_offset){
        av_free(chunk_size);
        av_free(audio_size);
        av_free(chunk_offset);
        return AVERROR(ENOMEM);
    }

    /** read offset and size tables */
    for(i=0; i < frame_count;i++)
        chunk_size[i] = get_le32(pb);
    for(i=0; i < frame_count;i++)
        chunk_offset[i] = get_le32(pb);
    for(i=0; i < frame_count;i++)
        audio_size[i] = get_le32(pb) & 0xFFFF;

    /** build the sample index */
    for(i=0;i<frame_count;i++){
        if(chunk_size[i] < 0 || audio_size[i] > chunk_size[i]){
            ret = AVERROR_INVALIDDATA;
            break;
        }

        if(sound_rate && audio_size[i]){
            av_add_index_entry(s->streams[1], chunk_offset[i],
                audio_frame_counter,audio_size[i], 0, AVINDEX_KEYFRAME);
            audio_frame_counter += audio_size[i] / channels;
        }
        av_add_index_entry(s->streams[0], chunk_offset[i] + audio_size[i],
            video_frame_counter,chunk_size[i]-audio_size[i],0,AVINDEX_KEYFRAME);
        ++video_frame_counter;
    }


    av_free(chunk_size);
    av_free(audio_size);
    av_free(chunk_offset);

    return ret;
}

/**
 * read a single audio or video packet
 * @param s demuxer context
 * @param pkt the packet to be filled
 * @return 0 on success, AVERROR otherwise
 */
static int rl2_read_packet(AVFormatContext *s,
                            AVPacket *pkt)
{
    Rl2DemuxContext *rl2 = s->priv_data;
    ByteIOContext *pb = s->pb;
    AVIndexEntry *sample = NULL;
    int i;
    int ret = 0;
    int stream_id = -1;
    int64_t pos = INT64_MAX;

    /** check if there is a valid video or audio entry that can be used */
    for(i=0; i<s->nb_streams; i++){
        if(rl2->index_pos[i] < s->streams[i]->nb_index_entries
              && s->streams[i]->index_entries[ rl2->index_pos[i] ].pos < pos){
            sample = &s->streams[i]->index_entries[ rl2->index_pos[i] ];
            pos= sample->pos;
            stream_id= i;
        }
    }

    if(stream_id == -1)
        return AVERROR(EIO);

    ++rl2->index_pos[stream_id];

    /** position the stream (will probably be there anyway) */
    url_fseek(pb, sample->pos, SEEK_SET);

    /** fill the packet */
    ret = av_get_packet(pb, pkt, sample->size);
    if(ret != sample->size){
        av_free_packet(pkt);
        return AVERROR(EIO);
    }

    pkt->stream_index = stream_id;
    pkt->pts = sample->timestamp;

    return ret;
}

/**
 * seek to a new timestamp
 * @param s demuxer context
 * @param stream_index index of the stream that should be seeked
 * @param timestamp wanted timestamp
 * @param flags direction and seeking mode
 * @return 0 on success, -1 otherwise
 */
static int rl2_read_seek(AVFormatContext *s, int stream_index, int64_t timestamp, int flags)
{
    AVStream *st = s->streams[stream_index];
    Rl2DemuxContext *rl2 = s->priv_data;
    int i;
    int index = av_index_search_timestamp(st, timestamp, flags);
    if(index < 0)
        return -1;

    rl2->index_pos[stream_index] = index;
    timestamp = st->index_entries[index].timestamp;

    for(i=0; i < s->nb_streams; i++){
        AVStream *st2 = s->streams[i];
        index = av_index_search_timestamp(st2,
                    av_rescale_q(timestamp, st->time_base, st2->time_base),
                    flags | AVSEEK_FLAG_BACKWARD);

        if(index < 0)
            index = 0;

        rl2->index_pos[i] = index;
    }

    return 0;
}

AVInputFormat rl2_demuxer = {
    "rl2",
    NULL_IF_CONFIG_SMALL("RL2 format"),
    sizeof(Rl2DemuxContext),
    rl2_probe,
    rl2_read_header,
    rl2_read_packet,
    NULL,
    rl2_read_seek,
};