当前位置：首页 > news >正文

音视频编解码全流程之用Extractor后Muxer生成MP4

news 2025/9/30 10:51:47

系列文章：

音视频编解码全流程之提取器Extractor

音视频编解码全流程之复用器Muxer

根据前面叙述的音视频编解码全流程之Extractor 和音视频编解码全流程之复用器Muxer 两篇文章可知：

媒体文件提取数据包的初始化及提取流程；

向目标文件写入数据的初始化及写入流程。

依此为基础现在本篇文章中实现“从媒体文件中Extractor提取数据包后，用复用器Muxer向目标媒体文件写入数据包生成MP4文件的操作”。

一 .FFmpeg中提取媒体文件数据包后，写入目标文件：

1.FFmpeg交叉编译：

首先的是FFmpeg交叉编译，在之前的博客中有介绍交叉编译的全过程，感兴趣的可以查看博客：

Android Liunx ffmpeg交叉编译

macOs上交叉编译ffmpeg及安装ffmpeg工具

2.提取媒体文件数据后写入目标文件的全流程：

具体的流程细节分解在音视频编解码全流程之复用器Muxer 已经进行了描述，这里不再赘述。

在本人个人的GitHub：https://github.com/wangyongyao1989/FFmpegPractices

中codecTraningLib模块的CopyMeidaFile.cpp类中：

//
// Created by wangyao on 2025/8/17.
//#include "includes/CopyMeidaFile.h"CopyMeidaFile::CopyMeidaFile() {}CopyMeidaFile::~CopyMeidaFile() {}string CopyMeidaFile::copyMediaFile(const char *srcPath, const char *destPath) {// 打开音视频文件int ret = avformat_open_input(&in_fmt_ctx, srcPath, nullptr, nullptr);if (ret < 0) {LOGE("Can't open file %s.\n", srcPath);copyInfo = "Can't open file ：" + string(srcPath);return copyInfo;}LOGI("Success open input_file %s.\n", srcPath);copyInfo = copyInfo + "\nSuccess open input_file :" + string(srcPath);// 查找音视频文件中的流信息ret = avformat_find_stream_info(in_fmt_ctx, nullptr);if (ret < 0) {LOGE("Can't find stream information.\n");copyInfo = "\nCan't find stream information. ";return copyInfo;}AVStream *src_video = nullptr;// 找到视频流的索引int video_index = av_find_best_stream(in_fmt_ctx, AVMEDIA_TYPE_VIDEO, -1, -1, nullptr, 0);if (video_index >= 0) {src_video = in_fmt_ctx->streams[video_index];}AVStream *src_audio = nullptr;// 找到音频流的索引int audio_index = av_find_best_stream(in_fmt_ctx, AVMEDIA_TYPE_AUDIO, -1, -1, nullptr, 0);if (audio_index >= 0) {src_audio = in_fmt_ctx->streams[audio_index];}AVFormatContext *out_fmt_ctx; // 输出文件的封装器实例// 分配音视频文件的封装实例ret = avformat_alloc_output_context2(&out_fmt_ctx, nullptr, nullptr, destPath);if (ret < 0) {LOGE("Can't alloc output_file %s.\n", destPath);copyInfo = "\nCan't alloc output_file : " + string(destPath);return copyInfo;}// 打开输出流ret = avio_open(&out_fmt_ctx->pb, destPath, AVIO_FLAG_READ_WRITE);if (ret < 0) {LOGE("Can't open output_file %s.\n", destPath);copyInfo = "\nCan't open output_file: " + string(destPath);return copyInfo;}LOGI("Success open output_file %s.\n", destPath);copyInfo = copyInfo + "\nSuccess open output_file :" + string(destPath);if (video_index >= 0) { // 源文件有视频流，就给目标文件创建视频流AVStream *dest_video = avformat_new_stream(out_fmt_ctx, nullptr); // 创建数据流// 把源文件的视频参数原样复制过来avcodec_parameters_copy(dest_video->codecpar, src_video->codecpar);dest_video->time_base = src_video->time_base;dest_video->codecpar->codec_tag = 0;}if (audio_index >= 0) { // 源文件有音频流，就给目标文件创建音频流AVStream *dest_audio = avformat_new_stream(out_fmt_ctx, nullptr); // 创建数据流// 把源文件的音频参数原样复制过来avcodec_parameters_copy(dest_audio->codecpar, src_audio->codecpar);dest_audio->codecpar->codec_tag = 0;}ret = avformat_write_header(out_fmt_ctx, nullptr); // 写文件头if (ret < 0) {LOGE("write file_header occur error %d.\n", ret);copyInfo = "\n write file_header occur error: " + to_string(ret);return copyInfo;}LOGI("Success write file_header.\n");copyInfo = copyInfo + "\nSuccess write file_header.";AVPacket *packet = av_packet_alloc(); // 分配一个数据包while (av_read_frame(in_fmt_ctx, packet) >= 0) { // 轮询数据包// 有的文件视频流没在第一路，需要调整到第一路，因为目标的视频流默认第一路if (packet->stream_index == video_index) { // 视频包packet->stream_index = 0;ret = av_write_frame(out_fmt_ctx, packet); // 往文件写入一个数据包} else { // 音频包packet->stream_index = 1;ret = av_write_frame(out_fmt_ctx, packet); // 往文件写入一个数据包}if (ret < 0) {LOGE("write frame occur error %d.\n", ret);copyInfo = copyInfo + "\n write frame occur error: " + to_string(ret);break;}av_packet_unref(packet); // 清除数据包}av_write_trailer(out_fmt_ctx); // 写文件尾LOGI("Success copy file.\n");copyInfo = copyInfo + "\n Success copy file.";av_packet_free(&packet); // 释放数据包资源avio_close(out_fmt_ctx->pb); // 关闭输出流avformat_free_context(out_fmt_ctx); // 释放封装器的实例avformat_close_input(&in_fmt_ctx); // 关闭音视频文件return copyInfo;
}

二.NDK中提取媒体文件数据包后，写入目标文件：

在这里使用了NDK中的NdkMediaExtractor和NdkMediaMuxer两个类，来实现“一”中类似FFmpeg的功能。

1.初始化提取器Extractor：

创建 AMediaExtractor_new() ——> 设置Extractor的fd AMediaExtractor_setDataSourceFd(extractor, input_fd, 0, fileSize)

该函数代码如下：

// 初始化提取器
bool MediaTransMuxer::initExtractor() {extractor = AMediaExtractor_new();if (!extractor) {LOGE("Failed to create media extractor ");callbackInfo ="Failed to create media extractor \n";PostStatusMessage(callbackInfo.c_str());return false;}LOGE("inputPath:%c", sSrcPath.c_str());FILE *inputFp = fopen(sSrcPath.c_str(), "rb");if (!inputFp) {LOGE("Unable to open output file :%s", sSrcPath.c_str());callbackInfo ="Unable to open output file :" + sSrcPath + "\n";PostStatusMessage(callbackInfo.c_str());return false;}struct stat buf;stat(sSrcPath.c_str(), &buf);size_t fileSize = buf.st_size;int32_t input_fd = fileno(inputFp);LOGE("input_fd:%d", input_fd);media_status_t status = AMediaExtractor_setDataSourceFd(extractor, input_fd, 0, fileSize);if (status != AMEDIA_OK) {LOGE("Failed to set data source: %d", status);callbackInfo ="Failed to set data source :" + to_string(status) + "\n";PostStatusMessage(callbackInfo.c_str());return false;}LOGI("Extractor initialized successfully");return true;
}

2.选择提取器Extractor中的轨道：

——> AMediaExtractor_getTrackCount()获取取器Extractor中的轨道数

——> AMediaExtractor_getTrackFormat(extractor, i) 遍历轨道从中获取每个轨道中的AMediaFormat

——> 筛选出音频轨道和视频轨道，分别获取音频轨道的audioTrackIndex值和视频轨道videoTrackIndex值。

该函数代码如下：

// 选择轨道
bool MediaTransMuxer::selectTracks() {size_t trackCount = AMediaExtractor_getTrackCount(extractor);LOGI("Total tracks: %zu", trackCount);callbackInfo ="Total tracks:" + to_string(trackCount) + "\n";PostStatusMessage(callbackInfo.c_str());for (size_t i = 0; i < trackCount; i++) {AMediaFormat *format = AMediaExtractor_getTrackFormat(extractor, i);if (!format) continue;const char *mime;if (AMediaFormat_getString(format, AMEDIAFORMAT_KEY_MIME, &mime)) {LOGI("Track %zu: MIME=%s", i, mime);if (strncmp(mime, "video/", 6) == 0 && videoTrackIndex == -1) {videoTrackIndex = i;hasVideo = true;LOGI("Selected video track: %d", videoTrackIndex);callbackInfo ="Selected video track:" + to_string(videoTrackIndex) + "\n";PostStatusMessage(callbackInfo.c_str());} else if (strncmp(mime, "audio/", 6) == 0 && audioTrackIndex == -1) {audioTrackIndex = i;hasAudio = true;LOGI("Selected audio track: %d", audioTrackIndex);callbackInfo ="Selected audio track:" + to_string(audioTrackIndex) + "\n";PostStatusMessage(callbackInfo.c_str());}}AMediaFormat_delete(format);}return hasVideo || hasAudio;
}

3.初始化复用器Muxer：

——> 创建AMediaMuxer_new(output_fd, AMEDIAMUXER_OUTPUT_FORMAT_MPEG_4)

——> 切换提取器Extractor至视频轨道 AMediaExtractor_selectTrack(extractor, videoTrackIndex) 后给复用器Muxer添加轨道 AMediaMuxer_addTrack(muxer, videoFormat)

——> 切换提取器Extractor至音频轨道 AMediaExtractor_selectTrack(extractor, audioTrackIndex) 后给复用器Muxer添加轨道 AMediaMuxer_addTrack(muxer, audioFormat)

——> 启动复用器AMediaMuxer_addTrack(muxer, audioFormat)。

该函数代码如下：

// 初始化复用器
bool MediaTransMuxer::initMuxer() {LOGE("outputPath:%c", sOutPath.c_str());FILE *outputFp = fopen(sOutPath.c_str(), "w+b");if (!outputFp) {LOGE("Unable to open output file :%s", sOutPath.c_str());callbackInfo ="Unable to open output file :" + sOutPath + "\n";PostStatusMessage(callbackInfo.c_str());return false;}int32_t output_fd = fileno(outputFp);muxer = AMediaMuxer_new(output_fd, AMEDIAMUXER_OUTPUT_FORMAT_MPEG_4);if (!muxer) {LOGE("Failed to create media muxer");callbackInfo ="Failed to create media muxer \n";PostStatusMessage(callbackInfo.c_str());return false;}// 添加视频轨道if (hasVideo) {AMediaExtractor_selectTrack(extractor, videoTrackIndex);AMediaFormat *videoFormat = AMediaExtractor_getTrackFormat(extractor, videoTrackIndex);muxerVideoTrackIndex = AMediaMuxer_addTrack(muxer, videoFormat);AMediaFormat_delete(videoFormat);if (muxerVideoTrackIndex < 0) {LOGE("Failed to add video track to muxer");callbackInfo ="Failed to add video track to muxer \n";PostStatusMessage(callbackInfo.c_str());return false;}LOGI("Muxer video track index: %d", muxerVideoTrackIndex);callbackInfo ="Muxer video track index:" + to_string(muxerVideoTrackIndex) + "\n";PostStatusMessage(callbackInfo.c_str());}// 添加音频轨道if (hasAudio) {AMediaExtractor_selectTrack(extractor, audioTrackIndex);AMediaFormat *audioFormat = AMediaExtractor_getTrackFormat(extractor, audioTrackIndex);muxerAudioTrackIndex = AMediaMuxer_addTrack(muxer, audioFormat);AMediaFormat_delete(audioFormat);if (muxerAudioTrackIndex < 0) {LOGE("Failed to add audio track to muxer");callbackInfo ="Failed to add audio track to muxer \n";PostStatusMessage(callbackInfo.c_str());return false;}LOGI("Muxer audio track index: %d", muxerAudioTrackIndex);callbackInfo ="Muxer audio track index:" + to_string(muxerAudioTrackIndex) + "\n";PostStatusMessage(callbackInfo.c_str());}// 启动复用器media_status_t status = AMediaMuxer_start(muxer);if (status != AMEDIA_OK) {LOGE("Failed to start muxer: %d", status);callbackInfo ="Failed to start muxer:" + to_string(status) + "\n";PostStatusMessage(callbackInfo.c_str());return false;}LOGI("Muxer initialized successfully");callbackInfo ="Muxer initialized successfully \n";PostStatusMessage(callbackInfo.c_str());return true;
}

4.执行转封装：

——> 重新选择所有轨道以重置读取位置 AMediaExtractor_selectTrack(extractor, videoTrackIndex) 和 AMediaExtractor_selectTrack(extractor, audioTrackIndex)；

——> 设置读取位置到开始 AMediaExtractor_seekTo(extractor, 0, AMEDIAEXTRACTOR_SEEK_CLOSEST_SYNC)；

——> 遍历获取当前样本可用的轨道 AMediaExtractor_getSampleTrackIndex(extractor)；

——> 遍历分别设置结构AMediaCodecBufferInfo四个成员：

——> 设置样本大小： info.size = AMediaExtractor_getSampleSize(extractor)；

——> 设置样本偏移值：info.offset = 0；

——> 设置样本标志： info.flags = AMediaExtractor_getSampleFlags(extractor)；

——> 设置样本现在的时间：info.presentationTimeUs = AMediaExtractor_getSampleTime(extractor)；

——> 遍历读取样本数据：AMediaExtractor_readSampleData(extractor, sampleData, info.size)；

——> 遍历分别写入视频和音频的样本数据：AMediaMuxer_writeSampleData(muxer, muxerVideoTrackIndex, sampleData, &info) 和 AMediaMuxer_writeSampleData(muxer, muxerAudioTrackIndex, sampleData, &info)；

——> 遍历前进到下一个样本：AMediaExtractor_advance(extractor)；

该函数代码如下：

// 执行转封装
bool MediaTransMuxer::transmux() {if (!extractor || !muxer) {LOGE("Extractor or muxer not initialized");callbackInfo ="Extractor or muxer not initialized \n";PostStatusMessage(callbackInfo.c_str());return false;}AMediaCodecBufferInfo info;bool sawEOS = false;int64_t lastVideoPts = -1;int64_t lastAudioPts = -1;// 重新选择所有轨道以重置读取位置if (hasVideo) AMediaExtractor_selectTrack(extractor, videoTrackIndex);if (hasAudio) AMediaExtractor_selectTrack(extractor, audioTrackIndex);// 设置读取位置到开始AMediaExtractor_seekTo(extractor, 0, AMEDIAEXTRACTOR_SEEK_CLOSEST_SYNC);while (!sawEOS) {ssize_t trackIndex = AMediaExtractor_getSampleTrackIndex(extractor);if (trackIndex < 0) {sawEOS = true;break;}// 获取样本信息info.size = AMediaExtractor_getSampleSize(extractor);info.offset = 0;info.flags = AMediaExtractor_getSampleFlags(extractor);info.presentationTimeUs = AMediaExtractor_getSampleTime(extractor);// 读取样本数据uint8_t *sampleData = new uint8_t[info.size];ssize_t bytesRead = AMediaExtractor_readSampleData(extractor, sampleData, info.size);if (bytesRead < 0) {LOGE("Error reading sample data: %zd", bytesRead);callbackInfo ="Error reading sample data:" + to_string(bytesRead) + "\n";PostStatusMessage(callbackInfo.c_str());delete[] sampleData;break;}// 写入到复用器if (trackIndex == videoTrackIndex && hasVideo) {// 检查时间戳是否有效（避免重复或倒退的时间戳）if (info.presentationTimeUs > lastVideoPts) {media_status_t status = AMediaMuxer_writeSampleData(muxer,muxerVideoTrackIndex,sampleData,&info);if (status != AMEDIA_OK) {LOGE("Failed to write video sample: %d", status);callbackInfo ="Failed to write video sample:" + to_string(status) + "\n";PostStatusMessage(callbackInfo.c_str());}callbackInfo ="AMediaMuxer_writeSampleData video size:" + to_string(info.size) + "\n";PostStatusMessage(callbackInfo.c_str());lastVideoPts = info.presentationTimeUs;}} else if (trackIndex == audioTrackIndex && hasAudio) {// 检查时间戳是否有效if (info.presentationTimeUs > lastAudioPts) {media_status_t status = AMediaMuxer_writeSampleData(muxer,muxerAudioTrackIndex,sampleData,&info);if (status != AMEDIA_OK) {LOGE("Failed to write audio sample : %d", status);callbackInfo ="Failed to write audio sample:" + to_string(status) + "\n";PostStatusMessage(callbackInfo.c_str());}callbackInfo ="AMediaMuxer_writeSampleData audio size:" + to_string(info.size) + "\n";PostStatusMessage(callbackInfo.c_str());lastAudioPts = info.presentationTimeUs;}}delete[] sampleData;// 前进到下一个样本if (!AMediaExtractor_advance(extractor)) {sawEOS = true;}}LOGI("Transmuxing completed");callbackInfo ="Transmuxing completed \n";PostStatusMessage(callbackInfo.c_str());return true;
}

5.完整的代码：

以上的代码放在本人的GitHub项目：https://github.com/wangyongyao1989/FFmpegPractices

中hwCodecLib模块的MediaTransMuxer.cpp类中：

//  Author : wangyongyao https://github.com/wangyongyao1989
// Created by MMM on 2025/9/29.
//#include <sys/stat.h>
#include "includes/MediaTransMuxer.h"MediaTransMuxer::MediaTransMuxer(JNIEnv *env, jobject thiz) {mEnv = env;env->GetJavaVM(&mJavaVm);mJavaObj = env->NewGlobalRef(thiz);
}MediaTransMuxer::~MediaTransMuxer() {mEnv->DeleteGlobalRef(mJavaObj);if (mEnv) {mEnv = nullptr;}if (mJavaVm) {mJavaVm = nullptr;}if (mJavaObj) {mJavaObj = nullptr;}release();}void MediaTransMuxer::startMediaTransMuxer(const char *inputPath, const char *outputPath) {sSrcPath = inputPath;sOutPath = outputPath;LOGI("sSrcPath :%s \n sOutPath: %s ", sSrcPath.c_str(), sOutPath.c_str());callbackInfo ="sSrcPath:" + sSrcPath + "\n";PostStatusMessage(callbackInfo.c_str());// 1. 初始化提取器if (!initExtractor()) {LOGE("Failed to initialize extractor");callbackInfo ="Failed to initialize extractor \n";PostStatusMessage(callbackInfo.c_str());return;}// 2. 选择轨道if (!selectTracks()) {LOGE("No valid tracks found");callbackInfo ="No valid tracks found \n";PostStatusMessage(callbackInfo.c_str());return;}// 3. 初始化复用器if (!initMuxer()) {LOGE("Failed to initialize muxer");callbackInfo ="Failed to initialize muxer \n";PostStatusMessage(callbackInfo.c_str());return;}// 4. 执行转封装if (!transmux()) {LOGE("Transmuxing failed");callbackInfo ="Transmuxing failed \n";PostStatusMessage(callbackInfo.c_str());return;}// 释放资源release();
}// 初始化提取器
bool MediaTransMuxer::initExtractor() {extractor = AMediaExtractor_new();if (!extractor) {LOGE("Failed to create media extractor ");callbackInfo ="Failed to create media extractor \n";PostStatusMessage(callbackInfo.c_str());return false;}LOGE("inputPath:%c", sSrcPath.c_str());FILE *inputFp = fopen(sSrcPath.c_str(), "rb");if (!inputFp) {LOGE("Unable to open output file :%s", sSrcPath.c_str());callbackInfo ="Unable to open output file :" + sSrcPath + "\n";PostStatusMessage(callbackInfo.c_str());return false;}struct stat buf;stat(sSrcPath.c_str(), &buf);size_t fileSize = buf.st_size;int32_t input_fd = fileno(inputFp);LOGE("input_fd:%d", input_fd);media_status_t status = AMediaExtractor_setDataSourceFd(extractor, input_fd, 0, fileSize);if (status != AMEDIA_OK) {LOGE("Failed to set data source: %d", status);callbackInfo ="Failed to set data source :" + to_string(status) + "\n";PostStatusMessage(callbackInfo.c_str());return false;}LOGI("Extractor initialized successfully");return true;
}// 选择轨道
bool MediaTransMuxer::selectTracks() {size_t trackCount = AMediaExtractor_getTrackCount(extractor);LOGI("Total tracks: %zu", trackCount);callbackInfo ="Total tracks:" + to_string(trackCount) + "\n";PostStatusMessage(callbackInfo.c_str());for (size_t i = 0; i < trackCount; i++) {AMediaFormat *format = AMediaExtractor_getTrackFormat(extractor, i);if (!format) continue;const char *mime;if (AMediaFormat_getString(format, AMEDIAFORMAT_KEY_MIME, &mime)) {LOGI("Track %zu: MIME=%s", i, mime);if (strncmp(mime, "video/", 6) == 0 && videoTrackIndex == -1) {videoTrackIndex = i;hasVideo = true;LOGI("Selected video track: %d", videoTrackIndex);callbackInfo ="Selected video track:" + to_string(videoTrackIndex) + "\n";PostStatusMessage(callbackInfo.c_str());} else if (strncmp(mime, "audio/", 6) == 0 && audioTrackIndex == -1) {audioTrackIndex = i;hasAudio = true;LOGI("Selected audio track: %d", audioTrackIndex);callbackInfo ="Selected audio track:" + to_string(audioTrackIndex) + "\n";PostStatusMessage(callbackInfo.c_str());}}AMediaFormat_delete(format);}return hasVideo || hasAudio;
}// 初始化复用器
bool MediaTransMuxer::initMuxer() {LOGE("outputPath:%c", sOutPath.c_str());FILE *outputFp = fopen(sOutPath.c_str(), "w+b");if (!outputFp) {LOGE("Unable to open output file :%s", sOutPath.c_str());callbackInfo ="Unable to open output file :" + sOutPath + "\n";PostStatusMessage(callbackInfo.c_str());return false;}int32_t output_fd = fileno(outputFp);muxer = AMediaMuxer_new(output_fd, AMEDIAMUXER_OUTPUT_FORMAT_MPEG_4);if (!muxer) {LOGE("Failed to create media muxer");callbackInfo ="Failed to create media muxer \n";PostStatusMessage(callbackInfo.c_str());return false;}// 添加视频轨道if (hasVideo) {AMediaExtractor_selectTrack(extractor, videoTrackIndex);AMediaFormat *videoFormat = AMediaExtractor_getTrackFormat(extractor, videoTrackIndex);muxerVideoTrackIndex = AMediaMuxer_addTrack(muxer, videoFormat);AMediaFormat_delete(videoFormat);if (muxerVideoTrackIndex < 0) {LOGE("Failed to add video track to muxer");callbackInfo ="Failed to add video track to muxer \n";PostStatusMessage(callbackInfo.c_str());return false;}LOGI("Muxer video track index: %d", muxerVideoTrackIndex);callbackInfo ="Muxer video track index:" + to_string(muxerVideoTrackIndex) + "\n";PostStatusMessage(callbackInfo.c_str());}// 添加音频轨道if (hasAudio) {AMediaExtractor_selectTrack(extractor, audioTrackIndex);AMediaFormat *audioFormat = AMediaExtractor_getTrackFormat(extractor, audioTrackIndex);muxerAudioTrackIndex = AMediaMuxer_addTrack(muxer, audioFormat);AMediaFormat_delete(audioFormat);if (muxerAudioTrackIndex < 0) {LOGE("Failed to add audio track to muxer");callbackInfo ="Failed to add audio track to muxer \n";PostStatusMessage(callbackInfo.c_str());return false;}LOGI("Muxer audio track index: %d", muxerAudioTrackIndex);callbackInfo ="Muxer audio track index:" + to_string(muxerAudioTrackIndex) + "\n";PostStatusMessage(callbackInfo.c_str());}// 启动复用器media_status_t status = AMediaMuxer_start(muxer);if (status != AMEDIA_OK) {LOGE("Failed to start muxer: %d", status);callbackInfo ="Failed to start muxer:" + to_string(status) + "\n";PostStatusMessage(callbackInfo.c_str());return false;}LOGI("Muxer initialized successfully");callbackInfo ="Muxer initialized successfully \n";PostStatusMessage(callbackInfo.c_str());return true;
}// 执行转封装
bool MediaTransMuxer::transmux() {if (!extractor || !muxer) {LOGE("Extractor or muxer not initialized");callbackInfo ="Extractor or muxer not initialized \n";PostStatusMessage(callbackInfo.c_str());return false;}AMediaCodecBufferInfo info;bool sawEOS = false;int64_t lastVideoPts = -1;int64_t lastAudioPts = -1;// 重新选择所有轨道以重置读取位置if (hasVideo) AMediaExtractor_selectTrack(extractor, videoTrackIndex);if (hasAudio) AMediaExtractor_selectTrack(extractor, audioTrackIndex);// 设置读取位置到开始AMediaExtractor_seekTo(extractor, 0, AMEDIAEXTRACTOR_SEEK_CLOSEST_SYNC);while (!sawEOS) {ssize_t trackIndex = AMediaExtractor_getSampleTrackIndex(extractor);if (trackIndex < 0) {sawEOS = true;break;}// 获取样本信息info.size = AMediaExtractor_getSampleSize(extractor);info.offset = 0;info.flags = AMediaExtractor_getSampleFlags(extractor);info.presentationTimeUs = AMediaExtractor_getSampleTime(extractor);// 读取样本数据uint8_t *sampleData = new uint8_t[info.size];ssize_t bytesRead = AMediaExtractor_readSampleData(extractor, sampleData, info.size);if (bytesRead < 0) {LOGE("Error reading sample data: %zd", bytesRead);callbackInfo ="Error reading sample data:" + to_string(bytesRead) + "\n";PostStatusMessage(callbackInfo.c_str());delete[] sampleData;break;}// 写入到复用器if (trackIndex == videoTrackIndex && hasVideo) {// 检查时间戳是否有效（避免重复或倒退的时间戳）if (info.presentationTimeUs > lastVideoPts) {media_status_t status = AMediaMuxer_writeSampleData(muxer,muxerVideoTrackIndex,sampleData,&info);if (status != AMEDIA_OK) {LOGE("Failed to write video sample: %d", status);callbackInfo ="Failed to write video sample:" + to_string(status) + "\n";PostStatusMessage(callbackInfo.c_str());}callbackInfo ="AMediaMuxer_writeSampleData video size:" + to_string(info.size) + "\n";PostStatusMessage(callbackInfo.c_str());lastVideoPts = info.presentationTimeUs;}} else if (trackIndex == audioTrackIndex && hasAudio) {// 检查时间戳是否有效if (info.presentationTimeUs > lastAudioPts) {media_status_t status = AMediaMuxer_writeSampleData(muxer,muxerAudioTrackIndex,sampleData,&info);if (status != AMEDIA_OK) {LOGE("Failed to write audio sample : %d", status);callbackInfo ="Failed to write audio sample:" + to_string(status) + "\n";PostStatusMessage(callbackInfo.c_str());}callbackInfo ="AMediaMuxer_writeSampleData audio size:" + to_string(info.size) + "\n";PostStatusMessage(callbackInfo.c_str());lastAudioPts = info.presentationTimeUs;}}delete[] sampleData;// 前进到下一个样本if (!AMediaExtractor_advance(extractor)) {sawEOS = true;}}LOGI("Transmuxing completed");callbackInfo ="Transmuxing completed \n";PostStatusMessage(callbackInfo.c_str());return true;
}// 释放资源
void MediaTransMuxer::release() {if (muxer) {AMediaMuxer_stop(muxer);AMediaMuxer_delete(muxer);muxer = nullptr;}if (extractor) {AMediaExtractor_delete(extractor);extractor = nullptr;}LOGI("Resources released");
}JNIEnv *MediaTransMuxer::GetJNIEnv(bool *isAttach) {JNIEnv *env;int status;if (nullptr == mJavaVm) {LOGD("GetJNIEnv mJavaVm == nullptr");return nullptr;}*isAttach = false;status = mJavaVm->GetEnv((void **) &env, JNI_VERSION_1_6);if (status != JNI_OK) {status = mJavaVm->AttachCurrentThread(&env, nullptr);if (status != JNI_OK) {LOGD("GetJNIEnv failed to attach current thread");return nullptr;}*isAttach = true;}return env;
}void MediaTransMuxer::PostStatusMessage(const char *msg) {bool isAttach = false;JNIEnv *pEnv = GetJNIEnv(&isAttach);if (pEnv == nullptr) {return;}jobject javaObj = mJavaObj;jmethodID mid = pEnv->GetMethodID(pEnv->GetObjectClass(javaObj), "CppStatusCallback","(Ljava/lang/String;)V");jstring pJstring = pEnv->NewStringUTF(msg);pEnv->CallVoidMethod(javaObj, mid, pJstring);if (isAttach) {JavaVM *pJavaVm = mJavaVm;pJavaVm->DetachCurrentThread();}
}

三.FFmpeg/NDK向目标文件写入数据包的操作的对比：

FFmpeg：

——> 打开输入文件获取输入文件的封装实例AVFormatContext ：avformat_open_input(&in_fmt_ctx, srcPath, nullptr, nullptr)；

——> 查找音视频文件中的流信息:avformat_find_stream_info(in_fmt_ctx, nullptr);

——> 分别查找音频和视频的索引：av_find_best_stream(in_fmt_ctx, AVMEDIA_TYPE_VIDEO, -1, -1, nullptr, 0) / av_find_best_stream(in_fmt_ctx, AVMEDIA_TYPE_AUDIO, -1, -1, nullptr, 0) ；

——> 分别查找流封装的实例AVStream：src_video = in_fmt_ctx->streams[video_index]

/ src_audio = in_fmt_ctx->streams[audio_index] ;

——> 分配音视频文件的封装实例AVFormatContext：avformat_alloc_output_context2(&out_fmt_ctx, nullptr, nullptr, destPath)；

——> 打开输出流：avio_open(&out_fmt_ctx->pb, destPath, AVIO_FLAG_READ_WRITE)；

——> 创建数据流并把源文件的音视频参数原样复制过来：

——> AVStream *dest_video = avformat_new_stream(out_fmt_ctx, nullptr)

——> avcodec_parameters_copy(dest_video->codecpar, src_video->codecpar)

——> 写入文件头：avformat_write_header(out_fmt_ctx, nullptr)

——> 分配一个数据包/轮询数据包/轮询往文件写入数据包

——> AVPacket *packet = av_packet_alloc()

——> while (av_read_frame(in_fmt_ctx, packet) >= 0) ){}

——> av_write_frame(out_fmt_ctx, packet);

——> 写入文件尾： av_write_trailer(out_fmt_ctx)；

NDK：

——> 初始化提取器:

——> 创建提取器：AMediaExtractor_new()

——> 设置Extractor的fd : AMediaExtractor_setDataSourceFd(extractor, input_fd, 0, fileSize)

——> 选择提取器Extractor中的轨道:

——> AMediaExtractor_getTrackCount()获取取器Extractor中的轨道数

——> AMediaExtractor_getTrackFormat(extractor, i) 遍历轨道从中获取每个轨道中的AMediaFormat

——> 筛选出音频轨道和视频轨道，分别获取音频轨道的audioTrackIndex值和视频轨道videoTrackIndex值。

——> 初始化复用器Muxer:

——> 创建AMediaMuxer_new(output_fd, AMEDIAMUXER_OUTPUT_FORMAT_MPEG_4)

——> 切换提取器Extractor至视频轨道 AMediaExtractor_selectTrack(extractor, videoTrackIndex) 后给复用器Muxer添加轨道 AMediaMuxer_addTrack(muxer, videoFormat)

——> 切换提取器Extractor至音频轨道 AMediaExtractor_selectTrack(extractor, audioTrackIndex) 后给复用器Muxer添加轨道 AMediaMuxer_addTrack(muxer, audioFormat)

——> 启动复用器AMediaMuxer_addTrack(muxer, audioFormat)。

——> 执行转封装：

——> 重新选择所有轨道以重置读取位置 AMediaExtractor_selectTrack(extractor, videoTrackIndex) 和 AMediaExtractor_selectTrack(extractor, audioTrackIndex)；

——> 设置读取位置到开始 AMediaExtractor_seekTo(extractor, 0, AMEDIAEXTRACTOR_SEEK_CLOSEST_SYNC)；

——> 遍历获取当前样本可用的轨道 AMediaExtractor_getSampleTrackIndex(extractor)；

——> 遍历分别设置结构AMediaCodecBufferInfo四个成员：

——> 设置样本大小： info.size = AMediaExtractor_getSampleSize(extractor)；

——> 设置样本偏移值：info.offset = 0；

——> 设置样本标志： info.flags = AMediaExtractor_getSampleFlags(extractor)；

——> 设置样本现在的时间：info.presentationTimeUs = AMediaExtractor_getSampleTime(extractor)；

——> 遍历读取样本数据：AMediaExtractor_readSampleData(extractor, sampleData, info.size)；

——> 遍历前进到下一个样本：AMediaExtractor_advance(extractor)；

效果展示：

以上的代码放在本人的GitHub项目：https://github.com/wangyongyao1989/FFmpegPractices

查看全文

http://www.dtcms.com/a/423773.html

高德地图实现经纬度及获取编码、所属行政区、GIS

wordpress 扁平化新站seo快速排名排名

2025年在招投标及竞品信息采集机器人领域，主流RPA全面解析

电子商务网站建设与管理期末考试网站开发方案案例

Node.js命令行工具开发

《面向物理交互任务的触觉传感阵列仿真》2020AIM论文解读

未来最紧缺的十大专业seo优化师

OCP证书考试难度怎么样？

Vue3 defineModel === 实现原理

唐山营销型网站建设2023新闻头条最新消息今天

计算机网络---传输层

如何在阿里云上做网站制作软件的手机软件

深入理解 Java 虚拟机：从原理到实践的全方位剖析

网站谷歌seo做哪些凌点视频素材网

手机app应用网站C语言做网站需要创建窗口吗

uniapp 安卓FTP上传下载操作原生插件

国外知名平面设计网站黄骅打牌吧

C++ I/O流与文件操作速查

网站制作哪家好又便宜做电商网站的流程

网络边界突围：运营商QoS限速策略

【笔记】在WPF中Decorator是什么以及何时优先考虑 Decorator 派生类

[算法练习]Day 4:定长滑动窗口

外汇交易网站开发做网站前端后台

小红书网站建设目的优化师简历

集群的概述和分类和负载均衡集群

专业的商城网站开发搜索引擎优化不包括

哈尔滨市延寿建设局网站wordpress 主题添加

技术实践指南：多模态RAG从数据预处理到生成响应的完整流程

新中地三维GIS开发智慧城市效果和应用场景

做产品封面的网站赵公口网站建设公司

一 .FFmpeg中提取媒体文件数据包后，写入目标文件：

1.FFmpeg交叉编译：

2.提取媒体文件数据后写入目标文件的全流程：

二.NDK中提取媒体文件数据包后，写入目标文件：

1.初始化提取器Extractor：

2.选择提取器Extractor中的轨道：

3.初始化复用器Muxer：

4.执行转封装：

5.完整的代码：

三.FFmpeg/NDK向目标文件写入数据包的操作的对比：

FFmpeg：

NDK：

效果展示：

相关文章：