create project

This commit is contained in:
2026-01-16 17:30:40 +08:00
commit effac6b017
157 changed files with 45997 additions and 0 deletions

612
infrastructure/external/ffmpeg/ffmpeg.go vendored Normal file
View File

@@ -0,0 +1,612 @@
package ffmpeg
import (
"fmt"
"io"
"net/http"
"os"
"os/exec"
"path/filepath"
"strings"
"time"
"github.com/drama-generator/backend/pkg/logger"
)
type FFmpeg struct {
log *logger.Logger
tempDir string
}
func NewFFmpeg(log *logger.Logger) *FFmpeg {
tempDir := filepath.Join(os.TempDir(), "drama-video-merge")
os.MkdirAll(tempDir, 0755)
return &FFmpeg{
log: log,
tempDir: tempDir,
}
}
type VideoClip struct {
URL string
Duration float64
StartTime float64
EndTime float64
Transition map[string]interface{}
}
type MergeOptions struct {
OutputPath string
Clips []VideoClip
}
func (f *FFmpeg) MergeVideos(opts *MergeOptions) (string, error) {
if len(opts.Clips) == 0 {
return "", fmt.Errorf("no video clips to merge")
}
f.log.Infow("Starting video merge with trimming", "clips_count", len(opts.Clips))
// 下载并裁剪所有视频片段
trimmedPaths := make([]string, 0, len(opts.Clips))
downloadedPaths := make([]string, 0, len(opts.Clips))
for i, clip := range opts.Clips {
// 下载原始视频
downloadPath := filepath.Join(f.tempDir, fmt.Sprintf("download_%d_%d.mp4", time.Now().Unix(), i))
localPath, err := f.downloadVideo(clip.URL, downloadPath)
if err != nil {
f.cleanup(downloadedPaths)
f.cleanup(trimmedPaths)
return "", fmt.Errorf("failed to download clip %d: %w", i, err)
}
downloadedPaths = append(downloadedPaths, localPath)
// 裁剪视频片段根据StartTime和EndTime
trimmedPath := filepath.Join(f.tempDir, fmt.Sprintf("trimmed_%d_%d.mp4", time.Now().Unix(), i))
err = f.trimVideo(localPath, trimmedPath, clip.StartTime, clip.EndTime)
if err != nil {
f.cleanup(downloadedPaths)
f.cleanup(trimmedPaths)
return "", fmt.Errorf("failed to trim clip %d: %w", i, err)
}
trimmedPaths = append(trimmedPaths, trimmedPath)
f.log.Infow("Clip trimmed",
"index", i,
"start", clip.StartTime,
"end", clip.EndTime,
"duration", clip.EndTime-clip.StartTime)
}
// 清理下载的原始文件
f.cleanup(downloadedPaths)
// 确保输出目录存在
outputDir := filepath.Dir(opts.OutputPath)
if err := os.MkdirAll(outputDir, 0755); err != nil {
f.cleanup(trimmedPaths)
return "", fmt.Errorf("failed to create output directory: %w", err)
}
// 合并裁剪后的视频片段(支持转场效果)
err := f.concatenateVideosWithTransitions(trimmedPaths, opts.Clips, opts.OutputPath)
// 清理裁剪后的临时文件
f.cleanup(trimmedPaths)
if err != nil {
return "", fmt.Errorf("failed to concatenate videos: %w", err)
}
f.log.Infow("Video merge completed", "output", opts.OutputPath)
return opts.OutputPath, nil
}
func (f *FFmpeg) downloadVideo(url, destPath string) (string, error) {
f.log.Infow("Downloading video", "url", url, "dest", destPath)
resp, err := http.Get(url)
if err != nil {
return "", fmt.Errorf("failed to download: %w", err)
}
defer resp.Body.Close()
if resp.StatusCode != http.StatusOK {
return "", fmt.Errorf("bad status: %s", resp.Status)
}
out, err := os.Create(destPath)
if err != nil {
return "", fmt.Errorf("failed to create file: %w", err)
}
defer out.Close()
_, err = io.Copy(out, resp.Body)
if err != nil {
return "", fmt.Errorf("failed to save file: %w", err)
}
return destPath, nil
}
func (f *FFmpeg) trimVideo(inputPath, outputPath string, startTime, endTime float64) error {
f.log.Infow("Trimming video",
"input", inputPath,
"output", outputPath,
"start", startTime,
"end", endTime)
// 如果startTime和endTime都为0或者endTime <= startTime复制整个视频
// 使用重新编码而非-c copy以确保输出文件完整性
if (startTime == 0 && endTime == 0) || endTime <= startTime {
f.log.Infow("No valid trim range, re-encoding entire video")
cmd := exec.Command("ffmpeg",
"-i", inputPath,
"-c:v", "libx264",
"-preset", "fast",
"-crf", "23",
"-c:a", "aac",
"-b:a", "128k",
"-movflags", "+faststart",
"-y",
outputPath,
)
output, err := cmd.CombinedOutput()
if err != nil {
f.log.Errorw("FFmpeg re-encode failed", "error", err, "output", string(output))
return fmt.Errorf("ffmpeg re-encode failed: %w, output: %s", err, string(output))
}
f.log.Infow("Video re-encoded successfully", "output", outputPath)
return nil
}
// 使用FFmpeg裁剪视频
// -ss: 开始时间(秒)
// -to/-t: 结束时间或持续时间
// 使用重新编码而非-c copy以确保输出文件完整性避免Windows环境下流信息丢失
var cmd *exec.Cmd
if endTime > 0 {
// 有明确的结束时间
cmd = exec.Command("ffmpeg",
"-i", inputPath,
"-ss", fmt.Sprintf("%.2f", startTime),
"-to", fmt.Sprintf("%.2f", endTime),
"-c:v", "libx264",
"-preset", "fast",
"-crf", "23",
"-c:a", "aac",
"-b:a", "128k",
"-movflags", "+faststart",
"-y",
outputPath,
)
} else {
// 只有开始时间,裁剪到视频末尾
cmd = exec.Command("ffmpeg",
"-i", inputPath,
"-ss", fmt.Sprintf("%.2f", startTime),
"-c:v", "libx264",
"-preset", "fast",
"-crf", "23",
"-c:a", "aac",
"-b:a", "128k",
"-movflags", "+faststart",
"-y",
outputPath,
)
}
output, err := cmd.CombinedOutput()
if err != nil {
f.log.Errorw("FFmpeg trim failed", "error", err, "output", string(output))
return fmt.Errorf("ffmpeg trim failed: %w, output: %s", err, string(output))
}
f.log.Infow("Video trimmed successfully", "output", outputPath)
return nil
}
func (f *FFmpeg) concatenateVideosWithTransitions(inputPaths []string, clips []VideoClip, outputPath string) error {
if len(inputPaths) == 0 {
return fmt.Errorf("no input paths")
}
// 如果只有一个视频,直接复制
if len(inputPaths) == 1 {
f.log.Infow("Only one clip, copying directly")
return f.copyFile(inputPaths[0], outputPath)
}
// 检查是否有转场效果
hasTransitions := false
for _, clip := range clips {
if clip.Transition != nil && len(clip.Transition) > 0 {
hasTransitions = true
break
}
}
// 如果没有转场效果,使用简单拼接
if !hasTransitions {
f.log.Infow("No transitions, using simple concatenation")
return f.concatenateVideos(inputPaths, outputPath)
}
// 使用xfade滤镜添加转场效果
f.log.Infow("Merging with transitions", "clips_count", len(inputPaths))
return f.mergeWithXfade(inputPaths, clips, outputPath)
}
func (f *FFmpeg) concatenateVideos(inputPaths []string, outputPath string) error {
// 创建文件列表
listFile := filepath.Join(f.tempDir, fmt.Sprintf("filelist_%d.txt", time.Now().Unix()))
defer os.Remove(listFile)
var content strings.Builder
for _, path := range inputPaths {
content.WriteString(fmt.Sprintf("file '%s'\n", path))
}
if err := os.WriteFile(listFile, []byte(content.String()), 0644); err != nil {
return fmt.Errorf("failed to create file list: %w", err)
}
// 使用FFmpeg合并视频
// -f concat: 使用concat demuxer
// -safe 0: 允许不安全的文件路径
// -i: 输入文件列表
// -c copy: 直接复制流,不重新编码(速度快)
cmd := exec.Command("ffmpeg",
"-f", "concat",
"-safe", "0",
"-i", listFile,
"-c", "copy",
"-y", // 覆盖输出文件
outputPath,
)
output, err := cmd.CombinedOutput()
if err != nil {
f.log.Errorw("FFmpeg failed", "error", err, "output", string(output))
return fmt.Errorf("ffmpeg execution failed: %w, output: %s", err, string(output))
}
f.log.Infow("FFmpeg concatenation completed", "output", outputPath)
return nil
}
func (f *FFmpeg) mergeWithXfade(inputPaths []string, clips []VideoClip, outputPath string) error {
// 使用xfade滤镜进行转场
// 构建输入参数
args := []string{}
for _, path := range inputPaths {
args = append(args, "-i", path)
}
// 检测每个视频是否有音频流
audioStreams := make([]bool, len(inputPaths))
hasAnyAudio := false
for i, path := range inputPaths {
audioStreams[i] = f.hasAudioStream(path)
if audioStreams[i] {
hasAnyAudio = true
}
f.log.Infow("Audio stream detection", "index", i, "path", path, "has_audio", audioStreams[i])
}
f.log.Infow("Overall audio detection", "has_any_audio", hasAnyAudio, "audio_streams", audioStreams)
// 检测视频分辨率,找到最大分辨率作为目标分辨率
maxWidth := 0
maxHeight := 0
for i, path := range inputPaths {
width, height := f.getVideoResolution(path)
if width > maxWidth {
maxWidth = width
}
if height > maxHeight {
maxHeight = height
}
f.log.Infow("Video resolution detection", "index", i, "width", width, "height", height)
}
f.log.Infow("Target resolution", "width", maxWidth, "height", maxHeight)
// 为每个视频流添加缩放滤镜,统一分辨率
var scaleFilters []string
for i := 0; i < len(inputPaths); i++ {
// 使用scale滤镜缩放到目标分辨率pad添加黑边保持长宽比
scaleFilters = append(scaleFilters,
fmt.Sprintf("[%d:v]scale=%d:%d:force_original_aspect_ratio=decrease,pad=%d:%d:(ow-iw)/2:(oh-ih)/2[v%d]",
i, maxWidth, maxHeight, maxWidth, maxHeight, i))
}
// 构建filter_complex
// 例如: [0:v][1:v]xfade=transition=fade:duration=1:offset=5[v01];[v01][2:v]xfade=transition=fade:duration=1:offset=10[out]
// 构建转场滤镜,使用缩放后的视频流
var transitionFilters []string
var offset float64 = 0
for i := 0; i < len(inputPaths)-1; i++ {
// 获取当前片段的时长
clipDuration := clips[i].Duration
if clips[i].EndTime > 0 && clips[i].StartTime >= 0 {
clipDuration = clips[i].EndTime - clips[i].StartTime
}
// 获取转场类型和时长
transitionType := "fade" // 默认淡入淡出
transitionDuration := 1.0 // 默认转场时长为1秒
if clips[i].Transition != nil {
// 读取转场类型
if tType, ok := clips[i].Transition["type"].(string); ok && tType != "" {
transitionType = f.mapTransitionType(tType)
f.log.Infow("Using transition type", "type", tType, "mapped", transitionType)
}
// 读取转场时长
if tDuration, ok := clips[i].Transition["duration"].(float64); ok && tDuration > 0 {
transitionDuration = tDuration
}
}
// 计算转场开始的时间点
// 转场在两个片段的交界处,从前一个片段结束前 transitionDuration/2 开始
// 这样转场效果会平均分布在两个片段的交界处
offset += clipDuration - (transitionDuration / 2)
if offset < 0 {
offset = 0
}
f.log.Infow("Transition settings",
"clip_index", i,
"type", transitionType,
"duration", transitionDuration,
"offset", offset,
"clip_duration", clipDuration)
var inputLabel, outputLabel string
if i == 0 {
inputLabel = fmt.Sprintf("[v0][v1]")
} else {
inputLabel = fmt.Sprintf("[vx%02d][v%d]", i-1, i+1)
}
if i == len(inputPaths)-2 {
outputLabel = "[outv]"
} else {
outputLabel = fmt.Sprintf("[vx%02d]", i)
}
filterPart := fmt.Sprintf("%sxfade=transition=%s:duration=%.1f:offset=%.1f%s",
inputLabel, transitionType, transitionDuration, offset, outputLabel)
transitionFilters = append(transitionFilters, filterPart)
}
// 合并缩放和转场滤镜
var videoFilters []string
videoFilters = append(videoFilters, scaleFilters...)
videoFilters = append(videoFilters, transitionFilters...)
filterComplex := strings.Join(videoFilters, ";")
// 音频处理:如果有任何视频包含音频流,则处理音频
var fullFilter string
if hasAnyAudio {
// 为没有音频的视频生成静音轨道,确保所有输入音频流一致
var silenceFilters []string
for i := 0; i < len(inputPaths); i++ {
if !audioStreams[i] {
// 计算该视频的时长
clipDuration := clips[i].Duration
if clips[i].EndTime > 0 && clips[i].StartTime >= 0 {
clipDuration = clips[i].EndTime - clips[i].StartTime
}
// anullsrc是源滤镜不接受输入使用duration参数指定时长
silenceFilters = append(silenceFilters,
fmt.Sprintf("anullsrc=channel_layout=stereo:sample_rate=44100:duration=%.2f[a%d]", clipDuration, i))
}
}
// 拼接所有音频流(包括生成的静音流)
var audioConcat strings.Builder
for i := 0; i < len(inputPaths); i++ {
if audioStreams[i] {
audioConcat.WriteString(fmt.Sprintf("[%d:a]", i))
} else {
audioConcat.WriteString(fmt.Sprintf("[a%d]", i))
}
}
audioConcat.WriteString(fmt.Sprintf("concat=n=%d:v=0:a=1[outa]", len(inputPaths)))
// 构建完整滤镜:先生成静音流,再拼接音频
if len(silenceFilters) > 0 {
fullFilter = filterComplex + ";" + strings.Join(silenceFilters, ";") + ";" + audioConcat.String()
} else {
fullFilter = filterComplex + ";" + audioConcat.String()
}
} else {
// 所有视频都无音频流,只处理视频
fullFilter = filterComplex
}
// 构建完整命令
args = append(args,
"-filter_complex", fullFilter,
"-map", "[outv]",
)
// 仅在有任何音频时映射音频输出
if hasAnyAudio {
args = append(args, "-map", "[outa]")
}
args = append(args,
"-c:v", "libx264",
"-preset", "medium",
"-crf", "23",
)
// 仅在有任何音频时设置音频编码参数
if hasAnyAudio {
args = append(args,
"-c:a", "aac",
"-b:a", "128k",
)
}
args = append(args,
"-y",
outputPath,
)
f.log.Infow("Running FFmpeg with transitions", "filter", fullFilter, "has_any_audio", hasAnyAudio)
cmd := exec.Command("ffmpeg", args...)
output, err := cmd.CombinedOutput()
if err != nil {
f.log.Errorw("FFmpeg xfade failed", "error", err, "output", string(output))
return fmt.Errorf("ffmpeg xfade failed: %w, output: %s", err, string(output))
}
f.log.Infow("Video merged with transitions successfully")
return nil
}
func (f *FFmpeg) mapTransitionType(transType string) string {
// 将前端传入的转场类型映射为FFmpeg xfade支持的类型
// FFmpeg xfade支持的完整转场列表: https://ffmpeg.org/ffmpeg-filters.html#xfade
switch strings.ToLower(transType) {
// 淡入淡出类
case "fade", "fadein", "fadeout":
return "fade"
case "fadeblack":
return "fadeblack"
case "fadewhite":
return "fadewhite"
case "fadegrays":
return "fadegrays"
// 滑动类
case "slideleft":
return "slideleft"
case "slideright":
return "slideright"
case "slideup":
return "slideup"
case "slidedown":
return "slidedown"
// 擦除类
case "wipeleft":
return "wipeleft"
case "wiperight":
return "wiperight"
case "wipeup":
return "wipeup"
case "wipedown":
return "wipedown"
// 圆形类
case "circleopen":
return "circleopen"
case "circleclose":
return "circleclose"
// 矩形打开/关闭类
case "horzopen":
return "horzopen"
case "horzclose":
return "horzclose"
case "vertopen":
return "vertopen"
case "vertclose":
return "vertclose"
// 其他特效
case "dissolve":
return "dissolve"
case "distance":
return "distance"
case "pixelize":
return "pixelize"
default:
return "fade" // 默认淡入淡出
}
}
func (f *FFmpeg) hasAudioStream(videoPath string) bool {
cmd := exec.Command("ffprobe",
"-v", "error",
"-select_streams", "a:0",
"-show_entries", "stream=codec_type",
"-of", "default=noprint_wrappers=1:nokey=1",
videoPath,
)
output, err := cmd.CombinedOutput()
if err != nil {
return false
}
result := strings.TrimSpace(string(output))
return result == "audio"
}
func (f *FFmpeg) getVideoResolution(videoPath string) (int, int) {
cmd := exec.Command("ffprobe",
"-v", "error",
"-select_streams", "v:0",
"-show_entries", "stream=width,height",
"-of", "csv=p=0",
videoPath,
)
output, err := cmd.CombinedOutput()
if err != nil {
f.log.Warnw("Failed to get video resolution", "path", videoPath, "error", err)
return 1920, 1080 // 默认分辨率
}
result := strings.TrimSpace(string(output))
parts := strings.Split(result, ",")
if len(parts) != 2 {
f.log.Warnw("Invalid resolution format", "output", result)
return 1920, 1080
}
var width, height int
fmt.Sscanf(parts[0], "%d", &width)
fmt.Sscanf(parts[1], "%d", &height)
if width <= 0 || height <= 0 {
return 1920, 1080
}
return width, height
}
func (f *FFmpeg) copyFile(src, dst string) error {
cmd := exec.Command("cp", src, dst)
output, err := cmd.CombinedOutput()
if err != nil {
f.log.Errorw("File copy failed", "error", err, "output", string(output))
return fmt.Errorf("copy failed: %w", err)
}
return nil
}
func (f *FFmpeg) cleanup(paths []string) {
for _, path := range paths {
if err := os.Remove(path); err != nil {
f.log.Warnw("Failed to cleanup file", "path", path, "error", err)
}
}
}
func (f *FFmpeg) CleanupTempDir() error {
return os.RemoveAll(f.tempDir)
}