mirror of
https://github.com/crunchy-labs/crunchy-cli.git
synced 2026-01-21 04:02:00 -06:00
393 lines
11 KiB
Go
393 lines
11 KiB
Go
package crunchyroll
|
|
|
|
import (
|
|
"bytes"
|
|
"context"
|
|
"crypto/aes"
|
|
"crypto/cipher"
|
|
"fmt"
|
|
"github.com/grafov/m3u8"
|
|
"io"
|
|
"io/ioutil"
|
|
"math"
|
|
"net/http"
|
|
"os"
|
|
"os/exec"
|
|
"path/filepath"
|
|
"strings"
|
|
"sync"
|
|
"sync/atomic"
|
|
"time"
|
|
)
|
|
|
|
// NewDownloader creates a downloader with default settings which should
|
|
// fit the most needs
|
|
func NewDownloader(context context.Context, writer io.Writer, goroutines int, onSegmentDownload func(segment *m3u8.MediaSegment, current, total int, file *os.File) error) Downloader {
|
|
tmp, _ := os.MkdirTemp("", "crunchy_")
|
|
|
|
return Downloader{
|
|
Writer: writer,
|
|
TempDir: tmp,
|
|
DeleteTempAfter: true,
|
|
Context: context,
|
|
Goroutines: goroutines,
|
|
OnSegmentDownload: onSegmentDownload,
|
|
}
|
|
}
|
|
|
|
// Downloader is used to download Format's
|
|
type Downloader struct {
|
|
// The output is all written to Writer
|
|
Writer io.Writer
|
|
|
|
// TempDir is the directory where the temporary segment files should be stored.
|
|
// The files will be placed directly into the root of the directory.
|
|
// If empty a random temporary directory on the system's default tempdir
|
|
// will be created.
|
|
// If the directory does not exist, it will be created
|
|
TempDir string
|
|
// If DeleteTempAfter is true, the temp directory gets deleted afterwards.
|
|
// Note that in case of a hard signal exit (os.Interrupt, ...) the directory
|
|
// will NOT be deleted. In such situations try to catch the signal and
|
|
// cancel Context
|
|
DeleteTempAfter bool
|
|
|
|
// Context to control the download process with.
|
|
// There is a tiny delay when canceling the context and the actual stop of the
|
|
// process. So it is not recommend stopping the program immediately after calling
|
|
// the cancel function. It's better when canceling it and then exit the program
|
|
// when Format.Download throws an error. See the signal handling section in
|
|
// cmd/crunchyroll-go/cmd/download.go for an example
|
|
Context context.Context
|
|
|
|
// Goroutines is the number of goroutines to download segments with
|
|
Goroutines int
|
|
|
|
// A method to call when a segment was downloaded.
|
|
// Note that the segments are downloaded asynchronously (depending on the count of
|
|
// Goroutines) and the function gets called asynchronously too, so for example it is
|
|
// first called on segment 1, then segment 254, then segment 3 and so on
|
|
OnSegmentDownload func(segment *m3u8.MediaSegment, current, total int, file *os.File) error
|
|
// If LockOnSegmentDownload is true, only one OnSegmentDownload function can be called at
|
|
// once. Normally (because of the use of goroutines while downloading) multiple could get
|
|
// called simultaneously
|
|
LockOnSegmentDownload bool
|
|
|
|
// If FFmpegOpts is not nil, ffmpeg will be used to merge and convert files.
|
|
// The given opts will be used as ffmpeg parameters while merging.
|
|
//
|
|
// If Writer is *os.File and -f (which sets the output format) is not specified, the output
|
|
// format will be retrieved by its file ending. If this is not the case and -f is not given,
|
|
// the output format will be mpegts / mpeg transport stream.
|
|
// Execute 'ffmpeg -muxers' to see all available output formats.
|
|
FFmpegOpts []string
|
|
}
|
|
|
|
// download's the given format
|
|
func (d Downloader) download(format *Format) error {
|
|
if err := format.InitVideo(); err != nil {
|
|
return err
|
|
}
|
|
|
|
if _, err := os.Stat(d.TempDir); os.IsNotExist(err) {
|
|
if err := os.Mkdir(d.TempDir, 0700); err != nil {
|
|
return err
|
|
}
|
|
}
|
|
if d.DeleteTempAfter {
|
|
defer os.RemoveAll(d.TempDir)
|
|
}
|
|
|
|
files, err := d.downloadSegments(format)
|
|
if err != nil {
|
|
return err
|
|
}
|
|
if d.FFmpegOpts == nil {
|
|
return d.mergeSegments(files)
|
|
} else {
|
|
return d.mergeSegmentsFFmpeg(files)
|
|
}
|
|
}
|
|
|
|
// mergeSegments reads every file in tempDir and writes their content to Downloader.Writer.
|
|
// The given output file gets created or overwritten if already existing
|
|
func (d Downloader) mergeSegments(files []string) error {
|
|
for _, file := range files {
|
|
select {
|
|
case <-d.Context.Done():
|
|
return d.Context.Err()
|
|
default:
|
|
f, err := os.Open(file)
|
|
if err != nil {
|
|
return err
|
|
}
|
|
if _, err = io.Copy(d.Writer, f); err != nil {
|
|
f.Close()
|
|
return err
|
|
}
|
|
f.Close()
|
|
}
|
|
}
|
|
return nil
|
|
}
|
|
|
|
// mergeSegmentsFFmpeg reads every file in tempDir and merges their content to the outputFile
|
|
// with ffmpeg (https://ffmpeg.org/).
|
|
// The given output file gets created or overwritten if already existing
|
|
func (d Downloader) mergeSegmentsFFmpeg(files []string) error {
|
|
list, err := os.Create(filepath.Join(d.TempDir, "list.txt"))
|
|
if err != nil {
|
|
return err
|
|
}
|
|
|
|
for _, file := range files {
|
|
if _, err = fmt.Fprintf(list, "file '%s'\n", file); err != nil {
|
|
list.Close()
|
|
return err
|
|
}
|
|
}
|
|
list.Close()
|
|
|
|
// predefined options ... custom options ... predefined output filename
|
|
command := []string{
|
|
"-y",
|
|
"-f", "concat",
|
|
"-safe", "0",
|
|
"-i", list.Name(),
|
|
"-c", "copy",
|
|
}
|
|
if d.FFmpegOpts != nil {
|
|
command = append(command, d.FFmpegOpts...)
|
|
}
|
|
|
|
var tmpfile string
|
|
if _, ok := d.Writer.(*io.PipeWriter); ok {
|
|
if file, ok := d.Writer.(*os.File); ok {
|
|
tmpfile = filepath.Base(file.Name())
|
|
}
|
|
}
|
|
if filepath.Ext(tmpfile) == "" {
|
|
// checks if the -f flag is set (overwrites the output format)
|
|
var hasF bool
|
|
for _, opts := range d.FFmpegOpts {
|
|
if strings.TrimSpace(opts) == "-f" {
|
|
hasF = true
|
|
break
|
|
}
|
|
}
|
|
if !hasF {
|
|
command = append(command, "-f", "matroska")
|
|
f, err := os.CreateTemp(d.TempDir, "")
|
|
if err != nil {
|
|
return err
|
|
}
|
|
f.Close()
|
|
tmpfile = f.Name()
|
|
}
|
|
}
|
|
command = append(command, tmpfile)
|
|
|
|
var errBuf bytes.Buffer
|
|
cmd := exec.CommandContext(d.Context, "ffmpeg",
|
|
command...)
|
|
cmd.Stderr = &errBuf
|
|
|
|
if err = cmd.Run(); err != nil {
|
|
if errBuf.Len() > 0 {
|
|
return fmt.Errorf(errBuf.String())
|
|
} else {
|
|
return err
|
|
}
|
|
}
|
|
file, err := os.Open(tmpfile)
|
|
if err != nil {
|
|
return err
|
|
}
|
|
defer file.Close()
|
|
_, err = io.Copy(d.Writer, file)
|
|
return err
|
|
}
|
|
|
|
// downloadSegments downloads every mpeg transport stream segment to a given
|
|
// directory (more information below).
|
|
// After every segment download onSegmentDownload will be called with:
|
|
// the downloaded segment, the current position, the total size of segments to download,
|
|
// the file where the segment content was written to an error (if occurred).
|
|
// The filename is always <number of downloaded segment>.ts
|
|
//
|
|
// Short explanation:
|
|
// The actual crunchyroll video is split up in multiple segments (or video files) which
|
|
// have to be downloaded and merged after to generate a single video file.
|
|
// And this function just downloads each of this segment into the given directory.
|
|
// See https://en.wikipedia.org/wiki/MPEG_transport_stream for more information
|
|
func (d Downloader) downloadSegments(format *Format) ([]string, error) {
|
|
if err := format.InitVideo(); err != nil {
|
|
return nil, err
|
|
}
|
|
|
|
var wg sync.WaitGroup
|
|
var lock sync.Mutex
|
|
chunkSize := int(math.Ceil(float64(format.Video.Chunklist.Count()) / float64(d.Goroutines)))
|
|
|
|
// when a onSegmentDownload call returns an error, this context will be set cancelled and stop all goroutines
|
|
ctx, cancel := context.WithCancel(context.Background())
|
|
defer cancel()
|
|
|
|
// receives the decrypt block and iv from the first segment.
|
|
// in my tests, only the first segment has specified this data, so the decryption data from this first segments will be used in every other segment too
|
|
block, iv, err := getCrypt(format, format.Video.Chunklist.Segments[0])
|
|
if err != nil {
|
|
return nil, err
|
|
}
|
|
|
|
var total int32
|
|
for i := 0; i < int(format.Video.Chunklist.Count()); i += chunkSize {
|
|
wg.Add(1)
|
|
end := i + chunkSize
|
|
if end > int(format.Video.Chunklist.Count()) {
|
|
end = int(format.Video.Chunklist.Count())
|
|
}
|
|
i := i
|
|
|
|
go func() {
|
|
defer wg.Done()
|
|
|
|
for j, segment := range format.Video.Chunklist.Segments[i:end] {
|
|
select {
|
|
case <-d.Context.Done():
|
|
case <-ctx.Done():
|
|
return
|
|
default:
|
|
var file *os.File
|
|
for k := 0; k < 3; k++ {
|
|
filename := filepath.Join(d.TempDir, fmt.Sprintf("%d.ts", i+j))
|
|
file, err = d.downloadSegment(format, segment, filename, block, iv)
|
|
if err == nil {
|
|
break
|
|
}
|
|
if k == 2 {
|
|
cancel()
|
|
return
|
|
}
|
|
select {
|
|
case <-d.Context.Done():
|
|
case <-ctx.Done():
|
|
return
|
|
case <-time.After(5 * time.Duration(k) * time.Second):
|
|
// sleep if an error occurs. very useful because sometimes the connection times out
|
|
}
|
|
}
|
|
if d.OnSegmentDownload != nil {
|
|
if d.LockOnSegmentDownload {
|
|
lock.Lock()
|
|
}
|
|
|
|
if err = d.OnSegmentDownload(segment, int(atomic.AddInt32(&total, 1)), int(format.Video.Chunklist.Count()), file); err != nil {
|
|
if d.LockOnSegmentDownload {
|
|
lock.Unlock()
|
|
}
|
|
file.Close()
|
|
return
|
|
}
|
|
if d.LockOnSegmentDownload {
|
|
lock.Unlock()
|
|
}
|
|
}
|
|
file.Close()
|
|
}
|
|
}
|
|
}()
|
|
}
|
|
wg.Wait()
|
|
|
|
select {
|
|
case <-d.Context.Done():
|
|
return nil, d.Context.Err()
|
|
case <-ctx.Done():
|
|
return nil, err
|
|
default:
|
|
var files []string
|
|
for i := 0; i < int(total); i++ {
|
|
files = append(files, filepath.Join(d.TempDir, fmt.Sprintf("%d.ts", i)))
|
|
}
|
|
|
|
return files, nil
|
|
}
|
|
}
|
|
|
|
// getCrypt extracts the key and iv of a m3u8 segment and converts it into a cipher.Block and an iv byte sequence
|
|
func getCrypt(format *Format, segment *m3u8.MediaSegment) (block cipher.Block, iv []byte, err error) {
|
|
var resp *http.Response
|
|
|
|
resp, err = format.crunchy.Client.Get(segment.Key.URI)
|
|
if err != nil {
|
|
return nil, nil, err
|
|
}
|
|
defer resp.Body.Close()
|
|
key, err := ioutil.ReadAll(resp.Body)
|
|
|
|
block, err = aes.NewCipher(key)
|
|
if err != nil {
|
|
return nil, nil, err
|
|
}
|
|
iv = []byte(segment.Key.IV)
|
|
if len(iv) == 0 {
|
|
iv = key
|
|
}
|
|
|
|
return block, iv, nil
|
|
}
|
|
|
|
// downloadSegment downloads a segment, decrypts it and names it after the given index
|
|
func (d Downloader) downloadSegment(format *Format, segment *m3u8.MediaSegment, filename string, block cipher.Block, iv []byte) (*os.File, error) {
|
|
// every segment is aes-128 encrypted and has to be decrypted when downloaded
|
|
content, err := d.decryptSegment(format.crunchy.Client, segment, block, iv)
|
|
if err != nil {
|
|
return nil, err
|
|
}
|
|
|
|
file, err := os.Create(filename)
|
|
if err != nil {
|
|
return nil, err
|
|
}
|
|
defer file.Close()
|
|
if _, err = file.Write(content); err != nil {
|
|
return nil, err
|
|
}
|
|
|
|
return file, nil
|
|
}
|
|
|
|
// https://github.com/oopsguy/m3u8/blob/4150e93ec8f4f8718875a02973f5d792648ecb97/tool/crypt.go#L25
|
|
func (d Downloader) decryptSegment(client *http.Client, segment *m3u8.MediaSegment, block cipher.Block, iv []byte) ([]byte, error) {
|
|
req, err := http.NewRequest(http.MethodGet, segment.URI, nil)
|
|
if err != nil {
|
|
return nil, err
|
|
}
|
|
req.WithContext(d.Context)
|
|
|
|
resp, err := client.Do(req)
|
|
if err != nil {
|
|
return nil, err
|
|
}
|
|
defer resp.Body.Close()
|
|
|
|
raw, err := ioutil.ReadAll(resp.Body)
|
|
if err != nil {
|
|
return nil, err
|
|
}
|
|
|
|
blockMode := cipher.NewCBCDecrypter(block, iv[:block.BlockSize()])
|
|
decrypted := make([]byte, len(raw))
|
|
blockMode.CryptBlocks(decrypted, raw)
|
|
raw = d.pkcs5UnPadding(decrypted)
|
|
|
|
return raw, nil
|
|
}
|
|
|
|
// https://github.com/oopsguy/m3u8/blob/4150e93ec8f4f8718875a02973f5d792648ecb97/tool/crypt.go#L47
|
|
func (d Downloader) pkcs5UnPadding(origData []byte) []byte {
|
|
length := len(origData)
|
|
unPadding := int(origData[length-1])
|
|
return origData[:(length - unPadding)]
|
|
}
|