private async Task <string> GenrateSubTitleAsync(string id, string Path, string Name)
{
try
{
string FullPath = $@"{Path}\\{ Name.ValidNameForWindows()}.srt";
var trackInfos = await client.GetVideoClosedCaptionTrackInfosAsync(id);
var trackInfo = trackInfos.First(t => t.Language.Code == "en");
var track = await client.GetClosedCaptionTrackAsync(trackInfo);
using StreamWriter file =
new StreamWriter(FullPath);
int line = 1;
foreach (var item in track.Captions)
{
string from = $"{item.Offset.Hours.ToString("00")}:{item.Offset.Minutes.ToString("00")}:{item.Offset.Seconds.ToString("00")},{item.Offset.Milliseconds.ToString("000")}";
TimeSpan ToSpaon = item.Offset.Add(item.Duration);
string to = $"{ToSpaon.Hours.ToString("00")}:{ToSpaon.Minutes.ToString("00")}:{ToSpaon.Seconds.ToString("00")},{ToSpaon.Milliseconds.ToString("000")}";
file.WriteLine(line);
file.WriteLine($"{from} --> {to}");
file.WriteLine(item.Text);
file.WriteLine();
line++;
}
return($"subtitle Write success to file : {FullPath}");
}
catch (Exception ex)
{
throw ex;
}
}
/*public bool GetVideoThumbnail(string path, string saveThumbnailTo, int seconds)
* {
* string parameters = $"-ss {seconds} -i {path} -f image2 -vframes 1 -y {saveThumbnailTo}";
*
* var processInfo = new ProcessStartInfo();
* processInfo.FileName = pathToConvertor;
* processInfo.Arguments = parameters;
* processInfo.CreateNoWindow = true;
* processInfo.UseShellExecute = false;
*
* File.Delete(saveThumbnailTo);
*
* using (var process = new Process())
* {
* process.StartInfo = processInfo;
* process.Start();
* process.WaitForExit();
* }
*
* return File.Exists(saveThumbnailTo);
* }*/
private static void Test()
{
var client = new YoutubeClient();
client.GetChannelUploadsAsync("UCZqh6VE-OYFz2RCWDbZQOqw").ContinueWith(task =>
{
var video = task.Result.FirstOrDefault(v => v.Title.Contains("listening practice test"));
client.GetVideoAsync("").ContinueWith(t =>
{
//t.Result.
//t.Result.
});
client.GetVideoClosedCaptionTrackInfosAsync(video.Id).ContinueWith(task1 =>
{
//task1.Result[0].
var captionTrackInfo = task1.Result.FirstOrDefault(info => info.Language.Code == "en");
client.GetClosedCaptionTrackAsync(captionTrackInfo).ContinueWith(task2 =>
{
var captionTrack = task2.Result;
foreach (var captionTrackCaption in captionTrack.Captions)
{
Console.WriteLine($"{captionTrackCaption.Offset}: {captionTrackCaption.Text}");
}
});
});
});
}
public async Task <ClosedCaptionTrack> Subtitles(string video, int n)
{
var client = new YoutubeClient();
var info = await client.GetVideoClosedCaptionTrackInfosAsync(video);
return(await client.GetClosedCaptionTrackAsync(info[n]));
}
static async Task Load()
{
var client = new YoutubeClient();
var list = await client.GetChannelUploadsAsync("UC4V3oCikXeSqYQr0hBMARwg");
foreach (var video in list)
{
var tubeUrl = video.GetUrl();
if (await(await videoCollection.FindAsync(x => x.YoutubeUrl == tubeUrl)).AnyAsync())
{
continue;
}
var info = await client.GetVideoClosedCaptionTrackInfosAsync(video.Id);
if (info.Count <= 0)
{
continue;
}
var minfo = await client.GetVideoMediaStreamInfosAsync(video.Id);
var subVideo = new SubVideo
{
Title = video.Title,
YoutubeUrl = tubeUrl,
Thumbnail = video.Thumbnails.MediumResUrl,
SourceUrl = minfo.Muxed[0].Url,
Published = video.UploadDate.DateTime
};
await videoCollection.InsertOneAsync(subVideo);
var captions = await client.GetClosedCaptionTrackAsync(info.First());
var mergedCaptions = new List <VideoText>();
for (int i = 0; i < captions.Captions.Count; i++)
{
for (int size = 1; size < 4; size++)
{
var cap = captions.Captions.Skip(i).Take(size).Aggregate((f, s) =>
new ClosedCaption($"{f.Text} {s.Text}", f.Offset, f.Duration + s.Duration));
mergedCaptions.Add(new VideoText
{
VideoId = subVideo.Id,
OffSet = cap.Offset,
Text = cap.Text,
Duration = cap.Duration
});
}
}
await textCollection.InsertManyAsync(mergedCaptions);
}
}
public async Task YoutubeClient_GetClosedCaptionTrackAsync_Normal_Test()
{
var client = new YoutubeClient();
var videoInfo = await client.GetVideoInfoAsync("_QdPW8JrYzQ");
var trackInfo = videoInfo.ClosedCaptionTracks.First();
var track = await client.GetClosedCaptionTrackAsync(trackInfo);
Assert.That.IsSet(track);
}
private async Task <IReadOnlyList <ClosedCaption> > LoadCaptions(string videoId, string lang)
{
var client = new YoutubeClient();
var track = await client.GetVideoClosedCaptionTrackInfosAsync(videoId);
var captionTrackInfo = track.FirstOrDefault(info => info.Language.Code == lang);
var closedCaptionsTrack = await client.GetClosedCaptionTrackAsync(captionTrackInfo);
return(closedCaptionsTrack.Captions);
}
public async Task YoutubeClient_GetClosedCaptionTrackAsync_Test(string videoId)
{
var client = new YoutubeClient();
var closedCaptionTrackInfos = await client.GetVideoClosedCaptionTrackInfosAsync(videoId);
var trackInfo = closedCaptionTrackInfos.First();
var track = await client.GetClosedCaptionTrackAsync(trackInfo);
Assert.That(track, Is.Not.Null);
}
private async Task GetClosedCaptionTrackAsync(string id)
{
var client = new YoutubeClient();
var trackInfos = await client.GetVideoClosedCaptionTrackInfosAsync(id);
var trackInfo = trackInfos.First(t => t.Language.Code == "en");
var track = await client.GetClosedCaptionTrackAsync(trackInfo);
var caption = track.Captions;
//var text = caption.Text; // "And the game was afoot."
}
public async Task YoutubeClient_GetClosedCaptionTrackAsync_Test()
{
string id = (string)TestContext.DataRow["Id"];
var client = new YoutubeClient();
var videoInfo = await client.GetVideoInfoAsync(id);
var trackInfo = videoInfo.ClosedCaptionTracks.First();
var track = await client.GetClosedCaptionTrackAsync(trackInfo);
Assert.That.IsSet(track);
}
/// <summary>
/// Download closed captions for a youtube video
/// </summary>
/// <param name="videoUrl">Youtube url (e.g. https://www.youtube.com/watch?v=VIDEO_ID)</param>
/// <returns>A ClosedCaptionTrack object.</returns>
public async Task <ClosedCaptionTrack> DownloadClosedCaptions(string videoUrl = null)
{
videoUrl = videoUrl ?? youtubeUrl;
var videoId = GetVideoId(videoUrl);
var trackInfos = await youtubeClient.GetVideoClosedCaptionTrackInfosAsync(videoId);
if (trackInfos?.Count == 0)
{
return(null);
}
var trackInfo = trackInfos.FirstOrDefault(t => t.Language.Code == "en") ?? trackInfos.First();
return(await youtubeClient.GetClosedCaptionTrackAsync(trackInfo));
}
public async Task <string> GetAndUpdateVideoCaptions(string channelId, string videoId, ILogger log)
{
IReadOnlyList <ClosedCaptionTrackInfo> tracks;
try {
tracks = await ytScaper.GetVideoClosedCaptionTrackInfosAsync(videoId);
}
catch (Exception ex) {
log.Warning(ex, "Unable to get captions for {VideoID}: {Error}", videoId, ex.Message);
return(null);
}
var en = tracks.FirstOrDefault(t => t.Language.Code == "en");
if (en == null)
{
return(null);
}
ClosedCaptionTrack track;
try {
track = await Policy.Handle <HttpRequestException>()
.RetryWithBackoff()
.ExecuteAsync(() => ytScaper.GetClosedCaptionTrackAsync(en));
}
catch (Exception ex) {
log.Warning(ex, "Unable to get captions for {VideoID}: {Error}", videoId, ex.Message);
return(null);
}
var text = track.Captions.Select(c => c.Text).Join("\n");
if (text != null)
{
var path = StringPath.Relative("VideoCaptions", channelId, $"{videoId}.txt");
try {
await Store.Save(path, text.AsStream());
}
catch (Exception ex) {
log.Warning(ex, "Error when saving captions {Path}", path);
}
}
return(text);
}
/// <summary>
/// Download closed captions for a youtube video
/// </summary>
/// <param name="videoUrl">Youtube url (e.g. https://www.youtube.com/watch?v=VIDEO_ID)</param>
/// <returns>A ClosedCaptionTrack object.</returns>
public async Task <ClosedCaptionTrack> DownloadClosedCaptions(string videoUrl = null)
{
if (overrideURLWithInputField && !string.IsNullOrEmpty(m_inputField.text))
{
youtubeUrl = m_inputField.text;
}
videoUrl = videoUrl ?? youtubeUrl;
var videoId = GetVideoId(videoUrl);
var trackInfos = await youtubeClient.GetVideoClosedCaptionTrackInfosAsync(videoId);
if (trackInfos?.Count == 0)
{
return(null);
}
var trackInfo = trackInfos.FirstOrDefault(t => t.Language.Code == "en") ?? trackInfos.First();
return(await youtubeClient.GetClosedCaptionTrackAsync(trackInfo));
}
public async Task <SubtitledVideo> GetSubtitledVideoAsync(string videoId)
{
var client = new YoutubeClient();
var video = await client.GetVideoAsync(videoId);
var captionTrackInfos = await client.GetVideoClosedCaptionTrackInfosAsync(videoId);
var mediaStreamInfoSet = await client.GetVideoMediaStreamInfosAsync(videoId);
var captions = await client.GetClosedCaptionTrackAsync(captionTrackInfos.First());
var subVideo = new SubtitledVideo
{
Title = video.Title,
YoutubeUrl = video.GetUrl(),
YoutubeId = videoId,
Thumbnail = video.Thumbnails.MediumResUrl,
VideoSourceUrl = mediaStreamInfoSet.Muxed[0].Url,
Published = video.UploadDate.DateTime,
TextBody = string.Join(" ", captions.Captions.Select(x => x.Text))
};
var strHtml = $"<span onclick=\"goto(0)\">{captions.Captions[0].Text}</span>";
for (int i = 1; i < captions.Captions.Count; i++)
{
var diff = captions.Captions[i].Offset - (captions.Captions[i - 1].Offset + captions.Captions[i - 1].Duration);
if (diff >= TimeSpan.FromMilliseconds(150))
{
strHtml += ".</br>";
}
strHtml += $"<span onclick=\"goto({(int)captions.Captions[i].Offset.TotalSeconds})\">{captions.Captions[i].Text}</span>";
}
subVideo.HtmlBody = strHtml;
return(subVideo);
}
static async Task Main(string[] args)
{
var client = new YoutubeClient();
// The full json with all relevant data is loaded. At this point in the pipeline it contains the information about each channel plus the id of the top three
// videos from each channel.
// The task of this script is to download the captions for each video
using (StreamReader file = File.OpenText("../../output/unlabeled_data/U_channelDataWithScrapedTopVideos.json"))
using (JsonTextReader reader = new JsonTextReader(file))
{
JObject channelData = (JObject)JToken.ReadFrom(reader);
foreach (KeyValuePair <string, JToken> entry in channelData)
{
// if (entry.Key == "UCtD9a-aXIYS6-e-8s7DISiw"){
Console.WriteLine("Now processing " + entry.Key);
foreach (JToken topVideo in entry.Value["top3videos"])
{
// JToken videoId = topVideo["VideoId"];
JToken videoId = topVideo;
string captions = null;
try {
var trackInfos = await client.GetVideoClosedCaptionTrackInfosAsync((string)videoId);
var trackInfo = trackInfos.First(t => t.Language.Code == "en");
var track = await client.GetClosedCaptionTrackAsync(trackInfo);
var captionList = track.Captions;
foreach (var line in captionList)
{
captions = captions + " " + line;
}
} catch (System.InvalidOperationException e1) {
Console.WriteLine("no captions found for: " + videoId);
} catch (YoutubeExplode.Exceptions.VideoUnavailableException e2) {
Console.WriteLine("Video deleted: " + videoId + " Channel: " + entry.Key);
} catch (System.ArgumentNullException e3) {
Console.WriteLine("Caption could not be correctly loaded: " + videoId);
}
var captionData = new Dictionary <string, string>()
{
{ "VideoId", (string)videoId },
{ "Captions", captions },
{ "Info", "downloaded" }
};
var filePath = "../Data/Captions/" + entry.Key + "/" + (string)videoId + ".jsonl";
if (!File.Exists(filePath + ".gz"))
{
Console.WriteLine("################ " + filePath);
System.IO.FileInfo dirCheck = new System.IO.FileInfo(filePath);
dirCheck.Directory.Create(); // If the directory already exists, this method does nothing.
using (StreamWriter jsonFile = File.CreateText(filePath))
{
JsonSerializer serializer = new JsonSerializer();
//serialize object directly into file stream
serializer.Serialize(jsonFile, captionData);
}
var bytes = File.ReadAllBytes(filePath);
using (FileStream fs = new FileStream(filePath + ".gz", FileMode.CreateNew))
using (GZipStream zipStream = new GZipStream(fs, CompressionMode.Compress, false))
{
zipStream.Write(bytes, 0, bytes.Length);
}
// Get rid of the unzipped json
File.Delete(filePath);
}
else
{
Console.WriteLine("already scraped: " + videoId);
}
}
}
}
}