TuneRipper/TuneRipper.Api/Services/VideoService.cs

using TuneRipper.Shared.Models;
using System.Text.RegularExpressions;
using System.Net.Http;
using System.Threading.Tasks;
using System.Text.RegularExpressions;
using System.Diagnostics;
using System.Text;

namespace TuneRipper.Api.Services;

public interface IVideoService
{
    Task<YouTubeVideoDetails?> GetVideoDetailsAsync(string url);
    bool IsValidYouTubeUrl(string url);
    string ExtractVideoId(string url);
    Task<string?> ProcessVideoWithProgressAsync(string url, string format, string jobId, string sessionId, Action<int, string> progressCallback);
}

public class VideoService : IVideoService
{
    private readonly ILogger<VideoService> _logger;
    private readonly HttpClient _httpClient;
    private readonly IWebHostEnvironment _environment;
    private readonly ISessionService _sessionService;

    public VideoService(ILogger<VideoService> logger, IWebHostEnvironment environment, IHttpClientFactory httpClientFactory, ISessionService sessionService)
    {
        _logger = logger;
        _httpClient = httpClientFactory.CreateClient();
        _environment = environment;
        _sessionService = sessionService;
    }

    public async Task<YouTubeVideoDetails?> GetVideoDetailsAsync(string url)
    {
        try
        {
            if (!IsValidYouTubeUrl(url))
            {
                _logger.LogWarning("Invalid YouTube URL: {Url}", url);
                return null;
            }

            var videoId = ExtractVideoId(url);
            var thumbnailUrl = $"https://img.youtube.com/vi/{videoId}/maxresdefault.jpg";
            string? title = null;

            // Fetch HTML and parse title
            try
            {
                var html = await _httpClient.GetStringAsync(url);
                // Try OpenGraph title first
                var ogTitleMatch = Regex.Match(html, "<meta property='og:title' content='(.*?)'", RegexOptions.IgnoreCase);
                if (!ogTitleMatch.Success)
                {
                    ogTitleMatch = Regex.Match(html, "<meta property=\"og:title\" content=\"(.*?)\"", RegexOptions.IgnoreCase);
                }
                if (ogTitleMatch.Success)
                {
                    title = ogTitleMatch.Groups[1].Value;
                }
                else
                {
                    // Fallback to <title>
                    var titleMatch = Regex.Match(html, "<title>(.*?)</title>", RegexOptions.IgnoreCase);
                    if (titleMatch.Success)
                    {
                        title = titleMatch.Groups[1].Value.Replace(" - YouTube", "").Trim();
                    }
                }
            }
            catch (Exception ex)
            {
                _logger.LogWarning(ex, "Failed to fetch or parse video title for {Url}", url);
                title = $"YouTube Video {videoId}";
            }

            var videoDetails = new YouTubeVideoDetails
            {
                Id = videoId,
                Title = title,
                ThumbnailUrl = thumbnailUrl
            };

            _logger.LogInformation("Fetched image and title for video {VideoId}", videoId);
            return videoDetails;
        }
        catch (Exception ex)
        {
            _logger.LogError(ex, "Error fetching video details for URL: {Url}", url);
            return null;
        }
    }

    public async Task<string?> ProcessVideoWithProgressAsync(string url, string format, string jobId, string sessionId, Action<int, string> progressCallback)
    {
        try
        {
            var ytDlpPath = Path.Combine(_environment.WebRootPath, "yt-dlp.exe");
            
            if (!File.Exists(ytDlpPath))
            {
                throw new FileNotFoundException($"yt-dlp.exe not found at {ytDlpPath}");
            }

            // Get session-specific download directory
            var downloadDir = _sessionService.GetSessionDownloadsPath(sessionId);
            Directory.CreateDirectory(downloadDir);

            // Get video details first for better filename
            var videoDetails = await GetVideoDetailsAsync(url);
            
            // Generate filename with format: jobId_title.mp3
            var safeTitle = SanitizeFileName(videoDetails?.Title ?? "video");
            var customFileName = $"{jobId}_{safeTitle}";
            var outputTemplate = Path.Combine(downloadDir, $"{customFileName}.%(ext)s");

            _logger.LogInformation("Processing video with filename pattern: {FileName} in session: {SessionId}", customFileName, sessionId);

            // Build yt-dlp arguments
            var arguments = new StringBuilder();
            arguments.Append($"--extract-audio ");
            arguments.Append($"--audio-format {format} ");
            arguments.Append($"--audio-quality 0 ");
            arguments.Append("--no-playlist ");
            arguments.Append("--no-warnings ");
            arguments.Append("--newline ");
            arguments.Append($"--output \"{outputTemplate}\" ");
            arguments.Append($"\"{url}\"");

            _logger.LogInformation("Starting yt-dlp process with arguments: {Args}", arguments.ToString());

            var processInfo = new ProcessStartInfo
            {
                FileName = ytDlpPath,
                Arguments = arguments.ToString(),
                UseShellExecute = false,
                RedirectStandardOutput = true,
                RedirectStandardError = true,
                CreateNoWindow = true,
                WorkingDirectory = downloadDir
            };

            using var process = Process.Start(processInfo);
            if (process == null)
            {
                throw new Exception("Failed to start yt-dlp process");
            }

            var outputBuilder = new StringBuilder();
            var errorBuilder = new StringBuilder();
            
            // Monitor progress
            process.OutputDataReceived += (sender, e) =>
            {
                if (!string.IsNullOrEmpty(e.Data))
                {
                    outputBuilder.AppendLine(e.Data);
                    _logger.LogDebug("yt-dlp output: {Output}", e.Data);
                    
                    // Parse progress from yt-dlp output
                    ParseProgressFromOutput(e.Data, progressCallback);
                }
            };

            process.ErrorDataReceived += (sender, e) =>
            {
                if (!string.IsNullOrEmpty(e.Data))
                {
                    errorBuilder.AppendLine(e.Data);
                    _logger.LogWarning("yt-dlp error: {Error}", e.Data);
                }
            };

            process.BeginOutputReadLine();
            process.BeginErrorReadLine();

            // Wait for process completion with timeout
            var completed = await Task.Run(() => process.WaitForExit(300000)); // 5 minute timeout
            
            if (!completed)
            {
                process.Kill();
                throw new TimeoutException("yt-dlp process timed out after 5 minutes");
            }

            if (process.ExitCode != 0)
            {
                var error = errorBuilder.ToString();
                throw new Exception($"yt-dlp failed with exit code {process.ExitCode}. Error: {error}");
            }

            // Find the downloaded file with our job ID pattern in the session directory
            var downloadedFiles = Directory.GetFiles(downloadDir, $"{jobId}_*.*")
                .Where(f => !f.EndsWith(".part"))
                .OrderByDescending(System.IO.File.GetCreationTime)
                .ToArray();

            if (downloadedFiles.Length == 0)
            {
                throw new Exception($"No downloaded file found for job {jobId} in session {sessionId} after processing");
            }

            var downloadedFile = downloadedFiles[0];
            var finalFileName = Path.GetFileName(downloadedFile);
            
            _logger.LogInformation("Successfully processed video. Job: {JobId}, Session: {SessionId}, File: {FileName}", jobId, sessionId, finalFileName);
            
            progressCallback(100, "Processing completed successfully");
            
            return finalFileName;
        }
        catch (Exception ex)
        {
            _logger.LogError(ex, "Error processing video with yt-dlp for job {JobId} in session {SessionId}: {Url}", jobId, sessionId, url);
            progressCallback(0, $"Processing failed: {ex.Message}");
            throw;
        }
    }

    private string SanitizeFileName(string fileName)
    {
        if (string.IsNullOrWhiteSpace(fileName))
        {
            return "video";
        }

        // Remove or replace invalid filename characters
        var invalidChars = Path.GetInvalidFileNameChars();
        var sanitized = new StringBuilder();

        foreach (char c in fileName.Take(50)) // Limit length to 50 characters
        {
            if (invalidChars.Contains(c))
            {
                // Replace invalid characters with underscores or safe alternatives
                switch (c)
                {
                    case '/':
                    case '\\':
                    case '?':
                    case '*':
                        sanitized.Append('_');
                        break;
                    case ':':
                    case '|':
                        sanitized.Append('-');
                        break;
                    case '<':
                        sanitized.Append('(');
                        break;
                    case '>':
                        sanitized.Append(')');
                        break;
                    case '"':
                        sanitized.Append('\'');
                        break;
                    default:
                        sanitized.Append('_');
                        break;
                }
            }
            else if (char.IsWhiteSpace(c))
            {
                sanitized.Append('_');
            }
            else
            {
                sanitized.Append(c);
            }
        }

        var result = sanitized.ToString().Trim('_');
        
        // Ensure we have a valid filename
        if (string.IsNullOrWhiteSpace(result))
        {
            result = "video";
        }

        return result;
    }

    private void ParseProgressFromOutput(string output, Action<int, string> progressCallback)
    {
        try
        {
            // Parse different yt-dlp progress patterns
            if (output.Contains("[download]"))
            {
                // Example: "[download]  45.2% of 3.45MiB at 1.23MiB/s ETA 00:02"
                var percentMatch = Regex.Match(output, @"\[download\]\s+(\d+(?:\.\d+)?)%");
                if (percentMatch.Success)
                {
                    if (double.TryParse(percentMatch.Groups[1].Value, out var percent))
                    {
                        var progressPercent = (int)Math.Min(percent * 0.7, 70); // Download is 70% of total progress
                        progressCallback(progressPercent, "Downloading audio...");
                        return;
                    }
                }
                
                if (output.Contains("100%") || output.Contains("has already been downloaded"))
                {
                    progressCallback(70, "Download complete, converting...");
                    return;
                }
            }
            else if (output.Contains("[ffmpeg]"))
            {
                progressCallback(85, "Converting to MP3...");
                return;
            }
            else if (output.Contains("Deleting original file"))
            {
                progressCallback(95, "Finalizing...");
                return;
            }
        }
        catch (Exception ex)
        {
            _logger.LogWarning(ex, "Error parsing progress from output: {Output}", output);
        }
    }

    public bool IsValidYouTubeUrl(string url)
    {
        if (string.IsNullOrWhiteSpace(url))
            return false;

        var patterns = new[]
        {
            @"^https?://(www\.)?youtube\.com/watch\?v=[\w-]+",
            @"^https?://(www\.)?youtu\.be/[\w-]+",
            @"^https?://(www\.)?youtube\.com/embed/[\w-]+",
            @"^https?://(www\.)?youtube\.com/v/[\w-]+"
        };

        return patterns.Any(pattern => Regex.IsMatch(url, pattern, RegexOptions.IgnoreCase));
    }

    public string ExtractVideoId(string url)
    {
        if (string.IsNullOrWhiteSpace(url))
            return string.Empty;

        try
        {
            var patterns = new Dictionary<string, string>
            {
                { @"youtu\.be/([^?&]+)", "$1" },
                { @"youtube\.com/watch\?v=([^&]+)", "$1" },
                { @"youtube\.com/embed/([^?&]+)", "$1" },
                { @"youtube\.com/v/([^?&]+)", "$1" }
            };

            foreach (var pattern in patterns)
            {
                var match = Regex.Match(url, pattern.Key, RegexOptions.IgnoreCase);
                if (match.Success)
                {
                    return match.Groups[1].Value;
                }
            }
        }
        catch (Exception ex)
        {
            _logger.LogError(ex, "Error extracting video ID from URL: {Url}", url);
        }

        return string.Empty;
    }
}