summarizer/TranscriptFileService.cs

using System.Text;
using YoutubeSummarizer.Models;

namespace YoutubeSummarizer.Services;

/// <summary>
/// Saves video metadata and timestamped transcript to a plain text file.
/// The file is formatted with metadata at the top followed by the transcript
/// organized by timestamps.
/// </summary>
public static class TranscriptFileService
{
    /// <summary>
    /// Saves the transcript and metadata to a text file in the specified directory.
    /// Returns the full path to the saved file.
    /// </summary>
    public static async Task<string> SaveAsync(
        VideoMetadata metadata,
        VideoTranscript transcript,
        string? summaryText = null,
        string? outputDirectory = null,
        CancellationToken ct = default)
    {
        outputDirectory ??= Environment.CurrentDirectory;
        Directory.CreateDirectory(outputDirectory);

        // Build a safe filename from the video title
        var safeTitle = SanitizeFileName(metadata.Title);
        var fileName = $"{safeTitle}_{metadata.VideoId}.txt";
        var filePath = Path.Combine(outputDirectory, fileName);

        var sb = new StringBuilder();

        // ── Metadata section ─────────────────────────────────────────────────
        sb.AppendLine("════════════════════════════════════════════════════════════════");
        sb.AppendLine("  VIDEO METADATA");
        sb.AppendLine("════════════════════════════════════════════════════════════════");
        sb.AppendLine();
        sb.AppendLine($"  Title:       {metadata.Title}");
        sb.AppendLine($"  Channel:     {metadata.ChannelTitle}");
        sb.AppendLine($"  Published:   {metadata.PublishedAt:MMMM d, yyyy}");
        sb.AppendLine($"  Duration:    {metadata.FormattedDuration}");
        sb.AppendLine($"  Video ID:    {metadata.VideoId}");
        sb.AppendLine($"  URL:         https://youtu.be/{metadata.VideoId}");

        if (!string.IsNullOrWhiteSpace(summaryText))
        {
            sb.AppendLine();
            sb.AppendLine("  ── SUMMARY ──────────────────────────────────────────────");
            sb.AppendLine();
            foreach (var wrappedLine in WordWrap(summaryText, maxWidth: 72))
            {
                sb.AppendLine($"    {wrappedLine}");
            }
        }

        sb.AppendLine();

        // ── Transcript source ────────────────────────────────────────────────
        var sourceLabel = transcript.Source switch
        {
            TranscriptSource.OwnerPublished       => "Owner-published captions",
            TranscriptSource.CommunityContributed => "Community-contributed captions",
            TranscriptSource.AutoGenerated        => "Auto-generated (ASR)",
            TranscriptSource.MetadataOnly         => "Metadata only (no captions)",
            _                                     => "Unknown"
        };
        sb.AppendLine($"  Transcript Source: {sourceLabel}");
        sb.AppendLine($"  Word Count:        {transcript.WordCount:N0}");
        sb.AppendLine($"  Saved:             {DateTimeOffset.UtcNow:yyyy-MM-dd HH:mm} UTC");
        sb.AppendLine();

        // ── Transcript section ───────────────────────────────────────────────
        sb.AppendLine("════════════════════════════════════════════════════════════════");
        sb.AppendLine("  TRANSCRIPT");
        sb.AppendLine("════════════════════════════════════════════════════════════════");
        sb.AppendLine();

        if (transcript.Segments.Count > 0)
        {
            // Group segments into blocks by time intervals for readability
            // Each block groups consecutive segments within ~30 seconds
            var blocks = GroupSegmentsByInterval(transcript.Segments, intervalSeconds: 30);

            foreach (var block in blocks)
            {
                var firstTimestamp = block[0].FormattedTimestamp;
                sb.AppendLine($"  [{firstTimestamp}]");

                // Combine the text for segments in this time block
                var blockText = string.Join(" ", block.Select(s => s.Text));
                foreach (var wrappedLine in WordWrap(blockText, maxWidth: 72))
                {
                    sb.AppendLine($"    {wrappedLine}");
                }
                sb.AppendLine();
            }
        }
        else
        {
            // No timestamps available — write plain text
            sb.AppendLine("  (No timestamp data available)");
            sb.AppendLine();
            foreach (var wrappedLine in WordWrap(transcript.Text, maxWidth: 72))
            {
                sb.AppendLine($"    {wrappedLine}");
            }
            sb.AppendLine();
        }

        sb.AppendLine("════════════════════════════════════════════════════════════════");
        sb.AppendLine("  END OF TRANSCRIPT");
        sb.AppendLine("════════════════════════════════════════════════════════════════");

        await File.WriteAllTextAsync(filePath, sb.ToString(), ct);
        return filePath;
    }

    // ─────────────────────────────────────────────────────────────────────────
    // Helpers
    // ─────────────────────────────────────────────────────────────────────────

    /// <summary>
    /// Groups timestamped segments into blocks based on a time interval.
    /// This produces readable chunks (e.g. every 30 seconds) instead of
    /// one line per subtitle cue.
    /// </summary>
    private static List<List<TimestampedSegment>> GroupSegmentsByInterval(
        IReadOnlyList<TimestampedSegment> segments,
        int intervalSeconds)
    {
        var blocks = new List<List<TimestampedSegment>>();
        if (segments.Count == 0) return blocks;

        var currentBlock = new List<TimestampedSegment> { segments[0] };
        var blockStartTime = segments[0].Start;

        for (int i = 1; i < segments.Count; i++)
        {
            if ((segments[i].Start - blockStartTime).TotalSeconds >= intervalSeconds)
            {
                blocks.Add(currentBlock);
                currentBlock = new List<TimestampedSegment>();
                blockStartTime = segments[i].Start;
            }
            currentBlock.Add(segments[i]);
        }

        if (currentBlock.Count > 0)
            blocks.Add(currentBlock);

        return blocks;
    }

    /// <summary>
    /// Removes characters that are invalid in file names.
    /// Truncates to a reasonable length to avoid path-length issues.
    /// </summary>
    private static string SanitizeFileName(string title)
    {
        var invalid = Path.GetInvalidFileNameChars();
        var sb = new StringBuilder(title.Length);

        foreach (var ch in title)
        {
            if (Array.IndexOf(invalid, ch) < 0)
                sb.Append(ch);
            else
                sb.Append('_');
        }

        // Replace runs of spaces/underscores with a single underscore
        var result = System.Text.RegularExpressions.Regex.Replace(
            sb.ToString().Trim(), @"[\s_]+", "_");

        // Truncate to keep file paths manageable
        return result.Length > 80 ? result[..80] : result;
    }

    /// <summary>
    /// Word-wraps text at the specified width, breaking at word boundaries.
    /// </summary>
    private static IEnumerable<string> WordWrap(string text, int maxWidth)
    {
        foreach (var paragraph in text.Split('\n'))
        {
            if (string.IsNullOrWhiteSpace(paragraph))
            {
                yield return string.Empty;
                continue;
            }

            var words = paragraph.Split(' ', StringSplitOptions.RemoveEmptyEntries);
            var current = new StringBuilder();

            foreach (var word in words)
            {
                if (current.Length + word.Length + 1 > maxWidth)
                {
                    yield return current.ToString();
                    current.Clear();
                }

                if (current.Length > 0) current.Append(' ');
                current.Append(word);
            }

            if (current.Length > 0)
                yield return current.ToString();
        }
    }
}