URLNotesGrabberCore/URLNotesGrabberCORE/OutputMode.cs

using Microsoft.Extensions.Configuration;

namespace URLNotesGrabberCORE
{
    // Port of ThreeTxtFileHelper RunOutputMode + WritePostToFile + RenameExistingTxtFilesToBak.
    // For each Blog with a TTFolderPath, renames any existing .txt files in that folder to .bak,
    // then writes one .txt per PostType containing all posts of that type (date-sorted, fixed
    // field order). Reads from TL.db via DataAccess.GetAllPostsForBlog.
    public static class OutputMode
    {
        public static int Run(IConfiguration config, string[]? args = null)
        {
            DataAccess.EnsureTTFileHelperColumnsExist();

            string dbPath = DataAccess.GetActiveDbPath();
            Console.WriteLine($"Database: {Path.GetFullPath(dbPath)}");

            if (!RefreshPaths(config, args ?? Array.Empty<string>()))
                return 1;

            var blogs = DataAccess.GetAllBlogsWithTTFolderPath();
            int activeBlogs = DataAccess.CountActiveBlogs();
            Console.WriteLine($"{blogs.Count} of {activeBlogs} active blog(s) have a TTFolderPath.");

            if (blogs.Count == 0)
            {
                Console.WriteLine($"\nNothing to export: no blog in {Path.GetFullPath(dbPath)} has a TTFolderPath.");
                Console.WriteLine("Point --output at a TumblThree root so it can populate them: --output <root>,");
                Console.WriteLine("or set appSettings:PathTTRoot so the refresh runs automatically.");
                return 1;
            }

            int missingFolderCount = 0;
            int writtenCount = 0;

            foreach (var (blogName, folder) in blogs)
            {
                Console.WriteLine($"\nProcessing blog: {blogName}");

                // A stored path that this machine cannot see means the value was written on
                // another machine -- re-running --updatepaths locally is the fix, so say so
                // rather than lumping it in with "not set".
                if (!Directory.Exists(folder))
                {
                    Console.WriteLine($"  TTFolderPath folder not found: {folder}. Skipping.");
                    missingFolderCount++;
                    continue;
                }

                Console.WriteLine($"  TTFolderPath: {folder}");
                writtenCount++;

                try
                {
                    foreach (var bakFile in Directory.GetFiles(folder, "*.bak"))
                        File.Delete(bakFile);
                }
                catch (Exception ex)
                {
                    Console.WriteLine($"  Error deleting .bak files: {ex.Message}");
                }

                RenameExistingTxtFilesToBak(folder);

                var posts = DataAccess.GetAllPostsForBlog(blogName);
                Console.WriteLine($"  Found {posts.Count} post(s) for this blog.");

                // A post's type becomes a filename, so only a recognized type may be written. The
                // old `PostType ?? "Unknown"` invented Unknown.txt for untyped rows, which --ingest
                // then read back as a type named "Unknown" -- the two regenerated each other.
                // Untyped rows are skipped instead: after the backfill these are only rows with no
                // content signal at all, so nothing meaningful is lost, and nothing is invented.
                var typed = posts.Where(p => PostTypes.Normalize(p.PostType) != null).ToList();
                int untyped = posts.Count - typed.Count;
                if (untyped > 0)
                    Console.WriteLine($"    Skipping {untyped} post(s) with no recognized PostType.");

                var grouped = typed.GroupBy(p => PostTypes.Normalize(p.PostType)!);
                foreach (var typeGroup in grouped)
                {
                    string postType = typeGroup.Key;
                    string outputFilePath = Path.Combine(folder, $"{postType}.txt");
                    var ordered = typeGroup.OrderBy(p => p.Date).ToList();
                    Console.WriteLine($"    Writing {ordered.Count} post(s) to {postType}.txt");

                    using var writer = new StreamWriter(outputFilePath, false, System.Text.Encoding.UTF8);
                    bool isFirst = true;
                    foreach (var post in ordered)
                    {
                        if (!isFirst)
                        {
                            writer.WriteLine();
                            writer.WriteLine();
                        }
                        WritePostToFile(writer, post);
                        isFirst = false;
                    }
                }
            }

            Console.WriteLine($"\nOutput mode complete. {writtenCount} blog(s) exported, {missingFolderCount} skipped for a missing folder.");

            if (writtenCount == 0)
                Console.WriteLine("Every TTFolderPath points at a folder this machine cannot see. The paths were most likely written on another machine -- re-run --updatepaths <root> here so they match local drive letters.");

            return 0;
        }

        // Re-reads the TumblThree Index metadata into Blogs.TTFolderPath before exporting.
        // A TL.db synced between machines cannot hold one absolute path that is valid on
        // both, so the stored paths are only trustworthy on the machine that wrote them --
        // which makes this refresh part of a normal export rather than a separate chore.
        // Returns false only when the run should stop.
        private static bool RefreshPaths(IConfiguration config, string[] args)
        {
            var settings = config.GetSection("appSettings");

            if (args.Any(a => string.Equals(a, "--norefresh", StringComparison.OrdinalIgnoreCase)))
            {
                Console.WriteLine("Path refresh skipped (--norefresh); exporting to whatever paths TL.db already holds.");
                return true;
            }

            string? root = args.FirstOrDefault(a => !a.StartsWith("--", StringComparison.Ordinal))
                           ?? settings.GetValue<string>("PathTTRoot");

            var result = UpdateBlogPathsRunner.Scan(root, verbose: false);

            switch (result.Outcome)
            {
                case UpdateBlogPathsRunner.ScanOutcome.NoRootConfigured:
                    Console.WriteLine("No TumblThree root configured (appSettings:PathTTRoot is empty and none was passed),");
                    Console.WriteLine("so TTFolderPath was not refreshed. Pass one as --output <root> to refresh it.");
                    return true;

                case UpdateBlogPathsRunner.ScanOutcome.IndexFolderMissing:
                    // Silently exporting stale paths here would defeat the point of folding
                    // the refresh in, so a bad root is a hard stop.
                    Console.WriteLine($"Index folder not found at: {result.IndexPath}");
                    Console.WriteLine("Fix the root (or pass --norefresh to export the paths already in TL.db).");
                    return false;

                default:
                    Console.WriteLine($"Refreshed paths from {result.IndexPath}: " +
                                      $"{result.MetadataFiles} metadata file(s), {result.Written} written, " +
                                      $"{result.Unchanged} already correct, {result.NoLocation} without a location, " +
                                      $"{result.NoMatchingRow} without a blog row, {result.Errors} error(s).");
                    return true;
            }
        }

        private static void RenameExistingTxtFilesToBak(string folderPath)
        {
            try
            {
                foreach (var txtFile in Directory.GetFiles(folderPath, "*.txt"))
                {
                    string bakPath = Path.ChangeExtension(txtFile, ".bak");
                    if (File.Exists(bakPath)) File.Delete(bakPath);
                    File.Move(txtFile, bakPath, overwrite: true);
                }
            }
            catch (Exception ex)
            {
                Console.WriteLine($"    Error renaming txt files to .bak: {ex.Message}");
            }
        }

        private static void WritePostToFile(StreamWriter writer, TTPostRecord post)
        {
            var startColumns = new[] { "Post ID", "Date", "Post URL", "Slug", "Reblog Key", "Reblog URL", "Reblog Name", "Title", "Body" };
            var endColumns = new[] { "Tags", "Downloaded Files" };

            var columns = new Dictionary<string, string>();
            if (!string.IsNullOrWhiteSpace(post.PostId)) columns["Post ID"] = post.PostId;
            if (!string.IsNullOrWhiteSpace(post.Date)) columns["Date"] = post.Date!;
            if (!string.IsNullOrWhiteSpace(post.PostUrl)) columns["Post URL"] = post.PostUrl!;
            if (!string.IsNullOrWhiteSpace(post.Slug)) columns["Slug"] = post.Slug!;
            if (!string.IsNullOrWhiteSpace(post.ReblogKey)) columns["Reblog Key"] = post.ReblogKey!;
            if (!string.IsNullOrWhiteSpace(post.ReblogUrl)) columns["Reblog URL"] = post.ReblogUrl!;
            if (!string.IsNullOrWhiteSpace(post.ReblogName)) columns["Reblog Name"] = post.ReblogName!;
            if (!string.IsNullOrWhiteSpace(post.Title)) columns["Title"] = post.Title!;
            if (!string.IsNullOrWhiteSpace(post.Body)) columns["Body"] = post.Body!;
            if (!string.IsNullOrWhiteSpace(post.HasImage)) columns["Has Image"] = post.HasImage!;
            if (!string.IsNullOrWhiteSpace(post.Summary)) columns["Summary"] = post.Summary!;
            if (!string.IsNullOrWhiteSpace(post.Quote)) columns["Quote"] = post.Quote!;
            if (!string.IsNullOrWhiteSpace(post.Link)) columns["Link"] = post.Link!;
            if (!string.IsNullOrWhiteSpace(post.PhotoUrl)) columns["Photo URL"] = post.PhotoUrl!;
            if (!string.IsNullOrWhiteSpace(post.PhotoCaption)) columns["Photo Caption"] = post.PhotoCaption!;
            if (!string.IsNullOrWhiteSpace(post.AudioCaption)) columns["Audio Caption"] = post.AudioCaption!;
            if (!string.IsNullOrWhiteSpace(post.Question)) columns["Question"] = post.Question!;
            if (!string.IsNullOrWhiteSpace(post.Answer)) columns["Answer"] = post.Answer!;
            if (!string.IsNullOrWhiteSpace(post.Tags)) columns["Tags"] = post.Tags!;
            if (!string.IsNullOrWhiteSpace(post.DownloadedFiles)) columns["Downloaded Files"] = post.DownloadedFiles!;

            foreach (var col in startColumns)
            {
                if (columns.ContainsKey(col))
                {
                    writer.WriteLine($"{col}: {columns[col]}");
                    columns.Remove(col);
                }
            }

            var remaining = columns.Keys.Where(k => !endColumns.Contains(k)).OrderBy(k => k).ToList();
            foreach (var col in remaining)
                writer.WriteLine($"{col}: {columns[col]}");

            foreach (var col in endColumns)
            {
                if (columns.ContainsKey(col))
                    writer.WriteLine($"{col}: {columns[col]}");
            }
        }
    }
}