public async Task ConsumeThread(Thread thread, string board)
{
var dbCommands = GetPreparedStatements(board);
var hashObject = new ThreadHashObject(board, thread.OriginalPost.PostNumber);
if (!ThreadHashes.TryGetValue(hashObject, out var threadHashes))
{
// Rebuild hashes from database, if they exist
var hashes = await dbCommands.WithAccess(connection => dbCommands.GetHashesOfThread(hashObject.ThreadId));
threadHashes = new SortedList <ulong, int>();
foreach (var hashPair in hashes)
{
threadHashes.Add(hashPair.Key, hashPair.Value);
}
}
List <Post> postsToAdd = new List <Post>(thread.Posts.Length);
foreach (var post in thread.Posts)
{
if (threadHashes.TryGetValue(post.PostNumber, out int existingHash))
{
if (post.GenerateAsagiHash() != existingHash)
{
// Post has changed since we last saved it to the database
Program.Log($"[Asagi] Post /{board}/{post.PostNumber} has been modified");
await dbCommands.WithAccess(() => dbCommands.UpdatePost(post, false));
threadHashes[post.PostNumber] = post.GenerateAsagiHash();
}
else
{
// Post has not changed
}
}
else
{
// Post has not yet been inserted into the database
if (post.FileMd5 != null)
{
string timestampString = post.TimestampedFilename.ToString();
string radixString = Path.Combine(timestampString.Substring(0, 4), timestampString.Substring(4, 2));
if (Config.FullImagesEnabled)
{
Directory.CreateDirectory(Path.Combine(ImageDownloadLocation, board, radixString));
string fullImageFilename = Path.Combine(ImageDownloadLocation, radixString, post.TimestampedFilenameFull);
string fullImageUrl = $"https://i.4cdn.org/{board}/{post.TimestampedFilenameFull}";
await DownloadFile(fullImageUrl, fullImageFilename);
}
if (Config.ThumbnailsEnabled)
{
Directory.CreateDirectory(Path.Combine(ThumbDownloadLocation, board, radixString));
string thumbFilename = Path.Combine(ThumbDownloadLocation, radixString, $"{post.TimestampedFilename}s.jpg");
string thumbUrl = $"https://i.4cdn.org/{board}/{post.TimestampedFilename}s.jpg";
await DownloadFile(thumbUrl, thumbFilename);
}
}
postsToAdd.Add(post);
}
}
if (threadHashes.Count == 0)
{
// We are inserting the thread for the first time.
await dbCommands.WithAccess(() => dbCommands.InsertPosts(thread.Posts));
}
else
{
if (postsToAdd.Count > 0)
{
await dbCommands.WithAccess(() => dbCommands.InsertPosts(postsToAdd));
}
}
foreach (var post in postsToAdd)
{
threadHashes[post.PostNumber] = post.GenerateAsagiHash();
}
Program.Log($"[Asagi] {postsToAdd.Count} posts have been inserted from thread /{board}/{thread.OriginalPost.PostNumber}");
List <ulong> postNumbersToDelete = new List <ulong>(thread.Posts.Length);
foreach (var postNumber in threadHashes.Keys)
{
if (thread.Posts.All(x => x.PostNumber != postNumber))
{
// Post has been deleted
Program.Log($"[Asagi] Post /{board}/{postNumber} has been deleted");
await dbCommands.WithAccess(() => dbCommands.DeletePostOrThread(postNumber));
postNumbersToDelete.Add(postNumber);
}
}
// workaround for not being able to remove from a collection while enumerating it
foreach (var postNumber in postNumbersToDelete)
{
threadHashes.Remove(postNumber, out _);
}
if (thread.OriginalPost.Archived == true)
{
// We don't need the hashes if the thread is archived, since it will never change
// If it does change, we can just grab a new set from the database
ThreadHashes.TryRemove(hashObject, out _);
}
}
/// <inheritdoc/>
public async Task <IList <QueuedImageDownload> > ConsumeThread(Thread thread, string board)
{
var hashObject = new ThreadPointer(board, thread.OriginalPost.PostNumber);
List <QueuedImageDownload> imageDownloads = new List <QueuedImageDownload>(thread.Posts.Length);
async Task ProcessImages(Post post)
{
if (!Config.FullImagesEnabled && !Config.ThumbnailsEnabled)
{
return; // skip the DB check since we're not even bothering with images
}
if (post.FileMd5 != null)
{
MediaInfo mediaInfo = await GetMediaInfo(post.FileMd5, board);
if (mediaInfo?.Banned == true)
{
Program.Log($"[Asagi] Post /{board}/{post.PostNumber} contains a banned image; skipping");
return;
}
if (Config.FullImagesEnabled)
{
string fullImageName = mediaInfo?.MediaFilename ?? post.TimestampedFilenameFull;
string radixString = Path.Combine(fullImageName.Substring(0, 4), fullImageName.Substring(4, 2));
string radixDirectory = Path.Combine(ImageDownloadLocation, board, radixString);
Directory.CreateDirectory(radixDirectory);
string fullImageFilename = Path.Combine(radixDirectory, fullImageName);
string fullImageUrl = $"https://i.4cdn.org/{board}/{post.TimestampedFilenameFull}";
imageDownloads.Add(new QueuedImageDownload(new Uri(fullImageUrl), fullImageFilename));
}
if (Config.ThumbnailsEnabled)
{
string thumbImageName;
if (post.ReplyPostNumber == 0) // is OP
{
thumbImageName = mediaInfo?.PreviewOpFilename ?? $"{post.TimestampedFilename}s.jpg";
}
else
{
thumbImageName = mediaInfo?.PreviewReplyFilename ?? $"{post.TimestampedFilename}s.jpg";
}
string radixString = Path.Combine(thumbImageName.Substring(0, 4), thumbImageName.Substring(4, 2));
string radixDirectory = Path.Combine(ThumbDownloadLocation, board, radixString);
Directory.CreateDirectory(radixDirectory);
string thumbFilename = Path.Combine(radixDirectory, thumbImageName);
string thumbUrl = $"https://i.4cdn.org/{board}/{post.TimestampedFilename}s.jpg";
imageDownloads.Add(new QueuedImageDownload(new Uri(thumbUrl), thumbFilename));
}
}
}
if (!ThreadHashes.TryGetValue(hashObject, out var threadHashes))
{
// Rebuild hashes from database, if they exist
var hashes = await GetHashesOfThread(hashObject.ThreadId, board);
threadHashes = new SortedList <ulong, int>();
foreach (var hashPair in hashes)
{
threadHashes.Add(hashPair.Key, hashPair.Value);
var currentPost = thread.Posts.FirstOrDefault(post => post.PostNumber == hashPair.Key);
if (currentPost != null)
{
await ProcessImages(currentPost);
}
}
ThreadHashes.TryAdd(hashObject, threadHashes);
}
List <Post> postsToAdd = new List <Post>(thread.Posts.Length);
foreach (var post in thread.Posts)
{
if (threadHashes.TryGetValue(post.PostNumber, out int existingHash))
{
int hash = CalculateAsagiHash(post, true);
if (hash != existingHash)
{
// Post has changed since we last saved it to the database
Program.Log($"[Asagi] Post /{board}/{post.PostNumber} has been modified");
await UpdatePost(post, board, false);
threadHashes[post.PostNumber] = hash;
}
else
{
// Post has not changed
if (post.ReplyPostNumber == 0)
{
// OP post
await UpdatePostExif(post, board);
}
}
}
else
{
// Post has not yet been inserted into the database
postsToAdd.Add(post);
await ProcessImages(post);
}
}
if (threadHashes.Count == 0)
{
// We are inserting the thread for the first time.
await InsertPosts(thread.Posts, board);
}
else
{
if (postsToAdd.Count > 0)
{
await InsertPosts(postsToAdd, board);
}
}
foreach (var post in postsToAdd)
{
threadHashes[post.PostNumber] = CalculateAsagiHash(post, true);
}
Program.Log($"[Asagi] {postsToAdd.Count} posts have been inserted from thread /{board}/{thread.OriginalPost.PostNumber} ({imageDownloads.Count} media items enqueued)");
List <ulong> postNumbersToDelete = new List <ulong>(thread.Posts.Length);
foreach (var postNumber in threadHashes.Keys)
{
if (thread.Posts.All(x => x.PostNumber != postNumber))
{
// Post has been deleted
Program.Log($"[Asagi] Post /{board}/{postNumber} has been deleted");
await SetUntracked(postNumber, board, true);
postNumbersToDelete.Add(postNumber);
}
}
// workaround for not being able to remove from a collection while enumerating it
foreach (var postNumber in postNumbersToDelete)
{
threadHashes.Remove(postNumber, out _);
}
threadHashes.TrimExcess();
if (thread.OriginalPost.Archived == true)
{
// We don't need the hashes if the thread is archived, since it will never change
// If it does change, we can just grab a new set from the database
ThreadHashes.TryRemove(hashObject, out _);
}
return(imageDownloads);
}
