Get parsing working, and HTML generation from resulting objects

This commit is contained in:
Saphire committed 2022-02-21 02:27:27 +07:00
1 parent a7250ceacf
commit 38f317afd3
9 files changed
+171 -146

No files matched your search

+30 -30
View File
@@ -9,12 +9,16 @@ public class PostsSource
{
public List<ThreadPost> Posts { get; set; }
public List<ThreadPost> Accepted { get; set; }
public HashSet<ThreadPost> Accepted { get; set; }
public Metadata Metadata { get; set; }
public PostsSource(string questName, string basePath)
public string BasePath { get; set; }
public PostsSource(string questName)
{
BasePath = $"quests/{questName}";
var options = new JsonSerializerOptions
{
PropertyNamingPolicy = JsonNamingPolicy.CamelCase,
@@ -22,45 +26,41 @@ public class PostsSource
WriteIndented = true
};
using var fileStream = File.OpenRead(Path.Combine(basePath, "metadata.json"));
using var fileStream = File.OpenRead(Path.Combine(BasePath, "metadata.json"));
Metadata = JsonSerializer.Deserialize<Metadata>(fileStream, options)
?? throw new InvalidDataException("Empty deserialisation result for quest metadata");
fileStream.Dispose();
Console.Out.WriteLine($"Loaded metadata: {Metadata}");
Posts = Metadata.Threads.SelectMany(tId =>
{
using var fileStream = File.OpenRead(Path.Combine(basePath, $"thread_{tId}.json"));
var threadData = JsonSerializer.Deserialize<List<ThreadPost>>(fileStream, options)
?? throw new InvalidDataException("Empty deserialisation result for thread data");
fileStream.Dispose();
Posts = Metadata.Threads
.SelectMany(tId => QuestdenParse.GetThread(tId, BasePath).Result)
.ToList();
return threadData;
}).ToList();
using var postsListStream = File.OpenRead(Path.Combine(basePath, "accepted.json"));
using var postsListStream = File.OpenRead(Path.Combine(BasePath, "accepted.json"));
var ids = JsonSerializer.Deserialize<List<int>>(postsListStream, options)
?? throw new InvalidDataException("Empty deserialisation result for quest metadata");
Accepted = Posts.Where(p => ids.Contains(p.Id)).ToList();
Accepted = Posts.Where(p => ids.Contains(p.Id)).ToHashSet();
foreach (var chapter in Metadata.Chapters)
{
var post = Accepted.Single(p => p.Id == (chapter.Announce ?? chapter.Start));
post.IsChapterAnnounce = true;
post.Chapter = chapter;
}
Console.Out.WriteLine($"Loaded a list of {Accepted.Count} posts, referencing {Accepted.Where(a => a.File is not null).Count()} files");
var rx = new Regex(@"data-post-ref=""(\d+)""",
RegexOptions.Compiled | RegexOptions.IgnoreCase);
foreach (var post in Posts)
foreach (var post in Accepted)
{
var matches = rx.Matches(post.RawHtml);
if (!matches.Any())
continue;
post.RepliesTo = new List<int>();
foreach (Match match in matches)
{
var replyId = int.Parse(match.Groups[1].Value);
var found = Posts.FirstOrDefault(p => p.Id == replyId);
if (found is null)
continue;
post.RepliesTo.Add(replyId);
}
post.AuthorPost = true;
if (post.ParsedContent is null || post.ParsedContent.Version < QuestdenParse.LatestCompatibleVersion)
throw new NotImplementedException("Repairing missing post content or updating it is not implemented yet");
}
var referenced = Accepted.SelectMany(p => p.ParsedContent!.GetReferences());
Accepted.UnionWith(Posts.Where(p => referenced.Contains(p.Id)));
Accepted = Accepted.OrderBy(p => p.Id).ToHashSet();
Console.Out.WriteLine($"Done loading with {Accepted.Count} posts, referencing {Accepted.Where(a => a.File is not null).Count()} files");
}
}