static TitlingResult HandleHtml(
TitlingRequest request,
HtmlPage page,
Func <TitlingRequest, string, TitlingResult> handler)
{
const int maxTitleLength = 1024;
ReportCharsets(request, page);
string htmlTitle = WebTools.GetTitle(page.Content);
if (string.IsNullOrWhiteSpace(htmlTitle))
{
request.AddMessage("No <title> found, or title element was empty/whitespace.");
return(request.CreateResult(false));
}
if (htmlTitle.Length > maxTitleLength)
{
request.AddMessage("HTML title length was in excess of 1024 characters, assuming spam.");
return(request.CreateResult(false));
}
// If defined and not of ridiculous length make it available to TitleBuilder.
request.IrcTitle.HtmlTitle = htmlTitle;
return(handler(request, page.Content));
}
public TitlingResult WikipediaSummarize(TitlingRequest req, string htmlDoc)
{
WikipediaArticle article = WikipediaTools.Parse(htmlDoc);
// The paragraph to be summarized.
string p = null;
// Check if the URL has an anchor to a specific section in the article.
int anchorIndex = req.Url.IndexOf("#", StringComparison.OrdinalIgnoreCase);
if (anchorIndex >= 0 && (anchorIndex + 1) < req.Url.Length)
{
var anchorId = req.Url.Substring(anchorIndex + 1);
p = article.GetFirstParagraph(anchorId);
}
// If no anchor or if we couldn't extract a paragraph for the specific anchor,
// get first paragraph of the article.
if (p == null && article.SummaryParagraphs.Length > 0)
{
p = article.SummaryParagraphs[0];
}
if (!string.IsNullOrWhiteSpace(p))
{
string summary = Format.Shorten(p, MaxCharacters, ContinuationSymbol);
req.IrcTitle.SetFormat("[ {0} ]", summary);
}
return(req.CreateResult(true));
}
// To conform to the signature we accept the HTML document. We don't need it.
TitlingResult GenericHandler(TitlingRequest req, string htmlDoc)
{
// Because the similarity can only be 1 max, allow all titles to be printed if Threshold is set to 1 or
// higher. The similarity would always be equal to or less than 1.
if (Threshold >= 1)
{
return(req.CreateResult(true));
}
// If Threshold is set to 0 that would still mean that titles that had 0 similarity with their URLs would
// get printed. Set to a negative value to never print any title.
if (Threshold < 0)
{
return(req.CreateResult(false));
}
double urlTitleSimilarity = UrlTitle.Similarity(req.Url, req.IrcTitle.HtmlTitle);
req.AddMessage(
string.Format("URL-Title Similarity: {0} [Threshold: {1}]", urlTitleSimilarity, Threshold)
);
if (urlTitleSimilarity <= Threshold)
{
return(req.CreateResult(true));
}
return(req.CreateResult(false));
}
static void ReportCharsets(TitlingRequest req, HtmlPage page)
{
var encInfo = string.Format("(HTTP) \"{0}\" -> {1} ; (HTML) \"{2}\" -> {3}",
page.HeadersCharset, page.EncHeaders,
page.HtmlCharset, page.EncHtml);
req.AddMessage(encInfo);
}
public static TitlingResult YoutubeWithDuration(TitlingRequest req, string htmlDoc)
{
// If duration can be found, change the html info to include that.
var ytTime = WebTools.GetYoutubeTime(htmlDoc);
req.IrcTitle.SetHtmlTitle().AppendTime(ytTime);
return(req.CreateResult(true));
}
public TitlingResult HandleRequest(TitlingRequest req)
{
if (Supports(req))
{
return(ThreadTopicToIrc(req));
}
return(null);
}
public TitlingResult HandleRequest(TitlingRequest req)
{
if (req.Url.Contains("gelbooru.com/index.php?page=post&s=view&id=",
StringComparison.OrdinalIgnoreCase))
{
return(PostToIrc(req));
}
return(null);
}
public TitlingResult HandleRequest(TitlingRequest req)
{
if (req.Url.Contains("donmai.us/posts/",
StringComparison.OrdinalIgnoreCase))
{
return(PostToIrc(req));
}
return(null);
}
public static bool Supports(TitlingRequest req)
{
Source?src = GetSource(req.Url);
if (src == null)
{
return(false);
}
else
{
return(true);
}
}
public TitlingResult WebInfo(TitlingRequest request)
{
if (request == null)
{
throw new ArgumentNullException(nameof(request));
}
// TitlingRequest ensures that what we get passed is an absolute URI with a scheme we support. Most
// importantly this relieves the individual handlers of checking for those conditions.
foreach (var handler in preHtmlHandlers)
{
var result = handler(request);
if (result != null)
{
return(result);
}
}
foreach (var instruction in urlInstructions)
{
if (instruction.Match(request.Uri))
{
// Set options as per instructions.
urlFetcher.MaxSizeHtml = instruction.FetchSize;
urlFetcher.FollowMetaRefreshes = instruction.FollowMetaRefreshes;
var result = urlFetcher.Load(request.Uri);
request.Resource = result.Page;
// HTML handling.
if (result.IsHtml)
{
return(HandleHtml(
request, result.Page,
instruction.Handler ?? GenericHandler
));
}
// Media/Binary handling.
if (ParseMedia && result.Bytes.Success)
{
return(BinaryHandler.BinaryToIrc(request, result.Bytes));
}
return(request.CreateResult(false));
}
}
throw new InvalidOperationException(
"Reached end of method, this should not happen. There should be a catch-all in urlInstructions.");
}
public TitlingResult PostToIrc(TitlingRequest req)
{
BooruPost postInfo = GelboTools.GetPostInfo(req.Url);
req.Resource = postInfo;
if (postInfo.Success)
{
string warning = ConstructWarning(postInfo.Tags);
if (!string.IsNullOrEmpty(warning))
{
FormatMessage(req.IrcTitle, postInfo.Rated, warning, postInfo.PostNo);
return(req.CreateResult(true));
}
}
return(req.CreateResult(false));
}
public static TitlingResult BinaryToIrc(TitlingRequest req, WebBytes wb)
{
req.Resource = wb;
if (!wb.Success)
{
return(req.CreateResult(false));
}
var media = MediaDispatch.Parse(wb.Data);
string type;
switch (media.Type)
{
case MediaType.Jpeg:
type = "JPEG";
break;
case MediaType.Png:
type = "PNG";
break;
case MediaType.Gif:
type = "GIF";
break;
case MediaType.Matroska:
type = "Matroska";
break;
case MediaType.Webm:
type = "WebM";
break;
default:
type = wb.ContentType;
req.AddMessage("Binary format not supported.");
break;
}
FormatBinaryInfo(req.IrcTitle, type, media, wb.ContentLength);
return(req.CreateResult(true));
}
public TitlingResult WebInfo(Uri uri)
{
if (uri == null)
{
throw new ArgumentNullException(nameof(uri));
}
if (!uri.IsAbsoluteUri)
{
throw new ArgumentException("Uri must be absolute: " + uri, nameof(uri));
}
if (TitlingRequest.IsSchemeSupported(uri))
{
return(WebInfo(new TitlingRequest(uri)));
}
else
{
var ex = new NotSupportedException("Unsupported scheme: " + uri.Scheme);
return(TitlingResult.Failure(uri.OriginalString, ex));
}
}
public TitlingResult PostToIrc(TitlingRequest req)
{
DanboPost postInfo = DanboTools.GetPostInfo(req.Url);
req.Resource = postInfo;
if (postInfo.Success)
{
string warning = ConstructWarning(postInfo.GeneralTags, postInfo.MetaTags);
// If image has no character, copyright or artist tags, return just the post ID, rating and
// possible warning.
if (postInfo.CopyrightTags.Length == 0 &&
postInfo.CharacterTags.Length == 0 &&
postInfo.ArtistTags.Length == 0)
{
FormatMessage(req.IrcTitle, postInfo.Rated, warning, postInfo.PostNo);
return(req.CreateResult(true));
}
DanboTools.CleanupCharacterTags(postInfo.CharacterTags, postInfo.CopyrightTags);
// Convert to string and limit the number of tags as specified in `MaxTagCount`.
// Also colourize the tags if set to true.
var characters = TagArrayToString(postInfo.CharacterTags, CharacterCode);
var copyrights = TagArrayToString(postInfo.CopyrightTags, CopyrightCode);
var artists = TagArrayToString(postInfo.ArtistTags, ArtistCode);
FormatMessage(req.IrcTitle, postInfo.Rated, warning);
FormatDanboInfo(req.IrcTitle, characters, copyrights, artists);
return(req.CreateResult(true));
}
else
{
return(req.CreateResult(false));
}
}
public TitlingResult ThreadTopicToIrc(TitlingRequest req)
{
ChanPost post;
switch (GetSource(req.Url))
{
case Source.FourChan:
post = FourChan.GetPost(req.Url);
break;
case Source.ArchiveMoe:
post = ArchiveMoe.GetPost(req.Url);
break;
default:
// Fail loudly for now, this might change in the future.
throw new NotSupportedException("Passed TitlingRequest not supported.");
}
req.Resource = post;
if (post.Success)
{
string topic = ConstructTopic(post, req.Url);
if (topic == null)
{
req.AddMessage("Post contained neither subject or comment.");
return(req.CreateResult(false));
}
req.IrcTitle.SetFormat("[ /{0}/ - {1} ] [ {2} ]", post.Board, post.BoardName, topic);
return(req.CreateResult(true));
}
else
{
return(req.CreateResult(false));
}
}
