private static void SetTitleBody(TelegramPost post, string txt)
{
if (post == null)
{
return;
}
int brIndex = txt.IndexOf("<br", StringComparison.Ordinal);
if (brIndex > -1)
{
post.PossibleTitle = txt.Substring(0, brIndex).RemoveHtmlTags();
if (post.PostType == TelegramPostType.Text)
{
int afterBrIndex = brIndex + 4; // <br> length = 4
post.Body = txt.Substring(afterBrIndex, Math.Max(txt.Length - afterBrIndex, 0));
}
else
{
post.Body = txt;
}
}
else
{
var possibleTitle = txt.RemoveHtmlTags();
if (possibleTitle.Length > MAX_TITLE_LENGTH)
{
post.PossibleTitle = possibleTitle.Substring(0, MAX_TITLE_LENGTH);
post.Body = txt;
}
else
{
post.PossibleTitle = possibleTitle;
post.Body = post.PostType == TelegramPostType.Text ? txt : "";
}
}
}
private void SetAttachmentUri(TelegramPost post, HtmlNode photoNode, HtmlNode videoNode, HtmlNode stickerNode)
{
string url = null;
switch (post.PostType)
{
case TelegramPostType.Photo:
url = ParsePhotoUrl(photoNode);
break;
case TelegramPostType.Video:
url = videoNode?.Attributes["src"]?.Value;
break;
case TelegramPostType.Sticker:
url = ParseStickerUrl(stickerNode);
break;
}
if (!string.IsNullOrWhiteSpace(url) && Uri.TryCreate(url, UriKind.Absolute, out var uri))
{
post.AttachmentUri = uri;
}
}
protected TelegramPost Parse(HtmlDocument htmlDocument)
{
if (htmlDocument == null)
{
return(null);
}
var divText = htmlDocument.DocumentNode.Descendants("div")
.LastOrDefault(m => m.HasClass("tgme_widget_message_text"));
var video = htmlDocument.DocumentNode.Descendants("video")?
.LastOrDefault();
var aPhoto = htmlDocument.DocumentNode.Descendants("a")
.FirstOrDefault(m => m.HasClass("tgme_widget_message_photo_wrap"))
?? htmlDocument.DocumentNode.Descendants("i")
.FirstOrDefault(m => m.HasClass("link_preview_image"));
var iSticker = htmlDocument.DocumentNode.Descendants("i")
.FirstOrDefault(m => m.HasClass("tgme_widget_message_sticker"));
var aDocument = htmlDocument.DocumentNode.Descendants("a")
.FirstOrDefault(m => m.HasClass("tgme_widget_message_document_wrap"));
if (aPhoto == null && divText == null && video == null && iSticker == null && aDocument == null)
{
// log ("Post {0} has no photo an no message Text.", postId);
return(null);
}
// log "Parsing post {0}. ", postId;
var spans = htmlDocument.DocumentNode.Descendants("span")?.ToList();
var athuorNode = spans?.LastOrDefault(m => m.HasClass("tgme_widget_message_from_author"));
var dateNode = spans?
.SingleOrDefault(s => s.HasClass("tgme_widget_message_meta"))?.Descendants("a")?
.SingleOrDefault(s => s.HasClass("tgme_widget_message_date"))?.Descendants("time")?
.FirstOrDefault();
var viewNode = spans?.LastOrDefault(m => m.HasClass("tgme_widget_message_views"));
var post = new TelegramPost
{
Id = ParsePostId(htmlDocument),
PostType = video != null ? TelegramPostType.Video :
aPhoto != null ? TelegramPostType.Photo :
iSticker != null ? TelegramPostType.Sticker :
aDocument != null ? TelegramPostType.File : TelegramPostType.Text,
WebRaw = htmlDocument.ParsedText,
TextRaw = divText?.InnerText,
Author = athuorNode?.InnerText ?? string.Empty,
DateString = dateNode?.InnerText,
ViewCount = viewNode?.InnerText
};
SetAttachmentUri(post, aPhoto, video, iSticker);
SetTitleBody(post, divText?.InnerHtml);
if (DateTime.TryParse(dateNode?.Attributes["datetime"]?.Value, out var date))
{
post.Date = date;
}
return(post);
}
