public static ScheduleItem Parse(CQ element)
{
var speaker = element.Find(".before_hour_text").Text();
string speakerUrl = !string.IsNullOrWhiteSpace(speaker) ? element.Find(".event_header").Attr("href") : null;
string title = element.Find(".event_header").Text();
string startTimeText = element.Find(".top_hour span").First().Text();
string endTimeText = element.Find(".bottom_hour span").First().Text();
var startTime = TimeSpan.ParseExact(startTimeText, "g", CultureInfo.InvariantCulture);
var endTime = TimeSpan.ParseExact(endTimeText, "g", CultureInfo.InvariantCulture);
return new ScheduleItem(title, speaker, speakerUrl, startTime, endTime);
}
private static IEnumerable<ScheduleItem> FromTable(CQ tableElement)
{
var itemElements = tableElement.Find("tbody tr .event .event_container");
for (int i = 0; i < itemElements.Length; i++)
{
var item = itemElements.Eq(i);
yield return ScheduleItem.Parse(item);
}
}
public static IHtmlString CustomXhtmlString(this HtmlHelper helper, XhtmlString value)
{
var htmlString = helper.XhtmlString(value).ToHtmlString();
var headline = LocalizationService.Current
.GetString("/views/common/linklist/headline");
var cq = new CQ(htmlString, HtmlParsingMode.Document);
cq.Find("ul.link-list").Each(list =>
{
var box = CQ.CreateFragment(
"<div class=\"link-list\">"
+ "<h4><span class=\"glyphicon glyphicon-link\"></span> "
+ headline
+ "</h4></div>");
list.RemoveClass("link-list");
box.Append(list.OuterHTML);
list.Cq().ReplaceWith(box);
});
return new MvcHtmlString(cq.Find("body").Html());
}
public static Dictionary<string, string> GetInputFields(CQ dom)
{
Dictionary<string, string> result = new Dictionary<string, string>();
foreach (var v in dom.Find("input"))
{
var value = v.Cq().Attr("value");
var key = v.Cq().Attr("name");
if (!string.IsNullOrWhiteSpace(value) && !string.IsNullOrWhiteSpace(key))
{
if (!result.ContainsKey(key))
{
result.Add(key, value);
}
else
{
result[key] = value;
}
}
}
// Get all selects
foreach (var s in dom.Select("select"))
{
var select = s.Cq();
var key = select.Attr("name");
foreach (var option in select.Children("option"))
{
var opt = option.Cq();
if (!string.IsNullOrWhiteSpace(opt.Attr("selected")))
{
if (!result.ContainsKey(key))
{
result.Add(key, opt.Val());
}
else
{
result[key] = opt.Val();
}
}
}
}
return result;
}
/// <summary>
/// Parses volume information based on its title element.
/// </summary>
/// <param name="hEl"></param>
public Volume(CQ ul)
{
ul.Find("li").Each((el) =>
{
var link = new CQ(el).Find("a").First();
if (link.Count() == 0) return;
var name = Regex.Replace(link.Attr("href"), "^.+title=", "");
if (name.Contains("Illustrations") && _illustrationsPage == null)
{
_illustrationsPage = new IllustrationsPage(name);
}
else
{
_chapters.Add(new Chapter(name, link.Text()));
}
});
}
private static ParsedRSSFeedLink ParseLink(IDomObject linkNode)
{
ParsedRSSFeedLink parsedRSSFeedLink;
CQ linkCq = new CQ(linkNode.Clone());
CQ aCq = linkCq.Find("a");
if (aCq.Any())
{
ParsedRSSFeedLink parsedRSSFeedLink1 = new ParsedRSSFeedLink
{
LinkURL = aCq.Attr<string>("href"),
LinkTitle = aCq.First().Text()
};
ParsedRSSFeedLink link = parsedRSSFeedLink1;
aCq.FirstElement().Remove();
link.Description = linkCq.Text().Trim();
parsedRSSFeedLink = link;
}
else
{
parsedRSSFeedLink = null;
}
return parsedRSSFeedLink;
}
private string PrepareHtml(string html)
{
// Make title
if (!Notitle)
{
html = "<h2>" + Title + "</h2>" + html;
}
html = "<span class=\"invisible chapterstart\">" + (Noheader ? "" : Title) + "</span>" + html;
// Make sure page break is set
if (Pagebreak)
{
html = "<span class=\"invisible pagebreak\"></span>" + html;
}
CQ dom = CQ.CreateFragment("<div class=\"content\">" + html + "</div>");
// Remove next/prev table
dom.Find("table:contains('Forward'):contains('Back')").Last().Remove();
// Find images
foreach(IDomElement aNode in dom.Find("a.image")) {
var a = new CQ(aNode);
var img = new CQ(a.Find("img"));
var src = img.Attr("src").Replace("/thumb", "");
src = Regex.Replace(src, @"[.](jpg|png|gif)\/.+$", @".$1", RegexOptions.IgnoreCase);
var image = new Image(src, new Uri(Wiki)) {Sashie = true};
CQ node = a.Closest(".thumb").Add(a).First();
if (_images.Count == 0 && EntryPicture)
{
// We can view it as a full-fledged image since we don't need to worry about text-flow
image.Sashie = false;
dom.Before(image.Html);
}
else
{
node.Before(image.Html);
//node.After("<span class=\"image-stopper\"></span>");
}
node.Remove();
_images.Add(image);
}
// Catch references
foreach (IDomElement supNode in dom.Find("sup.reference"))
{
var sup = new CQ(supNode);
CQ footnote = "<span class=\"fn\"></span>";
CQ oldFootnote = dom.Find("#" + sup.Attr("id").Replace("_ref-", "_note-"));
footnote.Html(oldFootnote.Find(".reference-text").Html());
oldFootnote.Remove();
sup.Before(footnote).Remove();
}
// Remove possible reference title
dom.Find(".references").Prev(":header").Remove();
// Remove edit links
dom.Find(".editsection, #toc").Remove();
// Make smart quotes
dom.Find("p:contains('\"'), p:contains(\"'\"), li:contains('\"'), li:contains(\"'\")").Each((el) =>
{
CQ p = new CQ(el);
string pHtml = p.Html();
// Replace quotes
if (Regex.Matches(pHtml, """).Count % 2 == 0)
{
pHtml = Regex.Replace(pHtml, ""(.+?)"", "“$1”");
}
else
{
Console.WriteLine("NOTICE: possible quotes problem ({0})", pHtml.Trim());
}
// Replace single quotes (\b doesn't work)
pHtml = Regex.Replace(pHtml, "(?<!\\w)'(.+?)'(?!\\w)", "‘$1’");
// Replace apostrophes
pHtml = Regex.Replace(pHtml, "'", "’");
p.Html(pHtml);
});
// Parse Ruby
dom.Find("span > span > span").Each(el =>
{
var rubySpan = new CQ(el);
if(rubySpan.Css("position") == "relative" && rubySpan.Css("left") == "-50%") {
var textSpan = rubySpan.Parent().Siblings("span");
var containerSpan = textSpan.Parent();
if (textSpan.Length == 1 && containerSpan.Css("white-space") == "nowrap")
{
// Okay, this is ruby.
var ruby = new CQ("<ruby>");
ruby.Html(textSpan.Html());
ruby.Append(new CQ("<rp>(</rp>"));
ruby.Append(new CQ("<rt>").Html(rubySpan.Html()));
ruby.Append(new CQ("<rp>)</rp>"));
containerSpan.ReplaceWith(
ruby
);
}
}
});
// Hakomari specific
foreach (IDomElement star in dom.Find("p:contains(✵)"))
{
star.InnerHTML = "<img src=\"" + (new Uri( Helper.GetAssetsPath() + "blackstar.jpg" )) + "\">";
}
return dom.Render();
}
private bool DetermineIfMessageIsCloseVoteRequest(CQ messageContentsNode)
{
//message contains "cv-pls", "cv-plz", or "close", and has a link
var triggerPhrases = new[]
{
"cv-pls",
"cv-plz",
"close"
};
var text = messageContentsNode.Text();
var hasTriggerPhrase = triggerPhrases
.Select(x => text.Contains(x))
.Any(x => x);
var hasLink = messageContentsNode.Find("a").Any();
return hasTriggerPhrase && hasLink;
}
private static IEnumerable<ScheduleTableDay> GetScheduleTableDays(CQ pageQuery)
{
var titleElement = pageQuery.Find("h1");
var tableElement = pageQuery.Find("#all-events table.tt_timetable");
yield return new ScheduleTableDay { TitleElement = titleElement, TableElement = tableElement };
}
private static IEnumerable<string> GetHeaders(CQ tableElement)
{
var headers = tableElement.Find("thead tr th").Skip(1).ToList();
var headerTexts = headers.Select(x => (x.InnerText ?? string.Empty).Trim());
return headerTexts;
}
static string ParseBlockTitle(CQ head)
{
return head.Find("div.block-events-head").Children().Remove().End().Text().Trim();
}
