/// <summary>
/// Extracts categories from the article text; removes duplicate categories, cleans whitespace and underscores
/// </summary>
/// <param name="articleText"></param>
/// <param name="articleTitle"></param>
/// <returns></returns>
public string RemoveCats(ref string articleText, string articleTitle)
{
List <string> categoryList = new List <string>();
// http://en.wikipedia.org/wiki/Wikipedia_talk:AutoWikiBrowser/Bugs#Comments_get_removed_from_between_categories
// allow comments between categories, and keep them in the same place, but don't grab any comment just after the last category
Regex r = new Regex("<!-- ? ?\\[\\[" + Variables.NamespacesCaseInsensitive[Namespace.Category]
+ ".*?(\\]\\]|\\|.*?\\]\\]).*?-->|\\[\\["
+ Variables.NamespacesCaseInsensitive[Namespace.Category]
+ ".*?(\\]\\]|\\|.*?\\]\\])( {0,4}⌊⌊⌊⌊[0-9]{1,4}⌋⌋⌋⌋|\\s*<!--.*?-->(?=\r\n\\[\\[))?", RegexOptions.Singleline);
MatchCollection matches = r.Matches(articleText);
foreach (Match m in matches)
{
if (!Regex.IsMatch(m.Value, "\\[\\[Category:(Pages|Categories|Articles) for deletion\\]\\]"))
{
categoryList.Add(m.Value);
}
}
articleText = Tools.RemoveMatches(articleText, matches);
if (AddCatKey)
{
categoryList = CatKeyer(categoryList, articleTitle);
}
if (CatCommentRegex.IsMatch(articleText))
{
string catComment = CatCommentRegex.Match(articleText).Value;
articleText = articleText.Replace(catComment, "");
categoryList.Insert(0, catComment);
}
MatchCollection mc = WikiRegexes.Defaultsort.Matches(articleText);
if (mc.Count > 1)
{
throw new ArgumentException("Page contains multiple {{DEFAULTSORTS}} tags. Metadata sorting cancelled");
}
string defaultSort = "";
if (mc.Count > 0)
{
defaultSort = mc[0].Value;
}
if (!string.IsNullOrEmpty(defaultSort))
{
articleText = articleText.Replace(defaultSort, "");
}
if (!string.IsNullOrEmpty(defaultSort) && defaultSort.ToUpper().Contains("DEFAULTSORT"))
{
defaultSort = TalkPageHeaders.FormatDefaultSort(defaultSort);
}
if (!string.IsNullOrEmpty(defaultSort))
{
defaultSort += "\r\n";
}
// on en-wiki find any {{Lifetime}} template and move directly after categories
string lifetime = "";
if (Variables.LangCode == LangCodeEnum.en)
{
lifetime = WikiRegexes.Lifetime.Match(articleText).Value;
if (!string.IsNullOrEmpty(lifetime))
{
articleText = articleText.Replace(lifetime, "");
// http://en.wikipedia.org/wiki/Wikipedia_talk:AutoWikiBrowser/Bugs#Blank_lines_after_Lifetime
lifetime += "\r\n";
}
}
return(defaultSort + ListToString(categoryList) + lifetime);
}
/// <summary>
/// Extracts categories from the article text; removes duplicate categories, cleans whitespace and underscores
/// </summary>
/// <param name="articleText">The wiki text of the article.</param>
/// <param name="articleTitle">Title of the article</param>
/// <returns>The cleaned page categories in a single string</returns>
public string RemoveCats(ref string articleText, string articleTitle)
{
List <string> categoryList = new List <string>();
string originalArticleText = articleText;
// allow comments between categories, and keep them in the same place, but don't grab any comment just after the last category
Regex r = new Regex(@"<!-- [^<>]*?\[\[\s*" + Variables.NamespacesCaseInsensitive[Namespace.Category]
+ @".*?(\]\]|\|.*?\]\]).*?-->|\[\["
+ Variables.NamespacesCaseInsensitive[Namespace.Category]
+ @".*?(\]\]|\|.*?\]\])(\s*⌊⌊⌊⌊\d{1,4}⌋⌋⌋⌋|\s*<!--.*?-->(?=\r\n\[\[\s*" + Variables.NamespacesCaseInsensitive[Namespace.Category]
+ @"))?", RegexOptions.Singleline);
MatchCollection matches = r.Matches(articleText);
foreach (Match m in matches)
{
if (!Regex.IsMatch(m.Value, @"\[\[Category:(Pages|Categories|Articles) for deletion\]\]"))
{
categoryList.Add(m.Value);
}
}
articleText = Tools.RemoveMatches(articleText, matches);
// if category tidying has changed comments/nowikis return with no changes – we've pulled a cat from a comment
if (!UnformattedTextNotChanged(originalArticleText, articleText))
{
articleText = originalArticleText;
return("");
}
if (AddCatKey)
{
categoryList = CatKeyer(categoryList, articleTitle);
}
if (CatCommentRegex.IsMatch(articleText))
{
string catComment = CatCommentRegex.Match(articleText).Value;
articleText = articleText.Replace(catComment, "");
categoryList.Insert(0, catComment);
}
MatchCollection mc = WikiRegexes.Defaultsort.Matches(articleText);
if (mc.Count > 1)
{
throw new ArgumentException("Page contains multiple {{DEFAULTSORTS}} tags. Metadata sorting cancelled");
}
string defaultSort = "";
// ignore commented out DEFAULTSORT – http://en.wikipedia.org/wiki/Wikipedia_talk:AutoWikiBrowser/Bugs#Moving_DEFAULTSORT_in_HTML_comments
if (mc.Count > 0 && WikiRegexes.Defaultsort.Matches(WikiRegexes.Comments.Replace(articleText, "")).Count == mc.Count)
{
defaultSort = mc[0].Value;
}
if (!string.IsNullOrEmpty(defaultSort))
{
articleText = articleText.Replace(defaultSort, "");
}
if (!string.IsNullOrEmpty(defaultSort) && defaultSort.ToUpper().Contains("DEFAULTSORT"))
{
defaultSort = TalkPageHeaders.FormatDefaultSort(defaultSort);
}
if (!string.IsNullOrEmpty(defaultSort))
{
defaultSort += "\r\n";
}
return(defaultSort + ListToString(categoryList));
}
public string removeCats(ref string ArticleText, string ArticleTitle)
{
List <string> categoryList = new List <string>();
string x = "";
Regex r = new Regex("<!-- ? ?\\[\\[" + Variables.NamespacesCaseInsensitive[14] + ".*?(\\]\\]|\\|.*?\\]\\]).*?-->|\\[\\[" + Variables.NamespacesCaseInsensitive[14] + ".*?(\\]\\]|\\|.*?\\]\\])( {0,4}⌊⌊⌊⌊[0-9]{1,4}⌋⌋⌋⌋)?");
MatchCollection matches = r.Matches(ArticleText);
foreach (Match m in matches)
{
x = m.Value;
//add to array, replace underscores with spaces, ignore=
if (!Regex.IsMatch(x, "\\[\\[Category:(Pages|Categories|Articles) for deletion\\]\\]"))
{
categoryList.Add(x.Replace("_", " "));
}
}
ArticleText = Tools.RemoveMatches(ArticleText, matches);
if (parser.addCatKey)
{
categoryList = catKeyer(categoryList, ArticleTitle);
}
if (CatCommentRegex.IsMatch(ArticleText))
{
string catComment = CatCommentRegex.Match(ArticleText).Value;
ArticleText = ArticleText.Replace(catComment, "");
categoryList.Insert(0, catComment);
}
MatchCollection mc = WikiRegexes.Defaultsort.Matches(ArticleText);
if (mc.Count > 1)
{
throw new ArgumentException("Page contains multiple {{DEFAULTSORTS}} tags. Metadata sorting cancelled");
}
string defaultSort = "";
if (mc.Count > 0)
{
defaultSort = mc[0].Value;
}
if (!string.IsNullOrEmpty(defaultSort))
{
ArticleText = ArticleText.Replace(defaultSort, "");
}
if (defaultSort.ToUpper().Contains("DEFAULTSORT"))
{
defaultSort = TalkPageHeaders.FormatDefaultSort(defaultSort);
}
if (!string.IsNullOrEmpty(defaultSort))
{
defaultSort += "\r\n";
}
return(defaultSort + ListToString(categoryList));
}
