| public RemoveCats ( string &articleText, string articleTitle ) : string | ||
| articleText | string | The wiki text of the article. |
| articleTitle | string | Title of the article |
| return | string | |
public string RemoveCats(ref string articleText, string articleTitle)
{
// don't pull category from redirects to a category e.g. page Hello is #REDIRECT[[Category:Hello]]
string rt = Tools.RedirectTarget(articleText);
if (rt.Length > 0 && WikiRegexes.Category.IsMatch(@"[[" + rt + @"]]"))
return "";
List<string> categoryList = new List<string>();
string originalArticleText = articleText;
string articleTextNoComments = Tools.ReplaceWithSpaces(articleText, WikiRegexes.Comments.Matches(articleText));
// don't operate on pages with (incorrectly) multiple defaultsorts
// ignore commented out DEFAULTSORT – https://en.wikipedia.org/wiki/Wikipedia_talk:AutoWikiBrowser/Bugs/Archive_12#Moving_DEFAULTSORT_in_HTML_comments
MatchCollection mc = WikiRegexes.Defaultsort.Matches(articleTextNoComments);
if (mc.Count > 1)
{
Tools.WriteDebug("RemoveCats", "Page " + articleTitle + " has multiple DEFAULTSORTs");
return "";
}
string defaultSort = "";
bool defaultSortRemoved = false;
// allow comments between categories, and keep them in the same place, only grab any comment after the last category if on same line
// whitespace: remove all whitespace after, but leave a blank newline before a heading (rare case where category not in last section)
// performance: apply regex on portion of article containing category links rather than whole text
Match cq = WikiRegexes.CategoryQuick.Match(articleTextNoComments);
if (cq.Success)
{
int cutoff = Math.Max(0, cq.Index - 500);
string cut = articleText.Substring(cutoff);
cut = WikiRegexes.RemoveCatsAllCats.Replace(cut, m => {
if (!CatsForDeletion.IsMatch(m.Value))
categoryList.Add(m.Value.Trim());
// if category not at start of line, leave newline, otherwise text on next line moved up
if (m.Index > 2 && !cut.Substring(m.Index - 2, 2).Trim().Equals(""))
return "\r\n";
return "";
});
// if category tidying has changed comments/nowikis return with no changes – we've pulled a cat from a comment
if (!Tools.UnformattedTextNotChanged(originalArticleText.Substring(cutoff), cut))
{
articleText = originalArticleText;
return "";
}
if (AddCatKey)
categoryList = CatKeyer(categoryList, articleTitle);
// now refresh defaultsort to pick up any comment on same line after it
if (mc.Count > 0)
mc = Regex.Matches(articleText, WikiRegexes.Defaultsort.ToString() + @"(?: *<!--[^<>]*-->)?");
// remove defaultsort now if we can, faster to remove from cut than whole articleText
if (mc.Count > 0 && cut.Contains(mc[0].Value))
{
cut = cut.Replace(mc[0].Value, "");
defaultSortRemoved = true;
}
articleText = articleText.Substring(0, cutoff) + cut;
if (CatCommentRegex.IsMatch(cut))
articleText = CatCommentRegex.Replace(articleText, m =>
{
categoryList.Insert(0, m.Value);
return "";
}, 1);
}
if (Variables.LangCode.Equals("sl") && LifeTime.IsMatch(articleText))
{
defaultSort = LifeTime.Match(articleText).Value;
}
else if (mc.Count > 0)
defaultSort = mc[0].Value;
if (!string.IsNullOrEmpty(defaultSort))
{
// if defaultsort wasn't in the cut area before the categories, remove now
if (!defaultSortRemoved)
articleText = articleText.Replace(defaultSort, "");
if (defaultSort.ToUpper().Contains("DEFAULTSORT"))
defaultSort = TalkPageFixes.FormatDefaultSort(defaultSort);
defaultSort += "\r\n";
}
// Extract any {{uncategorized}} template, but not uncat stub templates
// remove exact duplicates
string uncat = "";
if (TemplateExists(Parsers.GetAllTemplates(originalArticleText), WikiRegexes.Uncat) && WikiRegexes.Uncat.IsMatch(articleTextNoComments))
{
articleText = WikiRegexes.Uncat.Replace(articleText, uncatm =>
{
if (WikiRegexes.PossiblyCommentedStub.IsMatch(uncatm.Value))
return uncatm.Value;
// remove exact duplicates
if (!uncat.Contains(uncatm.Value))
uncat += uncatm.Value + "\r\n";
return "";
});
}
return uncat + defaultSort + ListToString(categoryList);
}
/// <summary>
/// determines whether the article is about a person by looking for persondata/birth death categories, bio stub etc. for en wiki only
/// Should only return true if the article is the principle article about the individual (not early life/career/discography etc.)
/// </summary>
/// <param name="articleText">The wiki text of the article.</param>
/// <param name="articleTitle">Title of the article</param>
/// <param name="parseTalkPage"></param>
/// <returns></returns>
public static bool IsArticleAboutAPerson(string articleText, string articleTitle, bool parseTalkPage)
{
#if DEBUG || UNITTEST
if (Globals.UnitTestMode)
parseTalkPage = false;
#endif
// fix for duplicate living people categories being miscounted as article about multiple people
MetaDataSorter MDS = new MetaDataSorter();
string cats = MDS.RemoveCats(ref articleText, articleTitle);
articleText += cats;
if (!Variables.LangCode.Equals("en")
|| Namespace.Determine(articleTitle).Equals(Namespace.Category)
|| articleTitle.StartsWith(@"List of ")
|| articleTitle.StartsWith(@"Murder of ")
|| articleTitle.StartsWith(@"Deaths ")
|| articleTitle.EndsWith("discography") || articleTitle.EndsWith(" murders")
|| articleText.Contains(@"[[Category:Multiple people]]")
|| articleText.Contains(@"[[Category:Married couples")
|| articleText.Contains(@"[[Category:Fictional")
|| Regex.IsMatch(articleText, @"\[\[Category:\d{4} animal")
|| articleText.Contains(@"[[fictional character")
|| InUniverse.IsMatch(articleText)
|| articleText.Contains(@"[[Category:Presidencies")
|| articleText.Contains(@"[[Category:Military careers")
|| Regex.IsMatch(articleText, @"[[Category:[^\[\]\|]*[nN]oble families")
|| CategoryCharacters.IsMatch(articleText)
|| Tools.NestedTemplateRegex("Infobox cricketer tour biography").IsMatch(articleText)
|| WikiRegexes.Disambigs.IsMatch(articleText)
|| WikiRegexes.DeathsOrLivingCategory.Matches(articleText).Count > 1
|| WikiRegexes.InfoBox.Match(articleText).Groups[1].Value.ToLower().Contains("organization")
)
return false;
string MABackground =
Tools.GetTemplateParameterValue(IMA.Match(articleText).Value,
"Background", true);
if(MABackground.Contains("band") || MABackground.Contains("classical_ensemble") || MABackground.Contains("temporary"))
return false;
string CLSA = Tools.NestedTemplateRegex(@"Infobox Chinese-language singer and actor").Match(articleText).Value;
if(CLSA.Length > 0)
{
if (Tools.GetTemplateParameterValue(CLSA, "currentmembers").Length > 0
|| Tools.GetTemplateParameterValue(CLSA, "pastmembers").Length > 0)
return false;
}
string zerothSection = WikiRegexes.ZerothSection.Match(articleText).Value;
// not about a person if it's not the principle article on the subject
if (SeeAlsoOrMain.IsMatch(zerothSection))
return false;
// not about one person if multiple different birth or death date templates
List<string> BD = new List<string>();
foreach(Match m in BirthDate.Matches(articleText))
{
if(BD.Count > 0 && !BD.Contains(m.Value))
return false;
BD.Add(m.Value);
}
List<string> DD = new List<string>();
foreach(Match m in DeathDate.Matches(articleText))
{
if(DD.Count > 0 && !DD.Contains(m.Value))
return false;
DD.Add(m.Value);
}
if(WikiRegexes.PeopleInfoboxTemplates.Matches(articleText).Count > 1)
return false;
if (WikiRegexes.Persondata.Matches(articleText).Count == 1
|| articleText.Contains(@"-bio-stub}}")
|| articleText.Contains(@"[[Category:Living people")
|| WikiRegexes.PeopleInfoboxTemplates.Matches(zerothSection).Count == 1)
return true;
// articles with bold linking to another article may be linking to the main article on the person the article is about
// e.g. '''military career of [[Napoleon Bonaparte]]'''
string zerothSectionNoTemplates = WikiRegexes.Template.Replace(zerothSection, "");
foreach(Match m in WikiRegexes.Bold.Matches(zerothSectionNoTemplates))
{
if(WikiRegexes.WikiLink.IsMatch(m.Value))
return false;
}
int dateBirthAndAgeCount =BirthDate.Matches(zerothSection).Count;
int dateDeathCount = DeathDate.Matches(zerothSection).Count;
if (dateBirthAndAgeCount == 1 || dateDeathCount == 1)
return true;
return WikiRegexes.DeathsOrLivingCategory.IsMatch(articleText)
|| WikiRegexes.LivingPeopleRegex2.IsMatch(articleText)
|| WikiRegexes.BirthsCategory.IsMatch(articleText)
|| WikiRegexes.BLPSources.IsMatch(BLPUnsourcedSection.Replace(articleText, ""))
|| RefImproveBLP.IsMatch(articleText);
/* || (!string.IsNullOrEmpty(articleTitle) && articleText.Length < 10000 && parseTalkPage &&
TryGetArticleText(Variables.Namespaces[Namespace.Talk] + articleTitle).Contains(@"{{WPBiography"))*/
}