public void AutoTag()
{
#if DEBUG
const string LongText =
@"Lorem ipsum dolor sit amet, consectetur adipiscing elit. Suspendisse dictum ultrices augue. Fusce sem diam, vestibulum sit amet, vehicula id, congue a, nisl. Phasellus pulvinar posuere purus. Donec elementum justo mattis nulla. Sed a purus dictum lacus pharetra adipiscing. Nam non dui non ante viverra iaculis. Fusce euismod lacus id nulla vulputate gravida. Suspendisse lectus pede, tempus sed, tristique id, pharetra eget, urna. Integer mattis libero vel quam accumsan suscipit. Vivamus molestie dapibus est. Quisque quis metus eget nisl accumsan aliquet. Donec tempus pellentesque tellus. Aliquam lacinia gravida justo. Aliquam erat volutpat. Class aptent taciti sociosqu ad litora torquent per conubia nostra, per inceptos himenaeos. Mauris ultricies suscipit urna. Ut mollis tempor leo. Pellentesque fringilla mattis enim. Proin sapien enim, congue non, aliquet et, sollicitudin nec, mauris. Sed porta.
Curabitur luctus mollis massa. Nullam consectetur mollis lacus. Suspendisse turpis. Fusce velit. Morbi egestas dui. Donec commodo ornare lorem. Vestibulum sodales. Curabitur egestas libero ut metus. Sed eget orci a ligula consectetur vestibulum. Cras sapien.
Sed libero. Ut volutpat massa. Donec nulla pede, porttitor eu, sodales et, consectetur nec, quam. Pellentesque vestibulum hendrerit est. Nulla facilisi. Pellentesque habitant morbi tristique senectus et netus et malesuada fames ac turpis egestas. Duis et nibh eu lacus iaculis pretium. Fusce sed turpis. In cursus. Etiam interdum augue. Morbi commodo auctor ligula. In imperdiet, neque nec hendrerit consequat, lacus purus tristique turpis, eu hendrerit ipsum ligula at libero. Duis varius nunc vel tortor. Praesent tempor. Nunc non pede at velit congue feugiat. Curabitur gravida, nisl quis mattis porttitor, purus nulla viverra dui, non suscipit augue nunc ac libero. Donec lacinia est non augue.
Nulla quam dui, tristique id, condimentum sed, sodales in, ante. Vestibulum vitae diam. Integer placerat ante non orci. Nulla gravida. Integer magna enim, iaculis ut, ornare dignissim, ultrices a, urna. Donec urna. Fusce fringilla, pede vitae pulvinar ullamcorper, est nisi eleifend ipsum, ac adipiscing odio massa vehicula neque. Sed blandit est. Morbi faucibus, nisl vel commodo vulputate, mi ipsum tincidunt sem, id ornare orci orci et velit. Morbi commodo sollicitudin ligula. Pellentesque vitae urna. Duis massa arcu, accumsan id, euismod eu, tincidunt et, odio. Phasellus purus leo, rhoncus sed, condimentum nec, vestibulum vel, lacus. In egestas, lectus vitae lacinia tristique, elit magna consequat risus, id sodales metus nulla ac pede. Suspendisse potenti.
Fusce massa. Nullam lacinia purus nec ipsum. Vestibulum ante ipsum primis in faucibus orci luctus et ultrices posuere cubilia Curae; Suspendisse potenti. Proin augue. Donec mi magna, interdum a, elementum quis, bibendum sit amet, felis. Donec vel libero eget magna hendrerit ultrices. Suspendisse potenti. Sed scelerisque lacinia nisi. Quisque elementum, nunc nec luctus iaculis, ante quam aliquet orci, et ullamcorper dui ipsum at mi. Vestibulum a dolor id tortor posuere elementum. Sed mauris nisl, ultrices a, malesuada non, convallis ac, velit. Sed aliquam elit id metus. Donec malesuada, lorem ut pharetra auctor, mi risus viverra enim, vitae pulvinar urna metus at lorem. Vivamus id lorem. Pellentesque habitant morbi tristique senectus et netus et malesuada fames ac turpis egestas. Nulla facilisi. Ut vel odio. Pellentesque habitant morbi tristique senectus et netus et malesuada fames ac turpis egestas. Pellentesque lobortis sem.
Proin in odio. Pellentesque [[habitant]] [[morbi]] [[tristique]] senectus et netus et malesuada fames ac turpis egestas. Vivamus bibendum arcu nec risus. Nulla iaculis ligula in purus. Etiam vulputate nibh sit amet lectus. Pellentesque habitant morbi tristique senectus et netus et malesuada fames ac turpis egestas. Suspendisse potenti. Suspendisse eleifend. Donec blandit nibh hendrerit turpis. Integer accumsan posuere odio. Ut commodo augue malesuada risus. Curabitur augue. Praesent volutpat nunc a diam. Nulla lobortis interdum dolor. Nunc imperdiet, ipsum ac tempor iaculis, nunc.
";
Article a = new Article("A", "ABC");
Parsers p = new Parsers(500, true);
a.AutoTag(p, true, true);
Assert.IsFalse(a.NoArticleTextChanged);
a = new Article("A", LongText);
Globals.UnitTestIntValue = 1;
Globals.UnitTestBoolValue = false;
a.AutoTag(p, true, true);
Assert.IsTrue(a.NoArticleTextChanged);
#endif
}
public string ProcessArticle(IAutoWikiBrowser sender, ProcessArticleEventArgs eventargs)
{
//If menu item is not checked, then return
if (!PluginEnabled || Settings.Images.Count == 0)
{
eventargs.Skip = false;
return eventargs.ArticleText;
}
eventargs.EditSummary = "";
string text = eventargs.ArticleText;
Parsers parse = new Parsers();
foreach (KeyValuePair<string, string> p in Settings.Images)
{
bool noChange;
if (p.Value.Length == 0)
{
text = parse.RemoveImage(p.Key, text, Settings.Comment, "", out noChange);
if (!noChange) eventargs.EditSummary += ", removed " + Variables.Namespaces[6] + p.Key;
}
else
{
text = parse.ReplaceImage(p.Key, p.Value, text, out noChange);
if (!noChange) eventargs.EditSummary += ", replaced: " + Variables.Namespaces[6]
+ p.Key + FindandReplace.Arrow + Variables.Namespaces[6] + p.Value;
}
if (!noChange) text = Regex.Replace(text, "<includeonly>[\\s\\r\\n]*\\</includeonly>", "");
}
eventargs.Skip = (text == eventargs.ArticleText) && Settings.Skip;
return text;
}
public void Unicodify()
{
Parsers Parser = new Parsers();
HideText RemoveText = new HideText(false, true, false);
Article a = new Article("a", @"'''test'''. z & a‡ †.
{{DEFAULTSORT:Hello test}}
[[Category:Test pages]]
");
a.Unicodify(true, Parser, RemoveText);
Assert.AreEqual(@"'''test'''. z & a‡ †.
{{DEFAULTSORT:Hello test}}
[[Category:Test pages]]
", a.ArticleText, "Text unicodified");
a = new Article("a", @"'''test'''. z & {{t|a‡ †}}.
{{DEFAULTSORT:Hello test}}
[[Category:Test pages]]
");
a.Unicodify(true, Parser, RemoveText);
Assert.AreEqual(@"'''test'''. z & {{t|a‡ †}}.
{{DEFAULTSORT:Hello test}}
[[Category:Test pages]]
", a.ArticleText, "Text unicodified, hidemore used");
a = new Article("a", @"ABC");
a.Unicodify(true, Parser, RemoveText);
Assert.AreEqual(@"ABC", a.ArticleText, "No change");
}
public void EmboldenTitles()
{
Parsers Parser = new Parsers();
Article a = new Article("a", @"A [[B]]");
a.EmboldenTitles(Parser, true);
Assert.IsTrue(a.ArticleText.Contains("[[B]]"));
}
public RequiresParser2()
{
parser2 = new Parsers();
}
/// <summary>
/// Returns the input date in the requested format (American or International). If another Locale is pasased in the input date is returned. For en-wiki only.
/// </summary>
/// <param name="inputDate">string representing a date, any format that C# can parse</param>
/// <param name="locale">Locale of output date required (American/International/ISO)</param>
/// <param name="AmericanInputDate">Whether the input date is in American MM/DD/YYYY format</param>
/// <returns>The English-language (American or International) date</returns>
public static string ConvertDate(string inputDate, Parsers.DateLocale locale, bool AmericanInputDate)
{
if (Variables.LangCode != "en" || YearMon.IsMatch(inputDate) || MonthYear.IsMatch(inputDate))
return inputDate;
DateTime dt;
try
{
dt = Convert.ToDateTime(inputDate, AmericanInputDate ? AmericanEnglish : BritishEnglish);
}
catch
{
return inputDate;
}
switch (locale)
{
case Parsers.DateLocale.American:
return dt.ToString("MMMM d, yyyy", BritishEnglish);
case Parsers.DateLocale.International:
return dt.ToString("d MMMM yyyy", BritishEnglish);
case Parsers.DateLocale.ISO:
return dt.ToString("yyyy-MM-dd", BritishEnglish);
default:
return inputDate;
}
}
/// <summary>
/// Expands (substitutes) template calls
/// </summary>
/// <param name="articleText">The text of the article</param>
/// <param name="articleTitle">The title of the artlce</param>
/// <param name="regexes">Dictionary of templates to substitute</param>
/// <param name="includeComment"></param>
/// <returns>The updated article text</returns>
public static string ExpandTemplate(string articleText, string articleTitle, Dictionary<Regex, string> regexes, bool includeComment)
{
foreach (KeyValuePair<Regex, string> p in regexes)
{
string originalArticleText = "";
while(!originalArticleText.Equals(articleText))
{
originalArticleText = articleText;
// avoid matching on previously commented out calls
Match m = p.Key.Match(Tools.ReplaceWithSpaces(articleText, WikiRegexes.Comments));
if(!m.Success)
continue;
string call = m.Value, result;
if(Globals.UnitTestMode)
result = "Expanded template test return";
else
{
string expandUri = Variables.URLApi + "?action=expandtemplates&format=xml&title=" + WikiEncode(articleTitle) + "&text=" + HttpUtility.UrlEncode(call);
try
{
string respStr = GetHTML(expandUri);
Match m1 = ExpandTemplatesRegex.Match(respStr);
if (!m.Success) continue;
result = HttpUtility.HtmlDecode(m1.Groups[1].Value);
}
catch
{
continue;
}
}
bool skipArticle;
result = new Parsers().Unicodify(result, out skipArticle);
if (includeComment)
result = result + "<!-- " + call + " -->";
articleText = articleText.Replace(call, result);
}
}
return articleText;
}
/// <summary>
///
/// </summary>
/// <param name="articleText">The text of the article</param>
/// <param name="articleTitle">The title of the artlce</param>
/// <param name="regexes"></param>
/// <param name="includeComment"></param>
/// <returns></returns>
public static string ExpandTemplate(string articleText, string articleTitle, Dictionary<Regex, string> regexes, bool includeComment)
{
foreach (KeyValuePair<Regex, string> p in regexes)
{
foreach (Match m in p.Key.Matches(articleText))
{
string call = m.Value;
string expandUri = Variables.URLApi + "?action=expandtemplates&format=xml&title=" + WikiEncode(articleTitle) + "&text=" + HttpUtility.UrlEncode(call);
string result;
try
{
string respStr = GetHTML(expandUri);
Match m1 = ExpandTemplatesRegex.Match(respStr);
if (!m.Success) continue;
result = HttpUtility.HtmlDecode(m1.Groups[1].Value);
}
catch
{
continue;
}
bool skipArticle;
result = new Parsers().Unicodify(result, out skipArticle);
if (includeComment)
result = result + "<!-- " + call + " -->";
articleText = articleText.Replace(call, result);
}
}
return articleText;
}
public LivingPerson(Parsers p)
{
parsers = p;
}
public HasSectionError(Parsers p)
{
parsers = p;
}
public HasHTMLEntities(Parsers p)
{
parsers = p;
}
public HasNoBoldTitle(Parsers p)
{
parsers = p;
}
/// <summary>
/// Sorts article meta data
/// </summary>
/// <param name="articleText">The wiki text of the article.</param>
/// <param name="articleTitle">Title of the article</param>
/// <param name="fixOptionalWhitespace">Whether to request optional excess whitespace to be fixed</param>
/// <returns>The updated article text</returns>
internal string Sort(string articleText, string articleTitle, bool fixOptionalWhitespace)
{
if (Namespace.Determine(articleTitle) == Namespace.Template) // Don't sort on templates
{
return(articleText);
}
string strSave = articleText;
try
{
articleText = Regex.Replace(articleText, "<!-- ?\\[\\[en:.*?\\]\\] ?-->", "");
string personData = Tools.Newline(RemovePersonData(ref articleText));
string disambig = Tools.Newline(RemoveDisambig(ref articleText));
string categories = Tools.Newline(RemoveCats(ref articleText, articleTitle));
string interwikis = Tools.Newline(Interwikis(ref articleText));
// Dablinks above orphan tags per [[WP:LAYOUT]]
if (Variables.LangCode.Equals("en"))
{
articleText = MoveMaintenanceTags(articleText);
}
articleText = MoveDablinks(articleText);
if (Variables.LangCode.Equals("en"))
{
articleText = MovePortalTemplates(articleText);
articleText = MoveTemplateToSeeAlsoSection(articleText, Tools.NestedTemplateRegex("Wikipedia-Books"));
articleText = MoveSisterlinks(articleText);
articleText = MoveTemplateToReferencesSection(articleText, WikiRegexes.Ibid);
articleText = MoveExternalLinks(articleText);
articleText = MoveSeeAlso(articleText);
}
// two newlines here per http://en.wikipedia.org/w/index.php?title=Wikipedia_talk:AutoWikiBrowser&oldid=243224092#Blank_lines_before_stubs
// http://en.wikipedia.org/wiki/Wikipedia_talk:AutoWikiBrowser/Bugs/Archive_11#Two_empty_lines_before_stub-templates
// Russian wiki uses only one newline
string strStub = Tools.Newline(RemoveStubs(ref articleText), Variables.LangCode.Equals("ru") ? 1 : 2);
//filter out excess white space and remove "----" from end of article
articleText = Parsers.RemoveWhiteSpace(articleText, fixOptionalWhitespace) + "\r\n";
articleText += disambig;
switch (Variables.LangCode)
{
case "de":
case "sl":
articleText += strStub + categories + personData;
// http://en.wikipedia.org/wiki/Wikipedia_talk:AutoWikiBrowser#Removal_of_blank_lines
// on de wiki a blank line is desired between persondata and interwikis
if (Variables.LangCode == "de" && personData.Length > 0 && interwikis.Length > 0)
{
articleText += "\r\n";
}
break;
case "pl":
case "ru":
case "simple":
articleText += personData + strStub + categories;
break;
case "it":
if (Variables.Project == ProjectEnum.wikiquote)
{
articleText += personData + strStub + categories;
}
else
{
articleText += personData + categories + strStub;
}
break;
default:
articleText += personData + categories + strStub;
break;
}
articleText = (articleText + interwikis);
if (Namespace.Determine(articleTitle) == Namespace.Category)
{
return(articleText.Trim());
}
else
{
return(articleText.TrimEnd());
}
}
catch (Exception ex)
{
if (!ex.Message.Contains("DEFAULTSORT"))
{
ErrorHandler.Handle(ex);
}
return(strSave);
}
}
public RequiresParser()
{
parser = new Parsers();
}
public MissingDefaultsort(Parsers p)
{
parsers = p;
}
/// <summary>
/// Merges multiple {{portal}} templates into a single one, removing any duplicates. En-wiki only.
/// Restricted to {{portal}} calls with one argument
/// Article must have existing {{portal}} and/or a 'see also' section
/// </summary>
/// <param name="articleText">The article text</param>
/// <returns>The updated article text</returns>
public static string MergePortals(string articleText)
{
if (!Variables.LangCode.Equals("en"))
{
return(articleText);
}
// return if {{portal}} within {{stack}}
List <string> stacks = Parsers.GetAllTemplateDetail(articleText).Where(t => StackTemplate.IsMatch(t)).ToList();
if (stacks.Any(s => WikiRegexes.PortalTemplate.IsMatch(s)))
{
return(articleText);
}
string originalArticleText = articleText;
List <string> Portals = new List <string>();
int firstPortal = WikiRegexes.PortalTemplate.Match(articleText).Index;
MatchCollection portalCalls = WikiRegexes.PortalTemplate.Matches(articleText);
foreach (Match m in portalCalls)
{
string thePortalCall = m.Value;
// Do not process portal templates with named arguments
if (!Tools.GetTemplateParameterValues(thePortalCall).Any())
{
for (int i = 1; i <= Tools.GetTemplateArgumentCount(thePortalCall); i++)
{
Portals.Add(Tools.GetTemplateArgument(thePortalCall, i).Trim());
}
articleText = Regex.Replace(articleText, Regex.Escape(thePortalCall) + @"\s*(?:\r\n)?", "");
}
}
// if no portal parameters, return now, removing empty portal template
if (!Portals.Any())
{
return(articleText);
}
// abort if not enough portals to merge, or all in one existing template
if (Portals.Count < 2 || portalCalls.Count == 1)
{
return(originalArticleText);
}
// generate portal string
string portalsToAdd = Tools.DeduplicateList(Portals).Aggregate("", (current, portal) => current + ("|" + portal));
// first merge to see also section
if (WikiRegexes.SeeAlso.Matches(articleText).Count == 1)
{
return(WikiRegexes.SeeAlso.Replace(articleText, "$0" + Tools.Newline(@"{{Portal" + portalsToAdd + @"}}")));
}
// otherwise merge to original location if all portals in same section
if (!WikiRegexes.Headings.IsMatch(originalArticleText) || Summary.ModifiedSection(originalArticleText, articleText).Length > 0)
{
return(articleText.Insert(firstPortal, @"{{Portal" + portalsToAdd + @"}}" + "\r\n"));
}
return(originalArticleText);
}
//TODO:Needs re-write
/// <summary>
/// If necessary, adds/removes wikify or stub tag
/// </summary>
/// <param name="articleText">The wiki text of the article.</param>
/// <param name="articleTitle">The article title.</param>
/// <param name="restrictOrphanTagging"></param>
/// <param name="summary"></param>
/// <returns>The tagged article.</returns>
public string Tagger(string articleText, string articleTitle, bool restrictOrphanTagging, ref string summary)
{
// don't tag redirects/outside article namespace/no tagging changes
if (!Namespace.IsMainSpace(articleTitle) || Tools.IsRedirect(articleText) || WikiRegexes.Wi.IsMatch(articleText))
return articleText;
tagsRemoved.Clear();
tagsAdded.Clear();
string commentsStripped = WikiRegexes.Comments.Replace(articleText, "");
string commentsCategoriesStripped = WikiRegexes.Category.Replace(commentsStripped, "");
Sorter.Interwikis(ref commentsStripped);
// bulleted or indented text should weigh less than simple text.
// for example, actor stubs may contain large filmographies
string crapStripped = WikiRegexes.BulletedText.Replace(commentsCategoriesStripped, "");
int words = (Tools.WordCount(commentsCategoriesStripped) + Tools.WordCount(crapStripped))/2;
// remove stub tags from long articles
if ((words > StubMaxWordCount) && WikiRegexes.Stub.IsMatch(commentsStripped))
{
articleText = WikiRegexes.Stub.Replace(articleText, StubChecker).Trim();
tagsRemoved.Add("stub");
}
// refresh
commentsStripped = WikiRegexes.Comments.Replace(articleText, "");
commentsCategoriesStripped = WikiRegexes.Category.Replace(commentsStripped, "");
// on en wiki, remove expand template when a stub template exists
// http://en.wikipedia.org/wiki/Wikipedia_talk:AutoWikiBrowser/Feature_requests/Archive_5#Remove_.7B.7Bexpand.7D.7D_when_a_stub_template_exists
if (Variables.LangCode == "en" && WikiRegexes.Stub.IsMatch(commentsCategoriesStripped) &&
WikiRegexes.Expand.IsMatch(commentsCategoriesStripped))
{
articleText = WikiRegexes.Expand.Replace(articleText, "");
tagsRemoved.Add("expand");
}
// refresh
commentsStripped = WikiRegexes.Comments.Replace(articleText, "");
commentsCategoriesStripped = WikiRegexes.Category.Replace(commentsStripped, "");
// do orphan tagging before template analysis for categorisation tags
articleText = TagOrphans(articleText, articleTitle, restrictOrphanTagging);
articleText = TagRefsIbid(articleText);
articleText = TagEmptySection(articleText);
int totalCategories;
int linkCount = Tools.LinkCount(commentsStripped);
#if DEBUG || UNITTEST
if (Globals.UnitTestMode)
{
totalCategories = Globals.UnitTestIntValue;
}
else
#endif
{
// stubs add non-hidden stub categories, don't count these in categories count
List<Article> Cats = CategoryProv.MakeList(new[] {articleTitle});
List<Article> CatsNotStubs = new List<Article>();
foreach (Article a in Cats)
{
if (!a.Name.EndsWith(" stubs") && !a.Name.EndsWith(":Stubs"))
CatsNotStubs.Add(a);
}
totalCategories = CatsNotStubs.Count;
}
if (linkCount > 0 && WikiRegexes.DeadEnd.IsMatch(articleText))
{
articleText = WikiRegexes.DeadEnd.Replace(articleText, new MatchEvaluator(SectionTagME));
if(!WikiRegexes.DeadEnd.IsMatch(articleText))
tagsRemoved.Add("deadend");
}
// discount persondata along with comments and categories from wikify and stub evaluation
int length = WikiRegexes.Persondata.Replace(commentsCategoriesStripped, "").Length + 1;
bool underlinked = (linkCount < 0.0025*length);
if (length <= 300 && !WikiRegexes.Stub.IsMatch(commentsCategoriesStripped) &&
!WikiRegexes.Disambigs.IsMatch(commentsCategoriesStripped) && !WikiRegexes.SIAs.IsMatch(commentsCategoriesStripped))
{
// add stub tag
articleText += Tools.Newline("{{stub}}", 3);
tagsAdded.Add("stub");
commentsStripped = WikiRegexes.Comments.Replace(articleText, "");
}
// http://en.wikipedia.org/wiki/Wikipedia_talk:AutoWikiBrowser/Archive_19#AWB_problems
// nl wiki doesn't use {{Uncategorized}} template
// prevent wictionary redirects from being tagged as uncategorised
if (words > 6 && totalCategories == 0
&& !WikiRegexes.Uncat.IsMatch(articleText)
&& Variables.LangCode != "nl"
&& !Tools.NestedTemplateRegex("cat improve").IsMatch(articleText)
// category count is from API; don't add uncat tag if genfixes added person categories
&& !WikiRegexes.DeathsOrLivingCategory.IsMatch(articleText)
&& !WikiRegexes.BirthsCategory.IsMatch(articleText))
{
if (WikiRegexes.Stub.IsMatch(commentsStripped))
{
// add uncategorized stub tag
articleText += Tools.Newline("{{Uncategorized stub|", 2) + WikiRegexes.DateYearMonthParameter +
@"}}";
tagsAdded.Add("[[CAT:UNCATSTUBS|uncategorised]]");
}
else
{
// add uncategorized tag
articleText += Tools.Newline("{{Uncategorized|", 2) + WikiRegexes.DateYearMonthParameter + @"}}";
tagsAdded.Add("[[CAT:UNCAT|uncategorised]]");
}
}
// remove {{Uncategorized}} if > 0 real categories (stub categories not counted)
// rename {{Uncategorized}} to {{Uncategorized stub}} if stub with zero categories (stub categories not counted)
if (WikiRegexes.Uncat.IsMatch(articleText))
{
if (totalCategories > 0)
{
articleText = WikiRegexes.Uncat.Replace(articleText, "");
tagsRemoved.Add("uncategorised");
}
else if (totalCategories == 0 && WikiRegexes.Stub.IsMatch(commentsStripped))
{
string uncatname = WikiRegexes.Uncat.Match(articleText).Groups[1].Value;
if (!uncatname.Contains("stub"))
articleText = Tools.RenameTemplate(articleText, uncatname, "Uncategorized stub");
}
}
if (linkCount == 0 && !WikiRegexes.DeadEnd.IsMatch(articleText) && Variables.LangCode != "sv"
&& !Regex.IsMatch(WikiRegexes.MultipleIssues.Match(articleText).Value.ToLower(), @"\bdead ?end\b"))
{
// add dead-end tag
articleText = "{{dead end|" + WikiRegexes.DateYearMonthParameter + "}}\r\n\r\n" + articleText;
tagsAdded.Add("[[:Category:Dead-end pages|deadend]]");
}
if (linkCount < 3 && underlinked && !WikiRegexes.Wikify.IsMatch(articleText)
&& !WikiRegexes.MultipleIssues.Match(articleText).Value.ToLower().Contains("wikify"))
{
// add wikify tag
articleText = "{{Wikify|" + WikiRegexes.DateYearMonthParameter + "}}\r\n\r\n" + articleText;
tagsAdded.Add("[[WP:WFY|wikify]]");
}
else if (linkCount > 3 && !underlinked &&
WikiRegexes.Wikify.IsMatch(articleText))
{
articleText = WikiRegexes.Wikify.Replace(articleText, new MatchEvaluator(SectionTagME));
if(!WikiRegexes.Wikify.IsMatch(articleText))
tagsRemoved.Add("wikify");
}
// rename unreferenced --> refimprove if has existing refs
if (WikiRegexes.Unreferenced.IsMatch(commentsCategoriesStripped)
&& WikiRegexes.Refs.Matches(commentsCategoriesStripped).Count > 0)
{
articleText = Tools.RenameTemplate(articleText, "unreferenced", "refimprove", true);
Match m = WikiRegexes.MultipleIssues.Match(articleText);
if(m.Success)
{
string newValue = Tools.RenameTemplateParameter(m.Value, "unreferenced", "refimprove");
if(!newValue.Equals(m.Value))
articleText = articleText.Replace(m.Value, newValue);
}
}
if (tagsAdded.Count > 0 || tagsRemoved.Count > 0)
{
Parsers p = new Parsers();
HideText ht = new HideText();
articleText = ht.HideUnformatted(articleText);
articleText = p.MultipleIssues(articleText);
articleText = Conversions(articleText);
articleText = ht.AddBackUnformatted(articleText);
// sort again in case tag removal requires whitespace cleanup
articleText = p.Sorter.Sort(articleText, articleTitle);
}
summary = PrepareTaggerEditSummary();
return articleText;
}
/// <summary>
/// Returns the input ISO date in the requested format (American or International). If another Locale is pasased in the input date is returned. For en-wiki only.
/// </summary>
/// <param name="ISODate">string representing ISO date</param>
/// <param name="locale">Locale of output date required (American or International)</param>
/// <returns>The English-language (American or International) date</returns>
public static string ISOToENDate(string ISODate, Parsers.DateLocale locale)
{
if (Variables.LangCode != "en")
return ISODate;
DateTime dt;
try
{
dt = Convert.ToDateTime(ISODate);
}
catch
{
return ISODate;
}
switch (locale)
{
case Parsers.DateLocale.American:
return dt.ToString("MMMM d, yyyy", English);
case Parsers.DateLocale.International:
return dt.ToString("d MMMM yyyy", English);
default:
return ISODate;
}
}
/// <summary>
/// First checks for a <references< missing '/' to correct, otherwise:
/// if the article uses cite references but has no recognised template to display the references, add {{Reflist}} in the appropriate place
/// </summary>
/// <param name="articleText">The wiki text of the article</param>
/// <returns>The updated article text</returns>
public static string AddMissingReflist(string articleText)
{
if (!IsMissingReferencesDisplay(articleText) || !Variables.LangCode.Equals("en"))
return articleText;
if (ReferencesMissingSlash.IsMatch(articleText))
return ReferencesMissingSlash.Replace(articleText, @"<references/>");
// Rename ==Links== to ==External links==
articleText = LinksHeading.Replace(articleText, "$1External links$2");
// add to any existing references section if present
if (ReferencesHeadingLevel2.IsMatch(articleText))
articleText = ReferencesHeadingLevelLower.Replace(articleText, "$1\r\n{{Reflist}}");
else
{
articleText += "\r\n==References==\r\n{{Reflist}}";
// now sort metadata in case Category at top of article
Parsers p = new Parsers();
articleText = p.SortMetaData(articleText, "A", false);
// try to move just above external links
if (ExternalLinksHeading.IsMatch(articleText))
articleText = ExternalLinksToReferences.Replace(articleText, "$2\r\n$1");
else if (Category.IsMatch(articleText))
// try to move just above categories
articleText = CategoryToReferences.Replace(articleText, "$3\r\n$1$2");
else // not moved, so extra blank line required before heading
articleText = articleText.Replace("\r\n==References==", "\r\n\r\n==References==");
}
return articleText;
}
/// <summary>
/// Returns the input date in the requested format (American or International). If another Locale is pasased in the input date is returned. For en-wiki only.
/// </summary>
/// <param name="inputDate">string representing a date, any format that C# can parse</param>
/// <param name="locale">Locale of output date required (American/International/ISO)</param>
/// <returns>The English-language (American or International) date</returns>
public static string ConvertDate(string inputDate, Parsers.DateLocale locale)
{
return ConvertDate(inputDate, locale, false);
}
/// <summary>
/// Removes various unwanted punctuation and comment characters from a derived reference name
/// </summary>
/// <param name="derivedName">the input reference name</param>
/// <returns>the cleaned reference name</returns>
private static string CleanDerivedReferenceName(string derivedName)
{
derivedName = WikiRegexes.PipedWikiLink.Replace(derivedName, "$2"); // piped wikilinks -> text value
derivedName = CommentOrFloorNumber.Replace(derivedName, "");
// rm comments from ref name, might be masked
derivedName = derivedName.Trim(CharsToTrim.ToCharArray());
derivedName = SequenceOfQuotesInDerivedName.Replace(derivedName, ""); // remove chars
derivedName = WhitespaceInDerivedName.Replace(derivedName, " "); // spacing fixes
derivedName = derivedName.Replace(@"–", "–");
Parsers p = new Parsers();
derivedName = p.FixDatesA(derivedName);
derivedName = p.FixDatesB(derivedName, false, false);
return DateRetrievedOrAccessed.IsMatch(derivedName) ? "" : derivedName;
}
public MainForm()
{
SplashScreen.Show(this);
RightToLeft = System.Globalization.CultureInfo.CurrentCulture.TextInfo.IsRightToLeft
? RightToLeft.Yes : RightToLeft.No;
InitializeComponent();
SplashScreen.SetProgress(5);
try
{
btntsShowHide.Image = Resources.Showhide;
btntsShowHideParameters.Image = Resources.Showhideparameters;
btntsSave.Image = Resources.Save;
btntsIgnore.Image = Resources.RightArrow;
btntsStop.Image = Resources.Stop;
btntsPreview.Image = Resources.preview;
btntsChanges.Image = Resources.changes;
btntsFalsePositive.Image = Resources.RollBack;
btntsStart.Image = Resources.Run;
btntsDelete.Image = Resources.Vista_trashcan_empty;
SplashScreen.SetProgress(10);
try
{
Parser = new Parsers(Properties.Settings.Default.StubMaxWordCount,
Properties.Settings.Default.AddHummanKeyToCats);
}
catch (Exception ex)
{
Parser = new Parsers();
ErrorHandler.Handle(ex);
}
toolStripComboOnLoad.SelectedIndex = 0;
cmboCategorise.SelectedIndex = 0;
cmboImages.SelectedIndex = 0;
Variables.User.UserNameChanged += UpdateUserName;
Variables.User.BotStatusChanged += UpdateBotStatus;
Variables.User.AdminStatusChanged += UpdateAdminStatus;
//Variables.User.WikiStatusChanged += UpdateWikiStatus;
Variables.User.WebBrowserLogin.DocumentCompleted += WebLoginCompleted;
Variables.User.WebBrowserLogin.Navigating += WebLoginStarting;
webBrowserEdit.Deleted += CaseWasDelete;
webBrowserEdit.Loaded += CaseWasLoad;
webBrowserEdit.Saved += CaseWasSaved;
webBrowserEdit.None += CaseWasNull;
webBrowserEdit.Fault += StartDelayedRestartTimer;
webBrowserEdit.StatusChanged += UpdateWebBrowserStatus;
listMaker.UserInputTextBox.ContextMenuStrip = mnuMakeFromTextBox;
listMaker.BusyStateChanged += SetProgressBar;
listMaker.NoOfArticlesChanged += UpdateButtons;
listMaker.StatusTextChanged += UpdateListStatus;
listMaker.cmboSourceSelect.SelectedIndexChanged += ListMakerSourceSelectHandler;
Profiles = new WikiFunctions.Profiles.AWBProfilesForm(webBrowserEdit);
Profiles.LoadProfile += LoadProfileSettings;
SplashScreen.SetProgress(15);
}
catch (Exception ex)
{
ErrorHandler.Handle(ex);
}
}
/// <summary>
/// Performs fixes to a given citation template call
/// </summary>
/// <param name="m"></param>
/// <returns></returns>
private static string FixCitationTemplatesME(Match m)
{
string newValue = Tools.RemoveExcessTemplatePipes(m.Value);
string templatename = m.Groups[2].Value;
Dictionary<string, string> paramsFound = new Dictionary<string, string>();
// remove duplicated fields, ensure the URL is not touched (may have pipes in)
newValue = Tools.RemoveDuplicateTemplateParameters(newValue, paramsFound);
string theURL,
id,
format,
theTitle,
TheYear,
lang,
TheDate,
TheMonth,
TheWork,
nopp,
TheIssue,
accessyear,
accessdate,
pages,
page,
ISBN,
origyear,
archiveurl,
contributionurl;
if(!paramsFound.TryGetValue("url", out theURL))
theURL = "";
if(!paramsFound.TryGetValue("id", out id))
id = "";
if(!paramsFound.TryGetValue("format", out format))
format = "";
if(!paramsFound.TryGetValue("title", out theTitle))
theTitle = "";
if(!paramsFound.TryGetValue("year", out TheYear))
TheYear = "";
if(!paramsFound.TryGetValue("date", out TheDate))
TheDate = "";
if(!paramsFound.TryGetValue("language", out lang))
lang = "";
if(!paramsFound.TryGetValue("month", out TheMonth))
TheMonth = "";
if(!paramsFound.TryGetValue("work", out TheWork))
TheWork = "";
if(!paramsFound.TryGetValue("nopp", out nopp))
nopp = "";
if(!paramsFound.TryGetValue("issue", out TheIssue))
TheIssue = "";
if(!paramsFound.TryGetValue("accessyear", out accessyear))
accessyear = "";
if(!paramsFound.TryGetValue("accessdate", out accessdate))
accessdate = "";
if(!paramsFound.TryGetValue("pages", out pages))
pages = "";
if(!paramsFound.TryGetValue("page", out page))
page = "";
if(!paramsFound.TryGetValue("origyear", out origyear))
origyear = "";
if(!paramsFound.TryGetValue("archiveurl", out archiveurl))
archiveurl = "";
if(!paramsFound.TryGetValue("contribution-url", out contributionurl))
contributionurl = "";
if(!paramsFound.TryGetValue("isbn", out ISBN) && !paramsFound.TryGetValue("ISBN", out ISBN))
ISBN = "";
string theURLoriginal = theURL;
// remove the unneeded 'format=HTML' field
// https://en.wikipedia.org/wiki/Wikipedia_talk:AutoWikiBrowser/Feature_requests#Remove_.22format.3DHTML.22_in_citation_templates
// remove format= field with null value when URL is HTML page
if(paramsFound.ContainsKey("format"))
{
if (format.TrimStart("[]".ToCharArray()).ToUpper().StartsWith("HTM")
||
(format.Length == 0 &&
theURL.ToUpper().TrimEnd('L').EndsWith("HTM")))
newValue = Tools.RemoveTemplateParameter(newValue, "format");
}
// newlines to spaces in title field if URL used, otherwise display broken
if (theURL.Length > 0 && theTitle.Contains("\r\n"))
{
theTitle = theTitle.Replace("\r\n", " ");
paramsFound.Remove("title");
paramsFound.Add("title", theTitle);
newValue = Tools.UpdateTemplateParameterValue(newValue, "title", theTitle);
}
// remove language=English on en-wiki
if (lang.Equals("english", StringComparison.OrdinalIgnoreCase) || lang.Equals("en", StringComparison.OrdinalIgnoreCase))
newValue = Tools.RemoveTemplateParameter(newValue, "language");
// remove italics for work field for book/periodical, but not website -- auto italicised by template
if (TheWork.Length > 0 && !TheWork.Contains("."))
newValue = WorkInItalics.Replace(newValue, "$1$2");
// remove quotes around title field: are automatically added by template markup
foreach (string dequoteParam in ParametersToDequote)
{
string quotetitle;
if(paramsFound.TryGetValue(dequoteParam, out quotetitle))
{
string before = quotetitle;
// convert curly quotes to straight quotes per [[MOS:PUNCT]], but » or › may be section delimeter
// so only change those when balanced. Note regular <> characters are not changed.
quotetitle = WikiRegexes.CurlyDoubleQuotes.Replace(quotetitle, @"""");
quotetitle = BalancedArrows.Replace(quotetitle, @"""$1$2""");
if (quotetitle.Contains(@"""") && !quotetitle.Trim('"').Contains(@""""))
quotetitle = quotetitle.Trim('"');
if(!before.Equals(quotetitle))
newValue = Tools.SetTemplateParameterValue(newValue, dequoteParam, quotetitle);
}
}
// page= and pages= fields don't need p. or pp. in them when nopp not set
if ((pages.Contains("p") || page.Contains("p")) && !templatename.Equals("cite journal", StringComparison.OrdinalIgnoreCase) && nopp.Length == 0)
{
newValue = CiteTemplatePagesPP.Replace(newValue, "");
pages = Tools.GetTemplateParameterValue(newValue, "pages");
}
// date = YYYY --> year = YYYY if year the same
// with Lua no need to rename date to year when date = YYYY
if (TheDate.Length == 4)
{
if(TheYear.Equals(TheDate))
newValue = Tools.RemoveTemplateParameter(newValue, "date");
}
// year = full date --> date = full date
if (TheYear.Length > 5)
{
string TheYearCorected = IncorrectCommaInternationalDates.Replace(TheYear, @"$1 $2");
TheYearCorected = IncorrectCommaAmericanDates.Replace(TheYearCorected, @"$1 $2, $3");
if(!TheYearCorected.Equals(TheYear))
{
newValue = Tools.UpdateTemplateParameterValue(newValue, "year", TheYearCorected);
TheYear = TheYearCorected;
}
}
if (TheYear.Length > 5 && (WikiRegexes.ISODates.IsMatch(TheYear) || WikiRegexes.InternationalDates.IsMatch(TheYear)
|| WikiRegexes.AmericanDates.IsMatch(TheYear)))
{
TheDate = TheYear;
TheYear = "";
newValue = Tools.RenameTemplateParameter(newValue, "year", "date");
}
// year=YYYY and date=...YYYY -> remove year; not for year=YYYYa
if (TheYear.Length == 4 && TheDate.Contains(TheYear) && YearOnly.IsMatch(TheYear))
{
Parsers p = new Parsers();
TheDate = p.FixDatesAInternal(TheDate);
if(WikiRegexes.InternationalDates.IsMatch(TheDate) || WikiRegexes.AmericanDates.IsMatch(TheDate)
|| WikiRegexes.ISODates.IsMatch(TheDate))
{
TheYear = "";
newValue = Tools.RemoveTemplateParameter(newValue, "year");
}
}
// month=Month and date=...Month... OR month=Month and date=same month (by conversion from ISO format)Ors month=nn and date=same month (by conversion to ISO format)
int num;
if ((TheMonth.Length > 2 && TheDate.Contains(TheMonth)) // named month within date
|| (TheMonth.Length > 2 && Tools.ConvertDate(TheDate, DateLocale.International).Contains(TheMonth))
|| (int.TryParse(TheMonth, out num) && Regex.IsMatch(Tools.ConvertDate(TheDate, DateLocale.ISO), @"\-0?" + TheMonth + @"\-")))
newValue = Tools.RemoveTemplateParameter(newValue, "month");
// date = Month DD and year = YYYY --> date = Month DD, YYYY
if (!YearOnly.IsMatch(TheDate) && YearOnly.IsMatch(TheYear))
{
if (!WikiRegexes.AmericanDates.IsMatch(TheDate) && WikiRegexes.AmericanDates.IsMatch(TheDate + ", " + TheYear))
{
if(!TheDate.Contains(TheYear))
newValue = Tools.SetTemplateParameterValue(newValue, "date", TheDate + ", " + TheYear);
newValue = Tools.RemoveTemplateParameter(newValue, "year");
}
else if (!WikiRegexes.InternationalDates.IsMatch(TheDate) && WikiRegexes.InternationalDates.IsMatch(TheDate + " " + TheYear))
{
if(!TheDate.Contains(TheYear))
newValue = Tools.SetTemplateParameterValue(newValue, "date", TheDate + " " + TheYear);
newValue = Tools.RemoveTemplateParameter(newValue, "year");
}
}
// correct volume=vol 7... and issue=no. 8 for {{cite journal}} only
if (templatename.Equals("cite journal", StringComparison.OrdinalIgnoreCase))
{
newValue = CiteTemplatesJournalVolume.Replace(newValue, "");
newValue = CiteTemplatesJournalIssue.Replace(newValue, "");
if (TheIssue.Length == 0)
newValue = CiteTemplatesJournalVolumeAndIssue.Replace(newValue, @"| issue = ");
}
// {{cite web}} for Google books -> {{Cite book}}
if (templatename.Contains("web") && newValue.Contains("http://books.google.")
&& TheWork.Length == 0)
newValue = Tools.RenameTemplate(newValue, templatename, "Cite book");
// remove leading zero in day of month
if(Regex.IsMatch(newValue, @"\b0[1-9]") && DateLeadingZero.IsMatch(newValue))
{
newValue = DateLeadingZero.Replace(newValue, @"$1$2$3$4$5");
newValue = DateLeadingZero.Replace(newValue, @"$1$2$3$4$5");
TheDate = Tools.GetTemplateParameterValue(newValue, "date");
accessdate = Tools.GetTemplateParameterValue(newValue, "accessdate");
}
if (Regex.IsMatch(templatename, @"[Cc]ite(?: ?web| book| news)"))
{
// remove any empty accessdaymonth, accessmonthday, accessmonth and accessyear
newValue = AccessDayMonthDay.Replace(newValue, "");
// merge accessdate of 'D Month' or 'Month D' and accessyear of 'YYYY' in cite web
if(accessyear.Length == 4)
newValue = AccessDateYear.Replace(newValue, @" $2$1$3");
}
// remove accessyear where accessdate is present and contains said year
if (accessyear.Length > 0 && accessdate.Contains(accessyear))
newValue = Tools.RemoveTemplateParameter(newValue, "accessyear");
// fix unspaced comma ranges, avoid pages=12,345 as could be valid page number
if (Regex.Matches(pages, @"\b\d{1,2},\d{3}\b").Count == 0)
{
while (UnspacedCommaPageRange.IsMatch(pages))
{
pages = UnspacedCommaPageRange.Replace(pages, "$1, $2");
}
newValue = Tools.UpdateTemplateParameterValue(newValue, "pages", pages);
paramsFound.Remove("pages");
paramsFound.Add("pages", pages);
}
// page range should have unspaced en-dash; validate that page is range not section link
newValue = FixPageRanges(newValue, paramsFound);
// page range or list should use 'pages' parameter not 'page'
if (CiteTemplatesPageRangeName.IsMatch(newValue))
{
newValue = CiteTemplatesPageRangeName.Replace(newValue, @"$1pages$2");
newValue = Tools.RemoveDuplicateTemplateParameters(newValue);
}
// remove ordinals from dates
if(Ordinal.IsMatch(TheDate) || Ordinal.IsMatch(accessdate))
{
if (OrdinalsInDatesInt.IsMatch(TheDate))
newValue = Tools.UpdateTemplateParameterValue(newValue, "date", OrdinalsInDatesInt.Replace(TheDate, "$1$2$3 $4"));
else if (OrdinalsInDatesAm.IsMatch(TheDate))
newValue = Tools.UpdateTemplateParameterValue(newValue, "date", OrdinalsInDatesAm.Replace(TheDate, "$1 $2$3"));
if (OrdinalsInDatesInt.IsMatch(accessdate))
newValue = Tools.UpdateTemplateParameterValue(newValue, "accessdate", OrdinalsInDatesInt.Replace(accessdate, "$1$2$3 $4"));
else if(OrdinalsInDatesAm.IsMatch(accessdate))
newValue = Tools.UpdateTemplateParameterValue(newValue, "accessdate", OrdinalsInDatesAm.Replace(accessdate, "$1 $2$3"));
}
// catch after any other fixes
newValue = IncorrectCommaAmericanDates.Replace(newValue, @"$1 $2, $3");
// URL starting www needs http://
if (theURL.StartsWith("www", StringComparison.OrdinalIgnoreCase))
theURL = "http://" + theURL;
if(archiveurl.StartsWith("www", StringComparison.OrdinalIgnoreCase))
newValue = Tools.UpdateTemplateParameterValue(newValue, "archiveurl", "http://" + archiveurl);
if(contributionurl.StartsWith("www", StringComparison.OrdinalIgnoreCase))
newValue = Tools.UpdateTemplateParameterValue(newValue, "contribution-url", "http://" + contributionurl);
// (part) wikilinked/external linked URL in cite template, don't change when named external link format
if(!theURL.Contains(" "))
theURL = theURL.Trim('[').Trim(']');
if(!theURLoriginal.Equals(theURL))
newValue = Tools.UpdateTemplateParameterValue(newValue, "url", theURL);
// {{dead link}} should be placed outside citation, not in format field per [[Template:Dead link]]
if (WikiRegexes.DeadLink.IsMatch(format))
{
string deadLink = WikiRegexes.DeadLink.Match(format).Value;
if (theURL.ToUpper().TrimEnd('L').EndsWith("HTM") && format.Equals(deadLink))
newValue = Tools.RemoveTemplateParameter(newValue, "format");
else
newValue = Tools.UpdateTemplateParameterValue(newValue, "format", format.Replace(deadLink, ""));
newValue += (" " + deadLink);
}
//id=ISBN fix
if (IdISBN.IsMatch(id) && ISBN.Length == 0)
{
newValue = Tools.RenameTemplateParameter(newValue, "id", "isbn");
newValue = Tools.SetTemplateParameterValue(newValue, "isbn", IdISBN.Match(id).Groups[1].Value.Trim());
}
//id=ASIN fix
if (IdASIN.IsMatch(id) && Tools.GetTemplateParameterValue(newValue, "asin").Length == 0 && Tools.GetTemplateParameterValue(newValue, "ASIN").Length == 0)
{
newValue = Tools.RenameTemplateParameter(newValue, "id", "asin");
newValue = Tools.SetTemplateParameterValue(newValue, "asin", IdASIN.Match(id).Groups[1].Value.Trim());
}
if(ISBN.Length > 0)
{
string ISBNbefore = ISBN;
// remove ISBN at start, but not if multiple ISBN
if(ISBN.IndexOf("isbn", StringComparison.OrdinalIgnoreCase) > -1
&& ISBN.Substring(4).IndexOf("isbn", StringComparison.OrdinalIgnoreCase) == -1)
ISBN = Regex.Replace(ISBN, @"^(?i)ISBN\s*", "");
// trim unneeded characters
ISBN = ISBN.Trim(".;,:".ToCharArray()).Trim();
// fix dashes: only hyphens allowed
while(ISBNDash.IsMatch(ISBN))
ISBN = ISBNDash.Replace(ISBN, @"$1-$2");
ISBN = ISBN.Replace('\x2010', '-');
ISBN = ISBN.Replace('\x2012', '-');
if(!ISBN.Equals(ISBNbefore))
{
if(paramsFound.ContainsKey("ISBN"))
newValue = Tools.UpdateTemplateParameterValue(newValue, "ISBN", ISBN);
else
newValue = Tools.UpdateTemplateParameterValue(newValue, "isbn", ISBN);
}
}
// origyear --> year when no year/date
if (TheYear.Length == 0 && TheDate.Length == 0 && origyear.Length == 4)
{
newValue = Tools.RenameTemplateParameter(newValue, "origyear", "year");
newValue = Tools.RemoveDuplicateTemplateParameters(newValue);
}
return newValue;
}
public void CiteTemplateDates()
{
Parsers Parser = new Parsers();
Article a = new Article("a", @"A [[B]]");
a.CiteTemplateDates(Parser, true);
Assert.IsTrue(a.ArticleText.Contains("[[B]]"));
}
/// <summary>
/// takes input string of date and age e.g. "11 May 1990 (age 21)" and converts to {{birth date and age|1990|5|11}}
/// </summary>
/// <param name="dateandage"></param>
/// <returns></returns>
public static string FormatToBDA(string dateandage)
{
Parsers p = new Parsers();
string original = dateandage;
// clean up date format if possible
dateandage = p.FixDateOrdinalsAndOf(" " + dateandage, "test");
// remove date wikilinks
dateandage = WikiRegexes.WikiLinksOnlyPossiblePipe.Replace(dateandage, "$1").Trim();
// string must end with (age xx)
if (!AgeBrackets.IsMatch(dateandage))
return original;
dateandage = AgeBrackets.Replace(dateandage, "");
bool AmericanDate = WikiRegexes.AmericanDates.IsMatch(dateandage);
string ISODate = Tools.ConvertDate(dateandage, DateLocale.ISO);
if (ISODate.Equals(dateandage) && !WikiRegexes.ISODates.IsMatch(dateandage))
return original;
// we have ISO date, convert with {{birth date and age}}, American date, set mf=y
return @"{{birth date and age|" + (AmericanDate ? "mf=y|" : "df=y|") + ISODate.Replace("-", "|") + @"}}";
}
public void FixPeopleCategories()
{
Parsers p = new Parsers();
Article a = new Article("A", "ABC");
a.FixPeopleCategories(p, true);
Assert.AreEqual("ABC", a.ArticleText);
}
//TODO:Needs re-write
/// <summary>
/// If necessary, adds/removes wikify or stub tag
/// </summary>
/// <param name="articleText">The wiki text of the article.</param>
/// <param name="articleTitle">The article title.</param>
/// <param name="restrictOrphanTagging"></param>
/// <param name="summary"></param>
/// <returns>The tagged article.</returns>
public string Tagger(string articleText, string articleTitle, bool restrictOrphanTagging, ref string summary)
{
if(!TaggerPermitted(articleText, articleTitle))
return articleText;
tagsRemoved.Clear();
tagsAdded.Clear();
int tagsrenamed = 0;
string commentsStripped = WikiRegexes.Comments.Replace(articleText, "");
string commentsCategoriesStripped = WikiRegexes.Category.Replace(commentsStripped, "");
Sorter.Interwikis(ref commentsStripped);
// bulleted or indented text should weigh less than simple text.
// for example, actor stubs may contain large filmographies
string crapStripped = BulletedText.Replace(commentsCategoriesStripped, "");
int words = (Tools.WordCount(commentsCategoriesStripped, 999) + Tools.WordCount(crapStripped, 999)) / 2;
// remove stub tags from long articles, don't move section stubs
if ((words > StubMaxWordCount) && WikiRegexes.Stub.IsMatch(commentsStripped))
{
articleText = WikiRegexes.Stub.Replace(articleText, StubChecker).Trim();
if (Variables.LangCode.Equals("ar"))
{
tagsRemoved.Add("بذرة");
}
else if (Variables.LangCode.Equals("arz"))
{
tagsRemoved.Add("تقاوى");
}
else if (Variables.LangCode.Equals("hy"))
{
tagsRemoved.Add("Անավարտ");
}
else
{
tagsRemoved.Add("stub");
}
}
// refresh
commentsStripped = WikiRegexes.Comments.Replace(articleText, "");
commentsCategoriesStripped = WikiRegexes.Category.Replace(commentsStripped, "");
//remove disambiguation if disambiguation cleanup exists (en-wiki only)
if (Variables.LangCode.Equals("en") && WikiRegexes.DisambigsCleanup.IsMatch(commentsStripped))
{
articleText = WikiRegexes.DisambigsGeneral.Replace(articleText, "").Trim();
}
// do orphan tagging before template analysis for categorisation tags
articleText = TagOrphans(articleText, articleTitle, restrictOrphanTagging);
articleText = TagRefsIbid(articleText);
articleText = TagEmptySection(articleText);
int totalCategories;
// ignore commented out wikilinks, and any in {{Proposed deletion/dated}}
int wikiLinkCount = Tools.LinkCount(ProposedDeletionDatedEndorsed.Replace(commentsStripped, ""));
#if DEBUG || UNITTEST
if (Globals.UnitTestMode)
{
totalCategories = Globals.UnitTestIntValue;
}
else
#endif
{
// stubs add non-hidden stub categories, don't count these in categories count
// also don't count "Proposed deletion..." cats
// limitation: in the unlikely event that the article has only redlinked cats then it is {{uncat}} but we won't tag it as such
totalCategories = RegularCategories(articleText).Count;
// templates may add categories to page that are not [[Category...]] links, so use API call for accurate Category count
if(totalCategories == 0)
totalCategories = RegularCategories(CategoryProv.MakeList(new[] { articleTitle })).Count;
}
// remove dead end if > 0 wikilinks on page
if (wikiLinkCount > 0 && WikiRegexes.DeadEnd.IsMatch(articleText))
{
if (Variables.LangCode.Equals("ar") || Variables.LangCode.Equals("arz"))
articleText = WikiRegexes.DeadEnd.Replace(articleText, "");
else
articleText = WikiRegexes.DeadEnd.Replace(articleText, m => Tools.IsSectionOrReasonTemplate(m.Value, articleText) ? m.Value : m.Groups[1].Value).TrimStart();
if (!WikiRegexes.DeadEnd.IsMatch(articleText))
{
if (Variables.LangCode.Equals("ar"))
{
tagsRemoved.Add("نهاية مسدودة");
}
else if (Variables.LangCode.Equals("arz"))
{
tagsRemoved.Add("نهاية مسدودة");
}
else
{
tagsRemoved.Add("deadend");
}
}
}
// discount persondata, comments, infoboxes and categories from wikify/underlinked and stub evaluation
string lengthtext = WikiRegexes.Persondata.Replace(commentsCategoriesStripped, "");
lengthtext = WikiRegexes.InfoBox.Replace(lengthtext, "");
lengthtext = Drugbox.Replace(lengthtext, "");
int length = lengthtext.Length + 1;
bool underlinked = (wikiLinkCount < 0.0025 * length);
if (length <= 300 && !WikiRegexes.Stub.IsMatch(commentsCategoriesStripped) &&
!WikiRegexes.Disambigs.IsMatch(commentsCategoriesStripped) && !WikiRegexes.SIAs.IsMatch(commentsCategoriesStripped) && !WikiRegexes.NonDeadEndPageTemplates.IsMatch(commentsCategoriesStripped))
{
// add stub tag. Exclude pages their title starts with "List of..."
if (!ListOf.IsMatch(articleTitle) && !WikiRegexes.MeaningsOfMinorPlanetNames.IsMatch(articleTitle))
{
if (Variables.LangCode.Equals("ar"))
{
articleText += Tools.Newline("{{بذرة}}", 3);
tagsAdded.Add("بذرة");
}
else if (Variables.LangCode.Equals("arz"))
{
articleText += Tools.Newline("{{تقاوى}}", 3);
tagsAdded.Add("تقاوى");
}
else if (Variables.LangCode.Equals("hy"))
{
articleText += Tools.Newline("{{Անավարտ}}", 3);
tagsAdded.Add("Անավարտ");
}
else
{
articleText += Tools.Newline("{{stub}}", 3);
tagsAdded.Add("stub");
}
commentsStripped = WikiRegexes.Comments.Replace(articleText, "");
}
}
// rename existing {{improve categories}} else add uncategorized tag
if (totalCategories == 0 && ImproveCategories.IsMatch(articleText))
articleText = Tools.RenameTemplate(articleText, "improve categories", "Uncategorized");
// https://en.wikipedia.org/wiki/Wikipedia_talk:AutoWikiBrowser/Archive_19#AWB_problems
// nl wiki doesn't use {{Uncategorized}} template
// prevent wictionary redirects from being tagged as uncategorised
if (words > 6 && totalCategories == 0
&& !WikiRegexes.Uncat.IsMatch(articleText)
&& Variables.LangCode != "nl")
{
if (WikiRegexes.Stub.IsMatch(commentsStripped))
{
// add uncategorized stub tag
if (Variables.LangCode.Equals("ar"))
{
articleText += Tools.Newline("{{بذرة غير مصنفة|", 2) + WikiRegexes.DateYearMonthParameter + @"}}";
tagsAdded.Add("[[تصنيف:مقالات غير مصنفة|غير مصنفة]]");
}
else if (Variables.LangCode.Equals("arz"))
{
articleText += Tools.Newline("{{تقاوى مش متصنفه|", 2) + WikiRegexes.DateYearMonthParameter + @"}}";
tagsAdded.Add("[[قالب:تقاوى مش متصنفه|تقاوى مش متصنفه]]");
}
else if(Variables.LangCode.Equals("hy")) // same template for uncat and uncat stub
{
articleText += Tools.Newline("{{Կատեգորիա չկա|", 2) + WikiRegexes.DateYearMonthParameter + @"}}";
tagsAdded.Add("Կատեգորիա չկա");
}
else if(Variables.LangCode.Equals("sv")) // same template for uncat and uncat stub
{
articleText += Tools.Newline("{{Okategoriserad|", 2) + WikiRegexes.DateYearMonthParameter + @"}}";
tagsAdded.Add("[[Mall:Okategoriserad|okategoriserad]]");
}
else
{
articleText += Tools.Newline("{{Uncategorized stub|", 2) + WikiRegexes.DateYearMonthParameter + @"}}";
tagsAdded.Add("[[CAT:UNCATSTUBS|uncategorised]]");
}
}
else
{
if (Variables.LangCode.Equals("ar"))
{
articleText += Tools.Newline("{{غير مصنفة|", 2) + WikiRegexes.DateYearMonthParameter + @"}}";
tagsAdded.Add("[[CAT:UNCAT|مقالات غير مصنفة]]");
}
else if (Variables.LangCode.Equals("arz"))
{
articleText += Tools.Newline("{{مش متصنفه|", 2) + WikiRegexes.DateYearMonthParameter + @"}}";
tagsAdded.Add("[[CAT:UNCAT|مش متصنفه]]");
}
else if(Variables.LangCode.Equals("el"))
{
articleText += Tools.Newline("{{Ακατηγοριοποίητο|", 2) + WikiRegexes.DateYearMonthParameter + @"}}";
tagsAdded.Add("[[Πρότυπο:Ακατηγοριοποίητο|ακατηγοριοποίητο]]");
}
else if(Variables.LangCode.Equals("hy"))
{
articleText += Tools.Newline("{{Կատեգորիա չկա|", 2) + WikiRegexes.DateYearMonthParameter + @"}}";
tagsAdded.Add("Կատեգորիա չկա");
}
else if(Variables.LangCode.Equals("sv"))
{
articleText += Tools.Newline("{{Okategoriserad|", 2) + WikiRegexes.DateYearMonthParameter + @"}}";
tagsAdded.Add("[[Mall:Okategoriserad|okategoriserad]]");
}
else
{
articleText += Tools.Newline("{{Uncategorized|", 2) + WikiRegexes.DateYearMonthParameter + @"}}";
tagsAdded.Add("[[CAT:UNCAT|uncategorised]]");
}
}
}
// remove {{Uncategorized}} if > 0 real categories (stub categories not counted)
// rename {{Uncategorized}} to {{Uncategorized stub}} if stub with zero categories (stub categories not counted)
if (WikiRegexes.Uncat.IsMatch(articleText))
{
if (totalCategories > 0)
{
articleText = WikiRegexes.Uncat.Replace(articleText, "").TrimStart();
if (Variables.LangCode.Equals("ar"))
tagsRemoved.Add("غير مصنفة");
else if (Variables.LangCode.Equals("arz"))
tagsRemoved.Add("مش متصنفه");
else
tagsRemoved.Add("uncategorised");
}
else if (totalCategories == 0 && WikiRegexes.Stub.IsMatch(commentsStripped))
{
// rename uncat to uncat stub if no uncat stub. If uncat and uncat stub, remove uncat.
bool uncatstub = false;
foreach(Match u in WikiRegexes.Uncat.Matches(articleText))
{
if(WikiRegexes.Stub.IsMatch(u.Value))
{
uncatstub = true;
break;
}
}
articleText = WikiRegexes.Uncat.Replace(articleText, u2 => {
if (!uncatstub) // rename
{
tagsrenamed++;
if (Variables.LangCode.Equals("ar"))
return Tools.RenameTemplate(u2.Value, "بذرة غير مصنفة");
else if (Variables.LangCode.Equals("arz"))
return Tools.RenameTemplate(u2.Value, "تقاوى مش متصنفه");
else if (Variables.LangCode.Equals("en") || Variables.LangCode.Equals("simple"))
return Tools.RenameTemplate(u2.Value, "Uncategorized stub");
}
else // already uncat stub so remove plain uncat
{
if(!WikiRegexes.Stub.IsMatch(u2.Value))
{
if (Variables.LangCode.Equals("ar"))
tagsRemoved.Add("غير مصنفة");
else if (Variables.LangCode.Equals("arz"))
tagsRemoved.Add("مش متصنفه");
else
tagsRemoved.Add("uncategorised");
return "";
}
}
return u2.Value;
});
}
}
if (wikiLinkCount == 0 &&
!WikiRegexes.DeadEnd.IsMatch(articleText) &&
!WikiRegexes.SIAs.IsMatch(articleText) &&
!WikiRegexes.NonDeadEndPageTemplates.IsMatch(articleText) &&
!WikiRegexes.MeaningsOfMinorPlanetNames.IsMatch(articleTitle)
)
{
// add dead-end tag
// no blank line between dead end and orphan tags for ar/arz
if (Variables.LangCode.Equals("ar"))
{
articleText = "{{نهاية مسدودة|" + WikiRegexes.DateYearMonthParameter + "}}\r\n" + (WikiRegexes.Orphan.IsMatch(articleText) ? "" : "\r\n") + articleText;
tagsAdded.Add("[[:تصنيف:مقالات نهاية مسدودة|نهاية مسدودة]]");
// if dead end then remove underlinked/wikify
if(WikiRegexes.Wikify.IsMatch(articleText))
{
articleText = WikiRegexes.Wikify.Replace(articleText, "").TrimStart();
tagsRemoved.Add("ويكي");
}
}
else if (Variables.LangCode.Equals("arz"))
{
articleText = "{{نهايه مسدوده|" + WikiRegexes.DateYearMonthParameter + "}}\r\n" + articleText;
tagsAdded.Add("[[:قالب:نهايه مسدوده|نهايه مسدوده]]");
// if dead end then remove underlinked
if(WikiRegexes.Wikify.IsMatch(articleText))
{
articleText = WikiRegexes.Wikify.Replace(articleText, "").TrimStart();
tagsRemoved.Add("ويكى");
}
}
else if (Variables.LangCode != "sv" && !WikiRegexes.Centuryinbox.IsMatch(articleText)
&& !Regex.IsMatch(WikiRegexes.MultipleIssues.Match(articleText).Value.ToLower(), @"\bdead ?end\b")
&& !MinorPlanetListFooter.IsMatch(articleText))
{
// Don't add excess newlines between new tags
articleText = "{{Dead end|" + WikiRegexes.DateYearMonthParameter + "}}" + (tagsAdded.Count > 0 ? "\r\n" : "\r\n\r\n") + articleText;
tagsAdded.Add("[[CAT:DE|deadend]]");
// if dead end then remove underlinked
if(articleText.IndexOf("underlinked", StringComparison.OrdinalIgnoreCase) > -1)
{
articleText = Tools.NestedTemplateRegex("underlinked").Replace(articleText, "").TrimStart();
tagsRemoved.Add("underlinked");
}
}
}
// add underlinked/wikify tag, don't add underlinked/wikify if {{dead end}} already present
// Dont' tag SIA pages, may create wikilinks from templates
else if (wikiLinkCount < 3 && underlinked && !WikiRegexes.Wikify.IsMatch(articleText)
&& !WikiRegexes.MultipleIssues.Match(articleText).Value.ToLower().Contains("wikify")
&& !WikiRegexes.DeadEnd.IsMatch(articleText)
&& !WikiRegexes.SIAs.IsMatch(articleText)
&& !WikiRegexes.NonDeadEndPageTemplates.IsMatch(articleText)
&& !WikiRegexes.MeaningsOfMinorPlanetNames.IsMatch(articleTitle))
{
// Avoid excess newlines between templates
string templateEnd = "}}\r\n" + (articleText.TrimStart().StartsWith(@"{{") ? "" : "\r\n");
if (Variables.LangCode.Equals("ar"))
{
articleText = "{{ويكي|" + WikiRegexes.DateYearMonthParameter + templateEnd + articleText.TrimStart();
tagsAdded.Add("[[وب:ويكي|ويكي]]");
}
else if (Variables.LangCode.Equals("arz"))
{
articleText = "{{ويكى|" + WikiRegexes.DateYearMonthParameter + templateEnd + articleText;
tagsAdded.Add("[[قالب:ويكى|ويكى]]");
}
else if (Variables.LangCode.Equals("sv"))
{
articleText = "{{Ickewiki|" + WikiRegexes.DateYearMonthParameter + templateEnd + articleText;
tagsAdded.Add("[[WP:PW|ickewiki]]");
}
else
{
articleText = "{{Underlinked|" + WikiRegexes.DateYearMonthParameter + templateEnd + articleText;
tagsAdded.Add("[[CAT:UL|underlinked]]");
}
}
else if (wikiLinkCount > 3 && !underlinked &&
WikiRegexes.Wikify.IsMatch(articleText))
{
if (Variables.LangCode.Equals("ar") || Variables.LangCode.Equals("arz"))
articleText = WikiRegexes.Wikify.Replace(articleText, "");
else
// remove wikify, except section templates or wikify tags with reason parameter specified
articleText = WikiRegexes.Wikify.Replace(articleText, m => Tools.IsSectionOrReasonTemplate(m.Value, articleText) ? m.Value : m.Groups[1].Value).TrimStart();
if (!WikiRegexes.Wikify.IsMatch(articleText))
{
if (Variables.LangCode.Equals("ar"))
{
tagsRemoved.Add("ويكي");
}
else if (Variables.LangCode.Equals("arz"))
{
tagsRemoved.Add("ويكى");
}
else
{
tagsRemoved.Add("underlinked");
}
}
}
// rename unreferenced --> refimprove if has existing refs, update date
if (WikiRegexes.Unreferenced.IsMatch(commentsCategoriesStripped)
&& (TotalRefsNotGrouped(commentsCategoriesStripped) + Tools.NestedTemplateRegex("sfn").Matches(articleText).Count) > 0)
{
articleText = Unreferenced.Replace(articleText, m2 => Tools.UpdateTemplateParameterValue(Tools.RenameTemplate(m2.Value, "refimprove"), "date", "{{subst:CURRENTMONTHNAME}} {{subst:CURRENTYEAR}}"));
// update tag in old-style multiple issues
Match m = WikiRegexes.MultipleIssues.Match(articleText);
if (m.Success && Tools.GetTemplateParameterValue(m.Value, "unreferenced").Length > 0)
{
string newValue = Tools.RenameTemplateParameter(m.Value, "unreferenced", "refimprove");
newValue = Tools.UpdateTemplateParameterValue(newValue, "refimprove", "{{subst:CURRENTMONTHNAME}} {{subst:CURRENTYEAR}}");
if (!newValue.Equals(m.Value))
articleText = articleText.Replace(m.Value, newValue);
}
}
if (tagsAdded.Count > 0 || tagsRemoved.Count > 0 || tagsrenamed > 0)
{
Parsers p = new Parsers();
HideText hider = new HideText();
articleText = hider.HideUnformatted(articleText);
articleText = p.MultipleIssues(articleText);
articleText = Conversions(articleText);
articleText = hider.AddBackUnformatted(articleText);
// sort again in case tag removal requires whitespace cleanup
// Don't sort interwikis, we can't specify the correct InterWikiSortOrder
p.SortInterwikis = false;
articleText = p.Sorter.Sort(articleText, articleTitle);
}
summary = PrepareTaggerEditSummary();
return articleText;
}
public void Parser()
{
Parsers p1 = new Parsers();
Parsers p2 = new Parsers();
// https://en.wikipedia.org/wiki/Wikipedia_talk:AutoWikiBrowser/Bugs/Archive_10#NullReferenceException_in_HideText.AddBackMore
string s1 = p1.HideText("<pre>foo bar</pre>");
string s2 = p2.HideText("<source>quux</source>");
Assert.AreEqual("<pre>foo bar</pre>", p1.AddBackText(s1));
Assert.AreEqual("<source>quux</source>", p2.AddBackText(s2));
// in the future, we may use parser objects for processing several wikis at once
//Assert.AreNotEqual(p1.StubMaxWordCount, p2.StubMaxWordCount);
}
public MainForm()
{
splash.Show(this);
RightToLeft = System.Globalization.CultureInfo.CurrentCulture.TextInfo.IsRightToLeft
? RightToLeft.Yes : RightToLeft.No;
InitializeComponent();
Tools.RegistryMigration();
splash.SetProgress(5);
try
{
Icon = WikiFunctions.Properties.Resources.AWBIcon;
lblUserName.Alignment = ToolStripItemAlignment.Right;
lblProject.Alignment = ToolStripItemAlignment.Right;
lblTimer.Alignment = ToolStripItemAlignment.Right;
lblEditsPerMin.Alignment = ToolStripItemAlignment.Right;
lblIgnoredArticles.Alignment = ToolStripItemAlignment.Right;
lblEditCount.Alignment = ToolStripItemAlignment.Right;
btntsShowHide.Image = Resources.Showhide;
btntsShowHideParameters.Image = Resources.Showhideparameters;
btntsSave.Image = Resources.Save;
btntsIgnore.Image = Resources.RightArrow;
btntsStop.Image = Resources.Stop;
btntsPreview.Image = Resources.preview;
btntsChanges.Image = Resources.changes;
btntsFalsePositive.Image = Resources.RollBack;
btntsStart.Image = Resources.Run;
btntsDelete.Image = Resources.Vista_trashcan_empty;
splash.SetProgress(10);
int stubcount = 500;
bool catkey = false;
try
{
stubcount = AutoWikiBrowser.Properties.Settings.Default.StubMaxWordCount;
catkey = AutoWikiBrowser.Properties.Settings.Default.AddHummanKeyToCats;
parsers = new Parsers(stubcount, catkey);
}
catch (Exception ex)
{
parsers = new Parsers();
ErrorHandler.Handle(ex);
}
toolStripComboOnLoad.SelectedIndex = 0;
cmboCategorise.SelectedIndex = 0;
cmboImages.SelectedIndex = 0;
lblStatusText.AutoSize = true;
lblBotTimer.AutoSize = true;
Variables.User.UserNameChanged += UpdateUserName;
Variables.User.BotStatusChanged += UpdateBotStatus;
Variables.User.AdminStatusChanged += UpdateAdminStatus;
Variables.User.WikiStatusChanged += UpdateWikiStatus;
Variables.User.webBrowserLogin.DocumentCompleted += web4Completed;
Variables.User.webBrowserLogin.Navigating += web4Starting;
webBrowserEdit.Deleted += CaseWasDelete;
webBrowserEdit.Loaded += CaseWasLoad;
webBrowserEdit.Saved += CaseWasSaved;
webBrowserEdit.None += CaseWasNull;
webBrowserEdit.Fault += StartDelayedRestartTimer;
webBrowserEdit.StatusChanged += UpdateWebBrowserStatus;
listMaker.UserInputTextBox.ContextMenuStrip = mnuMakeFromTextBox;
listMaker.BusyStateChanged += SetProgressBar;
listMaker.NoOfArticlesChanged += UpdateButtons;
listMaker.StatusTextChanged += UpdateListStatus;
listMaker.cmboSourceSelect.SelectedIndexChanged += new EventHandler(ListMakerSourceSelectHandler);
profiles = new WikiFunctions.Profiles.AWBProfilesForm(webBrowserEdit);
profiles.LoadProfile += LoadProfileSettings;
splash.SetProgress(15);
WikiFunctions.Profiles.AWBProfiles.ResetTempPassword();
}
catch (Exception ex)
{
ErrorHandler.Handle(ex);
}
}
public MainForm()
{
DiffScriptingAdapter = new JsAdapter(this);
Updater.UpdateUpdaterFile();
SplashScreen.Show(this);
RightToLeft = System.Globalization.CultureInfo.CurrentCulture.TextInfo.IsRightToLeft
? RightToLeft.Yes : RightToLeft.No;
SplashScreen.SetProgress(1);
InitializeComponent();
SplashScreen.SetProgress(5);
try
{
btntsShowHide.Image = Resources.Showhide;
btntsShowHideParameters.Image = Resources.Showhideparameters;
btntsSave.Image = Resources.Save;
btntsIgnore.Image = Resources.RightArrow;
btntsStop.Image = Resources.Stop;
btntsPreview.Image = Resources.preview;
btntsChanges.Image = Resources.changes;
btntsFalsePositive.Image = Resources.RollBack;
btntsStart.Image = Resources.Run;
btntsDelete.Image = Resources.Vista_trashcan_empty;
SplashScreen.SetProgress(10);
try
{
Parser = new Parsers(Properties.Settings.Default.StubMaxWordCount,
Properties.Settings.Default.AddHummanKeyToCats);
}
catch (Exception ex)
{
Parser = new Parsers();
ErrorHandler.Handle(ex);
}
addToWatchList.SelectedIndex = 3;
cmboCategorise.SelectedIndex = 0;
cmboImages.SelectedIndex = 0;
listMaker.UserInputTextBox.ContextMenuStrip = mnuMakeFromTextBox;
listMaker.BusyStateChanged += SetProgressBar;
listMaker.NoOfArticlesChanged += UpdateButtons;
listMaker.StatusTextChanged += UpdateListStatus;
listMaker.cmboSourceSelect.SelectedIndexChanged += ListMakerSourceSelectHandler;
TheSession = new Session(this);
CreateEditor();
Profiles = new WikiFunctions.Profiles.AWBProfilesForm(TheSession);
Profiles.LoggedIn += ProfileLoggedIn;
SplashScreen.SetProgress(15);
_pasteMoreItems = new[]
{
PasteMore1, PasteMore2, PasteMore3, PasteMore4, PasteMore5, PasteMore6, PasteMore7, PasteMore8, PasteMore9, PasteMore10
};
// to avoid saving to app data
saveXML.InitialDirectory = openXML.InitialDirectory =
Environment.GetFolderPath(Environment.SpecialFolder.MyDocuments);
}
catch (Exception ex)
{
ErrorHandler.Handle(ex);
}
}
/// <summary>
/// Sets persondata date of birth/death fields based on unformatted info in zeroth section of article, provided dates match existing birth/death categories
/// </summary>
/// <param name="personData">Persondata template call</param>
/// <param name="articletext">The article text</param>
/// <returns>The updated persondata template call</returns>
private static string CompletePersonDataDate(string personData, string articletext)
{
// get the existing values
string existingBirthYear = Tools.GetTemplateParameterValue(personData, "DATE OF BIRTH", true);
string existingDeathYear = Tools.GetTemplateParameterValue(personData, "DATE OF DEATH", true);
if (existingBirthYear.Length == 4 || existingDeathYear.Length == 4)
{
Parsers p = new Parsers();
string birthDateFound = "", deathDateFound = "";
string zerothSection = Tools.GetZerothSection(articletext);
// remove references, wikilinks, templates
zerothSection = WikiRegexes.Refs.Replace(zerothSection, " ");
zerothSection = WikiRegexes.SimpleWikiLink.Replace(zerothSection, " ");
if (WikiRegexes.CircaTemplate.IsMatch(zerothSection))
zerothSection = zerothSection.Substring(0, WikiRegexes.CircaTemplate.Match(zerothSection).Index);
zerothSection = Tools.NestedTemplateRegex("ndash").Replace(zerothSection, " –");
zerothSection = WikiRegexes.NestedTemplates.Replace(zerothSection, " ");
// clean up any format errors in birth/death dates we may want to use
zerothSection = p.FixDatesAInternal(zerothSection);
// look for date in bracketed text, check date matches existing value (from categories)
foreach (Match m in BracketedBirthDeathDate.Matches(zerothSection))
{
string bValue = m.Value;
if (!UncertainWordings.IsMatch(bValue) && !ReignedRuledUnsure.IsMatch(bValue) && !FloruitTemplate.IsMatch(bValue))
{
string bBorn, bDied = "";
// split on died/spaced dash
if (FreeFormatDied.IsMatch(bValue))
{
bBorn = bValue.Substring(0, FreeFormatDied.Match(bValue).Index);
bDied = bValue.Substring(FreeFormatDied.Match(bValue).Index);
}
else
bBorn = bValue;
// born
if (existingBirthYear.Length == 4)
{
if (WikiRegexes.AmericanDates.Matches(bBorn).Count == 1 && WikiRegexes.AmericanDates.Match(bBorn).Value.Contains(existingBirthYear))
birthDateFound = WikiRegexes.AmericanDates.Match(bBorn).Value;
else if (WikiRegexes.InternationalDates.Matches(bBorn).Count == 1 && WikiRegexes.InternationalDates.Match(bBorn).Value.Contains(existingBirthYear))
birthDateFound = WikiRegexes.InternationalDates.Match(bBorn).Value;
}
// died
if (existingDeathYear.Length == 4)
{
if (WikiRegexes.AmericanDates.Matches(bDied).Count == 1 && WikiRegexes.AmericanDates.Match(bDied).Value.Contains(existingDeathYear))
deathDateFound = WikiRegexes.AmericanDates.Match(bDied).Value;
else if (WikiRegexes.InternationalDates.Matches(bDied).Count == 1 && WikiRegexes.InternationalDates.Match(bDied).Value.Contains(existingDeathYear))
deathDateFound = WikiRegexes.InternationalDates.Match(bDied).Value;
}
if (birthDateFound.Length > 0 || deathDateFound.Length > 0)
break;
}
}
if (birthDateFound.Length > 4)
personData = Tools.SetTemplateParameterValue(personData, "DATE OF BIRTH", Tools.ConvertDate(birthDateFound, DeterminePredominantDateLocale(articletext, true)), false);
if (deathDateFound.Length > 4)
personData = Tools.SetTemplateParameterValue(personData, "DATE OF DEATH", Tools.ConvertDate(deathDateFound, DeterminePredominantDateLocale(articletext, true)), false);
}
return personData;
}
public void TestFixNonBreakingSpacesDE()
{
#if DEBUG
Variables.SetProjectLangCode("fr");
parser = new Parsers();
Assert.AreEqual(@"a 50.247 µm laser", parser.FixNonBreakingSpaces(@"a 50.247µm laser"));
Assert.AreEqual(@"a 50.247um laser", parser.FixNonBreakingSpaces(@"a 50.247um laser"));
Variables.SetProjectLangCode("en");
parser = new Parsers();
Assert.AreEqual(@"a 50.247 um laser", parser.FixNonBreakingSpaces(@"a 50.247um laser"));
#endif
}
/// <summary>
/// Hides images, external links, templates, headings
/// </summary>
public string HideMore(string articleText, bool hideOnlyTargetOfWikilink)
{
MoreHide.Clear();
string articleTextBefore;
do
{ // hide nested templates
articleTextBefore = articleText;
List <Match> matches = Parsers.GetTemplates(articleText, Parsers.EveryTemplate);
ReplaceMore(matches, ref articleText);
}while (!articleTextBefore.Equals(articleText));
ReplaceMore(WikiRegexes.Blockquote.Matches(articleText), ref articleText);
ReplaceMore(WikiRegexes.Poem.Matches(articleText), ref articleText);
ReplaceMore(WikiRegexes.Source.Matches(articleText), ref articleText);
ReplaceMore(WikiRegexes.Code.Matches(articleText), ref articleText);
if (HideExternalLinks)
{
ReplaceMore(WikiRegexes.ExternalLinks.Matches(articleText), ref articleText);
}
ReplaceMore(WikiRegexes.Headings.Matches(articleText), ref articleText);
ReplaceMore(WikiRegexes.UnFormattedText.Matches(articleText), ref articleText);
ReplaceMore(WikiRegexes.IndentedText.Matches(articleText), ref articleText);
// This hides internal wikilinks (with or without pipe) with extra word character(s) e.g. [[link]]age, which need hiding even if hiding for typo fixing
// http://en.wikipedia.org/wiki/Wikipedia_talk:AutoWikiBrowser/Feature_requests#Improve_HideText.HideMore.28.29
// place this as first wikilink rule as otherwise WikiLinksOnly will grab link without extra word character(s)
ReplaceMore(WikiRegexes.WikiLinksOnlyPlusWord.Matches(articleText), ref articleText);
// if HideOnlyTargetOfWikilink is not set, pipes of links e.g. [[target|pipe]] will be hidden
// if set then don't mask the pipe of a link so that typo fixing can be done on it
if (!hideOnlyTargetOfWikilink)
{
ReplaceMore(WikiRegexes.WikiLinksOnly.Matches(articleText), ref articleText);
ReplaceMore(WikiRegexes.SimpleWikiLink.Matches(articleText), ref articleText);
}
ReplaceMore(WikiRegexes.Cites.Matches(articleText), ref articleText);
ReplaceMore(WikiRegexes.Refs.Matches(articleText), ref articleText);
// this hides only the target of a link, leaving the pipe exposed
ReplaceMore(WikiRegexes.WikiLink.Matches(articleText), ref articleText);
//TODO: replace with gallery-only regex, all normal images should be hidden by now as simple wikilinks
ReplaceMore(WikiRegexes.Images.Matches(articleText), ref articleText);
// hide untemplated quotes between some form of quotation marks (most particularly for typo fixing)
ReplaceMore(WikiRegexes.UntemplatedQuotes.Matches(articleText), ref articleText);
ReplaceMore(WikiRegexes.Pstyles.Matches(articleText), ref articleText);
return(articleText);
}