public List <Article> MakeList(params string[] searchCriteria)
{
if (Visible)
{
return(null);
}
txtPages.Text = "";
List <Article> list = new List <Article>();
if (ShowDialog() == DialogResult.OK)
{
searchCriteria = txtPages.Text.Split(new [] { '|' });
ISpecialPageProvider item = (ISpecialPageProvider)cmboSourceSelect.SelectedItem;
if (!string.IsNullOrEmpty(txtPages.Text))
{
list = item.MakeList(Namespace.Determine(cboNamespace.Text), searchCriteria);
}
else if (item.PagesNeeded)
{
MessageBox.Show("Pages needed!");
}
else
{
list = item.MakeList(Namespace.Determine(cboNamespace.Text), new[] { "" });
}
}
Hide();
return(list);
}
// Covered by: RecategorizerTests.Addition()
/// <summary>
/// Adds the category to the article.
/// </summary>
/// <param name="articleText">The wiki text of the article.</param>
/// <param name="newCategory">The new category.</param>
/// <param name="articleTitle">Title of the article</param>
/// <returns>The article text.</returns>
public string AddCategory(string newCategory, string articleText, string articleTitle)
{
string oldText = articleText;
articleText = FixCategories(articleText);
if (Regex.IsMatch(articleText, @"\[\["
+ Variables.NamespacesCaseInsensitive[Namespace.Category]
+ Regex.Escape(newCategory) + @"[\|\]]"))
{
return(oldText);
}
string cat = Tools.Newline("[[" + Variables.Namespaces[Namespace.Category] + newCategory + "]]");
cat = Tools.ApplyKeyWords(articleTitle, cat);
if (Namespace.Determine(articleTitle) == Namespace.Template)
{
articleText += "<noinclude>" + cat + Tools.Newline("</noinclude>");
}
else
{
articleText += cat;
}
return(SortMetaData(articleText, articleTitle, false)); // Sort metadata ordering so general fixes do not need to be enabled
}
public List <Article> MakeList(params string[] searchCriteria)
{
txtPages.Text = "";
List <Article> list = new List <Article>();
if (ShowDialog() == DialogResult.OK)
{
searchCriteria = txtPages.Text.Split(new [] { '|' });
ISpecialPageProvider item = (ISpecialPageProvider)cmboSourceSelect.SelectedItem;
//TODO: There must be a better way to do the NS determination??
if (!string.IsNullOrEmpty(txtPages.Text))
{
list = item.MakeList(Namespace.Determine(cboNamespace.Text), searchCriteria);
}
else if (item.PagesNeeded)
{
MessageBox.Show("Pages needed!");
}
else
{
list = item.MakeList(Namespace.Determine(cboNamespace.Text), new[] { "" });
}
}
return(Tools.FilterSomeArticles(list));
}
public void DetermineDeviations()
{
Assert.AreEqual(Namespace.File, Namespace.Determine("File : foo"));
Assert.AreEqual(Namespace.User, Namespace.Determine("user:foo"));
Assert.AreEqual(Namespace.UserTalk, Namespace.Determine("user_talk:foo"));
Assert.AreEqual(Namespace.UserTalk, Namespace.Determine("user%20talk:foo"));
}
public void Determine()
{
Assert.AreEqual(0, Namespace.Determine("test"));
Assert.AreEqual(0, Namespace.Determine(":test"));
Assert.AreEqual(0, Namespace.Determine("test:test"));
Assert.AreEqual(0, Namespace.Determine("My Project:Foo"));
Assert.AreEqual(0, Namespace.Determine("User:"******"Talk:foo"));
Assert.AreEqual(Namespace.UserTalk, Namespace.Determine("User talk:bar"));
Assert.AreEqual(Namespace.File, Namespace.Determine("File:foo"));
Assert.AreEqual(Namespace.File, Namespace.Determine("Image:foo"));
Assert.AreEqual(Namespace.Project, Namespace.Determine("Wikipedia:Foo"));
Assert.AreEqual(Namespace.Project, Namespace.Determine("Project:Foo"));
}
/// <summary>
/// Checks the XML returned by the server for error codes and throws an appropriate exception
/// </summary>
/// <param name="xml">Server output</param>
/// <param name="action">The action performed, null if don't check</param>
private XmlDocument CheckForErrors(string xml, string action)
{
if (string.IsNullOrEmpty(xml))
{
throw new ApiBlankException(this);
}
var doc = new XmlDocument();
try
{
doc.Load(new StringReader(xml));
}
catch (XmlException xe)
{
Tools.WriteDebug("ApiEdit::CheckForErrors", xml);
string postParams = "";
if (lastPostParameters != null)
{
int length = lastPostParameters.GetUpperBound(0);
for (int i = 0; i <= length; i++)
{
if (lastPostParameters[i, 0].Contains("password") || lastPostParameters[i, 0].Contains("token"))
{
lastPostParameters[i, 1] = "<removed>";
}
}
postParams = BuildQuery(lastPostParameters);
}
throw new ApiXmlException(this, xe, lastGetUrl, postParams, xml);
}
//TODO: can't figure out the best time for this check
bool prevMessages = User.HasMessages;
User.Update(doc);
if (action != "login" &&
action != "userinfo" &&
NewMessageThrows &&
User.HasMessages &&
!prevMessages)
{
throw new NewMessagesException(this);
}
var errors = doc.GetElementsByTagName("error");
if (errors.Count > 0)
{
var error = errors[0];
string errorCode = error.Attributes["code"].Value;
string errorMessage = error.Attributes["info"].Value;
switch (errorCode.ToLower())
{
case "maxlag": //guessing
int maxlag;
int.TryParse(MaxLag.Match(xml).Groups[1].Value, out maxlag);
throw new MaxlagException(this, maxlag, 10);
case "wrnotloggedin":
throw new LoggedOffException(this);
case "spamdetected":
throw new SpamlistException(this, errorMessage);
//case "confirmemail":
//
default:
if (errorCode.Contains("disabled"))
{
throw new FeatureDisabledException(this, errorCode, errorMessage);
}
throw new ApiErrorException(this, errorCode, errorMessage);
}
}
if (string.IsNullOrEmpty(action))
{
return(doc); // no action to check
}
var api = doc["api"];
if (api == null)
{
return(doc);
}
var redirects = api.GetElementsByTagName("r");
if (action == "query" && redirects.Count >= 1) //We have redirects
{
// Workaround for https://bugzilla.wikimedia.org/show_bug.cgi?id=39492
if (Namespace.IsSpecial(Namespace.Determine(redirects[redirects.Count - 1].Attributes["to"].Value)))
{
throw new RedirectToSpecialPageException(this);
}
}
//FIXME: Awful code is awful
var page = api.GetElementsByTagName("page");
if (page.Count > 0 && page[0].Attributes != null && page[0].Attributes["invalid"] != null &&
page[0].Attributes["invalid"].Value == "")
{
throw new InvalidTitleException(this, page[0].Attributes["title"].Value);
}
if (api.GetElementsByTagName("interwiki").Count > 0)
{
throw new InterwikiException(this);
}
var actionElement = api[action];
if (actionElement == null)
{
return(doc); // or shall we explode?
}
if (actionElement.HasAttribute("assert"))
{
string what = actionElement.GetAttribute("assert");
if (what == "user")
{
throw new LoggedOffException(this);
}
throw new AssertionFailedException(this, what);
}
if (actionElement.HasAttribute("spamblacklist"))
{
throw new SpamlistException(this, actionElement.GetAttribute("spamblacklist"));
}
if (actionElement.GetElementsByTagName("captcha").Count > 0)
{
throw new CaptchaException(this);
}
string result = actionElement.GetAttribute("result");
if (!string.IsNullOrEmpty(result) && result != "Success")
{
throw new OperationFailedException(this, action, result, xml);
}
return(doc);
}
public override bool Check(ArticleInfo article)
{
return(Namespaces.Contains(Namespace.Determine(article.Title)));
}
/// <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="fixExcessWhitespace">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 really want to be fooling around with 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));
// http://en.wikipedia.org/wiki/Wikipedia_talk:AutoWikiBrowser/Feature_requests#Move_orphan_tags_on_the_top
// Dablinks above orphan tags per [[WP:LAYOUT]]
if (Variables.LangCode == "en")
{
articleText = MoveMaintenanceTags(articleText);
}
articleText = MoveDablinks(articleText);
// http://en.wikipedia.org/wiki/Wikipedia_talk:AutoWikiBrowser/Feature_requests#Placement_of_portal_template
// http://en.wikipedia.org/wiki/Wikipedia_talk:AutoWikiBrowser/Feature_requests.html#Move_nofootnotes_to_the_references_section
if (Variables.LangCode == "en")
{
articleText = MovePortalTemplates(articleText);
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 == "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;
default:
articleText += personData + categories + strStub;
break;
}
return((articleText + interwikis).TrimEnd());
}
catch (Exception ex)
{
if (!ex.Message.Contains("DEFAULTSORT"))
{
ErrorHandler.Handle(ex);
}
return(strSave);
}
}
/// <summary>
/// Performs some cleanup operations on dablinks
/// Merges some for & about dablinks
/// Merges multiple distinguish into one
/// </summary>
/// <param name="articleText">The article text</param>
/// <returns>The updated article text</returns>
public static string Dablinks(string articleText)
{
if (!Variables.LangCode.Equals("en"))
{
return(articleText);
}
string zerothSection = Tools.GetZerothSection(articleText);
string restOfArticle = articleText.Substring(zerothSection.Length);
articleText = zerothSection;
// conversions
// otheruses4 rename - Wikipedia only
if (Variables.IsWikipediaEN)
{
articleText = Tools.RenameTemplate(articleText, "otheruses4", "about");
}
// "{{about|about x..." --> "{{about|x..."
foreach (Match m in Tools.NestedTemplateRegex("about").Matches(articleText))
{
if (m.Groups[3].Value.TrimStart("| ".ToCharArray()).StartsWith("about", StringComparison.OrdinalIgnoreCase))
{
articleText = articleText.Replace(m.Value, m.Groups[1].Value + m.Groups[2].Value + Regex.Replace(m.Groups[3].Value, @"^\|\s*[Aa]bout\s*", "|"));
}
}
// merging
// multiple same about into one
string oldArticleText = "";
while (oldArticleText != articleText)
{
oldArticleText = articleText;
bool doneAboutMerge = false;
foreach (Match m in Tools.NestedTemplateRegex("about").Matches(articleText))
{
string firstarg = Tools.GetTemplateArgument(m.Value, 1);
foreach (Match m2 in Tools.NestedTemplateRegex("about").Matches(articleText))
{
if (m2.Value == m.Value)
{
continue;
}
// match when reason is the same, not matching on self
if (Tools.GetTemplateArgument(m2.Value, 1).Equals(firstarg))
{
// argument 2 length > 0
if (Tools.GetTemplateArgument(m.Value, 2).Length > 0 && Tools.GetTemplateArgument(m2.Value, 2).Length > 0)
{
articleText = articleText.Replace(m.Value, m.Value.TrimEnd('}') + @"|" + Tools.GetTemplateArgument(m2.Value, 2) + @"|" + Tools.GetTemplateArgument(m2.Value, 3) + @"}}");
doneAboutMerge = true;
}
// argument 2 is null
if (Tools.GetTemplateArgument(m.Value, 2).Length == 0 && Tools.GetTemplateArgument(m2.Value, 2).Length == 0)
{
articleText = articleText.Replace(m.Value, m.Value.TrimEnd('}') + @"|and|" + Tools.GetTemplateArgument(m2.Value, 3) + @"}}");
doneAboutMerge = true;
}
}
// match when reason of one is null, the other not
else if (Tools.GetTemplateArgument(m2.Value, 1).Length == 0)
{
// argument 2 length > 0
if (Tools.GetTemplateArgument(m.Value, 2).Length > 0 && Tools.GetTemplateArgument(m2.Value, 2).Length > 0)
{
articleText = articleText.Replace(m.Value, m.Value.TrimEnd('}') + @"|" + Tools.GetTemplateArgument(m2.Value, 2) + @"|" + Tools.GetTemplateArgument(m2.Value, 3) + @"}}");
doneAboutMerge = true;
}
}
if (doneAboutMerge)
{
articleText = articleText.Replace(m2.Value, "");
break;
}
}
if (doneAboutMerge)
{
break;
}
}
}
// multiple for into about: rename a 2-argument for into an about with no reason value
if (Tools.NestedTemplateRegex("for").Matches(articleText).Count > 1 && Tools.NestedTemplateRegex("about").Matches(articleText).Count == 0)
{
foreach (Match m in Tools.NestedTemplateRegex("for").Matches(articleText))
{
if (Tools.GetTemplateArgument(m.Value, 3).Length == 0)
{
articleText = articleText.Replace(m.Value, Tools.RenameTemplate(m.Value, "about|"));
break;
}
}
}
// for into existing about, when about has >=2 arguments
if (Tools.NestedTemplateRegex("about").Matches(articleText).Count == 1 &&
Tools.GetTemplateArgument(Tools.NestedTemplateRegex("about").Match(articleText).Value, 2).Length > 0)
{
foreach (Match m in Tools.NestedTemplateRegex("for").Matches(articleText))
{
string about = Tools.NestedTemplateRegex("about").Match(articleText).Value;
// about supports up to 9 arguments
if (Tools.GetTemplateArgument(about, 9).Length > 0)
{
continue;
}
string extra = "";
// where about has 2 arguments need extra pipe
if (Tools.GetTemplateArgument(Tools.NestedTemplateRegex("about").Match(articleText).Value, 3).Length == 0 &&
Tools.GetTemplateArgument(Tools.NestedTemplateRegex("about").Match(articleText).Value, 4).Length == 0)
{
extra = @"|";
}
// append {{for}} value to the {{about}}
if (Tools.GetTemplateArgument(m.Value, 3).Length == 0)
{
articleText = articleText.Replace(about, about.TrimEnd('}') + extra + m.Groups[3].Value);
}
else if (Tools.GetTemplateArgument(m.Value, 4).Length == 0) // where for has 3 arguments need extra and
{
articleText = articleText.Replace(about, about.TrimEnd('}') + extra + m.Groups[3].Value.Insert(m.Groups[3].Value.LastIndexOf('|') + 1, "and|"));
}
// if there are 4 arguments do nothing
// remove the old {{for}}
if (Tools.GetTemplateArgument(m.Value, 4).Length == 0)
{
articleText = articleText.Replace(m.Value, "");
}
}
// if for with blank first argument copied over then now need to put "other uses" as the argment
articleText = Tools.NestedTemplateRegex("about").Replace(articleText, m2 => {
string res = m2.Value;
if (Tools.GetTemplateArgument(res, 7).Length > 0 && Tools.GetTemplateArgument(res, 6).Length == 0)
{
res = res.Insert(Tools.GetTemplateArgumentIndex(res, 6), "other uses");
}
return(res);
});
}
// non-mainspace links need escaping in {{about}}
foreach (Match m in Tools.NestedTemplateRegex("about").Matches(articleText))
{
string aboutcall = m.Value;
for (int a = 1; a <= Tools.GetTemplateArgumentCount(m.Value); a++)
{
string arg = Tools.GetTemplateArgument(aboutcall, a);
if (arg.Length > 0 && Namespace.Determine(arg) != Namespace.Mainspace)
{
aboutcall = aboutcall.Replace(arg, @":" + arg);
}
}
if (!m.Value.Equals(aboutcall))
{
articleText = articleText.Replace(m.Value, aboutcall);
}
}
// multiple {{distinguish}} into one
oldArticleText = "";
while (oldArticleText != articleText)
{
oldArticleText = articleText;
bool doneDistinguishMerge = false;
foreach (Match m in Tools.NestedTemplateRegex("distinguish").Matches(articleText))
{
foreach (Match m2 in Tools.NestedTemplateRegex("distinguish").Matches(articleText))
{
if (m2.Value.Equals(m.Value))
{
continue;
}
articleText = articleText.Replace(m.Value, m.Value.TrimEnd('}') + m2.Groups[3].Value);
doneDistinguishMerge = true;
articleText = articleText.Replace(m2.Value, "");
break;
}
if (doneDistinguishMerge)
{
break;
}
}
}
return(articleText + restOfArticle);
}
/// <summary>
///
/// </summary>
/// <param name="articleText"></param>
/// <param name="articleTitle"></param>
/// <returns></returns>
internal string Sort(string articleText, string articleTitle)
{
if (Namespace.Determine(articleTitle) == 10) //Dont really want to be fooling around with templates
{
return(articleText);
}
string strSave = articleText;
try
{
articleText = Regex.Replace(articleText, "<!-- ?\\[\\[en:.*?\\]\\] ?-->", "");
string strPersonData = Newline(RemovePersonData(ref articleText));
string strDisambig = Newline(RemoveDisambig(ref articleText));
string strCategories = Newline(RemoveCats(ref articleText, articleTitle));
string strInterwikis = Newline(Interwikis(ref articleText));
// http://en.wikipedia.org/wiki/Wikipedia_talk:AutoWikiBrowser/Feature_requests#Move_orphan_tags_on_the_top
// Dablinks above orphan tags per [[WP:LAYOUT]]
if (Variables.LangCode == LangCodeEnum.en)
{
articleText = MoveOrphanTags(articleText);
}
articleText = MoveDablinks(articleText);
// http://en.wikipedia.org/wiki/Wikipedia_talk:AutoWikiBrowser/Feature_requests#Placement_of_portal_template
// http://en.wikipedia.org/wiki/Wikipedia_talk:AutoWikiBrowser/Feature_requests.html#Move_nofootnotes_to_the_references_section
if (Variables.LangCode == LangCodeEnum.en)
{
articleText = MovePortalTemplates(articleText);
articleText = MoveMoreNoFootnotes(articleText);
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#Two_empty_lines_before_stub-templates
// Russian wiki uses only one newline
string strStub = Newline(RemoveStubs(ref articleText), Variables.LangCode == LangCodeEnum.ru ? 1 : 2);
//filter out excess white space and remove "----" from end of article
articleText = Parsers.RemoveWhiteSpace(articleText) + "\r\n";
articleText += strDisambig;
switch (Variables.LangCode)
{
case LangCodeEnum.de:
case LangCodeEnum.sl:
articleText += strStub + strCategories + strPersonData;
// 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 == LangCodeEnum.de && strPersonData.Length > 0 && strInterwikis.Length > 0)
{
articleText += "\r\n";
}
break;
case LangCodeEnum.pl:
case LangCodeEnum.ru:
case LangCodeEnum.simple:
articleText += strPersonData + strStub + strCategories;
break;
default:
articleText += strPersonData + strCategories + strStub;
break;
}
return(articleText + strInterwikis);
}
catch (Exception ex)
{
if (!ex.Message.Contains("DEFAULTSORT"))
{
ErrorHandler.Handle(ex);
}
return(strSave);
}
}
public override bool Check(ref string ArticleText, ref string ArticleTitle, string ArticleTimestamp, string ArticleRestrictions)
{
return(namespaces.Contains(Namespace.Determine(ArticleTitle)));
}
/// <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);
}
// short pages monitor check for en-wiki: keep at very end of article if present
// See [[Template:Long comment/doc]]
// SPM regex quick check for performance on long pages
string shortPagesMonitor = "";
if (Variables.LangCode.Equals("en") && SPMQuick.IsMatch(articleText))
{
Match spm = WikiRegexes.ShortPagesMonitor.Match(articleText);
if (spm.Success)
{
articleText = WikiRegexes.ShortPagesMonitor.Replace(articleText, "").TrimEnd();
shortPagesMonitor = spm.Value.TrimEnd();
}
}
articleText = CommentedOutEnInterwiki.Replace(articleText, "");
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));
if (Namespace.IsMainSpace(articleTitle))
{
// maintenance templates above infoboxes etc., zeroth section only
if (Variables.LangCode.Equals("en"))
{
string zerothSection = WikiRegexes.ZerothSection.Match(articleText).Value;
string restOfArticle = articleText.Substring(zerothSection.Length);
articleText = MoveMaintenanceTags(zerothSection) + restOfArticle;
}
// Dablinks above maintance tags per [[WP:LAYOUT]]
articleText = MoveDablinks(articleText);
if (Variables.LangCode.Equals("en"))
{
articleText = MovePortalTemplates(articleText);
articleText = MoveTemplateToSeeAlsoSection(articleText, WikiRegexes.WikipediaBooks);
articleText = MoveSisterlinks(articleText);
articleText = MoveTemplateToReferencesSection(articleText, WikiRegexes.Ibid);
articleText = MoveExternalLinks(articleText);
articleText = MoveSeeAlso(articleText);
}
}
// two newlines here per https://en.wikipedia.org/w/index.php?title=Wikipedia_talk:AutoWikiBrowser&oldid=243224092#Blank_lines_before_stubs
// https://en.wikipedia.org/wiki/Wikipedia_talk:AutoWikiBrowser/Bugs/Archive_11#Two_empty_lines_before_stub-templates
// ru, sl, ar, arz wikis use only one newline
string strStub = "";
// Category: can use {{Verylargestub}}/{{popstub}} which is not a stub template, don't do stub sorting
if (!Namespace.Determine(articleTitle).Equals(Namespace.Category))
{
strStub = Tools.Newline(RemoveStubs(ref articleText), (Variables.LangCode.Equals("ru") || Variables.LangCode.Equals("sl") || Variables.LangCode.Equals("ar") || Variables.LangCode.Equals("arz")) ? 1 : 2);
}
// filter out excess white space and remove "----" from end of article
articleText = Parsers.RemoveWhiteSpace(articleText, fixOptionalWhitespace) + "\r\n";
articleText += disambig;
articleText = WikiRegexes.MultipleIssues.Replace(articleText, m => Regex.Replace(m.Value, "(\r\n)+", "\r\n"));
switch (Variables.LangCode)
{
case "de":
case "sl":
articleText += strStub + categories + personData;
// https://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.Equals("de") && personData.Length > 0 && interwikis.Length > 0)
{
articleText += "\r\n";
}
break;
case "ar":
case "arz":
case "cs":
case "el":
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);
// Only trim start on Category namespace, restore any saved short page monitor text
return((Namespace.Determine(articleTitle) == Namespace.Category ? articleText.Trim() : articleText.TrimEnd()) + shortPagesMonitor);
}
// Covered by: LinkTests.TestSimplifyLinks()
/// <summary>
/// Simplifies some links in article wiki text such as changing [[Dog|Dogs]] to [[Dog]]s
/// Fixes CHECKWIKI error 64
/// </summary>
/// <param name="articleText">The wiki text of the article.</param>
/// <returns>The simplified article text.</returns>
public static string SimplifyLinks(string articleText)
{
// Performance: first get a list of unique links to avoid processing duplicate links more than once
List <string> pipedLinks = Tools.DeduplicateList(GetAllWikiLinks(articleText)).FindAll(link => link.Contains("|"));
// Performance: second determine if any links with pipe whitespace to clean
string categoryNS = Variables.Namespaces.ContainsKey(Namespace.Category) ? Variables.Namespaces[Namespace.Category] : "Category:";
bool whitespaceTrimNeeded = pipedLinks.Any(s => ((s.Contains("| ") && !s.Contains(categoryNS)) || s.Contains(" |") || (!s.Contains("| ]]") && s.Contains(" ]]"))));
foreach (string pipedlink in pipedLinks)
{
Match m = WikiRegexes.PipedWikiLink.Match(pipedlink);
// don't process if only matched part of link eg link is [[Image:...]] link with nested wikilinks
if (pipedlink.Length != m.Length)
{
continue;
}
string a = m.Groups[1].Value.Trim(), b = m.Groups[2].Value;
// Must retain space after pipe in Category namespace
if (whitespaceTrimNeeded)
{
b = (Namespace.Determine(a) != Namespace.Category)
? m.Groups[2].Value.Trim()
: m.Groups[2].Value.TrimEnd(new[] { ' ' });
if (b.Length == 0)
{
continue;
}
}
string lb = Tools.TurnFirstToLower(b), la = Tools.TurnFirstToLower(a);
if (pipedlink.IndexOfAny("&%_".ToCharArray()) > -1) // check for performance
{
string cb = CanonicalizeTitle(b), ca = CanonicalizeTitle(a);
if (cb.Equals(a) || cb.Equals(la) || ca.Equals(b) || ca.Equals(lb)) // target and text the same after cleanup and case conversion e.g. [[A|a]] or [[Foo_bar|Foo bar]] etc.
{
articleText = articleText.Replace(pipedlink, "[[" + b.Replace("_", " ") + "]]");
}
}
foreach (string punct in new List <string>(new [] { ".", "," }))
{
if (lb.Equals(la + punct))
{
articleText = articleText.Replace(pipedlink, "[[" + b.Substring(0, b.Length - 1) + "]]" + punct);
}
}
// [[dog|(dog)]] --> ([[dog]])
if (lb.Equals("(" + la + ")"))
{
articleText = articleText.Replace(pipedlink, "([[" + b.Substring(1, b.Length - 2) + "]])");
}
if (lb.StartsWith(la, StringComparison.Ordinal)) // target is substring of text e.g. [[Dog|Dogs]] --> [[Dog]]s
{
bool doBreak = false;
foreach (char ch in b.Remove(0, a.Length))
{
if (!char.IsLower(ch))
{
doBreak = true;
break;
}
}
if (doBreak)
{
continue;
}
articleText = articleText.Replace(pipedlink, "[[" + b.Substring(0, a.Length) + "]]" + b.Substring(a.Length));
}
else if (whitespaceTrimNeeded) // whitespace trimming around the pipe to apply
{
string newlink = "[[" + a + "|" + b + "]]";
// fix all pipe whitespace in file/image links
if (WikiRegexes.FileNamespaceLink.IsMatch(newlink))
{
newlink = Regex.Replace(newlink, @" *\| *", @"|");
}
if (newlink != pipedlink)
{
articleText = articleText.Replace(pipedlink, newlink);
}
}
}
return(articleText);
}
// Partially covered by FixMainArticleTests.SelfLinkRemoval()
/// <summary>
/// Fixes link syntax, including removal of self links.
/// Underscores not removed from link where page in [[Category:Articles with underscores in the title]]
/// </summary>
/// <param name="articleText">The wiki text of the article.</param>
/// <param name="articleTitle">Title of the article</param>
/// <param name="noChange">Value that indicated whether no change was made.</param>
/// <returns>The modified article text.</returns>
public static string FixLinks(string articleText, string articleTitle, out bool noChange)
{
string articleTextAtStart = articleText;
string escTitle = Regex.Escape(articleTitle);
if (TemplateExists(GetAllTemplates(articleText), InfoBoxSingleAlbum))
{
articleText = FixLinksInfoBoxSingleAlbum(articleText, articleTitle);
}
// clean up wikilinks: replace underscores, percentages and URL encoded accents etc.
List <string> wikiLinks = Tools.DeduplicateList(GetAllWikiLinks(articleText));
// Replace {{!}} with a standard pipe
if (GetAllTemplates(articleText).Contains("!"))
{
foreach (string e in wikiLinks.Where(l => l.Contains(@"{{!}}") && !l.Contains("|")))
{
articleText = articleText.Replace(e, e.Replace(@"{{!}}", "|"));
}
}
// See if any self interwikis that need fixing later
bool hasAnySelfInterwikis = wikiLinks.Any(l => l.Contains(Variables.LangCode + ":"));
// Performance: on articles with lots of links better to filter down to those that could be changed by canonicalization, rather than running regex replace against all links
foreach (string l in wikiLinks.Where(link => link.IndexOfAny("&%_".ToCharArray()) > -1))
{
string res = WikiRegexes.WikiLink.Replace(l, FixLinksWikilinkCanonicalizeME);
if (res != l)
{
articleText = articleText.Replace(l, res);
}
}
// First check for performance, second to avoid (dodgy) apostrophe after link
if (wikiLinks.Any(link => link.Contains("|''")) && !articleText.Contains(@"']]'"))
{
articleText = WikiRegexes.PipedWikiLink.Replace(articleText, FixLinksWikilinkBoldItalicsME);
}
// fix excess trailing pipe, TrailingPipe regex for performance
if (wikiLinks.Any(link => link.Contains("|") && TrailingPipe.IsMatch(link)))
{
articleText = WikiRegexes.PipedWikiLink.Replace(articleText, m => (m.Groups[2].Value.Trim().EndsWith("|") ? "[[" + m.Groups[1].Value + "|" + m.Groups[2].Value.Trim().TrimEnd('|').Trim() + "]]" : m.Value));
}
// fix excess leading pipe in piped link e.g. [[|foo|bar]], avoid malformatted image links
foreach (string pl in wikiLinks.Where(link => link.Substring(2, 1).Equals("|") && link.Substring(3).Contains("|") &&
!link.Substring(3).StartsWith("thumb")))
{
articleText = articleText.Replace(pl, @"[[" + pl.Substring(3).TrimStart());
}
// https://en.wikipedia.org/wiki/Wikipedia_talk:AutoWikiBrowser/Bugs/Archive_11#Your_code_creates_page_errors_inside_imagemap_tags.
// don't apply if there's an imagemap on the page or some noinclude transclusion business
// https://en.wikipedia.org/wiki/Wikipedia_talk:AutoWikiBrowser/Bugs/Archive_11#Includes_and_selflinks
// TODO, better to not apply to text within imagemaps
if (Regex.IsMatch(articleText, @"\[\[\s*(" + Tools.CaseInsensitive(escTitle) + @")\s*(?:\]|\|)") &&
!WikiRegexes.ImageMap.IsMatch(articleText) &&
!WikiRegexes.IncludeonlyNoinclude.IsMatch(articleText) &&
!TaxoboxColour.IsMatch(articleText))
{
// remove any self-links, but not other links with different capitaliastion e.g. [[Foo]] vs [[FOO]]
articleText = Regex.Replace(articleText, @"\[\[\s*(" + Tools.CaseInsensitive(escTitle)
+ @")\s*\]\]", "$1");
// remove piped self links by leaving target, or removing when is a category with self category listed
articleText = Regex.Replace(articleText, @"\[\[\s*" + Tools.CaseInsensitive(escTitle)
+ @"\s*\|\s*([^\]]+)\s*\]\]", (Namespace.Determine(articleTitle) == Namespace.Category ? "" : "$1"));
}
// fix for self interwiki links, not for monolingual projects
if (hasAnySelfInterwikis && !Variables.IsWikimediaMonolingualProject)
{
articleText = FixSelfInterwikis(articleText);
}
noChange = articleText.Equals(articleTextAtStart);
return(articleText);
}
/// <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 = "";
// Category: can use {{Verylargestub}}/{{popstub}} which is not a stub template, don't do stub sorting
if (!Namespace.Determine(articleTitle).Equals(Namespace.Category))
{
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.Equals("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);
}
}
/// <summary>
///
/// </summary>
/// <param name="ArticleText"></param>
/// <param name="ArticleTitle"></param>
/// <returns></returns>
internal string Sort(string ArticleText, string ArticleTitle)
{
if (Namespace.Determine(ArticleTitle) == 10) //Dont really want to be fooling around with templates
{
return(ArticleText);
}
string strSave = ArticleText;
try
{
ArticleText = Regex.Replace(ArticleText, "<!-- ?\\[\\[en:.*?\\]\\] ?-->", "");
string strPersonData = Newline(removePersonData(ref ArticleText));
string strDisambig = Newline(removeDisambig(ref ArticleText));
string strCategories = Newline(removeCats(ref ArticleText, ArticleTitle));
string strInterwikis = Newline(interwikis(ref ArticleText));
ArticleText = moveDablinks(ArticleText);
// two newlines here per http://en.wikipedia.org/w/index.php?title=Wikipedia_talk:AutoWikiBrowser&oldid=243224092#Blank_lines_before_stubs
string strStub = Newline(removeStubs(ref ArticleText), 2);
//filter out excess white space and remove "----" from end of article
ArticleText = Parsers.RemoveWhiteSpace(ArticleText) + "\r\n";
ArticleText += strDisambig;
switch (Variables.LangCode)
{
case LangCodeEnum.de:
case LangCodeEnum.sl:
ArticleText += strStub + strCategories + strPersonData;
// 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 == LangCodeEnum.de && strPersonData.Length > 0 && strInterwikis.Length > 0)
{
ArticleText += "\r\n";
}
break;
case LangCodeEnum.pl:
case LangCodeEnum.ru:
case LangCodeEnum.simple:
ArticleText += strPersonData + strStub + strCategories;
break;
default:
ArticleText += strPersonData + strCategories + strStub;
break;
}
return(ArticleText + strInterwikis);
}
catch (Exception ex)
{
if (!ex.Message.Contains("DEFAULTSORT"))
{
ErrorHandler.Handle(ex);
}
return(strSave);
}
}
