/// <summary>
/// Fully processes a page, applying all needed changes
/// </summary>
/// <param name="theArticle">Page to process</param>
/// <param name="mainProcess">True if the page is being processed for save as usual,
/// otherwise (Re-parse in context menu, prefetch, etc) false</param>
private void ProcessPage(Article theArticle, bool mainProcess)
{
bool process = true;
TypoStats = null;
#if DEBUG
Variables.Profiler.Start("ProcessPage(\"" + theArticle.Name + "\")");
#endif
try
{
// Must be performed regardless of general fixes, otherwise there may be breakage
theArticle.AWBChangeArticleText("Fixes for Unicode compatibility",
Parser.FixUnicode(theArticle.ArticleText),
true);
if (NoParse.Contains(theArticle.Name))
process = false;
if (!IgnoreNoBots &&
!Parsers.CheckNoBots(theArticle.ArticleText, TheSession.User.Name))
{
theArticle.AWBSkip("Restricted by {{bots}}/{{nobots}}");
return;
}
Variables.Profiler.Profile("Initial skip checks");
if (CModule.ModuleUsable)
{
theArticle.SendPageToCustomModule(CModule.Module);
if (theArticle.SkipArticle) return;
}
Variables.Profiler.Profile("Custom module");
if (ExtProgram.ModuleEnabled)
{
theArticle.SendPageToCustomModule(ExtProgram);
if (theArticle.SkipArticle) return;
}
Variables.Profiler.Profile("External Program");
if (Plugin.AWBPlugins.Count > 0)
{
foreach (KeyValuePair<string, IAWBPlugin> a in Plugin.AWBPlugins)
{
theArticle.SendPageToPlugin(a.Value, this);
if (theArticle.SkipArticle) return;
}
}
Variables.Profiler.Profile("Plugins");
// unicodify whole article
if (process && chkUnicodifyWhole.Checked)
{
theArticle.HideMoreText(RemoveText);
Variables.Profiler.Profile("HideMoreText");
theArticle.Unicodify(Skip.SkipNoUnicode, Parser);
Variables.Profiler.Profile("Unicodify");
theArticle.UnHideMoreText(RemoveText);
Variables.Profiler.Profile("UnHideMoreText");
}
// find and replace before general fixes
if (chkFindandReplace.Checked)
{
theArticle.PerformFindAndReplace(FindAndReplace, SubstTemplates, RplcSpecial,
chkSkipWhenNoFAR.Checked, chkSkipOnlyMinorFaR.Checked, false);
Variables.Profiler.Profile("F&R");
theArticle.DoFaRSkips(FindAndReplace);
if (theArticle.SkipArticle)
return;
}
// RegexTypoFix
if (chkRegExTypo.Checked && RegexTypos != null && !BotMode && !Namespace.IsTalk(theArticle.NameSpaceKey))
{
if (!NoRetf.Contains(theArticle.Name))
{
theArticle.PerformTypoFixes(RegexTypos, chkSkipIfNoRegexTypo.Checked);
Variables.Profiler.Profile("Typos");
TypoStats = RegexTypos.GetStatistics();
}
else if (chkSkipIfNoRegexTypo.Checked)
TheArticle.Trace.AWBSkipped("No typo fixes (Title blacklisted from RegExTypoFix Typo Fixing)");
if (theArticle.SkipArticle)
{
if (mainProcess)
{
// update stats only if not called from e.g. 'Re-parse' than could be clicked repeatedly
OverallTypoStats.UpdateStats(TypoStats, true);
UpdateTypoCount();
}
}
}
// replace/add/remove categories
if (cmboCategorise.SelectedIndex != 0)
{
theArticle.Categorisation((WikiFunctions.Options.CategorisationOptions)
cmboCategorise.SelectedIndex, Parser, chkSkipNoCatChange.Checked,
txtNewCategory.Text.Trim(),
txtNewCategory2.Text.Trim(), chkRemoveSortKey.Checked);
if (theArticle.SkipArticle)
return;
else if (!chkGeneralFixes.Checked)
theArticle.AWBChangeArticleText("Fix categories",
Parsers.FixCategories(theArticle.ArticleText), true);
}
Variables.Profiler.Profile("Categories");
if (process)
{
if (chkGeneralFixes.Checked)
{
theArticle.PerformUniversalGeneralFixes();
Variables.Profiler.Profile("Universal Genfixes");
}
if (theArticle.CanDoGeneralFixes)
{
if (chkGeneralFixes.Checked)
{
if (!TemplateRedirectsLoaded)
{
LoadTemplateRedirects();
Variables.Profiler.Profile("LoadTemplateRedirects");
}
if (!DatedTemplatesLoaded)
{
LoadDatedTemplates();
Variables.Profiler.Profile("LoadDatedTemplates");
}
if (!RenamedTemplateParametersLoaded)
{
LoadRenameTemplateParameters();
Variables.Profiler.Profile("LoadRenameTemplateParameters");
}
theArticle.PerformGeneralFixes(Parser, RemoveText, Skip,
replaceReferenceTagsToolStripMenuItem.Checked,
restrictDefaultsortChangesToolStripMenuItem.Checked,
noMOSComplianceFixesToolStripMenuItem.Checked);
}
Variables.Profiler.Profile("Mainspace Genfixes");
// auto tag
if (chkAutoTagger.Checked)
{
theArticle.AutoTag(Parser, Skip.SkipNoTag, restrictOrphanTaggingToolStripMenuItem.Checked);
if (mainProcess && theArticle.SkipArticle)
return;
}
Variables.Profiler.Profile("Auto-tagger");
}
else if (chkGeneralFixes.Checked)
{
if (theArticle.NameSpaceKey == Namespace.UserTalk)
{
if (!UserTalkWarningsLoaded)
{
LoadUserTalkWarnings();
Variables.Profiler.Profile("loadUserTalkWarnings");
}
theArticle.PerformUserTalkGeneralFixes(RemoveText, UserTalkTemplatesRegex,
Skip.SkipNoUserTalkTemplatesSubstd);
}
else if (theArticle.CanDoTalkGeneralFixes)
{
theArticle.PerformTalkGeneralFixes(RemoveText);
}
Variables.Profiler.Profile("Talk Genfixes");
}
}
// find and replace after general fixes
if (chkFindandReplace.Checked)
{
theArticle.PerformFindAndReplace(FindAndReplace, SubstTemplates, RplcSpecial,
chkSkipWhenNoFAR.Checked, chkSkipOnlyMinorFaR.Checked, true);
theArticle.DoFaRSkips(FindAndReplace);
Variables.Profiler.Profile("F&R (2nd)");
if (theArticle.SkipArticle) return;
}
// append/prepend text
if (chkAppend.Checked)
{
// customized number of newlines
string newlines = "";
for (int i = 0; i < (int)udNewlineChars.Value; i++)
newlines += "\n";
if (rdoAppend.Checked)
theArticle.AWBChangeArticleText("Appended your message",
theArticle.ArticleText + newlines + Tools.ApplyKeyWords(theArticle.Name, txtAppendMessage.Text), false);
else
theArticle.AWBChangeArticleText("Prepended your message",
Tools.ApplyKeyWords(theArticle.Name, txtAppendMessage.Text) + newlines + theArticle.ArticleText, false);
if (chkAppendMetaDataSort.Checked)
theArticle.PerformMetaDataSort(Parser);
}
Variables.Profiler.Profile("Append Text");
// replace/remove/comment out images
if (cmboImages.SelectedIndex != 0)
{
theArticle.UpdateImages((WikiFunctions.Options.ImageReplaceOptions)cmboImages.SelectedIndex,
txtImageReplace.Text, txtImageWith.Text, chkSkipNoImgChange.Checked);
if (theArticle.SkipArticle)
return;
}
Variables.Profiler.Profile("Images");
// disambiguation
if (!preParseModeToolStripMenuItem.Checked && chkEnableDab.Checked && txtDabLink.Text.Trim().Length > 0 &&
txtDabVariants.Text.Trim().Length > 0)
{
if (theArticle.Disambiguate(TheSession, txtDabLink.Text.Trim(), txtDabVariants.Lines, BotMode,
(int)udContextChars.Value, chkSkipNoDab.Checked))
{
if (theArticle.SkipArticle)
return;
}
else
{
Abort = true;
Stop();
return;
}
}
Variables.Profiler.Profile("Disambiguate");
}
catch (Exception ex)
{
ErrorHandler.Handle(ex);
// don't remove page over regex error – page itself is not at fault
if (!ex.StackTrace.Contains("System.Text.RegularExpressions"))
theArticle.Trace.AWBSkipped("Exception:" + ex.Message);
else
Skippable = false;
Stop();
StopDelayedAutoSaveTimer();
}
finally
{
Variables.Profiler.Flush();
}
}
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");
}