| private FixDatesAInternal ( string textPortion ) : string | ||
| textPortion | string | |
| return | string | |
/// <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;
}
/// <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);
}
/// <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;
}
/// <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);
// fix cite params not in lower case, allowing for ISBN, DOI identifiers being uppercase, avoiding changing text within malformatted URL
foreach (
string notlowercaseCiteParam in
paramsFound.Keys
.Where(
p =>
(p.ToLower() != p) &&
!Regex.IsMatch(p,
@"(?:IS[BS]N|DOI|PMID|OCLC|PMC|LCCN|ASIN|ARXIV|ASIN\-TLD|BIBCODE|ID|ISBN13|JFM|JSTOR|MR|OL|OSTI|RFC|SSRN|URL|ZBL)")
&& !CiteUrl.Match(newValue).Value.Contains(p)))
{
newValue = Tools.RenameTemplateParameter(newValue, notlowercaseCiteParam,
notlowercaseCiteParam.ToLower());
}
string theURL,
id,
format,
theTitle,
TheYear,
lang,
TheDate,
TheMonth,
TheWork,
nopp,
TheIssue,
TheVolume,
accessdate,
pages,
page,
ISBN,
ISSN,
origyear,
origdate,
archiveurl,
contributionurl;
if (!paramsFound.TryGetValue("url", out theURL))
theURL = "";
if (!paramsFound.TryGetValue("id", out id) && !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("volume", out TheVolume))
TheVolume = "";
if (!paramsFound.TryGetValue("accessdate", out accessdate) &&
!paramsFound.TryGetValue("access-date", 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("origdate", out origdate))
origdate = "";
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 = "";
if (!paramsFound.TryGetValue("issn", out ISSN) && !paramsFound.TryGetValue("ISSN", out ISSN))
ISSN = "";
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");
}
if (paramsFound.ContainsKey("origdate") && origdate.Length == 0)
{
newValue = Tools.RemoveTemplateParameter(newValue, "origdate");
}
// newlines to spaces in all parameters
foreach(KeyValuePair<string, string> newlines in paramsFound.Where(p => p.Value.Contains("\r\n")))
{
newValue = Tools.UpdateTemplateParameterValue(newValue, newlines.Key, newlines.Value.Replace("\r\n", " "));
}
// {{sv icon}} -> sv in language=
if (lang.Contains("{{"))
{
newValue = LangTemplate.Replace(newValue, "$1$3");
lang = Tools.GetTemplateParameterValue(newValue, "language");
}
// remove italics for work field for book/periodical, but not website -- auto italicised by template
if (TheWork.Contains("''") && !TheWork.Contains("."))
newValue = WorkInItalics.Replace(newValue, "$1$2");
// format quotes in title fields, remove stray quotes
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""");
// trim stray quotes (but don't change title in quotes as this may be a title that is itself a quote)
if(!quotetitle.Trim('"').Contains(@""""))
{
if (quotetitle.StartsWith(@"""") && !quotetitle.EndsWith(@""""))
quotetitle = quotetitle.TrimStart('"');
else if (quotetitle.EndsWith(@"""") && !quotetitle.StartsWith(@""""))
quotetitle = quotetitle.TrimEnd('"');
}
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");
paramsFound.Remove("pages");
paramsFound.Add("pages", pages);
}
// with Lua no need to rename date to year when date = YYYY, just remove year and date duplicating each other
if (TheDate.Length == 4 && 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 (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
else 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) Or 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))
{
if(TheVolume.Length > 0)
newValue = CiteTemplatesJournalVolume.Replace(newValue, "");
if (TheIssue.Length > 0)
newValue = CiteTemplatesJournalIssue.Replace(newValue, "");
else
newValue = CiteTemplatesJournalVolumeAndIssue.Replace(newValue, @"| issue = ");
}
// {{cite web}} for Google books -> {{Cite book}}
else 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(paramsFound.Any(p => p.Key.Contains("date") && Regex.IsMatch(p.Value, @"\b0[1-9]")))
{
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 (paramsFound.Any(s => s.Key.Contains("access") && !s.Key.Contains("date")))
{
string accessyear;
if (!paramsFound.TryGetValue("accessyear", out accessyear))
accessyear = "";
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 (pages.Contains(",") && !Regex.IsMatch(pages, @"\b[0-9]{1,2},[0-9]{3}\b"))
{
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 (page.Length > 0 && 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
if(!IncorrectCommaAmericanDates.IsMatch(theURLoriginal))
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]]
Match deadLinkMatch = WikiRegexes.DeadLink.Match(format);
if (deadLinkMatch.Success)
{
string deadLink = deadLinkMatch.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);
}
if(id.Length > 0)
{
// get id param name, id or ID
string idParamName = paramsFound.FirstOrDefault(p => p.Key == "ID" || p.Key == "id").Key;
//id=ISBN fix
if (IdISBN.IsMatch(id) && ISBN.Length == 0)
{
newValue = Tools.RenameTemplateParameter(newValue, idParamName, "isbn");
newValue = Tools.SetTemplateParameterValue(newValue, "isbn", IdISBN.Match(id).Groups[1].Value.Trim());
}
//id=ASIN fix
if (IdASIN.IsMatch(id) && Tools.GetTemplateParameterValue(newValue, "asin", true).Length == 0)
{
newValue = Tools.RenameTemplateParameter(newValue, idParamName, "asin");
newValue = Tools.SetTemplateParameterValue(newValue, "asin", IdASIN.Match(id).Groups[1].Value.Trim());
}
//id=ISSN fix
Match IdISSNMatch = IdISSN.Match(id);
if (IdISSNMatch.Success && ISSN.Length == 0)
{
string newIssn = IdISSNMatch.Groups[1].Value + "-" + IdISSNMatch.Groups[2].Value; // 1234-5678 using standard hyphen
newValue = Tools.RenameTemplateParameter(newValue, idParamName, "issn");
newValue = Tools.SetTemplateParameterValue(newValue, "issn", newIssn);
}
}
// format ISSN: 1234-5678 with hyphen
if(ISSN.Length > 0)
{
string newISSN = Regex.Replace (ISSN, @"^([0-9]{4}) *[- –]* *([0-9]{3}[0-9X])$", "$1-$2");
if (!newISSN.Equals (ISSN))
newValue = Tools.UpdateTemplateParameterValue (newValue, paramsFound.FirstOrDefault (p => p.Key == "ISSN" || p.Key == "issn").Key, newISSN);
}
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 (origyear.Length == 4 && TheYear.Length == 0 && TheDate.Length == 0)
{
newValue = Tools.RenameTemplateParameter(newValue, "origyear", "year");
newValue = Tools.RemoveDuplicateTemplateParameters(newValue);
}
return newValue;
}
