5 // Jackson Harper (jackson@ximian.com)
6 // Atsushi Enomoto (atsushi@ximian.com)
7 // Marek Safar <marek.safar@gmail.com>
9 // (C) 2004-2005 Novell, Inc (http://www.novell.com)
10 // Copyright (C) 2012 Xamarin Inc (http://www.xamarin.com)
12 // Permission is hereby granted, free of charge, to any person obtaining
13 // a copy of this software and associated documentation files (the
14 // "Software"), to deal in the Software without restriction, including
15 // without limitation the rights to use, copy, modify, merge, publish,
16 // distribute, sublicense, and/or sell copies of the Software, and to
17 // permit persons to whom the Software is furnished to do so, subject to
18 // the following conditions:
20 // The above copyright notice and this permission notice shall be
21 // included in all copies or substantial portions of the Software.
23 // THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
24 // EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
25 // MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
26 // NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE
27 // LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION
28 // OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION
29 // WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
36 using System.Globalization;
37 using System.Text.RegularExpressions;
38 using System.Collections.Generic;
41 namespace Mono.Tools.LocaleBuilder
45 static readonly string data_root = Path.Combine ("CLDR", "common");
47 public static void Main (string[] args)
49 Driver d = new Driver ();
54 private static void ParseArgs (string[] args, Driver d)
56 for (int i = 0; i < args.Length; i++) {
57 if (args[i] == "--lang" && i + 1 < args.Length)
59 else if (args[i] == "--locales" && i + 1 < args.Length)
60 d.Locales = args[++i];
61 else if (args[i] == "--header" && i + 1 < args.Length)
62 d.HeaderFileName = args[++i];
63 else if (args[i] == "--compare")
64 d.OutputCompare = true;
69 private string locales;
70 private string header_name;
71 List<CultureInfoEntry> cultures;
72 Dictionary<string, string> region_currency;
73 Dictionary<string, string> currency_fractions;
74 Dictionary<string, string> extra_parent_locales;
76 // The lang is the language that display names will be displayed in
90 get { return locales; }
91 set { locales = value; }
94 public string HeaderFileName
98 if (header_name == null)
99 return "culture-info-tables.h";
102 set { header_name = value; }
105 public bool OutputCompare { get; set; }
109 cultures.Sort ((a, b) => int.Parse (a.LCID.Substring (2), NumberStyles.HexNumber).CompareTo (int.Parse (b.LCID.Substring (2), NumberStyles.HexNumber)));
111 var writer = Console.Out;
113 foreach (var c in cultures) {
114 writer.WriteLine ("Name: {0}, LCID {1}", c.OriginalName, c.LCID);
116 writer.WriteLine ("{0}: {1}", "DisplayName", c.DisplayName);
117 writer.WriteLine ("{0}: {1}", "EnglishName", c.EnglishName);
118 writer.WriteLine ("{0}: {1}", "NativeName", c.NativeName);
119 // writer.WriteLine ("{0}: {1}", "OptionalCalendars", c.OptionalCalendars);
120 writer.WriteLine ("{0}: {1}", "ThreeLetterISOLanguageName", c.ThreeLetterISOLanguageName);
121 writer.WriteLine ("{0}: {1}", "ThreeLetterWindowsLanguageName", c.ThreeLetterWindowsLanguageName);
122 writer.WriteLine ("{0}: {1}", "TwoLetterISOLanguageName", c.TwoLetterISOLanguageName);
123 writer.WriteLine ("{0}: {1}", "Calendar", GetCalendarType (c.CalendarType));
125 var df = c.DateTimeFormatEntry;
126 writer.WriteLine ("-- DateTimeFormat --");
127 Dump (writer, df.AbbreviatedDayNames, "AbbreviatedDayNames");
128 Dump (writer, df.AbbreviatedMonthGenitiveNames, "AbbreviatedMonthGenitiveNames");
129 Dump (writer, df.AbbreviatedMonthNames, "AbbreviatedMonthNames");
130 writer.WriteLine ("{0}: {1}", "AMDesignator", df.AMDesignator);
131 writer.WriteLine ("{0}: {1}", "CalendarWeekRule", (CalendarWeekRule) df.CalendarWeekRule);
132 writer.WriteLine ("{0}: {1}", "DateSeparator", df.DateSeparator);
133 Dump (writer, df.DayNames, "DayNames");
134 writer.WriteLine ("{0}: {1}", "FirstDayOfWeek", (DayOfWeek) df.FirstDayOfWeek);
135 // Dump (writer, df.GetAllDateTimePatterns (), "GetAllDateTimePatterns");
136 writer.WriteLine ("{0}: {1}", "LongDatePattern", df.LongDatePattern);
137 writer.WriteLine ("{0}: {1}", "LongTimePattern", df.LongTimePattern);
138 writer.WriteLine ("{0}: {1}", "MonthDayPattern", df.MonthDayPattern);
139 Dump (writer, df.MonthGenitiveNames, "MonthGenitiveNames");
140 Dump (writer, df.MonthNames, "MonthNames");
141 writer.WriteLine ("{0}: {1}", "NativeCalendarName", df.NativeCalendarName);
142 writer.WriteLine ("{0}: {1}", "PMDesignator", df.PMDesignator);
143 writer.WriteLine ("{0}: {1}", "ShortDatePattern", df.ShortDatePattern);
144 Dump (writer, df.ShortestDayNames, "ShortestDayNames");
145 writer.WriteLine ("{0}: {1}", "ShortTimePattern", df.ShortTimePattern);
146 writer.WriteLine ("{0}: {1}", "TimeSeparator", df.TimeSeparator);
147 writer.WriteLine ("{0}: {1}", "YearMonthPattern", df.YearMonthPattern);
149 var ti = c.TextInfoEntry;
150 writer.WriteLine ("-- TextInfo --");
151 writer.WriteLine ("{0}: {1}", "ANSICodePage", ti.ANSICodePage);
152 writer.WriteLine ("{0}: {1}", "EBCDICCodePage", ti.EBCDICCodePage);
153 writer.WriteLine ("{0}: {1}", "IsRightToLeft", ti.IsRightToLeft);
154 writer.WriteLine ("{0}: {1}", "ListSeparator", ti.ListSeparator);
155 writer.WriteLine ("{0}: {1}", "MacCodePage", ti.MacCodePage);
156 writer.WriteLine ("{0}: {1}", "OEMCodePage", ti.OEMCodePage);
158 var nf = c.NumberFormatEntry;
159 writer.WriteLine ("-- NumberFormat --");
160 writer.WriteLine ("{0}: {1}", "CurrencyDecimalDigits", nf.CurrencyDecimalDigits);
161 writer.WriteLine ("{0}: {1}", "CurrencyDecimalSeparator", nf.CurrencyDecimalSeparator);
162 writer.WriteLine ("{0}: {1}", "CurrencyGroupSeparator", nf.CurrencyGroupSeparator);
163 Dump (writer, nf.CurrencyGroupSizes, "CurrencyGroupSizes", true);
164 writer.WriteLine ("{0}: {1}", "CurrencyNegativePattern", nf.CurrencyNegativePattern);
165 writer.WriteLine ("{0}: {1}", "CurrencyPositivePattern", nf.CurrencyPositivePattern);
166 writer.WriteLine ("{0}: {1}", "CurrencySymbol", nf.CurrencySymbol);
167 writer.WriteLine ("{0}: {1}", "DigitSubstitution", nf.DigitSubstitution);
168 writer.WriteLine ("{0}: {1}", "NaNSymbol", nf.NaNSymbol);
169 Dump (writer, nf.NativeDigits, "NativeDigits");
170 writer.WriteLine ("{0}: {1}", "NegativeInfinitySymbol", nf.NegativeInfinitySymbol);
171 writer.WriteLine ("{0}: {1}", "NegativeSign", nf.NegativeSign);
172 writer.WriteLine ("{0}: {1}", "NumberDecimalDigits", nf.NumberDecimalDigits);
173 writer.WriteLine ("{0}: {1}", "NumberDecimalSeparator", nf.NumberDecimalSeparator);
174 writer.WriteLine ("{0}: {1}", "NumberGroupSeparator", nf.NumberGroupSeparator);
175 Dump (writer, nf.NumberGroupSizes, "NumberGroupSizes", true);
176 writer.WriteLine ("{0}: {1}", "NumberNegativePattern", nf.NumberNegativePattern);
177 writer.WriteLine ("{0}: {1}", "PercentNegativePattern", nf.PercentNegativePattern);
178 writer.WriteLine ("{0}: {1}", "PercentPositivePattern", nf.PercentPositivePattern);
179 writer.WriteLine ("{0}: {1}", "PercentSymbol", nf.PercentSymbol);
180 writer.WriteLine ("{0}: {1}", "PerMilleSymbol", nf.PerMilleSymbol);
181 writer.WriteLine ("{0}: {1}", "PositiveInfinitySymbol", nf.PositiveInfinitySymbol);
182 writer.WriteLine ("{0}: {1}", "PositiveSign", nf.PositiveSign);
184 if (c.RegionInfoEntry != null) {
185 var ri = c.RegionInfoEntry;
186 writer.WriteLine ("-- RegionInfo --");
187 writer.WriteLine ("{0}: {1}", "CurrencyEnglishName", ri.CurrencyEnglishName);
188 writer.WriteLine ("{0}: {1}", "CurrencyNativeName", ri.CurrencyNativeName);
189 writer.WriteLine ("{0}: {1}", "CurrencySymbol", ri.CurrencySymbol);
190 writer.WriteLine ("{0}: {1}", "DisplayName", ri.DisplayName);
191 writer.WriteLine ("{0}: {1}", "EnglishName", ri.EnglishName);
192 writer.WriteLine ("{0}: {1}", "GeoId", ri.GeoId);
193 writer.WriteLine ("{0}: {1}", "IsMetric", ri.IsMetric);
194 writer.WriteLine ("{0}: {1}", "ISOCurrencySymbol", ri.ISOCurrencySymbol);
195 writer.WriteLine ("{0}: {1}", "Name", ri.Name);
196 writer.WriteLine ("{0}: {1}", "NativeName", ri.NativeName);
197 writer.WriteLine ("{0}: {1}", "ThreeLetterISORegionName", ri.ThreeLetterISORegionName);
198 writer.WriteLine ("{0}: {1}", "ThreeLetterWindowsRegionName", ri.ThreeLetterWindowsRegionName);
199 writer.WriteLine ("{0}: {1}", "TwoLetterISORegionName", ri.TwoLetterISORegionName);
206 static Type GetCalendarType (CalendarType ct)
209 case CalendarType.Gregorian:
210 return typeof (GregorianCalendar);
211 case CalendarType.HijriCalendar:
212 return typeof (HijriCalendar);
213 case CalendarType.ThaiBuddhist:
214 return typeof (ThaiBuddhistCalendar);
215 case CalendarType.UmAlQuraCalendar:
216 return typeof (UmAlQuraCalendar);
218 throw new NotImplementedException ();
222 static void Dump<T> (TextWriter tw, IList<T> values, string name, bool stopOnNull = false) where T : class
227 for (int i = 0; i < values.Count; ++i) {
230 if (stopOnNull && v == null)
244 Regex locales_regex = null;
246 locales_regex = new Regex (Locales);
248 cultures = new List<CultureInfoEntry> ();
249 var regions = new List<RegionInfoEntry> ();
252 var supplemental = GetXmlDocument (Path.Combine (data_root, "supplemental", "supplementalData.xml"));
254 // Read currencies info
255 region_currency = new Dictionary<string, string> (StringComparer.OrdinalIgnoreCase);
256 foreach (XmlNode entry in supplemental.SelectNodes ("supplementalData/currencyData/region")) {
257 var child = entry.SelectSingleNode ("currency");
258 region_currency.Add (entry.Attributes["iso3166"].Value, child.Attributes["iso4217"].Value);
262 extra_parent_locales = new Dictionary<string, string> (StringComparer.OrdinalIgnoreCase);
263 foreach (XmlNode entry in supplemental.SelectNodes ("supplementalData/parentLocales/parentLocale")) {
264 var parent = entry.Attributes["parent"].Value;
266 if (parent == "root")
269 var locales = entry.Attributes["locales"].Value;
270 foreach (var locale in locales.Split (' '))
271 extra_parent_locales.Add (locale, parent);
274 var lcdids = GetXmlDocument ("lcids.xml");
275 foreach (XmlNode lcid in lcdids.SelectNodes ("lcids/lcid")) {
276 var name = lcid.Attributes["name"].Value;
278 if (locales_regex != null && !locales_regex.IsMatch (name))
281 var ci = new CultureInfoEntry ();
282 ci.LCID = lcid.Attributes["id"].Value;
283 ci.ParentLcid = lcid.Attributes["parent"].Value;
284 ci.TwoLetterISOLanguageName = lcid.Attributes["iso2"].Value;
285 ci.ThreeLetterISOLanguageName = lcid.Attributes["iso3"].Value;
286 ci.ThreeLetterWindowsLanguageName = lcid.Attributes["win"].Value;
287 ci.OriginalName = name.Replace ('_', '-');
288 ci.TextInfoEntry = new TextInfoEntry ();
289 ci.NumberFormatEntry = new NumberFormatEntry ();
291 if (!Import (ci, name))
297 var doc_english = GetXmlDocument (Path.Combine (data_root, "main", "en.xml"));
300 // Fill all EnglishName values from en.xml language file
302 foreach (var ci in cultures) {
303 var el = doc_english.SelectSingleNode (string.Format ("ldml/localeDisplayNames/languages/language[@type='{0}']", ci.Language));
305 ci.EnglishName = el.InnerText;
308 if (ci.Script != null) {
309 el = doc_english.SelectSingleNode (string.Format ("ldml/localeDisplayNames/scripts/script[@type='{0}']", ci.Script));
314 if (ci.Territory != null) {
315 el = doc_english.SelectSingleNode (string.Format ("ldml/localeDisplayNames/territories/territory[@type='{0}']", ci.Territory));
320 s = string.Join (", ", s, el.InnerText);
324 switch (ci.ThreeLetterWindowsLanguageName) {
334 ci.EnglishName = string.Format ("{0} ({1})", ci.EnglishName, s);
336 // Special case legacy chinese
337 if (ci.OriginalName == "zh-CHS" || ci.OriginalName == "zh-CHT")
338 ci.EnglishName += " Legacy";
340 // Mono is not localized and supports english only, hence the name will always be same
341 ci.DisplayName = ci.EnglishName;
345 // Fill culture hierarchy for easier data manipulation
347 foreach (var ci in cultures) {
348 foreach (var p in cultures.Where (l => ci.LCID == l.ParentLcid)) {
353 currency_fractions = new Dictionary<string, string> (StringComparer.OrdinalIgnoreCase);
354 foreach (XmlNode entry in supplemental.SelectNodes ("supplementalData/currencyData/fractions/info")) {
355 currency_fractions.Add (entry.Attributes["iso4217"].Value, entry.Attributes["digits"].Value);
358 var territory2dayofweek = new Dictionary<string, DayOfWeek> (StringComparer.OrdinalIgnoreCase);
359 foreach (XmlNode entry in supplemental.SelectNodes ("supplementalData/weekData/firstDay")) {
361 if (entry.Attributes ["alt"] != null)
365 switch (entry.Attributes["day"].Value) {
367 dow = DayOfWeek.Monday;
370 dow = DayOfWeek.Friday;
373 dow = DayOfWeek.Saturday;
376 dow = DayOfWeek.Sunday;
379 throw new NotImplementedException ();
382 var territories = entry.Attributes["territories"].Value.Split ();
383 foreach (var t in territories) {
384 territory2dayofweek.Add (t, dow);
388 var territory2wr = new Dictionary<string, CalendarWeekRule> (StringComparer.OrdinalIgnoreCase);
389 foreach (XmlNode entry in supplemental.SelectNodes ("supplementalData/weekData/minDays")) {
390 CalendarWeekRule rule;
392 switch (entry.Attributes["count"].InnerText) {
394 rule = CalendarWeekRule.FirstDay;
397 rule = CalendarWeekRule.FirstFourDayWeek;
400 throw new NotImplementedException ();
403 var territories = entry.Attributes["territories"].InnerText.Split ();
404 foreach (var t in territories)
405 territory2wr[t] = rule;
409 // Fill all territory speficic data where territory is available
411 var non_metric = new HashSet<string> ();
412 foreach (XmlNode entry in supplemental.SelectNodes ("supplementalData/measurementData/measurementSystem[@type='US']")) {
413 var territories = entry.Attributes["territories"].InnerText.Split ();
414 foreach (var t in territories)
418 foreach (var ci in cultures) {
419 if (ci.Territory == null)
423 if (territory2dayofweek.TryGetValue (ci.Territory, out value)) {
424 ci.DateTimeFormatEntry.FirstDayOfWeek = (int) value;
427 CalendarWeekRule rule;
428 if (territory2wr.TryGetValue (ci.Territory, out rule)) {
429 ci.DateTimeFormatEntry.CalendarWeekRule = (int) rule;
432 RegionInfoEntry region = regions.Where (l => l.Name == ci.Territory).FirstOrDefault ();
433 if (region == null) {
434 region = new RegionInfoEntry () {
435 CurrencySymbol = ci.NumberFormatEntry.CurrencySymbol,
436 EnglishName = ci.EnglishName,
437 NativeName = ci.NativeTerritoryName,
439 TwoLetterISORegionName = ci.Territory,
440 CurrencyNativeName = ci.NativeCurrencyName
443 var tc = supplemental.SelectSingleNode (string.Format ("supplementalData/codeMappings/territoryCodes[@type='{0}']", ci.Territory));
444 region.ThreeLetterISORegionName = tc.Attributes["alpha3"].Value;
445 region.ThreeLetterWindowsRegionName = region.ThreeLetterISORegionName;
447 var el = doc_english.SelectSingleNode (string.Format ("ldml/localeDisplayNames/territories/territory[@type='{0}']", ci.Territory));
448 region.EnglishName = el.InnerText;
449 region.DisplayName = region.EnglishName;
451 region.ISOCurrencySymbol = region_currency[ci.Territory];
453 el = doc_english.SelectSingleNode (string.Format ("ldml/numbers/currencies/currency[@type='{0}']/displayName", region.ISOCurrencySymbol));
454 region.CurrencyEnglishName = el.InnerText;
456 if (non_metric.Contains (ci.Territory))
457 region.IsMetric = false;
459 var lcdid_value = int.Parse (ci.LCID.Substring (2), NumberStyles.HexNumber);
460 Patterns.FillValues (lcdid_value, region);
461 regions.Add (region);
464 string fraction_value;
465 if (currency_fractions.TryGetValue (region.ISOCurrencySymbol, out fraction_value)) {
466 ci.NumberFormatEntry.CurrencyDecimalDigits = fraction_value;
469 ci.RegionInfoEntry = region;
473 // Fill neutral cultures territory data
475 foreach (var ci in cultures) {
476 var dtf = ci.DateTimeFormatEntry;
477 if (dtf.FirstDayOfWeek == null) {
480 dtf.FirstDayOfWeek = (int) DayOfWeek.Saturday;
485 dtf.FirstDayOfWeek = (int) DayOfWeek.Sunday;
492 dtf.FirstDayOfWeek = (int) DayOfWeek.Monday;
495 List<int?> all_fdow = new List<int?> ();
496 GetAllChildrenValues (ci, all_fdow, l => l.DateTimeFormatEntry.FirstDayOfWeek);
497 var children = all_fdow.Where (l => l != null).Distinct ().ToList ();
499 if (children.Count == 1) {
500 dtf.FirstDayOfWeek = children[0];
501 } else if (children.Count == 0) {
502 if (!ci.HasMissingLocale)
503 Console.WriteLine ("No week data for `{0}'", ci.Name);
506 dtf.FirstDayOfWeek = (int) DayOfWeek.Sunday;
508 // .NET has weird concept of territory data available for neutral cultures (e.g. en, es, pt)
509 // We have to manually disambiguate the correct entry (which is artofficial anyway)
510 throw new ApplicationException (string.Format ("Ambiguous week data for `{0}'", ci.Name));
517 if (dtf.CalendarWeekRule == null) {
526 dtf.CalendarWeekRule = (int) CalendarWeekRule.FirstDay;
529 List<int?> all_cwr = new List<int?> ();
530 GetAllChildrenValues (ci, all_cwr, l => l.DateTimeFormatEntry.CalendarWeekRule);
531 var children = all_cwr.Where (l => l != null).Distinct ().ToList ();
533 if (children.Count == 1) {
534 dtf.CalendarWeekRule = children[0];
535 } else if (children.Count == 0) {
536 if (!ci.HasMissingLocale)
537 Console.WriteLine ("No calendar week data for `{0}'", ci.Name);
540 // Default to FirstDay
541 dtf.CalendarWeekRule = (int) CalendarWeekRule.FirstDay;
543 // .NET has weird concept of territory data available for neutral cultures (e.g. en, es, pt)
544 // We have to manually disambiguate the correct entry (which is artofficial anyway)
545 throw new ApplicationException (string.Format ("Ambiguous calendar data for `{0}'", ci.Name));
552 var nfe = ci.NumberFormatEntry;
553 if (nfe.CurrencySymbol == null) {
556 nfe.CurrencySymbol = "ر.س.";
559 nfe.CurrencySymbol = "$";
562 nfe.CurrencySymbol = "KM";
569 nfe.CurrencySymbol = "€";
572 nfe.CurrencySymbol = "kn";
575 nfe.CurrencySymbol = "R$";
578 nfe.CurrencySymbol = "kr";
581 nfe.CurrencySymbol = "RM";
584 nfe.CurrencySymbol = "টা";
587 nfe.CurrencySymbol = "Дин.";
591 nfe.CurrencySymbol = "Din.";
595 nfe.CurrencySymbol = "¥";
598 nfe.CurrencySymbol = "HK$";
602 var all_currencies = new List<string> ();
603 GetAllChildrenValues (ci, all_currencies, l => l.NumberFormatEntry.CurrencySymbol);
604 var children = all_currencies.Where (l => l != null).Distinct ().ToList ();
606 if (children.Count == 1) {
607 nfe.CurrencySymbol = children[0];
608 } else if (children.Count == 0) {
609 if (!ci.HasMissingLocale)
610 Console.WriteLine ("No currency data for `{0}'", ci.Name);
614 // .NET has weird concept of territory data available for neutral cultures (e.g. en, es, pt)
615 // We have to manually disambiguate the correct entry (which is artofficial anyway)
616 throw new ApplicationException (string.Format ("Ambiguous currency data for `{0}'. Possible values '{1}'", ci.Name, string.Join (", ", children)));
623 if (nfe.CurrencyDecimalDigits == null) {
624 var all_digits = new List<string> ();
625 GetAllChildrenValues (ci, all_digits, l => l.NumberFormatEntry.CurrencyDecimalDigits);
626 var children = all_digits.Where (l => l != null).Distinct ().ToList ();
628 if (children.Count == 1) {
629 nfe.CurrencyDecimalDigits = children[0];
630 } else if (children.Count == 0) {
631 if (!ci.HasMissingLocale)
632 Console.WriteLine ("No currency decimal digits data for `{0}'", ci.Name);
634 nfe.CurrencyDecimalDigits = "2";
635 } else if (ci.IsNeutral) {
636 nfe.CurrencyDecimalDigits = "2";
638 // .NET has weird concept of territory data available for neutral cultures (e.g. en, es, pt)
639 // We have to manually disambiguate the correct entry (which is artofficial anyway)
640 throw new ApplicationException (string.Format ("Ambiguous currency decimal digits data for `{0}'", ci.Name));
648 regions.Sort (new RegionComparer ());
649 for (int i = 0; i < regions.Count; ++i)
650 regions[i].Index = i;
653 * Dump each table individually. Using StringBuilders
654 * because it is easier to debug, should switch to just
655 * writing to streams eventually.
657 using (StreamWriter writer = new StreamWriter (HeaderFileName, false, new UTF8Encoding (false, true))) {
658 writer.NewLine = "\n";
660 writer.WriteLine ("/* This is a generated file. Do not edit. See tools/locale-builder. */");
661 writer.WriteLine ("#ifndef MONO_METADATA_CULTURE_INFO_TABLES");
662 writer.WriteLine ("#define MONO_METADATA_CULTURE_INFO_TABLES 1");
663 writer.WriteLine ("\n");
665 writer.WriteLine ("#define NUM_CULTURE_ENTRIES {0}", cultures.Count);
666 writer.WriteLine ("#define NUM_REGION_ENTRIES {0}", regions.Count);
668 writer.WriteLine ("\n");
670 // Sort the cultures by lcid
671 cultures.Sort (new LcidComparer ());
673 StringBuilder builder = new StringBuilder ();
675 int count = cultures.Count;
676 for (int i = 0; i < count; i++) {
677 CultureInfoEntry ci = cultures[i];
678 if (ci.DateTimeFormatEntry == null)
680 ci.DateTimeFormatEntry.AppendTableRow (builder);
681 ci.DateTimeFormatEntry.Row = row++;
683 builder.Append (',');
684 builder.Append ('\n');
687 writer.WriteLine ("static const DateTimeFormatEntry datetime_format_entries [] = {");
688 writer.Write (builder);
689 writer.WriteLine ("};\n\n");
691 builder = new StringBuilder ();
693 for (int i = 0; i < count; i++) {
694 CultureInfoEntry ci = cultures[i];
695 if (ci.NumberFormatEntry == null)
697 ci.NumberFormatEntry.AppendTableRow (builder);
698 ci.NumberFormatEntry.Row = row++;
700 builder.Append (',');
701 builder.Append ('\n');
704 writer.WriteLine ("static const NumberFormatEntry number_format_entries [] = {");
705 writer.Write (builder);
706 writer.WriteLine ("};\n\n");
708 builder = new StringBuilder ();
710 for (int i = 0; i < count; i++) {
711 CultureInfoEntry ci = cultures[i];
712 ci.AppendTableRow (builder);
715 builder.Append (',');
716 builder.Append ('\n');
719 writer.WriteLine ("static const CultureInfoEntry culture_entries [] = {");
720 writer.Write (builder);
721 writer.WriteLine ("};\n\n");
723 cultures.Sort (new ExportNameComparer ()); // Sort based on name
724 builder = new StringBuilder ();
725 for (int i = 0; i < count; i++) {
726 CultureInfoEntry ci = cultures[i];
727 var name = ci.GetExportName ().ToLowerInvariant ();
728 builder.Append ("\t{" + Entry.EncodeStringIdx (name) + ", ");
729 builder.Append (ci.Row + "}");
731 builder.Append (',');
733 builder.AppendFormat ("\t /* {0} */", name);
734 builder.Append ('\n');
737 writer.WriteLine ("static const CultureInfoNameEntry culture_name_entries [] = {");
738 writer.Write (builder);
739 writer.WriteLine ("};\n\n");
741 builder = new StringBuilder ();
743 foreach (RegionInfoEntry r in regions) {
744 r.AppendTableRow (builder);
745 if (++rcount != regions.Count)
746 builder.Append (',');
748 builder.Append ('\n');
750 writer.WriteLine ("static const RegionInfoEntry region_entries [] = {");
751 writer.Write (builder);
752 writer.WriteLine ("};\n\n");
754 builder = new StringBuilder ();
756 foreach (RegionInfoEntry ri in regions) {
757 builder.Append ("\t{" + Entry.EncodeStringIdx (ri.TwoLetterISORegionName) + ", ");
758 builder.Append (ri.Index + "}");
759 if (++rcount != regions.Count)
760 builder.Append (',');
762 builder.AppendFormat ("\t /* {0} */", ri.TwoLetterISORegionName);
763 builder.Append ('\n');
766 writer.WriteLine ("static const RegionInfoNameEntry region_name_entries [] = {");
767 writer.Write (builder);
768 writer.WriteLine ("};\n\n");
770 writer.WriteLine ("static const char locale_strings [] = {");
771 writer.Write (Entry.GetStrings ());
772 writer.WriteLine ("};\n\n");
774 writer.WriteLine ("#endif\n");
778 static void GetAllChildrenValues<T> (CultureInfoEntry entry, List<T> values, Func<CultureInfoEntry, T> selector)
780 foreach (var e in entry.Children) {
784 values.Add (selector (e));
786 foreach (var e2 in e.Children) {
787 GetAllChildrenValues (e2, values, selector);
792 static XmlDocument GetXmlDocument (string path)
794 var doc = new XmlDocument ();
795 doc.Load (new XmlTextReader (path) { /*DtdProcessing = DtdProcessing.Ignore*/ } );
799 bool Import (CultureInfoEntry data, string locale)
802 var sep = locale.Split ('_');
803 data.Language = sep[0];
805 // CLDR strictly follow ISO names, .NET does not
806 // Replace names where non-iso2 is used, e.g. Norway
807 if (data.Language != data.TwoLetterISOLanguageName) {
808 locale = data.TwoLetterISOLanguageName;
809 if (sep.Length > 1) {
810 locale += string.Join ("_", sep.Skip (1));
814 // Convert broken Chinese names to correct one
823 locale = "zh_Hans_CN";
826 locale = "zh_Hant_HK";
829 locale = "zh_Hans_SG";
832 locale = "zh_Hant_TW";
835 locale = "zh_Hant_MO";
839 sep = locale.Split ('_');
841 string full_name = Path.Combine (data_root, "main", locale + ".xml");
842 if (!File.Exists (full_name)) {
843 Console.WriteLine ("Missing locale file for `{0}'", locale);
845 // We could fill default values but that's not as simple as it seems. For instance for non-neutral
846 // cultures the next part could be territory or not.
849 XmlDocument doc = null;
852 * Locale generation is done in several steps, first we
853 * read the root file which is the base invariant data
854 * then the supplemental root data,
855 * then the language file, the supplemental languages
856 * file then the locale file, then the supplemental
857 * locale file. Values in each descending file can
858 * overwrite previous values.
860 foreach (var part in sep) {
868 if (extra_parent_locales.TryGetValue (fname, out extra)) {
869 xml = GetXmlDocument (Path.Combine (data_root, "main", extra + ".xml"));
876 xml = GetXmlDocument (Path.Combine (data_root, "main", fname + ".xml"));
884 // Extract localized locale name from language xml file. Have to do it after both language and territory are read
886 var el = doc.SelectSingleNode (string.Format ("ldml/localeDisplayNames/languages/language[@type='{0}']", data.Language));
888 data.NativeName = el.InnerText;
890 if (data.Territory != null) {
891 el = doc.SelectSingleNode (string.Format ("ldml/localeDisplayNames/territories/territory[@type='{0}']", data.Territory));
893 // TODO: Should read <localePattern>
894 data.NativeName = string.Format ("{0} ({1})", data.NativeName, el.InnerText);
895 data.NativeTerritoryName = el.InnerText;
899 // We have territory now we have to run the process again to extract currency symbol
900 if (region_currency.TryGetValue (data.Territory, out currency)) {
903 var xml = GetXmlDocument (Path.Combine (data_root, "main", "root.xml"));
904 el = xml.SelectSingleNode (string.Format ("ldml/numbers/currencies/currency[@type='{0}']/symbol", currency));
906 data.NumberFormatEntry.CurrencySymbol = el.InnerText;
908 foreach (var part in sep) {
914 xml = GetXmlDocument (Path.Combine (data_root, "main", fname + ".xml"));
915 el = xml.SelectSingleNode (string.Format ("ldml/numbers/currencies/currency[@type='{0}']/symbol", currency));
917 data.NumberFormatEntry.CurrencySymbol = el.InnerText;
919 el = xml.SelectSingleNode (string.Format ("ldml/numbers/currencies/currency[@type='{0}']/displayName", currency));
921 data.NativeCurrencyName = el.InnerText;
927 // It looks like it never changes
928 data.DateTimeFormatEntry.TimeSeparator = ":";
930 // TODO: Don't have input data available but most values are 2 with few exceptions for 1 and 3
931 // We don't add 3 as it's for some arabic states only
932 switch (data.ThreeLetterISOLanguageName) {
934 data.NumberFormatEntry.NumberDecimalDigits = 1;
937 data.NumberFormatEntry.NumberDecimalDigits = 2;
941 // TODO: For now we capture only native name for default calendar
942 data.NativeCalendarNames[((int) data.CalendarType & 0xFF) - 1] = data.DateTimeFormatEntry.NativeCalendarName;
944 var lcdid_value = int.Parse (data.LCID.Substring (2), NumberStyles.HexNumber);
945 Patterns.FillValues (lcdid_value, data);
950 void Import (XmlDocument doc, CultureInfoEntry ci)
956 // Extract script & teritory
958 el = doc.SelectSingleNode ("ldml/identity/script");
960 ci.Script = el.Attributes["type"].Value;
962 el = doc.SelectSingleNode ("ldml/identity/territory");
964 ci.Territory = el.Attributes["type"].Value;
966 var df = ci.DateTimeFormatEntry;
969 // Default calendar is for now always "gregorian"
970 switch (ci.OriginalName) {
971 case "th": case "th-TH":
972 calendar = "buddhist";
973 ci.CalendarType = CalendarType.ThaiBuddhist; // typeof (ThaiBuddhistCalendar);
975 case "ar": case "ar-SA":
976 calendar = "islamic";
977 ci.CalendarType = CalendarType.UmAlQuraCalendar; // typeof (UmAlQuraCalendar);
979 case "ps": case "ps-AF": case "prs": case "prs-AF": case "dv": case "dv-MV":
980 calendar = "persian";
981 ci.CalendarType = CalendarType.HijriCalendar; // typeof (HijriCalendar);
984 calendar = "gregorian";
985 ci.CalendarType = CalendarType.Gregorian; // typeof (GregorianCalendar);
986 ci.GregorianCalendarType = GregorianCalendarTypes.Localized;
990 var node = doc.SelectSingleNode (string.Format ("ldml/dates/calendars/calendar[@type='{0}']", calendar));
992 el = doc.SelectSingleNode (string.Format ("ldml/localeDisplayNames/types/type[@type='{0}']", calendar));
994 df.NativeCalendarName = el.InnerText;
997 // Apply global rule first <alias source="locale" path="../../monthContext[@type='format']/monthWidth[@type='wide']"/>
998 nodes = node.SelectNodes ("months/monthContext[@type='format']/monthWidth[@type='wide']/month");
999 ProcessAllNodes (nodes, df.MonthNames, AddOrReplaceValue);
1000 nodes = node.SelectNodes ("months/monthContext[@type='stand-alone']/monthWidth[@type='wide']/month");
1001 ProcessAllNodes (nodes, df.MonthNames, AddOrReplaceValue);
1003 if (df.MonthNames != null) {
1004 if (ci.Name == "sv" || ci.Name == "sv-SE") {
1005 ToLower (df.MonthNames);
1009 // Apply global rule first <alias source="locale" path="../../monthContext[@type='format']/monthWidth[@type='abbreviated']"/>
1010 if (ci.Name == "ja" || ci.Name == "ja-JP") {
1011 // Use common number style
1013 nodes = node.SelectNodes ("months/monthContext[@type='format']/monthWidth[@type='abbreviated']/month");
1014 ProcessAllNodes (nodes, df.AbbreviatedMonthNames, AddOrReplaceValue);
1015 nodes = node.SelectNodes ("months/monthContext[@type='stand-alone']/monthWidth[@type='abbreviated']/month");
1016 ProcessAllNodes (nodes, df.AbbreviatedMonthNames, AddOrReplaceValue);
1019 if (df.AbbreviatedMonthNames != null) {
1020 if (ci.Name == "sv" || ci.Name == "sv-SE") {
1021 ToLower (df.AbbreviatedMonthNames);
1025 nodes = node.SelectNodes ("months/monthContext[@type='format']/monthWidth[@type='wide']/month");
1026 if (nodes != null) {
1027 ProcessAllNodes (nodes, df.MonthGenitiveNames, AddOrReplaceValue);
1030 // All values seem to match
1031 Array.Copy (df.AbbreviatedMonthNames, df.AbbreviatedMonthGenitiveNames, df.AbbreviatedMonthNames.Length);
1033 nodes = node.SelectNodes ("days/dayContext[@type='format']/dayWidth[@type='wide']/day");
1034 ProcessAllNodes (nodes, df.DayNames, AddOrReplaceDayValue);
1036 // Apply global rule first <alias source="locale" path="../../dayContext[@type='format']/dayWidth[@type='abbreviated']"/>
1037 nodes = node.SelectNodes ("days/dayContext[@type='format']/dayWidth[@type='abbreviated']/day");
1038 ProcessAllNodes (nodes, df.AbbreviatedDayNames, AddOrReplaceDayValue);
1039 nodes = node.SelectNodes ("days/dayContext[@type='stand-alone']/dayWidth[@type='abbreviated']/day");
1040 ProcessAllNodes (nodes, df.AbbreviatedDayNames, AddOrReplaceDayValue);
1042 if (df.AbbreviatedDayNames != null) {
1043 if (ci.Name == "sv" || ci.Name == "sv-SE") {
1044 ToLower (df.AbbreviatedDayNames);
1048 // TODO: This is not really ShortestDayNames as .NET uses it
1049 // Apply global rules first <alias source="locale" path="../../dayContext[@type='stand-alone']/dayWidth[@type='narrow']"/>
1050 nodes = node.SelectNodes ("days/dayContext[@type='format']/dayWidth[@type='narrow']/day");
1051 ProcessAllNodes (nodes, df.ShortestDayNames, AddOrReplaceDayValue);
1052 nodes = node.SelectNodes ("days/dayContext[@type='stand-alone']/dayWidth[@type='narrow']/day");
1053 ProcessAllNodes (nodes, df.ShortestDayNames, AddOrReplaceDayValue);
1055 Cannot really be used it's too different to .NET and most app rely on it
1057 el = node.SelectSingleNode ("dateFormats/dateFormatLength[@type='full']/dateFormat/pattern");
1059 df.LongDatePattern = ConvertDatePatternFormat (el.InnerText);
1061 // Medium is our short
1062 el = node.SelectSingleNode ("dateFormats/dateFormatLength[@type='medium']/dateFormat/pattern");
1064 df.ShortDatePattern = ConvertDatePatternFormat (el.InnerText);
1066 // Medium is our Long
1067 el = node.SelectSingleNode ("timeFormats/timeFormatLength[@type='medium']/timeFormat/pattern");
1069 df.LongTimePattern = ConvertTimePatternFormat (el.InnerText);
1071 el = node.SelectSingleNode ("timeFormats/timeFormatLength[@type='short']/timeFormat/pattern");
1073 df.ShortTimePattern = ConvertTimePatternFormat (el.InnerText);
1075 el = node.SelectSingleNode ("dateTimeFormats/availableFormats/dateFormatItem[@id='yyyyMMMM']");
1077 df.YearMonthPattern = ConvertDatePatternFormat (el.InnerText);
1079 el = node.SelectSingleNode ("dateTimeFormats/availableFormats/dateFormatItem[@id='MMMMdd']");
1081 df.MonthDayPattern = ConvertDatePatternFormat (el.InnerText);
1083 el = node.SelectSingleNode ("dayPeriods/dayPeriodContext/dayPeriodWidth[@type='abbreviated']/dayPeriod[@type='am']");
1085 // Apply global rule first <alias source="locale" path="../dayPeriodWidth[@type='wide']"/>
1086 el = node.SelectSingleNode ("dayPeriods/dayPeriodContext/dayPeriodWidth[@type='wide']/dayPeriod[@type='am']");
1088 // Manual edits for exact .net compatiblity
1091 df.AMDesignator = "AM";
1094 df.AMDesignator = "a.m.";
1098 df.AMDesignator = el.InnerText;
1102 el = node.SelectSingleNode ("dayPeriods/dayPeriodContext/dayPeriodWidth[@type='abbreviated']/dayPeriod[@type='pm']");
1104 // Apply global rule first <alias source="locale" path="../dayPeriodWidth[@type='wide']"/>
1105 el = node.SelectSingleNode ("dayPeriods/dayPeriodContext/dayPeriodWidth[@type='wide']/dayPeriod[@type='pm']");
1109 df.PMDesignator = "PM";
1112 df.PMDesignator = "p.m.";
1116 df.PMDesignator = el.InnerText;
1121 var ni = ci.NumberFormatEntry;
1123 node = doc.SelectSingleNode ("ldml/numbers/symbols");
1125 el = node.SelectSingleNode ("plusSign");
1127 ni.PositiveSign = el.InnerText;
1129 el = node.SelectSingleNode ("minusSign");
1131 // CLDR uses unicode negative sign for some culture (e.g sv, is, lt, don't kwnow why) but .net always
1132 // uses simple - sign
1133 if (el.InnerText == "\u2212") {
1134 ni.NegativeSign = "-";
1135 } else if (el.InnerText == "\u200F\u002D") {
1136 // Remove any right-to-left mark characters
1137 ni.NegativeSign = "-";
1139 ni.NegativeSign = el.InnerText;
1142 el = node.SelectSingleNode ("infinity");
1144 // We cannot use the value from CLDR because many broken
1145 // .NET serializers (e.g. JSON) use text value of NegativeInfinity
1146 // and different value would break interoperability with .NET
1147 var inf = GetInfinitySymbol (ci);
1149 ni.InfinitySymbol = inf;
1150 else if (el != null && el.InnerText != "∞") {
1151 ni.InfinitySymbol = el.InnerText;
1154 el = node.SelectSingleNode ("perMille");
1156 ni.PerMilleSymbol = el.InnerText;
1158 el = node.SelectSingleNode ("nan");
1160 ni.NaNSymbol = el.InnerText;
1162 el = node.SelectSingleNode ("percentSign");
1164 ni.PercentSymbol = el.InnerText;
1168 string value = null;
1170 // .net has incorrect separators for some countries and we want to be compatible
1174 // es-ES does not have group separator but .net has '.'
1179 el = node.SelectSingleNode ("group");
1181 value = el.InnerText;
1188 if (value != null) {
1189 ni.NumberGroupSeparator = ni.CurrencyGroupSeparator = value;
1193 static void ToLower (string[] values)
1198 for (int i = 0; i < values.Length; ++i) {
1199 if (values [i] == null)
1202 values [i] = values [i].ToLower ();
1206 string GetInfinitySymbol (CultureInfoEntry ci)
1209 switch (ci.TwoLetterISOLanguageName) {
1214 return "+nekonečno";
1216 return "+unendlich";
1231 return "+nieskończoność";
1234 return "бесконечность";
1236 return "neskončnost";
1250 static string ConvertDatePatternFormat (string format)
1253 // LDMR uses different characters for some fields
1254 // http://unicode.org/reports/tr35/#Date_Format_Patterns
1256 format = format.Replace ("EEEE", "dddd"); // The full name of the day of the week
1257 format = format.Replace ("LLLL", "MMMM"); // The full month name
1259 if (format.EndsWith (" y", StringComparison.Ordinal))
1265 static string ConvertTimePatternFormat (string format)
1267 format = format.Replace ("a", "tt"); // AM or PM
1271 static void ProcessAllNodes (XmlNodeList list, IList<string> values, Action<IList<string>, string, string> convertor)
1273 foreach (XmlNode entry in list) {
1274 var index = entry.Attributes["type"].Value;
1275 var value = entry.InnerText;
1276 convertor (values, index, value);
1280 // All text indexes are 1-based
1281 static void AddOrReplaceValue (IList<string> list, string oneBasedIndex, string value)
1283 int index = int.Parse (oneBasedIndex);
1284 AddOrReplaceValue (list, index - 1, value);
1287 static readonly string[] day_types = new string[] { "sun", "mon", "tue", "wed", "thu", "fri", "sat" };
1289 static void AddOrReplaceDayValue (IList<string> list, string dayType, string value)
1291 int index = Array.IndexOf (day_types, dayType);
1292 AddOrReplaceValue (list, index, value);
1295 static void AddOrReplaceValue (IList<string> list, int index, string value)
1297 if (list.Count <= index)
1298 ((List<string>) list).AddRange (new string[index - list.Count + 1]);
1300 list[index] = value;
1303 sealed class LcidComparer : IComparer<CultureInfoEntry>
1305 public int Compare (CultureInfoEntry x, CultureInfoEntry y)
1307 return x.LCID.CompareTo (y.LCID);
1311 sealed class ExportNameComparer : IComparer<CultureInfoEntry>
1313 public int Compare (CultureInfoEntry x, CultureInfoEntry y)
1315 return String.Compare (x.GetExportName (), y.GetExportName (), StringComparison.OrdinalIgnoreCase);
1319 class RegionComparer : IComparer<RegionInfoEntry>
1321 public int Compare (RegionInfoEntry x, RegionInfoEntry y)
1323 return x.TwoLetterISORegionName.CompareTo (y.TwoLetterISORegionName);