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}", "PercentDecimalDigits", nf.PercentDecimalDigits);
178 writer.WriteLine ("{0}: {1}", "PercentDecimalSeparator", nf.PercentDecimalSeparator);
179 writer.WriteLine ("{0}: {1}", "PercentGroupSeparator", nf.PercentGroupSeparator);
180 Dump (writer, nf.PercentGroupSizes, "PercentGroupSizes", true);
181 writer.WriteLine ("{0}: {1}", "PercentNegativePattern", nf.PercentNegativePattern);
182 writer.WriteLine ("{0}: {1}", "PercentPositivePattern", nf.PercentPositivePattern);
183 writer.WriteLine ("{0}: {1}", "PercentSymbol", nf.PercentSymbol);
184 writer.WriteLine ("{0}: {1}", "PerMilleSymbol", nf.PerMilleSymbol);
185 writer.WriteLine ("{0}: {1}", "PositiveInfinitySymbol", nf.PositiveInfinitySymbol);
186 writer.WriteLine ("{0}: {1}", "PositiveSign", nf.PositiveSign);
188 if (c.RegionInfoEntry != null) {
189 var ri = c.RegionInfoEntry;
190 writer.WriteLine ("-- RegionInfo --");
191 writer.WriteLine ("{0}: {1}", "CurrencyEnglishName", ri.CurrencyEnglishName);
192 writer.WriteLine ("{0}: {1}", "CurrencyNativeName", ri.CurrencyNativeName);
193 writer.WriteLine ("{0}: {1}", "CurrencySymbol", ri.CurrencySymbol);
194 writer.WriteLine ("{0}: {1}", "DisplayName", ri.DisplayName);
195 writer.WriteLine ("{0}: {1}", "EnglishName", ri.EnglishName);
196 writer.WriteLine ("{0}: {1}", "GeoId", ri.GeoId);
197 writer.WriteLine ("{0}: {1}", "IsMetric", ri.IsMetric);
198 writer.WriteLine ("{0}: {1}", "ISOCurrencySymbol", ri.ISOCurrencySymbol);
199 writer.WriteLine ("{0}: {1}", "Name", ri.Name);
200 writer.WriteLine ("{0}: {1}", "NativeName", ri.NativeName);
201 writer.WriteLine ("{0}: {1}", "ThreeLetterISORegionName", ri.ThreeLetterISORegionName);
202 writer.WriteLine ("{0}: {1}", "ThreeLetterWindowsRegionName", ri.ThreeLetterWindowsRegionName);
203 writer.WriteLine ("{0}: {1}", "TwoLetterISORegionName", ri.TwoLetterISORegionName);
210 static Type GetCalendarType (CalendarType ct)
213 case CalendarType.Gregorian:
214 return typeof (GregorianCalendar);
215 case CalendarType.HijriCalendar:
216 return typeof (HijriCalendar);
217 case CalendarType.ThaiBuddhist:
218 return typeof (ThaiBuddhistCalendar);
219 case CalendarType.UmAlQuraCalendar:
220 return typeof (UmAlQuraCalendar);
222 throw new NotImplementedException ();
226 static void Dump<T> (TextWriter tw, IList<T> values, string name, bool stopOnNull = false) where T : class
231 for (int i = 0; i < values.Count; ++i) {
234 if (stopOnNull && v == null)
248 Regex locales_regex = null;
250 locales_regex = new Regex (Locales);
252 cultures = new List<CultureInfoEntry> ();
253 var regions = new List<RegionInfoEntry> ();
256 var supplemental = GetXmlDocument (Path.Combine (data_root, "supplemental", "supplementalData.xml"));
258 // Read currencies info
259 region_currency = new Dictionary<string, string> (StringComparer.OrdinalIgnoreCase);
260 foreach (XmlNode entry in supplemental.SelectNodes ("supplementalData/currencyData/region")) {
261 var child = entry.SelectSingleNode ("currency");
262 region_currency.Add (entry.Attributes["iso3166"].Value, child.Attributes["iso4217"].Value);
266 extra_parent_locales = new Dictionary<string, string> (StringComparer.OrdinalIgnoreCase);
267 foreach (XmlNode entry in supplemental.SelectNodes ("supplementalData/parentLocales/parentLocale")) {
268 var parent = entry.Attributes["parent"].Value;
270 if (parent == "root")
273 var locales = entry.Attributes["locales"].Value;
274 foreach (var locale in locales.Split (' '))
275 extra_parent_locales.Add (locale, parent);
278 var lcdids = GetXmlDocument ("lcids.xml");
279 foreach (XmlNode lcid in lcdids.SelectNodes ("lcids/lcid")) {
280 var name = lcid.Attributes["name"].Value;
282 if (locales_regex != null && !locales_regex.IsMatch (name))
285 var ci = new CultureInfoEntry ();
286 ci.LCID = lcid.Attributes["id"].Value;
287 ci.ParentLcid = lcid.Attributes["parent"].Value;
288 ci.TwoLetterISOLanguageName = lcid.Attributes["iso2"].Value;
289 ci.ThreeLetterISOLanguageName = lcid.Attributes["iso3"].Value;
290 ci.ThreeLetterWindowsLanguageName = lcid.Attributes["win"].Value;
291 ci.OriginalName = name.Replace ('_', '-');
292 ci.TextInfoEntry = new TextInfoEntry ();
293 ci.NumberFormatEntry = new NumberFormatEntry ();
295 if (!Import (ci, name))
301 var doc_english = GetXmlDocument (Path.Combine (data_root, "main", "en.xml"));
304 // Fill all EnglishName values from en.xml language file
306 foreach (var ci in cultures) {
307 var el = doc_english.SelectSingleNode (string.Format ("ldml/localeDisplayNames/languages/language[@type='{0}']", ci.Language));
309 ci.EnglishName = el.InnerText;
312 if (ci.Script != null) {
313 el = doc_english.SelectSingleNode (string.Format ("ldml/localeDisplayNames/scripts/script[@type='{0}']", ci.Script));
318 if (ci.Territory != null) {
319 el = doc_english.SelectSingleNode (string.Format ("ldml/localeDisplayNames/territories/territory[@type='{0}']", ci.Territory));
324 s = string.Join (", ", s, el.InnerText);
328 switch (ci.ThreeLetterWindowsLanguageName) {
338 ci.EnglishName = string.Format ("{0} ({1})", ci.EnglishName, s);
340 // Special case legacy chinese
341 if (ci.OriginalName == "zh-CHS" || ci.OriginalName == "zh-CHT")
342 ci.EnglishName += " Legacy";
344 // Mono is not localized and supports english only, hence the name will always be same
345 ci.DisplayName = ci.EnglishName;
349 // Fill culture hierarchy for easier data manipulation
351 foreach (var ci in cultures) {
352 foreach (var p in cultures.Where (l => ci.LCID == l.ParentLcid)) {
357 currency_fractions = new Dictionary<string, string> (StringComparer.OrdinalIgnoreCase);
358 foreach (XmlNode entry in supplemental.SelectNodes ("supplementalData/currencyData/fractions/info")) {
359 currency_fractions.Add (entry.Attributes["iso4217"].Value, entry.Attributes["digits"].Value);
362 var territory2dayofweek = new Dictionary<string, DayOfWeek> (StringComparer.OrdinalIgnoreCase);
363 foreach (XmlNode entry in supplemental.SelectNodes ("supplementalData/weekData/firstDay")) {
366 switch (entry.Attributes["day"].Value) {
368 dow = DayOfWeek.Monday;
371 dow = DayOfWeek.Friday;
374 dow = DayOfWeek.Saturday;
377 dow = DayOfWeek.Sunday;
380 throw new NotImplementedException ();
383 var territories = entry.Attributes["territories"].Value.Split ();
384 foreach (var t in territories)
385 territory2dayofweek[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 string fraction_value;
433 if (currency_fractions.TryGetValue (ci.Territory, out fraction_value)) {
434 ci.NumberFormatEntry.CurrencyDecimalDigits = fraction_value;
437 RegionInfoEntry region = regions.Where (l => l.Name == ci.Territory).FirstOrDefault ();
438 if (region == null) {
439 region = new RegionInfoEntry () {
440 CurrencySymbol = ci.NumberFormatEntry.CurrencySymbol,
441 EnglishName = ci.EnglishName,
442 NativeName = ci.NativeTerritoryName,
444 TwoLetterISORegionName = ci.Territory,
445 CurrencyNativeName = ci.NativeCurrencyName
448 var tc = supplemental.SelectSingleNode (string.Format ("supplementalData/codeMappings/territoryCodes[@type='{0}']", ci.Territory));
449 region.ThreeLetterISORegionName = tc.Attributes["alpha3"].Value;
450 region.ThreeLetterWindowsRegionName = region.ThreeLetterISORegionName;
452 var el = doc_english.SelectSingleNode (string.Format ("ldml/localeDisplayNames/territories/territory[@type='{0}']", ci.Territory));
453 region.EnglishName = el.InnerText;
454 region.DisplayName = region.EnglishName;
456 region.ISOCurrencySymbol = region_currency[ci.Territory];
458 el = doc_english.SelectSingleNode (string.Format ("ldml/numbers/currencies/currency[@type='{0}']/displayName", region.ISOCurrencySymbol));
459 region.CurrencyEnglishName = el.InnerText;
461 if (non_metric.Contains (ci.Territory))
462 region.IsMetric = false;
464 var lcdid_value = int.Parse (ci.LCID.Substring (2), NumberStyles.HexNumber);
465 Patterns.FillValues (lcdid_value, region);
466 regions.Add (region);
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 = "$";
563 nfe.CurrencySymbol = "€";
566 nfe.CurrencySymbol = "R$";
569 nfe.CurrencySymbol = "kr";
572 nfe.CurrencySymbol = "RM";
575 nfe.CurrencySymbol = "টা";
578 nfe.CurrencySymbol = "Дин.";
582 nfe.CurrencySymbol = "Din.";
585 nfe.CurrencySymbol = "¥";
588 nfe.CurrencySymbol = "HK$";
592 var all_currencies = new List<string> ();
593 GetAllChildrenValues (ci, all_currencies, l => l.NumberFormatEntry.CurrencySymbol);
594 var children = all_currencies.Where (l => l != null).Distinct ().ToList ();
596 if (children.Count == 1) {
597 nfe.CurrencySymbol = children[0];
598 } else if (children.Count == 0) {
599 if (!ci.HasMissingLocale)
600 Console.WriteLine ("No currency data for `{0}'", ci.Name);
604 // .NET has weird concept of territory data available for neutral cultures (e.g. en, es, pt)
605 // We have to manually disambiguate the correct entry (which is artofficial anyway)
606 throw new ApplicationException (string.Format ("Ambiguous currency data for `{0}'", ci.Name));
613 if (nfe.CurrencyDecimalDigits == null) {
614 var all_digits = new List<string> ();
615 GetAllChildrenValues (ci, all_digits, l => l.NumberFormatEntry.CurrencyDecimalDigits);
616 var children = all_digits.Where (l => l != null).Distinct ().ToList ();
618 if (children.Count == 1) {
619 nfe.CurrencyDecimalDigits = children[0];
620 } else if (children.Count == 0) {
621 if (!ci.HasMissingLocale)
622 Console.WriteLine ("No currency decimal digits data for `{0}'", ci.Name);
624 nfe.CurrencyDecimalDigits = "2";
626 // .NET has weird concept of territory data available for neutral cultures (e.g. en, es, pt)
627 // We have to manually disambiguate the correct entry (which is artofficial anyway)
628 throw new ApplicationException (string.Format ("Ambiguous currency decimal digits data for `{0}'", ci.Name));
636 regions.Sort (new RegionComparer ());
637 for (int i = 0; i < regions.Count; ++i)
638 regions[i].Index = i;
641 * Dump each table individually. Using StringBuilders
642 * because it is easier to debug, should switch to just
643 * writing to streams eventually.
645 using (StreamWriter writer = new StreamWriter (HeaderFileName, false, new UTF8Encoding (false, true))) {
646 writer.NewLine = "\n";
648 writer.WriteLine ("/* This is a generated file. Do not edit. See tools/locale-builder. */");
649 writer.WriteLine ("#ifndef MONO_METADATA_CULTURE_INFO_TABLES");
650 writer.WriteLine ("#define MONO_METADATA_CULTURE_INFO_TABLES 1");
651 writer.WriteLine ("\n");
653 writer.WriteLine ("#define NUM_CULTURE_ENTRIES {0}", cultures.Count);
654 writer.WriteLine ("#define NUM_REGION_ENTRIES {0}", regions.Count);
656 writer.WriteLine ("\n");
658 // Sort the cultures by lcid
659 cultures.Sort (new LcidComparer ());
661 StringBuilder builder = new StringBuilder ();
663 int count = cultures.Count;
664 for (int i = 0; i < count; i++) {
665 CultureInfoEntry ci = cultures[i];
666 if (ci.DateTimeFormatEntry == null)
668 ci.DateTimeFormatEntry.AppendTableRow (builder);
669 ci.DateTimeFormatEntry.Row = row++;
671 builder.Append (',');
672 builder.Append ('\n');
675 writer.WriteLine ("static const DateTimeFormatEntry datetime_format_entries [] = {");
676 writer.Write (builder);
677 writer.WriteLine ("};\n\n");
679 builder = new StringBuilder ();
681 for (int i = 0; i < count; i++) {
682 CultureInfoEntry ci = cultures[i];
683 if (ci.NumberFormatEntry == null)
685 ci.NumberFormatEntry.AppendTableRow (builder);
686 ci.NumberFormatEntry.Row = row++;
688 builder.Append (',');
689 builder.Append ('\n');
692 writer.WriteLine ("static const NumberFormatEntry number_format_entries [] = {");
693 writer.Write (builder);
694 writer.WriteLine ("};\n\n");
696 builder = new StringBuilder ();
698 for (int i = 0; i < count; i++) {
699 CultureInfoEntry ci = cultures[i];
700 ci.AppendTableRow (builder);
703 builder.Append (',');
704 builder.Append ('\n');
707 writer.WriteLine ("static const CultureInfoEntry culture_entries [] = {");
708 writer.Write (builder);
709 writer.WriteLine ("};\n\n");
711 cultures.Sort (new ExportNameComparer ()); // Sort based on name
712 builder = new StringBuilder ();
713 for (int i = 0; i < count; i++) {
714 CultureInfoEntry ci = cultures[i];
715 var name = ci.GetExportName ().ToLowerInvariant ();
716 builder.Append ("\t{" + Entry.EncodeStringIdx (name) + ", ");
717 builder.Append (ci.Row + "}");
719 builder.Append (',');
721 builder.AppendFormat ("\t /* {0} */", name);
722 builder.Append ('\n');
725 writer.WriteLine ("static const CultureInfoNameEntry culture_name_entries [] = {");
726 writer.Write (builder);
727 writer.WriteLine ("};\n\n");
729 builder = new StringBuilder ();
731 foreach (RegionInfoEntry r in regions) {
732 r.AppendTableRow (builder);
733 if (++rcount != regions.Count)
734 builder.Append (',');
736 builder.Append ('\n');
738 writer.WriteLine ("static const RegionInfoEntry region_entries [] = {");
739 writer.Write (builder);
740 writer.WriteLine ("};\n\n");
742 builder = new StringBuilder ();
744 foreach (RegionInfoEntry ri in regions) {
745 builder.Append ("\t{" + Entry.EncodeStringIdx (ri.TwoLetterISORegionName) + ", ");
746 builder.Append (ri.Index + "}");
747 if (++rcount != regions.Count)
748 builder.Append (',');
750 builder.AppendFormat ("\t /* {0} */", ri.TwoLetterISORegionName);
751 builder.Append ('\n');
754 writer.WriteLine ("static const RegionInfoNameEntry region_name_entries [] = {");
755 writer.Write (builder);
756 writer.WriteLine ("};\n\n");
758 writer.WriteLine ("static const char locale_strings [] = {");
759 writer.Write (Entry.GetStrings ());
760 writer.WriteLine ("};\n\n");
762 writer.WriteLine ("#endif\n");
766 static void GetAllChildrenValues<T> (CultureInfoEntry entry, List<T> values, Func<CultureInfoEntry, T> selector)
768 foreach (var e in entry.Children) {
772 values.Add (selector (e));
774 foreach (var e2 in e.Children) {
775 GetAllChildrenValues (e2, values, selector);
780 static XmlDocument GetXmlDocument (string path)
782 var doc = new XmlDocument ();
783 doc.Load (new XmlTextReader (path) { /*DtdProcessing = DtdProcessing.Ignore*/ } );
787 bool Import (CultureInfoEntry data, string locale)
790 var sep = locale.Split ('_');
791 data.Language = sep[0];
793 // CLDR strictly follow ISO names, .NET does not
794 // Replace names where non-iso2 is used, e.g. Norway
795 if (data.Language != data.TwoLetterISOLanguageName) {
796 locale = data.TwoLetterISOLanguageName;
797 if (sep.Length > 1) {
798 locale += string.Join ("_", sep.Skip (1));
802 // Convert broken Chinese names to correct one
811 locale = "zh_Hans_CN";
814 locale = "zh_Hant_HK";
817 locale = "zh_Hans_SG";
820 locale = "zh_Hant_TW";
823 locale = "zh_Hant_MO";
827 sep = locale.Split ('_');
829 string full_name = Path.Combine (data_root, "main", locale + ".xml");
830 if (!File.Exists (full_name)) {
831 Console.WriteLine ("Missing locale file for `{0}'", locale);
833 // We could fill default values but that's not as simple as it seems. For instance for non-neutral
834 // cultures the next part could be territory or not.
837 XmlDocument doc = null;
840 * Locale generation is done in several steps, first we
841 * read the root file which is the base invariant data
842 * then the supplemental root data,
843 * then the language file, the supplemental languages
844 * file then the locale file, then the supplemental
845 * locale file. Values in each descending file can
846 * overwrite previous values.
848 foreach (var part in sep) {
856 if (extra_parent_locales.TryGetValue (fname, out extra)) {
857 xml = GetXmlDocument (Path.Combine (data_root, "main", extra + ".xml"));
864 xml = GetXmlDocument (Path.Combine (data_root, "main", fname + ".xml"));
872 // Extract localized locale name from language xml file. Have to do it after both language and territory are read
874 var el = doc.SelectSingleNode (string.Format ("ldml/localeDisplayNames/languages/language[@type='{0}']", data.Language));
876 data.NativeName = el.InnerText;
878 if (data.Territory != null) {
879 el = doc.SelectSingleNode (string.Format ("ldml/localeDisplayNames/territories/territory[@type='{0}']", data.Territory));
881 // TODO: Should read <localePattern>
882 data.NativeName = string.Format ("{0} ({1})", data.NativeName, el.InnerText);
883 data.NativeTerritoryName = el.InnerText;
887 // We have territory now we have to run the process again to extract currency symbol
888 if (region_currency.TryGetValue (data.Territory, out currency)) {
891 var xml = GetXmlDocument (Path.Combine (data_root, "main", "root.xml"));
892 el = xml.SelectSingleNode (string.Format ("ldml/numbers/currencies/currency[@type='{0}']/symbol", currency));
894 data.NumberFormatEntry.CurrencySymbol = el.InnerText;
896 foreach (var part in sep) {
902 xml = GetXmlDocument (Path.Combine (data_root, "main", fname + ".xml"));
903 el = xml.SelectSingleNode (string.Format ("ldml/numbers/currencies/currency[@type='{0}']/symbol", currency));
905 data.NumberFormatEntry.CurrencySymbol = el.InnerText;
907 el = xml.SelectSingleNode (string.Format ("ldml/numbers/currencies/currency[@type='{0}']/displayName", currency));
909 data.NativeCurrencyName = el.InnerText;
915 // It looks like it never changes
916 data.DateTimeFormatEntry.TimeSeparator = ":";
918 // TODO: Don't have input data available but most values are 2 with few exceptions for 1 and 3
919 // We don't add 3 as it's for some arabic states only
920 switch (data.ThreeLetterISOLanguageName) {
922 data.NumberFormatEntry.NumberDecimalDigits =
923 data.NumberFormatEntry.PercentDecimalDigits = 1;
926 data.NumberFormatEntry.NumberDecimalDigits =
927 data.NumberFormatEntry.PercentDecimalDigits = 2;
931 // TODO: For now we capture only native name for default calendar
932 data.NativeCalendarNames[((int) data.CalendarType & 0xFF) - 1] = data.DateTimeFormatEntry.NativeCalendarName;
934 var lcdid_value = int.Parse (data.LCID.Substring (2), NumberStyles.HexNumber);
935 Patterns.FillValues (lcdid_value, data);
940 void Import (XmlDocument doc, CultureInfoEntry ci)
946 // Extract script & teritory
948 el = doc.SelectSingleNode ("ldml/identity/script");
950 ci.Script = el.Attributes["type"].Value;
952 el = doc.SelectSingleNode ("ldml/identity/territory");
954 ci.Territory = el.Attributes["type"].Value;
956 var df = ci.DateTimeFormatEntry;
959 // Default calendar is for now always "gregorian"
961 case "th": case "th-TH":
962 calendar = "buddhist";
963 ci.CalendarType = CalendarType.ThaiBuddhist; // typeof (ThaiBuddhistCalendar);
965 case "ar": case "ar-SA":
966 calendar = "islamic";
967 ci.CalendarType = CalendarType.UmAlQuraCalendar; // typeof (UmAlQuraCalendar);
969 case "ps": case "ps-AF": case "prs": case "prs-AF": case "dv": case "dv-MV":
970 calendar = "persian";
971 ci.CalendarType = CalendarType.HijriCalendar; // typeof (HijriCalendar);
974 calendar = "gregorian";
975 ci.CalendarType = CalendarType.Gregorian; // typeof (GregorianCalendar);
976 ci.GregorianCalendarType = GregorianCalendarTypes.Localized;
980 var node = doc.SelectSingleNode (string.Format ("ldml/dates/calendars/calendar[@type='{0}']", calendar));
982 el = doc.SelectSingleNode (string.Format ("ldml/localeDisplayNames/types/type[@type='{0}']", calendar));
984 df.NativeCalendarName = el.InnerText;
987 // Apply global rule first <alias source="locale" path="../../monthContext[@type='format']/monthWidth[@type='wide']"/>
988 nodes = node.SelectNodes ("months/monthContext[@type='format']/monthWidth[@type='wide']/month");
989 ProcessAllNodes (nodes, df.MonthNames, AddOrReplaceValue);
990 nodes = node.SelectNodes ("months/monthContext[@type='stand-alone']/monthWidth[@type='wide']/month");
991 ProcessAllNodes (nodes, df.MonthNames, AddOrReplaceValue);
993 // Apply global rule first <alias source="locale" path="../../monthContext[@type='format']/monthWidth[@type='abbreviated']"/>
994 nodes = node.SelectNodes ("months/monthContext[@type='format']/monthWidth[@type='abbreviated']/month");
995 ProcessAllNodes (nodes, df.AbbreviatedMonthNames, AddOrReplaceValue);
996 nodes = node.SelectNodes ("months/monthContext[@type='stand-alone']/monthWidth[@type='abbreviated']/month");
997 ProcessAllNodes (nodes, df.AbbreviatedMonthNames, AddOrReplaceValue);
999 nodes = node.SelectNodes ("months/monthContext[@type='format']/monthWidth[@type='wide']/month");
1000 if (nodes != null) {
1001 ProcessAllNodes (nodes, df.MonthGenitiveNames, AddOrReplaceValue);
1004 // All values seem to match
1005 Array.Copy (df.AbbreviatedMonthNames, df.AbbreviatedMonthGenitiveNames, df.AbbreviatedMonthNames.Length);
1007 nodes = node.SelectNodes ("days/dayContext[@type='format']/dayWidth[@type='wide']/day");
1008 ProcessAllNodes (nodes, df.DayNames, AddOrReplaceDayValue);
1010 // Apply global rule first <alias source="locale" path="../../dayContext[@type='format']/dayWidth[@type='abbreviated']"/>
1011 nodes = node.SelectNodes ("days/dayContext[@type='format']/dayWidth[@type='abbreviated']/day");
1012 ProcessAllNodes (nodes, df.AbbreviatedDayNames, AddOrReplaceDayValue);
1013 nodes = node.SelectNodes ("days/dayContext[@type='stand-alone']/dayWidth[@type='abbreviated']/day");
1014 ProcessAllNodes (nodes, df.AbbreviatedDayNames, AddOrReplaceDayValue);
1016 // TODO: This is not really ShortestDayNames as .NET uses it
1017 // Apply global rules first <alias source="locale" path="../../dayContext[@type='stand-alone']/dayWidth[@type='narrow']"/>
1018 nodes = node.SelectNodes ("days/dayContext[@type='format']/dayWidth[@type='narrow']/day");
1019 ProcessAllNodes (nodes, df.ShortestDayNames, AddOrReplaceDayValue);
1020 nodes = node.SelectNodes ("days/dayContext[@type='stand-alone']/dayWidth[@type='narrow']/day");
1021 ProcessAllNodes (nodes, df.ShortestDayNames, AddOrReplaceDayValue);
1023 Cannot really be used it's too different to .NET and most app rely on it
1025 el = node.SelectSingleNode ("dateFormats/dateFormatLength[@type='full']/dateFormat/pattern");
1027 df.LongDatePattern = ConvertDatePatternFormat (el.InnerText);
1029 // Medium is our short
1030 el = node.SelectSingleNode ("dateFormats/dateFormatLength[@type='medium']/dateFormat/pattern");
1032 df.ShortDatePattern = ConvertDatePatternFormat (el.InnerText);
1034 // Medium is our Long
1035 el = node.SelectSingleNode ("timeFormats/timeFormatLength[@type='medium']/timeFormat/pattern");
1037 df.LongTimePattern = ConvertTimePatternFormat (el.InnerText);
1039 el = node.SelectSingleNode ("timeFormats/timeFormatLength[@type='short']/timeFormat/pattern");
1041 df.ShortTimePattern = ConvertTimePatternFormat (el.InnerText);
1043 el = node.SelectSingleNode ("dateTimeFormats/availableFormats/dateFormatItem[@id='yyyyMMMM']");
1045 df.YearMonthPattern = ConvertDatePatternFormat (el.InnerText);
1047 el = node.SelectSingleNode ("dateTimeFormats/availableFormats/dateFormatItem[@id='MMMMdd']");
1049 df.MonthDayPattern = ConvertDatePatternFormat (el.InnerText);
1051 el = node.SelectSingleNode ("dayPeriods/dayPeriodContext/dayPeriodWidth[@type='abbreviated']/dayPeriod[@type='am']");
1053 // Apply global rule first <alias source="locale" path="../dayPeriodWidth[@type='wide']"/>
1054 el = node.SelectSingleNode ("dayPeriods/dayPeriodContext/dayPeriodWidth[@type='wide']/dayPeriod[@type='am']");
1057 df.AMDesignator = el.InnerText;
1059 el = node.SelectSingleNode ("dayPeriods/dayPeriodContext/dayPeriodWidth[@type='abbreviated']/dayPeriod[@type='pm']");
1061 // Apply global rule first <alias source="locale" path="../dayPeriodWidth[@type='wide']"/>
1062 el = node.SelectSingleNode ("dayPeriods/dayPeriodContext/dayPeriodWidth[@type='wide']/dayPeriod[@type='pm']");
1066 df.PMDesignator = el.InnerText;
1069 var ni = ci.NumberFormatEntry;
1071 node = doc.SelectSingleNode ("ldml/numbers/symbols");
1073 el = node.SelectSingleNode ("decimal");
1075 ni.NumberDecimalSeparator =
1076 ni.PercentDecimalSeparator =
1077 ni.CurrencyDecimalSeparator = el.InnerText;
1080 el = node.SelectSingleNode ("plusSign");
1082 ni.PositiveSign = el.InnerText;
1084 el = node.SelectSingleNode ("minusSign");
1086 // CLDR uses unicode negative sign for some culture (e.g sv, is, lt, don't kwnow why) but .net always
1087 // uses simple - sign
1088 if (el.InnerText == "\u2212")
1089 ni.NegativeSign = "-";
1091 ni.NegativeSign = el.InnerText;
1094 el = node.SelectSingleNode ("infinity");
1096 // We cannot use the value from CLDR because many broken
1097 // .NET serializers (e.g. JSON) use text value of NegativeInfinity
1098 // and different value would break interoperability with .NET
1099 var inf = GetInfinitySymbol (ci);
1101 ni.InfinitySymbol = inf;
1102 else if (el != null && el.InnerText != "∞") {
1103 ni.InfinitySymbol = el.InnerText;
1106 el = node.SelectSingleNode ("perMille");
1108 ni.PerMilleSymbol = el.InnerText;
1110 el = node.SelectSingleNode ("nan");
1112 ni.NaNSymbol = el.InnerText;
1114 el = node.SelectSingleNode ("percentSign");
1116 ni.PercentSymbol = el.InnerText;
1118 el = node.SelectSingleNode ("group");
1120 ni.NumberGroupSeparator =
1121 ni.PercentGroupSeparator =
1122 ni.CurrencyGroupSeparator = el.InnerText;
1127 string GetInfinitySymbol (CultureInfoEntry ci)
1130 switch (ci.TwoLetterISOLanguageName) {
1135 return "+nekonečno";
1137 return "+unendlich";
1152 return "+nieskończoność";
1155 return "бесконечность";
1157 return "neskončnost";
1171 static string ConvertDatePatternFormat (string format)
1174 // LDMR uses different characters for some fields
1175 // http://unicode.org/reports/tr35/#Date_Format_Patterns
1177 format = format.Replace ("EEEE", "dddd"); // The full name of the day of the week
1178 format = format.Replace ("LLLL", "MMMM"); // The full month name
1180 if (format.EndsWith (" y", StringComparison.Ordinal))
1186 static string ConvertTimePatternFormat (string format)
1188 format = format.Replace ("a", "tt"); // AM or PM
1192 static void ProcessAllNodes (XmlNodeList list, IList<string> values, Action<IList<string>, string, string> convertor)
1194 foreach (XmlNode entry in list) {
1195 var index = entry.Attributes["type"].Value;
1196 var value = entry.InnerText;
1197 convertor (values, index, value);
1201 // All text indexes are 1-based
1202 static void AddOrReplaceValue (IList<string> list, string oneBasedIndex, string value)
1204 int index = int.Parse (oneBasedIndex);
1205 AddOrReplaceValue (list, index - 1, value);
1208 static readonly string[] day_types = new string[] { "sun", "mon", "tue", "wed", "thu", "fri", "sat" };
1210 static void AddOrReplaceDayValue (IList<string> list, string dayType, string value)
1212 int index = Array.IndexOf (day_types, dayType);
1213 AddOrReplaceValue (list, index, value);
1216 static void AddOrReplaceValue (IList<string> list, int index, string value)
1218 if (list.Count <= index)
1219 ((List<string>) list).AddRange (new string[index - list.Count + 1]);
1221 list[index] = value;
1224 sealed class LcidComparer : IComparer<CultureInfoEntry>
1226 public int Compare (CultureInfoEntry x, CultureInfoEntry y)
1228 return x.LCID.CompareTo (y.LCID);
1232 sealed class ExportNameComparer : IComparer<CultureInfoEntry>
1234 public int Compare (CultureInfoEntry x, CultureInfoEntry y)
1236 return String.Compare (x.GetExportName (), y.GetExportName (), StringComparison.OrdinalIgnoreCase);
1240 class RegionComparer : IComparer<RegionInfoEntry>
1242 public int Compare (RegionInfoEntry x, RegionInfoEntry y)
1244 return x.TwoLetterISORegionName.CompareTo (y.TwoLetterISORegionName);