Include timezone L10n data in QLocaleXML files

This makes the XML file bigger by a factor of roughly 8, at about 30
MB. Code to read the new data out of it shall follow in a later
commit.

Task-number: QTBUG-115158
Change-Id: I7b9b6abe88be2457fa6cf0e8d7b6a68845136770
Reviewed-by: Mate Barany <mate.barany@qt.io>

util/locale_database/cldr.py

@@ -37,6 +37,8 @@ class CldrReader (object):
         self.root = CldrAccess(root)
         self.whitter, self.grumble = whitter, grumble
         self.root.checkEnumData(grumble)
+        # TODO: can we do anything but ignore with the namings here ?
+        self.__bcp47Alias, ignore = self.root.bcp47Aliases()
 
     def likelySubTags(self):
         """Generator for likely subtag information.
@@ -105,7 +107,7 @@ class CldrReader (object):
         that are not mentioned in enumdata.territory_map, on any
         Windows IDs given in zonedata.windowsIdList that are no longer
         covered by the CLDR data."""
-        alias, ignored = self.root.bcp47Aliases()
+        alias = self.__bcp47Alias
         defaults, winIds = self.root.readWindowsTimeZones(alias)
         metamap, zones, territorial = self.root.readMetaZoneMap(alias)
 
@@ -296,6 +298,9 @@ class CldrReader (object):
         locale.update(scan.unitData()) # byte, kB, MB, GB, ..., KiB, MiB, GiB, ...
         locale.update(scan.calendarNames(calendars)) # Names of days and months
 
+        # Naming of timezones:
+        locale.update(scan.timeZoneNames(self.__bcp47Alias))
+
         return locale
 
 # Note: various caches assume this class is a singleton, so the

util/locale_database/ldml.py

@@ -441,7 +441,7 @@ class LocaleScanner (object):
         See CldrAccess.readMetaZoneMap() for the locale-independent
         data that stitches these pieces together."""
         stem, formats = 'dates/timeZoneNames', {}
-        # '+HH:mm;-HH:mm'
+        # '+HH:mm;-HH:mm' (and :ss is also supported, but nowhere used in v45).
         # Sometimes has single-digit hours
         hours = self.find(f'{stem}/hourFormat').split(';')
         assert all('H' in f and 'm' in f for f in hours), (hours, self.name)

util/locale_database/qlocalexml.py

@@ -505,14 +505,16 @@ class QLocaleXmlWriter (object):
         tuple of numeric IDs that corresponds to en_US (needed to
         provide fallbacks for the C locale)."""
 
+        def writeLocale(locale, cal = calendars, this = self):
+            this.__openTag('locale')
+            this.__writeLocale(locale, cal)
+            this.__writeLocaleZones(locale)
+            this.__closeTag('locale')
+
         self.__openTag('localeList')
-        self.__openTag('locale')
+        writeLocale(Locale.C(locales[en_US]))
-        self.__writeLocale(Locale.C(locales[en_US]), calendars)
-        self.__closeTag('locale')
         for key in sorted(locales.keys()):
-            self.__openTag('locale')
+            writeLocale(locales[key])
-            self.__writeLocale(locales[key], calendars)
-            self.__closeTag('locale')
         self.__closeTag('localeList')
 
     def inTag(self, tag, text, **attrs):
@@ -602,6 +604,35 @@ class QLocaleXmlWriter (object):
         self.__scripts.discard(locale.script_code)
         self.__territories.discard(locale.territory_code)
 
+    def __writeLocaleZones(self, locale):
+        self.__writeZoneForms('regionZoneFormats', locale.regionZoneFormats)
+        self.__writeZoneNaming('zoneNaming', locale.zoneNaming)
+        self.__writeZoneNaming('metaZoneNaming', locale.metaZoneNaming)
+
+    def __writeZoneNaming(self, group, naming):
+        if not naming:
+            return
+        self.__openTag(group)
+        for iana in sorted(naming.keys()):
+            data = naming[iana]
+            self.__openTag('zoneNames', name=iana)
+            if 'exemplarCity' in data:
+                self.inTag('exemplar', data['exemplarCity'])
+            for form in ('short', 'long'):
+                if form in data:
+                    self.__writeZoneForms(form, data[form])
+            self.__closeTag('zoneNames')
+        self.__closeTag(group)
+
+    def __writeZoneForms(self, group, forms):
+        if all(x is None for x in forms):
+            return
+        self.__openTag('zoneForms', name=group)
+        for i, tag in enumerate(('generic', 'standard', 'daylightSaving')):
+            if forms[i]:
+                self.safeInTag(tag, forms[i])
+        self.__closeTag('zoneForms')
+
     def __openTag(self, tag, **attrs):
         if attrs:
             text = ' '.join(f'{k}="{v}"' for k, v in attrs.items())
@@ -721,12 +752,18 @@ class Locale (object):
                     'longTimeFormat', 'shortTimeFormat',
                     'currencyIsoCode', 'currencySymbol', 'currencyDisplayName',
                     'currencyFormat', 'currencyNegativeFormat',
+                    'positiveOffsetFormat', 'negativeOffsetFormat',
+                    'gmtOffsetFormat', 'fallbackZoneFormat',
                     ) + tuple(self.propsMonthDay('days')) + tuple(
                 '_'.join((k, cal))
                 for k in self.propsMonthDay('months')
                 for cal in calendars):
             write(key, escape(get(key)))
 
+        # The regionZoneFormats, zoneNaming and metaZoneNaming members
+        # are handled by QLocaleXmlWriter.__writeLocaleZones(). Their
+        # elements hold sub-elements.
+
         write('groupSizes', ';'.join(str(x) for x in get('groupSizes')))
         for key in ('currencyDigits', 'currencyRounding'):
             write(key, get(key))

util/locale_database/qlocalexml.rnc

@@ -86,6 +86,19 @@ Digit = xsd:string { pattern = "\d" }
 Punctuation = xsd:string { pattern = "\p{P}" }
 GroupSizes = xsd:string { pattern = "\d;\d;\d" }
 
+ZoneForms = element zoneForms {
+  attribute name { text }, # 'regionFormats', 'short' or 'long'
+  element generic { text }?,
+  element standard { text }?,
+  element daylightSaving { text }?
+}
+
+ZoneNames = element zoneNames {
+  attribute name { text }, # IANA ID of zone, or CLDR metazone name
+  element exemplar { text }?, # metaZoneNaming omits exemplar
+  ZoneForms* # for name in {'short', 'long'}
+}
+
 Locale = element locale {
   element language { text },
   element script { text },
@@ -125,6 +138,14 @@ Locale = element locale {
   element currencyDisplayName { text },
   element currencyFormat { text },
   element currencyNegativeFormat { text },
+
+  # Timezone (and metazone) format data
+  element positiveOffsetFormat { text },
+  element negativeOffsetFormat { text },
+  element gmtOffsetFormat { text },
+  element fallbackZoneFormat { text },
+
+  # Day names
   element longDays { text },
   element standaloneLongDays { text },
   element shortDays { text },
@@ -154,5 +175,10 @@ Locale = element locale {
 
   element groupSizes { GroupSizes },
   element currencyDigits { xsd:nonNegativeInteger },
-  element currencyRounding { xsd:nonNegativeInteger }
+  element currencyRounding { xsd:nonNegativeInteger },
+
+  # Timezone (and metazone) naming data
+  ZoneForms?, # name is 'regionFormats'
+  element zoneNaming { ZoneNames+ }?,
+  element metaZoneNaming { ZoneNames+ }?
 }