1berry/qtbase
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

Rework cldr2qlocalexml.py's reading of CLDR data

Browse Source 
Move the code out to a CldrReader class in cldr.py, expand CldrAccess
with facilities that needs, expand ldml.py to include support for more
features, finally making xpathlite.py redundant. This initial commit
aims, though, to be bug-for-bug compatible with xpathlite in its
reading of the CLDR data.

It turns out we've been using draftier data than we were aware of
(which might not be a bad thing). The xpathlite code appeared to check
for draft attributes, but these only appear on leaf nodes and most
data were fetched by finding a parent and then scanning its children
without the draft check; only am/pm data was actually being excluded
based on draft values.  (We allowed contributed, for am/pm, in
addition to approved, which is all the xpathlite code allows
otherwise.) There are also some less equivocal bugs; I'll deal with
these in later commits.

Simplified number-system data look-ups; the old get_number_in_system()
was taking care of old LDML versions' placement of the number system
attribute; this is no longer needed. (It was also being used for a
currency value to which it was not appropriate, which is now handled
separately; this is one of the bugs mentioned above.) Ditched a
fall-back to nativeZeroDigit, which no longer exists in CLDR.

Change the command-line to take the root of the CLDR data tree, rather
than its common/main/ sub-directory. Support naming the file to which
to write output, as a second command-line argument, instead of always
writing to stdout (which remains the default) and leaving whoever runs
the script to redirect stdout.

Support (internally for now, while adding TODOs to give main() more
command-line options) separating the stderr output into its more and
less interesting parts; for now, continue producing both, but suppress
the least interesting entirely.

Task-number: QTBUG-81344
Change-Id: Ie611b47403a9452b51feaeeaaa0fbc8f7e84dc71
Reviewed-by: Cristian Maureira-Fredes <cristian.maureira-fredes@qt.io>
This commit is contained in:
Edward Welbourne 2020-02-27 13:58:58 +01:00 committed by Edward Welbourne
parent c834dbc6fb
commit be3dfd7a71
5 changed files with 972 additions and 898 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

498
util/locale_database/cldr.py
View File

@ -27,7 +27,8 @@
#############################################################################
"""Digesting the CLDR's data.
Provides two class:
Provides two classes:
CldrReader -- driver for reading CLDR data
CldrAccess -- used by the reader to access the tree of data files
The former should normally be all you need to access.
@ -38,9 +39,206 @@ from xml.dom import minidom
from weakref import WeakValueDictionary as CacheDict
import os
from localetools import Error
from ldml import Node, Supplement
from ldml import Error, Node, XmlScanner, Supplement, LocaleScanner
from qlocalexml import Locale
class CldrReader (object):
def __init__(self, root, grumble = lambda msg: None, whitter = lambda msg: None):
"""Set up a reader object for reading CLDR data.
Single parameter, root, is the file-system path to the root of
the unpacked CLDR archive; its common/ sub-directory should
contain dtd/, main/ and supplemental/ sub-directories.
Optional second argument, grumble, is a callable that logs
warnings and complaints, e.g. sys.stderr.write would be a
suitable callable. The default is a no-op that ignores its
single argument. Optional third argument is similar, used for
less interesting output; pass sys.stderr.write for it for
verbose output."""
self.root = CldrAccess(root)
self.whitter, self.grumble = whitter, grumble
def likelySubTags(self):
"""Generator for likely subtag information.
Yields pairs (have, give) of 4-tuples; if what you have
matches the left member, giving the right member is probably
sensible. Each 4-tuple's entries are the full names of a
language, a script, a country (strictly territory) and a
variant (currently ignored)."""
skips = []
for got, use in self.root.likelySubTags():
try:
have = self.__parseTags(got)
give = self.__parseTags(use)
except Error as e:
if ((use.startswith(got) or got.startswith('und_'))
and e.message.startswith('Unknown ') and ' code ' in e.message):
skips.append(use)
else:
self.grumble('Skipping likelySubtag "{}" -> "{}" ({})\n'.format(got, use, e.message))
continue
if all(code.startswith('Any') and code[3].isupper() for code in have[:-1]):
continue
give = (give[0],
# Substitute according to http://www.unicode.org/reports/tr35/#Likely_Subtags
have[1] if give[1] == 'AnyScript' else give[1],
have[2] if give[2] == 'AnyCountry' else give[2],
give[3]) # AnyVariant similarly ?
yield have, give
if skips:
# TODO: look at LDML's reserved locale tag names; they
# show up a lot in this, and may be grounds for filtering
# more out.
pass # self.__wrapped(self.whitter, 'Skipping likelySubtags (for unknown codes): ', skips)
def readLocales(self, calendars = ('gregorian',)):
locales = tuple(self.__allLocales(calendars))
return dict(((k.language_id, k.script_id, k.country_id, k.variant_code),
k) for k in locales)
def __allLocales(self, calendars):
def skip(locale, reason):
return 'Skipping defaultContent locale "{}" ({})\n'.format(locale, reason)
for locale in self.root.defaultContentLocales:
try:
language, script, country, variant = self.__splitLocale(locale)
except ValueError:
self.whitter(skip(locale, 'only language tag'))
continue
if not (script or country):
self.grumble(skip(locale, 'second tag is neither script nor territory'))
continue
if not (language and country):
continue
try:
yield self.__getLocaleData(self.root.locale(locale), calendars,
language, script, country, variant)
except Error as e:
self.grumble(skip(locale, e.message))
for locale in self.root.fileLocales:
try:
chain = self.root.locale(locale)
language, script, country, variant = chain.tagCodes()
assert language
# TODO: this skip should probably be based on likely
# sub-tags, instead of empty country: if locale has a
# likely-subtag expansion, that's what QLocale uses,
# and we'll be saving its data for the expanded locale
# anyway, so don't need to record it for itself.
# See also QLocaleXmlReader.loadLocaleMap's grumble.
if not country:
continue
yield self.__getLocaleData(chain, calendars, language, script, country, variant)
except Error as e:
self.grumble('Skipping file locale "{}" ({})\n'.format(locale, e.message))
import textwrap
@staticmethod
def __wrapped(writer, prefix, tokens, wrap = textwrap.wrap):
writer('\n'.join(wrap(prefix + ', '.join(tokens),
subsequent_indent=' ', width=80)) + '\n')
del textwrap
def __parseTags(self, locale):
tags = self.__splitLocale(locale)
language = tags.next()
script = country = variant = ''
try:
script, country, variant = tags
except ValueError:
pass
return tuple(p[1] for p in self.root.codesToIdName(language, script, country, variant))
def __splitLocale(self, name):
"""Generate (language, script, territory, variant) from a locale name
Ignores any trailing fields (with a warning), leaves script (a
capitalised four-letter token), territory (either a number or
an all-uppercase token) or variant (upper case and digits)
empty if unspecified. Only generates one entry if name is a
single tag (i.e. contains no underscores). Always yields 1 or
4 values, never 2 or 3."""
tags = iter(name.split('_'))
yield tags.next() # Language
tag = tags.next() # may raise StopIteration
# Script is always four letters, always capitalised:
if len(tag) == 4 and tag[0].isupper() and tag[1:].islower():
yield tag
try:
tag = tags.next()
except StopIteration:
tag = ''
else:
yield ''
# Territory is upper-case or numeric:
if tag and tag.isupper() or tag.isdigit():
yield tag
try:
tag = tags.next()
except StopIteration:
tag = ''
else:
yield ''
# Variant can be any mixture of upper-case and digits.
if tag and all(c.isupper() or c.isdigit() for c in tag):
yield tag
tag = ''
else:
yield ''
# If nothing is left, StopIteration will avoid the warning:
if not tag:
tag = tags.next()
self.grumble('Ignoring unparsed cruft {} in {}\n'.format('_'.join(tag + tuple(tags)), name))
def __getLocaleData(self, scan, calendars, language, script, country, variant):
ids, names = zip(*self.root.codesToIdName(language, script, country, variant))
assert ids[0] > 0 and ids[2] > 0, (language, script, country, variant)
locale = Locale(
language = names[0], language_code = language, language_id = ids[0],
script = names[1], script_code = script, script_id = ids[1],
country = names[2], country_code = country, country_id = ids[2],
variant_code = variant)
firstDay, weStart, weEnd = self.root.weekData(country)
assert all(day in ('mon', 'tue', 'wed', 'thu', 'fri', 'sat', 'sun')
for day in (firstDay, weStart, weEnd))
locale.update(firstDayOfWeek = firstDay,
weekendStart = weStart,
weekendEnd = weEnd)
iso, digits, rounding = self.root.currencyData(country)
locale.update(currencyIsoCode = iso,
currencyDigits = int(digits),
currencyRounding = int(rounding))
locale.update(scan.currencyData(iso))
locale.update(scan.numericData(self.root.numberSystem, self.whitter))
locale.update(scan.textPatternData())
locale.update(scan.endonyms(language, script, country, variant))
locale.update(scan.unitData()) # byte, kB, MB, GB, ..., KiB, MiB, GiB, ...
locale.update(scan.calendarNames(calendars)) # Names of days and months
return locale
# Note: various caches assume this class is a singleton, so the
# "default" value for a parameter no caller should pass can serve as
# the cache. If a process were to instantiate this class with distinct
# roots, each cache would be filled by the first to need it !
class CldrAccess (object):
def __init__(self, root):
"""Set up a master object for accessing CLDR data.
@ -50,6 +248,12 @@ class CldrAccess (object):
contain dtd/, main/ and supplemental/ sub-directories."""
self.root = root
def xml(self, *path):
"""Load a single XML file and return its root element as an XmlScanner.
The path is interpreted relative to self.root"""
return XmlScanner(Node(self.__xml(path)))
def supplement(self, name):
"""Loads supplemental data as a Supplement object.
@ -57,6 +261,117 @@ class CldrAccess (object):
"""
return Supplement(Node(self.__xml(('common', 'supplemental', name))))
def locale(self, name):
"""Loads all data for a locale as a LocaleScanner object.
The name should be a locale name; adding suffix '.xml' to it
should usually yield a file in common/main/. The returned
LocaleScanner object packages this file along with all those
from which it inherits; its methods know how to handle that
inheritance, where relevant."""
return LocaleScanner(name, self.__localeRoots(name), self.__rootLocale)
@property
def fileLocales(self, joinPath = os.path.join, listDirectory = os.listdir,
splitExtension = os.path.splitext):
"""Generator for locale IDs seen in file-names.
All *.xml other than root.xml in common/main/ are assumed to
identify locales."""
for name in listDirectory(joinPath(self.root, 'common', 'main')):
stem, ext = splitExtension(name)
if ext == '.xml' and stem != 'root':
yield stem
@property
def defaultContentLocales(self):
"""Generator for the default content locales."""
for name, attrs in self.supplement('supplementalMetadata.xml').find('metadata/defaultContent'):
try:
locales = attrs['locales']
except KeyError:
pass
else:
for locale in locales.split():
yield locale
def likelySubTags(self):
for ignore, attrs in self.supplement('likelySubtags.xml').find('likelySubtags'):
yield attrs['from'], attrs['to']
def numberSystem(self, system):
"""Get a description of a numbering system.
Returns a mapping, with keys u'digits', u'type' and u'id'; the
value for this last is system. Raises KeyError for unknown
number system, ldml.Error on failure to load data."""
try:
return self.__numberSystems[system]
except KeyError:
raise Error('Unsupported number system: {}'.format(system))
def weekData(self, country):
"""Data on the weekly cycle.
Returns a triple (W, S, E) of en's short names for week-days;
W is the first day of the week, S the start of the week-end
and E the end of the week-end. Where data for a country is
unavailable, the data for CLDR's territory 001 (The World) is
used."""
try:
return self.__weekData[country]
except KeyError:
return self.__weekData['001']
def currencyData(self, country):
"""Returns currency data for the given country code.
Return value is a tuple (ISO4217 code, digit count, rounding
mode). If CLDR provides no data for this country, ('', 2, 1)
is the default result.
"""
try:
return self.__currencyData[country]
except KeyError:
return '', 2, 1
def codesToIdName(self, language, script, country, variant = ''):
"""Maps each code to the appropriate ID and name.
Returns a 4-tuple of (ID, name) pairs corresponding to the
language, script, country and variant given. Raises a
suitable error if any of them is unknown, indicating all that
are unknown plus suitable names for any that could sensibly be
added to enumdata.py to make them known.
Until we implement variant support (QTBUG-81051), the fourth
member of the returned tuple is always 0 paired with a string
that should not be used."""
enum = self.__enumMap
try:
return (enum('language')[language],
enum('script')[script],
enum('country')[country],
enum('variant')[variant])
except KeyError:
pass
parts, values = [], [language, script, country, variant]
for index, key in enumerate(('language', 'script', 'country', 'variant')):
naming, enums = self.__codeMap(key), enum(key)
value = values[index]
if value not in enums:
text = '{} code {}'.format(key, value)
name = naming.get(value)
if name and value != 'POSIX':
text += u' (could add {})'.format(name)
parts.append(text)
if len(parts) > 1:
parts[-1] = 'and ' + parts[-1]
assert parts
raise Error('Unknown ' + ', '.join(parts),
language, script, country, variant)
def readWindowsTimeZones(self, lookup): # For use by cldr2qtimezone.py
"""Digest CLDR's MS-Win time-zone name mapping.
@ -138,12 +453,98 @@ class CldrAccess (object):
def __open(self, path, joinPath=os.path.join):
return open(joinPath(self.root, *path))
@property
def __rootLocale(self, cache = []):
if not cache:
cache.append(self.xml('common', 'main', 'root.xml'))
return cache[0]
@property
def __supplementalData(self, cache = []):
if not cache:
cache.append(self.supplement('supplementalData.xml'))
return cache[0]
@property
def __numberSystems(self, cache = {}, joinPath=os.path.join):
if not cache:
for ignore, attrs in self.supplement('numberingSystems.xml').find('numberingSystems'):
if ord(attrs.get('digits', u'\x10000')[0]) > 0xffff:
# FIXME, QTBUG-69324: make this redundant:
# omit number system if zero doesn't fit in single-char16 UTF-16 :-(
continue
cache[attrs['id']] = attrs
assert cache
return cache
@property
def __weekData(self, cache = {}):
if not cache:
firstDay, weStart, weEnd = self.__getWeekData()
# Massage those into an easily-consulted form:
# World defaults given for code '001':
mon, sat, sun = firstDay['001'], weStart['001'], weEnd['001']
lands = set(firstDay) | set(weStart) | set(weEnd)
cache.update((land,
(firstDay.get(land, mon), weStart.get(land, sat), weEnd.get(land, sun)))
for land in lands)
assert cache
return cache
def __getWeekData(self):
"""Scan for data on the weekly cycle.
Yields three mappings from locales to en's short names for
week-days; if a locale isn't a key of a given mapping, it
should use the '001' (world) locale's value. The first mapping
gives the day on which the week starts, the second gives the
day on which the week-end starts, the third gives the last day
of the week-end."""
source = self.__supplementalData
for key in ('firstDay', 'weekendStart', 'weekendEnd'):
result = {}
for ignore, attrs in source.find('weekData/' + key):
assert ignore == key
day = attrs['day']
assert day in ('mon', 'tue', 'wed', 'thu', 'fri', 'sat', 'sun'), day
if 'alt' in attrs:
continue
for loc in attrs.get('territories', '').split():
result[loc] = day
yield result
@property
def __currencyData(self, cache = {}):
if not cache:
source = self.__supplementalData
for elt in source.findNodes('currencyData/region'):
iso, digits, rounding = '', 2, 1
try:
country = elt.dom.attributes['iso3166'].nodeValue
except KeyError:
continue
for child in elt.findAllChildren('currency'):
try:
if child.dom.attributes['tender'].nodeValue == 'false':
continue
except KeyError:
pass
try:
child.dom.attributes['to'] # Is set if this element has gone out of date.
except KeyError:
iso = child.dom.attributes['iso4217'].nodeValue
break
if iso:
for tag, data in source.find(
'currencyData/fractions/info[iso4217={}]'.format(iso)):
digits = data['digits']
rounding = data['rounding']
cache[country] = iso, digits, rounding
assert cache
return cache
def __scanLdmlDtd(self, joinPath = os.path.join):
"""Scan the LDML DTD, record CLDR version."""
with self.__open(('common', 'dtd', 'ldml.dtd')) as dtd:
@ -151,7 +552,8 @@ class CldrAccess (object):
if line.startswith('<!ATTLIST '):
parts = line.split()
if parts[1:5] == ['version', 'cldrVersion', 'CDATA', '#FIXED']:
# parts[5] is the version, in quotes, although the final > might be stuck on its end:
# parts[5] is the version, in quotes, maybe
# with a final > attached to its end:
self.__cldrVersion = parts[5].split('"')[1]
break
@ -178,5 +580,93 @@ class CldrAccess (object):
return cache[key]
def __codeMap(self, key, cache = {},
# Maps our name for it to CLDR's name:
naming = {'language': 'languages', 'script': 'scripts',
'country': 'territories', 'variant': 'variants'}):
if not cache:
root = self.xml('common', 'main', 'en.xml').root.findUniqueChild('localeDisplayNames')
for dst, src in naming.items():
cache[dst] = dict(self.__codeMapScan(root.findUniqueChild(src)))
assert cache
return cache[key]
def __codeMapScan(self, node):
"""Get mapping from codes to element values.
Passed in node is a <languages>, <scripts>, <territories> or
<variants> node, each child of which is a <language>,
<script>, <territory> or <variant> node as appropriate, whose
type is a code (of the appropriate flavour) and content is its
full name. In some cases, two child nodes have the same type;
in these cases, one always has an alt attribute and we should
prefer the other. Yields all such type, content pairs found
in node's children (skipping any with an alt attribute, if
their type has been seen previously)."""
seen = set()
for elt in node.dom.childNodes:
try:
key, value = elt.attributes['type'].nodeValue, elt.childNodes[0].wholeText
except (KeyError, ValueError, TypeError):
pass
else:
if key not in seen or not elt.attributes.has_key('alt'):
yield key, value
seen.add(key)
# CLDR uses inheritance between locales to save repetition:
def __parentLocale(self, name, cache = {}):
# see http://www.unicode.org/reports/tr35/#Parent_Locales
if not cache:
for tag, attrs in self.__supplementalData.find('parentLocales'):
parent = attrs.get('parent', '')
for child in attrs['locales'].split():
cache[child] = parent
assert cache
return cache[name]
def __localeAsDoc(self, name, aliasFor = None,
joinPath = os.path.join, exists = os.path.isfile):
path = ('common', 'main', name + '.xml')
if exists(joinPath(self.root, *path)):
elt = self.__xml(path)
for child in Node(elt).findAllChildren('alias'):
try:
alias = child.dom.attributes['source'].nodeValue
except (KeyError, AttributeError):
pass
else:
return self.__localeAsDoc(alias, aliasFor or name)
# No alias child with a source:
return elt
if aliasFor:
raise Error('Fatal error: found an alias "{}" -> "{}", but found no file for the alias'
.format(aliasFor, name))
def __scanLocaleRoots(self, name):
while name and name != 'root':
doc = self.__localeAsDoc(name)
if doc is not None:
yield Node(doc)
try:
name = self.__parentLocale(name)
except KeyError:
try:
name, tail = name.rsplit('_', 1)
except ValueError: # No tail to discard: we're done
break
class __Seq (list): pass # No weakref for tuple and list, but list sub-class is ok.
def __localeRoots(self, name, cache = CacheDict()):
try:
chain = cache[name]
except KeyError:
cache[name] = chain = self.__Seq(self.__scanLocaleRoots(name))
return chain
# Unpolute the namespace: we don't need to export these.
del minidom, CacheDict, os

636
util/locale_database/cldr2qlocalexml.py
View File

@ -2,7 +2,7 @@
# coding=utf8
#############################################################################
##
## Copyright (C) 2018 The Qt Company Ltd.
## Copyright (C) 2020 The Qt Company Ltd.
## Contact: https://www.qt.io/licensing/
##
## This file is part of the test suite of the Qt Toolkit.
@ -31,15 +31,17 @@
The CLDR data can be downloaded from CLDR_, which has a sub-directory
for each version; you need the ``core.zip`` file for your version of
choice (typically the latest). This script has had updates to cope up
to v35; for later versions, we may need adaptations. Unpack the
choice (typically the latest). This script has had updates to cope up
to v35; for later versions, we may need adaptations. Unpack the
downloaded ``core.zip`` and check it has a common/main/ sub-directory:
pass the path of that sub-directory to this script as its single
command-line argument. Save its standard output (but not error) to a
file for later processing by ``./qlocalexml2cpp.py``
pass the path of that root of the download to this script as its first
command-line argument. Pass the name of the file in which to write
output as the second argument; either omit it or use '-' to select the
standard output. This file is the input needed by
``./qlocalexml2cpp.py``
When you update the CLDR data, be sure to also update
src/corelib/text/qt_attribution.json's entry for unicode-cldr. Check
src/corelib/text/qt_attribution.json's entry for unicode-cldr. Check
this script's output for unknown language, country or script messages;
if any can be resolved, use their entry in common/main/en.xml to
append new entries to enumdata.py's lists and update documentation in
@ -53,610 +55,62 @@ time zone names; see cldr2qtimezone.py for details.
"""
import os
import sys
import re
import textwrap
import enumdata
from localetools import Error
from xpathlite import DraftResolution, findAlias, findEntry, findTagsInFile, codeMapsFromFile, \
_findEntryInFile as findEntryInFile
from dateconverter import convert_date
from qlocalexml import Locale, QLocaleXmlWriter
# TODO: make calendars a command-line option
calendars = ['gregorian', 'persian', 'islamic'] # 'hebrew'
def wrappedwarn(err, prefix, tokens):
return err.write(
'\n'.join(textwrap.wrap(prefix + ', '.join(tokens),
subsequent_indent=' ', width=80)) + '\n')
def parse_number_format(patterns, data):
# this is a very limited parsing of the number format for currency only.
def skip_repeating_pattern(x):
p = x.replace('0', '#').replace(',', '').replace('.', '')
seen = False
result = ''
for c in p:
if c == '#':
if seen:
continue
seen = True
else:
seen = False
result = result + c
return result
patterns = patterns.split(';')
result = []
for pattern in patterns:
pattern = skip_repeating_pattern(pattern)
pattern = pattern.replace('#', "%1")
# according to http://www.unicode.org/reports/tr35/#Number_Format_Patterns
# there can be doubled or trippled currency sign, however none of the
# locales use that.
pattern = pattern.replace(u'\xa4', "%2")
pattern = pattern.replace("''", "###").replace("'", '').replace("###", "'")
pattern = pattern.replace('-', data['minus'])
pattern = pattern.replace('+', data['plus'])
result.append(pattern)
return result
cldr_dir = None
def raiseUnknownCode(code, form, cache={}):
"""Check whether an unknown code could be supported.
We declare a language, script or country code unknown if it's not
known to enumdata.py; however, if it's present in main/en.xml's
mapping of codes to names, we have the option of adding support.
This caches the necessary look-up (so we only read main/en.xml
once) and returns the name we should use if we do add support.
First parameter, code, is the unknown code. Second parameter,
form, is one of 'language', 'script' or 'country' to select the
type of code to look up. Do not pass further parameters (the next
will deprive you of the cache).
Raises localetools.Error with a suitable message, that includes
the unknown code's full name if found.
Relies on global cldr_dir being set before it's called; see tail
of this file.
"""
if not cache:
cache.update(codeMapsFromFile(os.path.join(cldr_dir, 'en.xml')))
name = cache[form].get(code)
msg = 'unknown %s code "%s"' % (form, code)
if name:
msg += ' - could use "%s"' % name
raise Error(msg)
def parse_list_pattern_part_format(pattern):
# This is a very limited parsing of the format for list pattern part only.
return pattern.replace("{0}", "%1").replace("{1}", "%2").replace("{2}", "%3")
def unit_quantifiers(find, path, stem, suffix, known,
# Stop at exa/exbi: 16 exbi = 2^{64} < zetta =
# 1000^7 < zebi = 2^{70}, the next quantifiers up:
si_quantifiers = ('kilo', 'mega', 'giga', 'tera', 'peta', 'exa')):
"""Work out the unit quantifiers.
Unfortunately, the CLDR data only go up to terabytes and we want
all the way to exabytes; but we can recognize the SI quantifiers
as prefixes, strip and identify the tail as the localized
translation for 'B' (e.g. French has 'octet' for 'byte' and uses
ko, Mo, Go, To from which we can extrapolate Po, Eo).
Should be called first for the SI quantifiers, with suffix = 'B',
then for the IEC ones, with suffix = 'iB'; the list known
(initially empty before first call) is used to let the second call
know what the first learned about the localized unit.
"""
if suffix == 'B': # first call, known = []
tail = suffix
for q in si_quantifiers:
it = find(path, stem % q)
# kB for kilobyte, in contrast with KiB for IEC:
q = q[0] if q == 'kilo' else q[0].upper()
if not it:
it = q + tail
elif it.startswith(q):
rest = it[1:]
tail = rest if all(rest == k for k in known) else suffix
known.append(rest)
yield it
else: # second call, re-using first's known
assert suffix == 'iB'
if known:
byte = known.pop()
if all(byte == k for k in known):
suffix = 'i' + byte
for q in si_quantifiers:
yield find(path, stem % q[:2],
# Those don't (yet, v31) exist in CLDR, so we always fall back to:
q[0].upper() + suffix)
def generateLocaleInfo(path):
if not path.endswith(".xml"):
return {}
# skip legacy/compatibility ones
alias = findAlias(path)
if alias:
raise Error('Alias to "{}"'.format(alias))
def code(tag):
return findEntryInFile(path, 'identity/' + tag, attribute="type")[0]
return _generateLocaleInfo(path, code('language'), code('script'),
code('territory'), code('variant'))
def getNumberSystems(cache={}):
"""Cached look-up of number system information.
Pass no arguments. Returns a mapping from number system names to,
for each system, a mapping with keys 'digits', 'type' and 'id'.
Relies on global cldr_dir being set before it's first called.\n"""
if not cache:
for ns in findTagsInFile(os.path.join(cldr_dir, '..', 'supplemental',
'numberingSystems.xml'),
'numberingSystems'):
# ns has form: [u'numberingSystem', [(u'digits', u'0123456789'), (u'type', u'numeric'), (u'id', u'latn')]]
entry = dict(ns[1])
name = entry[u'id']
if u'digits' in entry and ord(entry[u'digits'][0]) > 0xffff:
# FIXME, QTBUG-69324: make this redundant:
# omit number system if zero doesn't fit in single-char16 UTF-16 :-(
sys.stderr.write('skipping number system "%s" [can\'t represent its zero, U+%X]\n'
% (name, ord(entry[u'digits'][0])))
else:
cache[name] = entry
return cache
def _generateLocaleInfo(path, language_code, script_code, country_code, variant_code=""):
if not path.endswith(".xml"):
return {}
if language_code == 'root':
# just skip it
return {}
# we do not support variants
# ### actually there is only one locale with variant: en_US_POSIX
# does anybody care about it at all?
if variant_code:
raise Error('We do not support variants ("{}")'.format(variant_code))
language_id = enumdata.languageCodeToId(language_code)
if language_id <= 0:
raiseUnknownCode(language_code, 'language')
script_id = enumdata.scriptCodeToId(script_code)
if script_id == -1:
raiseUnknownCode(script_code, 'script')
# we should handle fully qualified names with the territory
if not country_code:
return {}
country_id = enumdata.countryCodeToId(country_code)
if country_id <= 0:
raiseUnknownCode(country_code, 'country')
# So we say we accept only those values that have "contributed" or
# "approved" resolution. see http://www.unicode.org/cldr/process.html
# But we only respect the resolution for new datas for backward
# compatibility.
draft = DraftResolution.contributed
result = dict(
language=enumdata.language_list[language_id][0],
language_code=language_code, language_id=language_id,
script=enumdata.script_list[script_id][0],
script_code=script_code, script_id=script_id,
country=enumdata.country_list[country_id][0],
country_code=country_code, country_id=country_id,
variant_code=variant_code)
(dir_name, file_name) = os.path.split(path)
def from_supplement(tag,
path=os.path.join(dir_name, '..', 'supplemental',
'supplementalData.xml')):
return findTagsInFile(path, tag)
currencies = from_supplement('currencyData/region[iso3166=%s]' % country_code)
result['currencyIsoCode'] = ''
result['currencyDigits'] = 2
result['currencyRounding'] = 1
if currencies:
for e in currencies:
if e[0] == 'currency':
t = [x[1] == 'false' for x in e[1] if x[0] == 'tender']
if t and t[0]:
pass
elif not any(x[0] == 'to' for x in e[1]):
result['currencyIsoCode'] = (x[1] for x in e[1] if x[0] == 'iso4217').next()
break
if result['currencyIsoCode']:
t = from_supplement("currencyData/fractions/info[iso4217=%s]"
% result['currencyIsoCode'])
if t and t[0][0] == 'info':
result['currencyDigits'] = (int(x[1]) for x in t[0][1] if x[0] == 'digits').next()
result['currencyRounding'] = (int(x[1]) for x in t[0][1] if x[0] == 'rounding').next()
numbering_system = None
try:
numbering_system = findEntry(path, "numbers/defaultNumberingSystem")
except Error:
pass
def findEntryDef(path, xpath, value=''):
try:
return findEntry(path, xpath)
except Error:
return value
def get_number_in_system(path, xpath, numbering_system):
if numbering_system:
try:
return findEntry(path, xpath + "[numberSystem=" + numbering_system + "]")
except Error:
# in CLDR 1.9 number system was refactored for numbers (but not for currency)
# so if previous findEntry doesn't work we should try this:
try:
return findEntry(path, xpath.replace("/symbols/", "/symbols[numberSystem=" + numbering_system + "]/"))
except Error:
# fallback to default
pass
return findEntry(path, xpath)
result['decimal'] = get_number_in_system(path, "numbers/symbols/decimal", numbering_system)
result['group'] = get_number_in_system(path, "numbers/symbols/group", numbering_system)
assert result['decimal'] != result['group']
result['list'] = get_number_in_system(path, "numbers/symbols/list", numbering_system)
result['percent'] = get_number_in_system(path, "numbers/symbols/percentSign", numbering_system)
try:
result['zero'] = getNumberSystems()[numbering_system][u"digits"][0]
except Exception as e:
sys.stderr.write("Native zero detection problem: %s\n" % repr(e))
result['zero'] = get_number_in_system(path, "numbers/symbols/nativeZeroDigit", numbering_system)
result['minus'] = get_number_in_system(path, "numbers/symbols/minusSign", numbering_system)
result['plus'] = get_number_in_system(path, "numbers/symbols/plusSign", numbering_system)
result['exp'] = get_number_in_system(path, "numbers/symbols/exponential", numbering_system).lower()
result['quotationStart'] = findEntry(path, "delimiters/quotationStart")
result['quotationEnd'] = findEntry(path, "delimiters/quotationEnd")
result['alternateQuotationStart'] = findEntry(path, "delimiters/alternateQuotationStart")
result['alternateQuotationEnd'] = findEntry(path, "delimiters/alternateQuotationEnd")
result['listPatternPartStart'] = parse_list_pattern_part_format(findEntry(path, "listPatterns/listPattern/listPatternPart[start]"))
result['listPatternPartMiddle'] = parse_list_pattern_part_format(findEntry(path, "listPatterns/listPattern/listPatternPart[middle]"))
result['listPatternPartEnd'] = parse_list_pattern_part_format(findEntry(path, "listPatterns/listPattern/listPatternPart[end]"))
result['listPatternPartTwo'] = parse_list_pattern_part_format(findEntry(path, "listPatterns/listPattern/listPatternPart[2]"))
result['am'] = findEntry(path, "dates/calendars/calendar[gregorian]/dayPeriods/dayPeriodContext[format]/dayPeriodWidth[wide]/dayPeriod[am]", draft)
result['pm'] = findEntry(path, "dates/calendars/calendar[gregorian]/dayPeriods/dayPeriodContext[format]/dayPeriodWidth[wide]/dayPeriod[pm]", draft)
result['longDateFormat'] = convert_date(findEntry(path, "dates/calendars/calendar[gregorian]/dateFormats/dateFormatLength[full]/dateFormat/pattern"))
result['shortDateFormat'] = convert_date(findEntry(path, "dates/calendars/calendar[gregorian]/dateFormats/dateFormatLength[short]/dateFormat/pattern"))
result['longTimeFormat'] = convert_date(findEntry(path, "dates/calendars/calendar[gregorian]/timeFormats/timeFormatLength[full]/timeFormat/pattern"))
result['shortTimeFormat'] = convert_date(findEntry(path, "dates/calendars/calendar[gregorian]/timeFormats/timeFormatLength[short]/timeFormat/pattern"))
endonym = None
if country_code and script_code:
endonym = findEntryDef(path, "localeDisplayNames/languages/language[type=%s_%s_%s]" % (language_code, script_code, country_code))
if not endonym and script_code:
endonym = findEntryDef(path, "localeDisplayNames/languages/language[type=%s_%s]" % (language_code, script_code))
if not endonym and country_code:
endonym = findEntryDef(path, "localeDisplayNames/languages/language[type=%s_%s]" % (language_code, country_code))
if not endonym:
endonym = findEntryDef(path, "localeDisplayNames/languages/language[type=%s]" % (language_code))
result['languageEndonym'] = endonym
result['countryEndonym'] = findEntryDef(path, "localeDisplayNames/territories/territory[type=%s]" % (country_code))
currency_format = get_number_in_system(path, "numbers/currencyFormats/currencyFormatLength/currencyFormat/pattern", numbering_system)
currency_format = parse_number_format(currency_format, result)
result['currencyFormat'] = currency_format[0]
result['currencyNegativeFormat'] = ''
if len(currency_format) > 1:
result['currencyNegativeFormat'] = currency_format[1]
result['currencySymbol'] = ''
result['currencyDisplayName'] = ''
if result['currencyIsoCode']:
result['currencySymbol'] = findEntryDef(path, "numbers/currencies/currency[%s]/symbol" % result['currencyIsoCode'])
result['currencyDisplayName'] = ';'.join(
findEntryDef(path, 'numbers/currencies/currency[' + result['currencyIsoCode']
+ ']/displayName' + tail)
for tail in ['',] + [
'[count=%s]' % x for x in ('zero', 'one', 'two', 'few', 'many', 'other')
]) + ';'
def findUnitDef(path, stem, fallback=''):
# The displayName for a quantified unit in en.xml is kByte
# instead of kB (etc.), so prefer any unitPattern provided:
for count in ('many', 'few', 'two', 'other', 'zero', 'one'):
try:
ans = findEntry(path, stem + 'unitPattern[count=%s]' % count)
except Error:
continue
# TODO: epxloit count-handling, instead of discarding placeholders
if ans.startswith('{0}'):
ans = ans[3:].lstrip()
if ans:
return ans
return findEntryDef(path, stem + 'displayName', fallback)
# First without quantifier, then quantified each way:
result['byte_unit'] = findEntryDef(
path, 'units/unitLength[type=long]/unit[type=digital-byte]/displayName',
'bytes')
stem = 'units/unitLength[type=short]/unit[type=digital-%sbyte]/'
known = [] # cases where we *do* have a given version:
result['byte_si_quantified'] = ';'.join(unit_quantifiers(findUnitDef, path, stem, 'B', known))
# IEC 60027-2
# http://physics.nist.gov/cuu/Units/binary.html
result['byte_iec_quantified'] = ';'.join(unit_quantifiers(findUnitDef, path, stem % '%sbi', 'iB', known))
# Used for month and day data:
namings = (
('standaloneLong', 'stand-alone', 'wide'),
('standaloneShort', 'stand-alone', 'abbreviated'),
('standaloneNarrow', 'stand-alone', 'narrow'),
('long', 'format', 'wide'),
('short', 'format', 'abbreviated'),
('narrow', 'format', 'narrow'),
)
# Month names for 12-month calendars:
for cal in calendars:
stem = 'dates/calendars/calendar[' + cal + ']/months/'
for (key, mode, size) in namings:
prop = 'monthContext[' + mode + ']/monthWidth[' + size + ']/'
result[key + 'Months_' + cal] = ';'.join(
findEntry(path, stem + prop + "month[%d]" % i)
for i in range(1, 13)) + ';'
# Day data (for Gregorian, at least):
stem = 'dates/calendars/calendar[gregorian]/days/'
days = ('sun', 'mon', 'tue', 'wed', 'thu', 'fri', 'sat')
for (key, mode, size) in namings:
prop = 'dayContext[' + mode + ']/dayWidth[' + size + ']/day'
result[key + 'Days'] = ';'.join(
findEntry(path, stem + prop + '[' + day + ']')
for day in days) + ';'
return Locale(result)
def integrateWeekData(filePath, locale_database):
if not filePath.endswith(".xml"):
return {}
def lookup(key):
return findEntryInFile(filePath, key, attribute='territories')[0].split()
days = ('mon', 'tue', 'wed', 'thu', 'fri', 'sat', 'sun')
firstDayByCountryCode = {}
for day in days:
for countryCode in lookup('weekData/firstDay[day=%s]' % day):
firstDayByCountryCode[countryCode] = day
weekendStartByCountryCode = {}
for day in days:
for countryCode in lookup('weekData/weekendStart[day=%s]' % day):
weekendStartByCountryCode[countryCode] = day
weekendEndByCountryCode = {}
for day in days:
for countryCode in lookup('weekData/weekendEnd[day=%s]' % day):
weekendEndByCountryCode[countryCode] = day
for (key, locale) in locale_database.iteritems():
countryCode = locale.country_code
if countryCode in firstDayByCountryCode:
locale.firstDayOfWeek = firstDayByCountryCode[countryCode]
else:
locale.firstDayOfWeek = firstDayByCountryCode["001"]
if countryCode in weekendStartByCountryCode:
locale.weekendStart = weekendStartByCountryCode[countryCode]
else:
locale.weekendStart = weekendStartByCountryCode["001"]
if countryCode in weekendEndByCountryCode:
locale.weekendEnd = weekendEndByCountryCode[countryCode]
else:
locale.weekendEnd = weekendEndByCountryCode["001"]
def splitLocale(name):
"""Split name into (language, script, territory) triple as generator.
Ignores any trailing fields (with a warning), leaves script (a capitalised
four-letter token) or territory (either a number or an all-uppercase token)
empty if unspecified, returns a single-entry generator if name is a single
tag (i.e. contains no underscores). Always yields 1 or 3 values, never 2."""
tags = iter(name.split('_'))
yield tags.next() # Language
tag = tags.next()
# Script is always four letters, always capitalised:
if len(tag) == 4 and tag[0].isupper() and tag[1:].islower():
yield tag
try:
tag = tags.next()
except StopIteration:
tag = ''
else:
yield ''
# Territory is upper-case or numeric:
if tag and tag.isupper() or tag.isdigit():
yield tag
tag = ''
else:
yield ''
# If nothing is left, StopIteration will avoid the warning:
tag = (tag if tag else tags.next(),)
sys.stderr.write('Ignoring unparsed cruft %s in %s\n' % ('_'.join(tag + tuple(tags)), name))
def _parseLocale(l):
language = "AnyLanguage"
script = "AnyScript"
country = "AnyCountry"
if l == "und":
raise Error('We treat unknown locale like C')
parsed = splitLocale(l)
language_code = parsed.next()
script_code = country_code = ''
try:
script_code, country_code = parsed
except ValueError:
pass
if language_code != "und":
language_id = enumdata.languageCodeToId(language_code)
if language_id == -1:
raise Error('Unknown language code "{}"'.format(language_code))
language = enumdata.language_list[language_id][0]
if script_code:
script_id = enumdata.scriptCodeToId(script_code)
if script_id == -1:
raise Error('Unknown script code "{}"'.format(script_code))
script = enumdata.script_list[script_id][0]
if country_code:
country_id = enumdata.countryCodeToId(country_code)
if country_id == -1:
raise Error('Unknown country code "{}"'.format(country_code))
country = enumdata.country_list[country_id][0]
return (language, script, country)
def likelySubtags(root, err):
skips = []
for ns in findTagsInFile(os.path.join(root, 'supplemental', 'likelySubtags.xml'), "likelySubtags"):
tmp = {}
for data in ns[1:][0]: # ns looks like this: [u'likelySubtag', [(u'from', u'aa'), (u'to', u'aa_Latn_ET')]]
tmp[data[0]] = data[1]
try:
from_language, from_script, from_country = _parseLocale(tmp[u"from"])
to_language, to_script, to_country = _parseLocale(tmp[u"to"])
except Error as e:
if (tmp['to'].startswith(tmp['from'])
and e.message == 'Unknown language code "{}"'.format(tmp['from'])):
skips.append(tmp['to'])
else:
sys.stderr.write('skipping likelySubtag "{}" -> "{}" ({})\n'.format(
tmp[u"from"], tmp[u"to"], e.message))
continue
# substitute according to http://www.unicode.org/reports/tr35/#Likely_Subtags
if to_country == "AnyCountry" and from_country != to_country:
to_country = from_country
if to_script == "AnyScript" and from_script != to_script:
to_script = from_script
yield ((from_language, from_script, from_country),
(to_language, to_script, to_country))
if skips:
wrappedwarn(err, 'skipping likelySubtags (for unknown language codes): ', skips)
from cldr import CldrReader
from qlocalexml import QLocaleXmlWriter
from enumdata import language_list, script_list, country_list
def usage(err, name, message = ''):
err.write("""Usage: {} <path-to-cldr-main> [out-file.xml]
""".format(name)) # TODO: expand
err.write("""Usage: {} path/to/cldr/common/main [out-file.xml]
""".format(name)) # TODO: expand command-line, improve help message
if message:
err.write('\n' + message + '\n')
def main(args, out, err):
# TODO: make calendars a command-line option
calendars = ['gregorian', 'persian', 'islamic'] # 'hebrew'
# TODO: make argument parsing more sophisticated
name = args.pop(0)
if len(args) < 1:
usage(err, name)
if not args:
usage(name, err, 'Where is your CLDR data tree ?')
return 1
global cldr_dir
cldr_dir = args.pop(0)
if not os.path.isdir(cldr_dir):
usage(err, name, 'Where did you unpack the CLDR data files ?')
root = args.pop(0)
if not os.path.exists(os.path.join(root, 'common', 'main', 'root.xml')):
usage(name, err,
'First argument is the root of the CLDR tree: found no common/main/root.xml under '
+ root)
return 1
if len(args) > 1:
usage(err, name, 'Too many arguments passed')
xml = args.pop(0) if args else None
if not xml or xml == '-':
emit = out
elif not xml.endswith('.xml'):
usage(name, err, 'Please use a .xml extension on your output file name, not ' + xml)
return 1
if args:
qxml = open(args.pop(0), 'w')
else:
qxml = out
getNumberSystems(cldr_dir)
cldr_files = os.listdir(cldr_dir)
locale_database = {}
# see http://www.unicode.org/reports/tr35/tr35-info.html#Default_Content
defaultContent_locales = []
for ns in findTagsInFile(os.path.join(cldr_dir, '..', 'supplemental',
'supplementalMetadata.xml'),
'metadata/defaultContent'):
for data in ns[1:][0]:
if data[0] == u"locales":
defaultContent_locales += data[1].split()
skips = []
for file in defaultContent_locales:
try:
language_code, script_code, country_code = splitLocale(file)
except ValueError:
sys.stderr.write('skipping defaultContent locale "' + file + '" [neither two nor three tags]\n')
continue
emit = open(xml, 'w')
except IOError as e:
usage(name, err, 'Failed to open "{}" to write output to it\n'.format(xml))
return 1
if not (script_code or country_code):
sys.stderr.write('skipping defaultContent locale "' + file + '" [second tag is neither script nor territory]\n')
continue
if args:
usage(name, err, 'Too many arguments - excess: ' + ' '.join(args))
return 1
try:
l = _generateLocaleInfo(cldr_dir + "/" + file + ".xml", language_code, script_code, country_code)
if not l:
skips.append(file)
continue
except Error as e:
sys.stderr.write('skipping defaultContent locale "{}" ({})\n'.format(file, e.message))
continue
# TODO - command line options to tune choice of grumble and whitter:
reader = CldrReader(root, err.write, err.write)
writer = QLocaleXmlWriter(emit.write)
locale_database[(l.language_id, l.script_id, l.country_id, l.variant_code)] = l
if skips:
wrappedwarn(err, 'skipping defaultContent locales [no locale info generated]: ', skips)
skips = []
for file in cldr_files:
try:
l = generateLocaleInfo(cldr_dir + "/" + file)
if not l:
skips.append(file)
continue
except Error as e:
sys.stderr.write('skipping file "{}" ({})\n'.format(file, e.message))
continue
locale_database[(l.language_id, l.script_id, l.country_id, l.variant_code)] = l
if skips:
wrappedwarn(err, 'skipping files [no locale info generated]: ', skips)
integrateWeekData(cldr_dir + "/../supplemental/supplementalData.xml", locale_database)
cldr_version = 'unknown'
with open(cldr_dir+"/../dtd/ldml.dtd", "r") as ldml:
for line in ldml:
if 'version cldrVersion CDATA #FIXED' in line:
cldr_version = line.split('"')[1]
xmlOut = QLocaleXmlWriter(qxml.write)
xmlOut.version(cldr_version)
xmlOut.enumData(enumdata.language_list,
enumdata.script_list,
enumdata.country_list)
xmlOut.likelySubTags(likelySubtags(os.path.split(cldr_dir)[0], err))
xmlOut.locales(locale_database, calendars)
xmlOut.close()
if qxml is not out:
qxml.close()
writer.version(reader.root.cldrVersion)
writer.enumData(language_list, script_list, country_list)
writer.likelySubTags(reader.likelySubTags())
writer.locales(reader.readLocales(calendars), calendars)
writer.close()
return 0
if __name__ == '__main__':

450
util/locale_database/ldml.py
View File

@ -39,10 +39,12 @@ returned by minidom.parse() and their child-nodes:
Node -- wraps any node in the DOM tree
XmlScanner -- wraps the root element of a stand-alone XML file
Supplement -- specializes XmlScanner for supplemental data files
LocaleScanner -- wraps a locale's inheritance-chain of file roots
See individual classes for further detail.
"""
from localetools import Error
from dateconverter import convert_date
class Node (object):
"""Wrapper for an arbitrary DOM node.
@ -51,11 +53,20 @@ class Node (object):
nodes are returned wrapped as Node objects. A Node exposes the
raw DOM node it wraps via its .dom attribute."""
def __init__(self, elt):
def __init__(self, elt, draft = 0):
"""Wraps a DOM node for ease of access.
Single argument, elt, is the DOM node to wrap."""
First argument, elt, is the DOM node to wrap. (Optional second
argument, draft, should only be supplied by this class's
creation of child nodes; it is the maximum draft score of any
ancestor of the new node.)"""
self.dom = elt
try:
attr = elt.attributes['draft'].nodeValue
except KeyError:
self.draft = draft
else:
self.draft = max(draft, self.draftScore(attr))
def findAllChildren(self, tag, wanted = None):
"""All children that do have the given tag and attributes.
@ -65,34 +76,60 @@ class Node (object):
Optional second argument, wanted, should either be None or map
attribute names to the values they must have. Only child nodes
with these attributes set to the given values are yielded."""
with thes attributes set to the given values are yielded."""
cutoff = 4 # Only accept approved, for now
for child in self.dom.childNodes:
if child.nodeType != child.ELEMENT_NODE:
continue
if child.nodeName != tag:
continue
try:
draft = child.attributes['draft']
except KeyError:
pass
else:
if self.__draftScores.get(draft, 0) < cutoff:
continue
if wanted is not None:
if wanted:
try:
if wanted and any(child.attributes[k].nodeValue != v for k, v in wanted.items()):
if any(child.attributes[k].nodeValue != v
for k, v in wanted.items()):
continue
except KeyError: # Some wanted attribute is missing
continue
yield Node(child)
yield Node(child, self.draft)
__draftScores = dict(true = 0, unconfirmed = 1, provisional = 2,
contributed = 3, approved = 4, false = 4)
def findUniqueChild(self, tag):
"""Returns the single child with the given nodeName.
Raises Error if there is no such child or there is more than
one."""
seq = self.findAllChildren(tag)
try:
node = seq.next()
except StopIteration:
raise Error('No child found where one was expected', tag)
for it in seq:
raise Error('Many children found where only one was expected', tag)
return node
@classmethod
def draftScore(cls, level):
"""Maps draft level names to numeric scores.
Single parameter, level, is the least sure value of the draft
attribute on a node that you're willing to accept; returns a
numeric value (lower is less drafty).
Tempting as it is to insist on low draft scores, there are
many locales in which pretty much every leaf is
unconfirmed. It may make sense to actually check each
XmlScanner object, or each node in each LocaleScanner's nodes
list, to see what its distribution of draft level looks like,
so as to set the acceptable draft score for its elements
accordingly. However, for the moment, we mostly just accept
all elements, regardless of draft values (the one exception is
am/pm indicators)."""
return cls.__draftScores.get(level, 5) if level else 0
# Implementation details:
__draftScores = dict(true = 4, unconfirmed = 3, provisional = 2,
contributed = 1, approved = 0, false = 0)
def _parseXPath(selector):
# Split "tag[attr=val][...]" into tag-name and attribute mapping
@ -129,7 +166,6 @@ class XmlScanner (object):
return elts
class Supplement (XmlScanner):
# Replaces xpathlite.findTagsInFile()
def find(self, xpath):
elts = self.findNodes(xpath)
for elt in _iterateEach(e.dom.childNodes if e.dom.childNodes else (e.dom,)
@ -138,3 +174,381 @@ class Supplement (XmlScanner):
yield (elt.nodeName,
dict((k, v if isinstance(v, basestring) else v.nodeValue)
for k, v in elt.attributes.items()))
class LocaleScanner (object):
def __init__(self, name, nodes, root):
self.name, self.nodes, self.base = name, nodes, root
def find(self, xpath, draft = None):
tags = xpath.split('/')
while True:
replace = None
for elt in self.nodes:
for selector in tags:
tag, attrs = _parseXPath(selector)
for elt in elt.findAllChildren(tag, attrs):
if draft is None or elt.draft <= draft:
break # and process the next selector
else:
break # no child, try next elt in self.nodes
else:
# processed all selectors
try:
return elt.dom.firstChild.nodeValue
except (AttributeError, KeyError):
pass # move on to next elt in self.nodes
# No match in self.nodes; check root
elt = self.base.root
for i, selector in enumerate(tags):
tag, attrs = _parseXPath(selector)
for alias in elt.findAllChildren('alias'):
if alias.dom.attributes['source'].nodeValue == 'locale':
replace = alias.dom.attributes['path'].nodeValue.split('/')
tags = self.__xpathJoin(tags[:i], replace, tags[i:])
break
else:
for elt in elt.findAllChildren(tag, attrs):
if draft is None or elt.draft <= draft:
break # and process the next selector
else:
break
if replace:
break
else:
# processed all selectors
try:
return elt.dom.firstChild.nodeValue
except (AttributeError, KeyError):
# No match
pass
if not replace:
break
sought = '/'.join(tags)
if sought != xpath:
sought += ' (for {})'.format(xpath)
raise Error('No {} in {}'.format(sought, self.name))
def findOr(self, xpath, fallback = ''):
"""Use a fall-back value if we don't find data.
Like find, but takes a fall-back value to return instead of
raising Error on failure."""
try:
return self.find(xpath)
except Error:
return fallback
def tagCodes(self):
"""Yields four tag codes
The tag codes are language, script, country and variant; an
empty value for any of them indicates that no value was
provided. The values are obtained from the primary file's
top-level <identity> element. An Error is raised if any
top-level <alias> element of this file has a non-empty source
attribute; that attribute value is mentioned in the error's
message."""
root = self.nodes[0]
for alias in root.findAllChildren('alias'):
try:
source = alias.dom.attributes['source'].nodeValue
except (KeyError, AttributeError):
pass
else:
raise Error('Alias to {}'.format(source))
ids = root.findUniqueChild('identity')
for code in ('language', 'script', 'territory', 'variant'):
for node in ids.findAllChildren(code):
try:
yield node.dom.attributes['type'].nodeValue
except (KeyError, AttributeError):
pass
else:
break # only want one value for each code
else: # No value for this code, use empty
yield ''
def currencyData(self, isoCode):
"""Fetches currency data for this locale.
Single argument, isoCode, is the ISO currency code for the
currency in use in the country. See also numericData, which
includes some currency formats.
"""
if isoCode:
stem = 'numbers/currencies/currency[{}]/'.format(isoCode)
symbol = self.findOr(stem + 'symbol')
name = ';'.join(
self.findOr(stem + 'displayName' + tail)
for tail in ('',) + tuple(
'[count={}]'.format(x) for x in ('zero', 'one', 'two', 'few', 'many', 'other')
)) + ';'
else:
symbol = name = ''
yield 'currencySymbol', symbol
yield 'currencyDisplayName', name
def numericData(self, lookup, complain = lambda text: None):
"""Generate assorted numeric data for the locale.
First argument, lookup, is a callable that maps a numbering
system's name to certain data about the system, as a mapping;
we expect this to have u'digits' as a key.
"""
system = self.find('numbers/defaultNumberingSystem')
stem = 'numbers/symbols[numberSystem={}]/'.format(system)
decimal = self.find(stem + 'decimal')
group = self.find(stem + 'group')
assert decimal != group, (self.name, system, decimal)
yield 'decimal', decimal
yield 'group', group
yield 'percent', self.find(stem + 'percentSign')
yield 'list', self.find(stem + 'list')
# FIXME: don't lower-case:
yield 'exp', self.find(stem + 'exponential').lower()
digits = lookup(system)['digits']
assert len(digits) == 10
zero = digits[0]
# Qt's number-formatting code assumes digits are consecutive:
assert all(ord(c) == i for i, c in enumerate(digits, ord(zero)))
yield 'zero', zero
plus = self.find(stem + 'plusSign')
minus = self.find(stem + 'minusSign')
yield 'plus', plus
yield 'minus', minus
# Currency formatting (currencyFormat may have a type field):
money = self.find('numbers/currencyFormats/currencyFormatLength/currencyFormat/pattern')
money = self.__currencyFormats(money, plus, minus)
yield 'currencyFormat', money.next()
neg = ''
for it in money:
assert not neg, 'There should be at most one more pattern'
neg = it
yield 'currencyNegativeFormat', neg
def textPatternData(self):
for key in ('quotationStart', 'alternateQuotationEnd',
'quotationEnd', 'alternateQuotationStart'):
yield key, self.find('delimiters/' + key)
for key in ('start', 'middle', 'end'):
yield ('listPatternPart' + key.capitalize(),
self.__fromLdmlListPattern(self.find(
'listPatterns/listPattern/listPatternPart[{}]'.format(key))))
yield ('listPatternPartTwo',
self.__fromLdmlListPattern(self.find(
'listPatterns/listPattern/listPatternPart[2]')))
stem = 'dates/calendars/calendar[gregorian]/'
# TODO: is wide really the right width to use here ?
# abbreviated might be an option ... or try both ?
meridiem = stem + 'dayPeriods/dayPeriodContext[format]/dayPeriodWidth[wide]/'
for key in ('am', 'pm'):
yield key, self.find(meridiem + 'dayPeriod[{}]'.format(key),
draft = Node.draftScore('contributed'))
for pair in (('long', 'full'), ('short', 'short')):
for key in ('time', 'date'):
yield (pair[0] + key.capitalize() + 'Format',
convert_date(self.find(
stem + '{}Formats/{}FormatLength[{}]/{}Format/pattern'.format(
key, key, pair[1], key))))
def endonyms(self, language, script, country, variant):
# TODO: take variant into account ?
for seq in ((language, script, country),
(language, script), (language, country), (language,)):
if not all(seq):
continue
try:
yield ('languageEndonym',
self.find('localeDisplayNames/languages/language[{}]'
.format('_'.join(seq))))
except Error:
pass
else:
break
else:
# grumble(failed to find endonym for language)
yield 'languageEndonym', ''
yield ('countryEndonym',
self.findOr('localeDisplayNames/territories/territory[{}]'
.format(country)))
def unitData(self):
yield ('byte_unit',
self.findOr('units/unitLength[long]/unit[digital-byte]/displayName',
'bytes'))
unit = self.__findUnit('', 'B')
cache = [] # Populated by the SI call, to give hints to the IEC call
yield ('byte_si_quantified',
';'.join(self.__unitCount('', unit, cache)))
# IEC 60027-2
# http://physics.nist.gov/cuu/Units/binary.html
yield ('byte_iec_quantified',
';'.join(self.__unitCount('bi', 'iB', cache)))
def calendarNames(self, calendars):
namings = self.__nameForms
for cal in calendars:
stem = 'dates/calendars/calendar[' + cal + ']/months/'
for key, mode, size in namings:
prop = 'monthContext[' + mode + ']/monthWidth[' + size + ']/'
yield (key + 'Months_' + cal,
';'.join(self.find(stem + prop + 'month[{}]'.format(i))
for i in range(1, 13)) + ';')
# Day data (for Gregorian, at least):
stem = 'dates/calendars/calendar[gregorian]/days/'
days = ('sun', 'mon', 'tue', 'wed', 'thu', 'fri', 'sat')
for (key, mode, size) in namings:
prop = 'dayContext[' + mode + ']/dayWidth[' + size + ']/day'
yield (key + 'Days',
';'.join(self.find(stem + prop + '[' + day + ']')
for day in days) + ';')
# Implementation details
__nameForms = (
('standaloneLong', 'stand-alone', 'wide'),
('standaloneShort', 'stand-alone', 'abbreviated'),
('standaloneNarrow', 'stand-alone', 'narrow'),
('long', 'format', 'wide'),
('short', 'format', 'abbreviated'),
('narrow', 'format', 'narrow'),
) # Used for month and day names
def __findUnit(self, keySuffix, quantify, fallback=''):
# The displayName for a quantified unit in en.xml is kByte
# (even for unitLength[narrow]) instead of kB (etc.), so
# prefer any unitPattern provided, but prune its placeholder:
for size in ('short', 'narrow'): # TODO: reverse order ?
stem = 'units/unitLength[{}]/unit[digital-{}byte]/'.format(size + keySuffix, quantify)
for count in ('many', 'few', 'two', 'other', 'zero', 'one'):
try:
ans = self.find(stem + 'unitPattern[count={}]'.format(count))
except Error:
continue
# TODO: do count-handling, instead of discarding placeholders
if False: # TODO: do it this way, instead !
ans = ans.replace('{0}', '').strip()
elif ans.startswith('{0}'):
ans = ans[3:].lstrip()
if ans:
return ans
try:
return self.find(stem + 'displayName')
except Error:
pass
return fallback
def __unitCount(self, keySuffix, suffix, cache,
# Stop at exa/exbi: 16 exbi = 2^{64} < zetta =
# 1000^7 < zebi = 2^{70}, the next quantifiers up:
siQuantifiers = ('kilo', 'mega', 'giga', 'tera', 'peta', 'exa')):
"""Work out the unit quantifiers.
Unfortunately, the CLDR data only go up to terabytes and we
want all the way to exabytes; but we can recognize the SI
quantifiers as prefixes, strip and identify the tail as the
localized translation for 'B' (e.g. French has 'octet' for
'byte' and uses ko, Mo, Go, To from which we can extrapolate
Po, Eo).
Should be called first for the SI quantifiers, with suffix =
'B', then for the IEC ones, with suffix = 'iB'; the list cache
(initially empty before first call) is used to let the second
call know what the first learned about the localized unit.
"""
if suffix == 'iB': # second call, re-using first's cache
if cache:
byte = cache.pop()
if all(byte == k for k in cache):
suffix = 'i' + byte
for q in siQuantifiers:
# Those don't (yet, v36) exist in CLDR, so we always get the fall-back:
yield self.__findUnit(keySuffix, q[:2], q[0].upper() + suffix)
else: # first call
tail = suffix = suffix or 'B'
for q in siQuantifiers:
it = self.__findUnit(keySuffix, q)
# kB for kilobyte, in contrast with KiB for IEC:
q = q[0] if q == 'kilo' else q[0].upper()
if not it:
it = q + tail
elif it.startswith(q):
rest = it[1:]
tail = rest if all(rest == k for k in cache) else suffix
cache.append(rest)
yield it
@staticmethod
def __currencyFormats(patterns, plus, minus):
for p in patterns.split(';'):
p = p.replace('0', '#').replace(',', '').replace('.', '')
try:
cut = p.find('#') + 1
except ValueError:
pass
else:
p = p[:cut] + p[cut:].replace('#', '')
p = p.replace('#', "%1")
# According to http://www.unicode.org/reports/tr35/#Number_Format_Patterns
# there can be doubled or trippled currency sign, however none of the
# locales use that.
p = p.replace(u'\xa4', "%2")
# Single quote goes away, but double goes to single:
p = p.replace("''", '###').replace("'", '').replace('###', "'")
# Use number system's signs:
p = p.replace('+', plus).replace('-', minus)
yield p
@staticmethod
def __fromLdmlListPattern(pattern):
# This is a very limited parsing of the format for list pattern part only.
return pattern.replace('{0}', '%1').replace('{1}', '%2').replace('{2}', '%3')
@staticmethod
def __fromLdmlPath(seq): # tool function for __xpathJoin()
"""Convert LDML's [@name='value'] to our [name=value] form."""
for it in seq:
# First dismember it:
attrs = it.split('[')
tag = attrs.pop(0)
if not attrs: # Short-cut the easy case:
yield it
continue
assert all(x.endswith(']') for x in attrs)
attrs = [x[:-1].split('=') for x in attrs]
# Then fix each attribute specification in it:
attrs = [(x[0][1:] if x[0].startswith('@') else x[0],
x[1][1:-1] if x[1].startswith("'") and x[1].endswith("'") else x[1])
for x in attrs]
# Finally, put it all back together:
attrs = ['='.join(x) + ']' for x in attrs]
attrs.insert(0, tag)
yield '['.join(attrs)
@classmethod
def __xpathJoin(cls, head, insert, tail):
"""Join three lists of XPath selectors.
Each of head, insert and tail is a sequence of selectors but
insert may start with some uses of '..', that we want to
resolve away, and may use LDML's attribute format, that we
want to convert to our format."""
while insert and insert[0] == '..':
insert.pop(0)
head.pop()
return head + list(cls.__fromLdmlPath(insert)) + tail

2
util/locale_database/qlocalexml2cpp.py
View File

@ -480,7 +480,7 @@ def main(args, out, err):
return 1
reader = QLocaleXmlReader(qlocalexml)
locale_map = dict(reader.loadLocaleMap(calendars, sys.stderr.write))
locale_map = dict(reader.loadLocaleMap(calendars, err.write))
locale_keys = locale_map.keys()
compareLocaleKeys.default_map = dict(reader.defaultMap())

284
util/locale_database/xpathlite.py
View File

@ -1,284 +0,0 @@
#!/usr/bin/env python
#############################################################################
##
## Copyright (C) 2016 The Qt Company Ltd.
## Contact: https://www.qt.io/licensing/
##
## This file is part of the test suite of the Qt Toolkit.
##
## $QT_BEGIN_LICENSE:GPL-EXCEPT$
## Commercial License Usage
## Licensees holding valid commercial Qt licenses may use this file in
## accordance with the commercial license agreement provided with the
## Software or, alternatively, in accordance with the terms contained in
## a written agreement between you and The Qt Company. For licensing terms
## and conditions see https://www.qt.io/terms-conditions. For further
## information use the contact form at https://www.qt.io/contact-us.
##
## GNU General Public License Usage
## Alternatively, this file may be used under the terms of the GNU
## General Public License version 3 as published by the Free Software
## Foundation with exceptions as appearing in the file LICENSE.GPL3-EXCEPT
## included in the packaging of this file. Please review the following
## information to ensure the GNU General Public License requirements will
## be met: https://www.gnu.org/licenses/gpl-3.0.html.
##
## $QT_END_LICENSE$
##
#############################################################################
import sys
import os
import xml.dom.minidom
from localetools import Error
class DraftResolution:
# See http://www.unicode.org/cldr/process.html for description
unconfirmed = 'unconfirmed'
provisional = 'provisional'
contributed = 'contributed'
approved = 'approved'
_values = { unconfirmed : 1, provisional : 2, contributed : 3, approved : 4 }
def __init__(self, resolution):
self.resolution = resolution
def toInt(self):
return DraftResolution._values[self.resolution]
doc_cache = {}
def parseDoc(file):
if not doc_cache.has_key(file):
doc_cache[file] = xml.dom.minidom.parse(file)
return doc_cache[file]
def findChild(parent, tag_name, arg_name=None, arg_value=None, draft=None):
for node in parent.childNodes:
if node.nodeType != node.ELEMENT_NODE:
continue
if node.nodeName != tag_name:
continue
if arg_value:
if not node.attributes.has_key(arg_name):
continue
if node.attributes[arg_name].nodeValue != arg_value:
continue
if draft:
if not node.attributes.has_key('draft'):
# if draft is not specified then it's approved
return node
value = node.attributes['draft'].nodeValue
value = DraftResolution(value).toInt()
exemplar = DraftResolution(draft).toInt()
if exemplar > value:
continue
return node
return False
def codeMapsFromFile(file):
"""Extract mappings of language, script and country codes to names.
The file shall typically be common/main/en.xml, which contains a
localeDisplayNames element with children languages, scripts and
territories; each element in each of these has a code as its type
attribute and its name as element content. This returns a mapping
withe keys 'language', 'script' and 'country', each of which
has, as value, a mapping of the relevant codes to names.
"""
parent = findChild(findChild(parseDoc(file), 'ldml'), 'localeDisplayNames')
keys, result = {'languages': 'language', 'scripts': 'script', 'territories': 'country'}, {}
for src, dst in keys.items():
child = findChild(parent, src)
data = result[dst] = {}
for elt in child.childNodes:
if elt.attributes and elt.attributes.has_key('type'):
key, value = elt.attributes['type'].value, elt.childNodes[0].wholeText
# Don't over-write previously-read data for an alt form:
if elt.attributes.has_key('alt') and data.has_key(key):
continue
data[key] = value
return result
def findTagsInFile(file, path):
doc = parseDoc(file)
elt = doc.documentElement
tag_spec_list = path.split("/")
last_entry = None
for tag_spec in tag_spec_list:
tag_name = tag_spec
arg_name = 'type'
arg_value = ''
left_bracket = tag_spec.find('[')
if left_bracket != -1:
tag_name = tag_spec[:left_bracket]
arg_value = tag_spec[left_bracket+1:-1].split("=")
if len(arg_value) == 2:
arg_name = arg_value[0]
arg_value = arg_value[1]
else:
arg_value = arg_value[0]
elt = findChild(elt, tag_name, arg_name, arg_value)
if not elt:
return None
ret = []
if elt.childNodes:
for node in elt.childNodes:
if node.attributes:
element = [node.nodeName, None]
element[1] = node.attributes.items()
ret.append(element)
else:
if elt.attributes:
element = [elt.nodeName, None]
element[1] = elt.attributes.items()
ret.append(element)
return ret
def _findEntryInFile(file, path, draft=None, attribute=None):
doc = parseDoc(file)
elt = doc.documentElement
tag_spec_list = path.split("/")
last_entry = None
for i in range(len(tag_spec_list)):
tag_spec = tag_spec_list[i]
tag_name = tag_spec
arg_name = 'type'
arg_value = ''
left_bracket = tag_spec.find('[')
if left_bracket != -1:
tag_name = tag_spec[:left_bracket]
arg_value = tag_spec[left_bracket+1:-1].split("=")
if len(arg_value) == 2:
arg_name = arg_value[0].replace("@", "").replace("'", "")
arg_value = arg_value[1]
else:
arg_value = arg_value[0]
alias = findChild(elt, 'alias')
if alias and alias.attributes['source'].nodeValue == 'locale':
path = alias.attributes['path'].nodeValue
aliaspath = tag_spec_list[:i] + path.split("/")
def resolve(x, y):
if y == '..':
return x[:-1]
return x + [y]
# resolve all dot-dot parts of the path
aliaspath = reduce(resolve, aliaspath, [])
# remove attribute specification that our xpathlite doesnt support
aliaspath = map(lambda x: x.replace("@type=", "").replace("'", ""), aliaspath)
# append the remaining path
aliaspath = aliaspath + tag_spec_list[i:]
aliaspath = "/".join(aliaspath)
# "locale" aliases are special - we need to start lookup from scratch
return (None, aliaspath)
elt = findChild(elt, tag_name, arg_name, arg_value, draft)
if not elt:
return ("", None)
if attribute is not None:
if elt.attributes.has_key(attribute):
return (elt.attributes[attribute].nodeValue, None)
return (None, None)
try:
return (elt.firstChild.nodeValue, None)
except:
pass
return (None, None)
def findAlias(file):
doc = parseDoc(file)
alias_elt = findChild(doc.documentElement, "alias")
if not alias_elt:
return False
if not alias_elt.attributes.has_key('source'):
return False
return alias_elt.attributes['source'].nodeValue
lookup_chain_cache = {}
parent_locales = {}
def _fixedLookupChain(dirname, name):
if lookup_chain_cache.has_key(name):
return lookup_chain_cache[name]
# see http://www.unicode.org/reports/tr35/#Parent_Locales
if not parent_locales:
for ns in findTagsInFile(dirname + "/../supplemental/supplementalData.xml", "parentLocales"):
tmp = {}
parent_locale = ""
for data in ns[1:][0]: # ns looks like this: [u'parentLocale', [(u'parent', u'root'), (u'locales', u'az_Cyrl bs_Cyrl en_Dsrt ..')]]
tmp[data[0]] = data[1]
if data[0] == u"parent":
parent_locale = data[1]
parent_locales[parent_locale] = tmp[u"locales"].split(" ")
items = name.split("_")
# split locale name into items and iterate through them from back to front
# example: az_Latn_AZ => [az_Latn_AZ, az_Latn, az]
items = list(reversed(map(lambda x: "_".join(items[:x+1]), range(len(items)))))
for i in range(len(items)):
item = items[i]
for parent_locale in parent_locales.keys():
for locale in parent_locales[parent_locale]:
if item == locale:
if parent_locale == u"root":
items = items[:i+1]
else:
items = items[:i+1] + _fixedLookupChain(dirname, parent_locale)
lookup_chain_cache[name] = items
return items
lookup_chain_cache[name] = items
return items
def _findEntry(base, path, draft=None, attribute=None):
if base.endswith(".xml"):
base = base[:-4]
(dirname, filename) = os.path.split(base)
items = _fixedLookupChain(dirname, filename)
for item in items:
file = dirname + "/" + item + ".xml"
if os.path.isfile(file):
alias = findAlias(file)
if alias:
# if alias is found we should follow it and stop processing current file
# see http://www.unicode.org/reports/tr35/#Common_Elements
aliasfile = os.path.dirname(file) + "/" + alias + ".xml"
if not os.path.isfile(aliasfile):
raise Error("findEntry: fatal error: found an alias '%s' to '%s', but the alias file couldn't be found" % (filename, alias))
# found an alias, recurse into parsing it
result = _findEntry(aliasfile, path, draft, attribute)
return result
(result, aliaspath) = _findEntryInFile(file, path, draft, attribute)
if aliaspath:
# start lookup again because of the alias source="locale"
return _findEntry(base, aliaspath, draft, attribute)
if result:
return result
return None
def findEntry(base, path, draft=None, attribute=None):
file = base
if base.endswith(".xml"):
file = base
base = base[:-4]
else:
file = base + ".xml"
(dirname, filename) = os.path.split(base)
result = None
while path:
result = _findEntry(base, path, draft, attribute)
if result:
return result
(result, aliaspath) = _findEntryInFile(dirname + "/root.xml", path, draft, attribute)
if result:
return result
if not aliaspath:
raise Error("findEntry: fatal error: %s: cannot find key %s" % (filename, path))
path = aliaspath
return result