qt5base-lts/util/local_database/xpathlite.py
Jyri Tahtela f9f395c28b Update licenseheader text in source files for qtbase Qt module
Updated version of LGPL and FDL licenseheaders.
Apply release phase licenseheaders for all source files.

Reviewed-by: Trust Me
2011-05-24 12:34:08 +03:00

250 lines
8.9 KiB
Python

#!/usr/bin/env python
#############################################################################
##
## Copyright (C) 2011 Nokia Corporation and/or its subsidiary(-ies).
## All rights reserved.
## Contact: Nokia Corporation (qt-info@nokia.com)
##
## This file is part of the test suite of the Qt Toolkit.
##
## $QT_BEGIN_LICENSE:LGPL$
## GNU Lesser General Public License Usage
## This file may be used under the terms of the GNU Lesser General Public
## License version 2.1 as published by the Free Software Foundation and
## appearing in the file LICENSE.LGPL included in the packaging of this
## file. Please review the following information to ensure the GNU Lesser
## General Public License version 2.1 requirements will be met:
## http://www.gnu.org/licenses/old-licenses/lgpl-2.1.html.
##
## In addition, as a special exception, Nokia gives you certain additional
## rights. These rights are described in the Nokia Qt LGPL Exception
## version 1.1, included in the file LGPL_EXCEPTION.txt in this package.
##
## GNU General Public License Usage
## Alternatively, this file may be used under the terms of the GNU General
## Public License version 3.0 as published by the Free Software Foundation
## and appearing in the file LICENSE.GPL included in the packaging of this
## file. Please review the following information to ensure the GNU General
## Public License version 3.0 requirements will be met:
## http://www.gnu.org/copyleft/gpl.html.
##
## Other Usage
## Alternatively, this file may be used in accordance with the terms and
## conditions contained in a signed written agreement between you and Nokia.
##
##
##
##
##
## $QT_END_LICENSE$
##
#############################################################################
import sys
import os
import xml.dom.minidom
doc_cache = {}
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]
class Error:
def __init__(self, msg):
self.msg = msg
def __str__(self):
return self.msg
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 findTagsInFile(file, path):
doc = False
if doc_cache.has_key(file):
doc = doc_cache[file]
else:
doc = xml.dom.minidom.parse(file)
doc_cache[file] = doc
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]
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 = False
if doc_cache.has_key(file):
doc = doc_cache[file]
else:
doc = xml.dom.minidom.parse(file)
doc_cache[file] = doc
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]
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)
return (elt.firstChild.nodeValue, None)
def findAlias(file):
if not doc_cache.has_key(file):
return False
doc = doc_cache[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
def _findEntry(base, path, draft=None, attribute=None):
file = base
if base.endswith(".xml"):
filename = base
base = base[:-4]
else:
file = base + ".xml"
(dirname, filename) = os.path.split(base)
items = filename.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 = reversed(map(lambda x: "_".join(items[:x+1]), range(len(items))))
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 couldnt 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: did not found key %s" % (filename, path))
path = aliaspath
return result