Attachment #303357 for bug #405865

Line Link Here

(-)file_not_specified_in_diff (-108 / +195 lines)
0	-- src/py/xml-rewrite-2.py	0	++ src/py/xml-rewrite-2.py
Lines 8-17 Link Here
8	# Authors:	8	# Authors:
9	# Saleem Abdulrasool <compnerd@compnerd.org>	9	# Saleem Abdulrasool <compnerd@compnerd.org>
10	# Petteri Räty <betelgeuse@gentoo.org>	10	# Petteri Räty <betelgeuse@gentoo.org>
		11	# Karsten Merkle <gentoo@karsten-merkle.de>
11	# Maintainer: Gentoo Java Herd <java@gentoo.org>	12	# Maintainer: Gentoo Java Herd <java@gentoo.org>
12	# Python based XML modifier	13	# Python based XML modifier
13		14
14	# ChangeLog	15	# ChangeLog
		16	# Karsten Merkle <gentoo@karsten-merkle.de>
		17	# Februar 13, 2012 - elements can now be deleted.
		18	# Februar 19, 2012 - paths for matches can now be defined
15	# Petteri Räty <betelgeuse@gentoo.org	19	# Petteri Räty <betelgeuse@gentoo.org
16	# December 06, 2006 - Changed to use xml.parsers.expat and basically rewrote the whole file	20	# December 06, 2006 - Changed to use xml.parsers.expat and basically rewrote the whole file
17	# December 29, 2006 - Added a SAX based implementation to handle entities etc ( test on dev-java/skinlf )	21	# December 29, 2006 - Added a SAX based implementation to handle entities etc ( test on dev-java/skinlf )
Lines 54-75 Link Here
54	be in StreamRewriterBase subclasses as they are much faster.	58	be in StreamRewriterBase subclasses as they are much faster.
55	"""	59	"""
56	from xml.dom import NotFoundErr	60	from xml.dom import NotFoundErr
		61	import re
57		62
58	def __init__(self, modifyElems, attributes, values=None, index=None):	63	def __init__(self, options):
59	self.modify = modifyElems	64	self.doAdd = options.doAdd
60	self.attributes = attributes	65	self.modify = options.elements
61	self.values = values	66	self.attributes = options.attributes
62	self.index = index	67	self.values = options.values
		68	self.index = options.index
		69	self.pathRe = re.compile('(?P<ex>(?P<root>//)?/?(?P<elem>[^/\[]+)(?P<attr>\[@(?P<attrName>[^=\]]+)(=\"?(?P<attrValue>[^\"]*)\"?)?\])?)')
		70	self.matches = options.matches
		71	self.qualifier = options.qualifier
63		72
64	def change_elem(self, elem):	73	def change_elem(self, elem):
65	for i,attr in enumerate(self.attributes):	74	if self.attributes:
66	if self.values:	75	for i,attr in enumerate(self.attributes):
67	elem.setAttribute(attr, self.values[i])	76	if self.doAdd:
68	else:	77	elem.setAttribute(attr, self.values[i])
69	try:	78	else:
70	elem.removeAttribute(attr)	79	try:
71	except DomRewriter.NotFoundErr:	80	elem.removeAttribute(attr)
		81	except DomRewriter.NotFoundErr:
		82	continue
		83	else:
		84	elem.parentNode.removeChild(elem)
		85
		86	def match_elem(self, position, index, parent):
		87	if len(self.matches[index]) == position:
		88	return True
		89	match = self.pathRe.match(self.matches[index], position)
		90	if not match:
		91	return False
		92	new_position = match.end()
		93	target = match.groupdict()
		94	t_elemet = target["elem"]
		95	t_attribute = target["attrName"]
		96	t_value = target["attrValue"]
		97	t_root = target["root"]
		98	if t_root and not position==0:
		99	error("missformed expression at "+ target["ex"] + " in self.matches[index]")
		100	if not t_elemet:
		101	error("Missing element in " + options.matches[index])
		102	for elem in parent.getElementsByTagName(t_elemet):
		103	if not position == 0 or t_root:
		104	if not elem.parentNode == parent:
		105	continue
		106	if t_attribute:
		107	if not elem.hasAttribute(t_attribute):
72	continue	108	continue
		109	if t_value:
		110	attr = elem.getAttributeNode(t_attribute)
		111	if not attr.value == t_value:
		112	continue
		113	if self.match_elem(new_position, index, elem) and self.qualify_elem(0, index, elem):
		114	if self.doAdd:
		115	if self.modify:
		116	elemNew = self.document.createElement(self.modify[index])
		117	elem.appendChild(elemNew)
		118	elem = elemNew
		119	if self.attributes:
		120	elem.setAttribute(self.attributes[index], self.values[index])
		121	else:
		122	if (t_attribute and self.values):
		123	elem.setAttribute(t_attribute, self.values[index])
		124	else:
		125	if t_attribute:
		126	elem.removeAttribute(t_attribute)
		127	else:
		128	elem.parentNode.removeChild(elem)
		129	return False;
		130
		131	def qualify_elem(self, position, index, parent):
		132	if not self.qualifier:
		133	return True
		134	if len(self.qualifier[index]) == position:
		135	return True
		136	match = self.pathRe.match(self.qualifier[index], position)
		137	if not match:
		138	return False
		139	new_position = match.end()
		140	target = match.groupdict()
		141	t_elemet = target["elem"]
		142	t_attribute = target["attrName"]
		143	t_value = target["attrValue"]
		144	t_root = target["root"]
		145	if t_root:
		146	error("qualifier must not start with root //. Qualifier uses always match path as root.")
		147	if not t_elemet:
		148	error("Missing element in " + options.matches[index])
		149	for elem in parent.getElementsByTagName(t_elemet):
		150	if not elem.parentNode == parent:
		151	continue
		152	if t_attribute:
		153	if not elem.hasAttribute(t_attribute):
		154	return False
		155	if t_value:
		156	attr = elem.getAttributeNode(t_attribute)
		157	if not attr.value == t_value:
		158	return False
		159	if self.qualify_elem(new_position, index, elem):
		160	return True
73		161
74	def process(self,in_stream,callback=None):	162	def process(self,in_stream,callback=None):
75	from xml.dom.minidom import parse	163	from xml.dom.minidom import parse
Lines 79-115 Link Here
79	if callback:	167	if callback:
80	callback(self.document)	168	callback(self.document)
81		169
82	if not self.modify:	170	if self.matches:
83	return	171	for i in range(len(self.matches)):
84		172	self.match_elem(0, i, self.document)
85	for tag in self.modify:	173	else:
86	matches = self.document.getElementsByTagName(tag)	174	if not self.modify:
87	if matches:	175	return
88	if self.index == None:	176	for tag in self.modify:
89	for match in matches:	177	matches = self.document.getElementsByTagName(tag)
90	self.change_elem(match)	178	if matches:
91	else:	179	if self.index == None:
92	self.change_elem(matches[self.index])	180	for match in matches:
		181	self.change_elem(match)
		182	else:
		183	self.change_elem(matches[self.index])
93		184
94	def write(self,stream):	185	def write(self,stream):
95	stream.write(self.document.toxml())	186	stream.write(self.document.toxml())
96		187
97	class StreamRewriterBase:	188	class StreamRewriterBase:
98		189
99	def __init__(self, elems, attributes, values, index,	190	def __init__(self, elems, attributes, values, index):
100	sourceElems = [], sourceAttributes = [], sourceValues = [],
101	targetElems = [], targetAttributes = [], targetValues = [] ):
102	self.buffer = StringIO.StringIO()	191	self.buffer = StringIO.StringIO()
103	self.__write = self.buffer.write	192	self.__write = self.buffer.write
104	self.elems = elems or []	193	self.elems = elems or []
105	self.attributes = attributes or []	194	self.attributes = attributes or []
106	self.values = values or []	195	self.values = values or []
107	self.sourceElems = sourceElems or []
108	self.sourceAttributes = sourceAttributes or []
109	self.sourceValues = sourceValues or []
110	self.targetElems = targetElems or []
111	self.targetAttributes = targetAttributes or []
112	self.targetValues = targetValues or []
113		196
114	def p(self,str):	197	def p(self,str):
115	self.__write(str.encode('utf8'))	198	self.__write(str.encode('utf8'))
Lines 126-179 Link Here
126	self.p(u'<%s ' % name)	209	self.p(u'<%s ' % name)
127		210
128	match = ( name in self.elems )	211	match = ( name in self.elems )
129	matchSource = ( name in self.sourceElems )
130	matchTarget = ( name in self.targetElems )
131		212
132	for a,v in attrs:	213	for a,v in attrs:
133	if not (	214	if not (match and a in self.attributes):
134	(match and a in self.attributes)
135	or (matchSource and a in self.sourceAttributes)
136	or (matchTarget and a in self.targetAttributes)
137	):
138	self.write_attr(a,v)	215	self.write_attr(a,v)
139		216
140	if matchSource:
141	for i, attr in enumerate(self.sourceAttributes):
142	self.write_attr(attr, self.sourceValues[i])
143
144	if matchTarget:
145	for i, attr in enumerate(self.targetAttributes):
146	self.write_attr(attr, self.targetValues[i])
147
148	if match:	217	if match:
149	for i, attr in enumerate(self.attributes):	218	for i, attr in enumerate(self.attributes):
150	self.write_attr(attr, self.values[i])	219	self.write_attr(attr, self.values[i])
151		220
152	self.p(u'>')	221	self.p(u'>')
153		222
154	class ExpatRewriter(StreamRewriterBase):	223	#class ExpatRewriter(StreamRewriterBase):
155	"""	224	# """
156	The only problem with this Expat based implementation is that it does not	225	# The only problem with this Expat based implementation is that it does not
157	handle entities doctypes etc properly so for example dev-java/skinlf fails.	226	# handle entities doctypes etc properly so for example dev-java/skinlf fails.
158	"""	227	# """
159	def process(self, in_stream):	228	# def process(self, in_stream):
160	from xml.parsers.expat import ParserCreate	229	# from xml.parsers.expat import ParserCreate
161	parser = ParserCreate()	230	# parser = ParserCreate()
162		231	#
163	parser.StartElementHandler = self.start_element	232	# parser.StartElementHandler = self.start_element
164	parser.EndElementHandler = self.end_element	233	# parser.EndElementHandler = self.end_element
165	parser.CharacterDataHandler = self.char_data	234	# parser.CharacterDataHandler = self.char_data
166	parser.ParseFile(in_stream)	235	# parser.ParseFile(in_stream)
167	self.p(u'\n')	236	# self.p(u'\n')
168		237	#
169	def start_element(self, name, attrs):	238	# def start_element(self, name, attrs):
170	StreamRewriterBase(self, name, attrs.iteritems())	239	# StreamRewriterBase(self, name, attrs.iteritems())
171		240	#
172	def end_element(self,name):	241	# def end_element(self,name):
173	self.p(u'</%s>' % name)	242	# self.p(u'</%s>' % name)
174		243	#
175	def char_data(self,data):	244	# def char_data(self,data):
176	self.p(escape(data))	245	# self.p(escape(data))
177		246
178	from xml.sax.saxutils import XMLGenerator	247	from xml.sax.saxutils import XMLGenerator
179	class SaxRewriter(XMLGenerator, StreamRewriterBase):	248	class SaxRewriter(XMLGenerator, StreamRewriterBase):
Lines 181-192 Link Here
181	Using Sax gives us the support for writing back doctypes and all easily	250	Using Sax gives us the support for writing back doctypes and all easily
182	and is only marginally slower than expat as it is just a tight layer over it	251	and is only marginally slower than expat as it is just a tight layer over it
183	"""	252	"""
184	def __init__(self, elems, attributes, values, index,	253	def __init__(self, elems, attributes, values, index):
185	sourceElems = [], sourceAttributes = [], sourceValues = [],	254	StreamRewriterBase.__init__(self, elems, attributes, values, index)
186	targetElems = [], targetAttributes = [], targetValues = []):
187	StreamRewriterBase.__init__(self, elems, attributes, values, index,
188	sourceElems, sourceAttributes, sourceValues,
189	targetElems, targetAttributes, targetValues)
190	XMLGenerator.__init__(self, self.buffer, 'UTF-8')	255	XMLGenerator.__init__(self, self.buffer, 'UTF-8')
191		256
192	def process(self, in_stream):	257	def process(self, in_stream):
Lines 198-226 Link Here
198	self.start_element(name, attrs.items())	263	self.start_element(name, attrs.items())
199		264
200	if __name__ == '__main__':	265	if __name__ == '__main__':
		266	import re
		267
201	usage = "XML Rewrite Python Module Version " + __version__ + "\n"	268	usage = "XML Rewrite Python Module Version " + __version__ + "\n"
202	usage += "Copyright 2004,2006,2007 Gentoo Foundation\n"	269	usage += "Copyright 2004,2006,2007 Gentoo Foundation\n"
203	usage += "Distributed under the terms of the GNU General Public Lincense v2\n"	270	usage += "Distributed under the terms of the GNU General Public License v2\n"
204	usage += "Please contact the Gentoo Java Team <java@gentoo.org> with problems.\n"	271	usage += "Please contact the Gentoo Java Team <java@gentoo.org> with problems.\n"
205	usage += "\n"	272	usage += "\n"
206	usage += "Usage:\n"	273	usage += "Usage:\n"
207	usage += " xml-rewrite.py [-f file] --delete [-g] -e tag [-e tag] -a attribute [-a attribute] [-i index]\n"	274	usage += " xml-rewrite.py [-f file]... --delete [-g] -e tag [-e tag]... [-a attribute]... [-i index] [-m <path> [-q <path>]]...\n"
208	usage += " xml-rewrite.py [-f file] --change [-g] -e tag [-e tag] -a attribute -v value [-a attribute -v value] \\\n"	275	usage += " xml-rewrite.py [-f file]... --change [-g] -e tag [-e tag]... -a attribute -v value [-a attribute -v value]... [-m <path> [-q <path>]]... \\\n"
209	usage += " [--source-element tag] [--source-attribute attribute --source-value value] \\\n"
210	usage += " [--target-element tag] [--target-attribute attribute --target-value value] [-i index]\n"
211	usage += "Or:\n"	276	usage += "Or:\n"
212	usage += " xml-rewrite.py [-f file] -g\n"	277	usage += " xml-rewrite.py [-f file] -g\n"
213	usage += "\n"	278	usage += "\n"
214	usage += "If the -f parameter is not utilized, the script will read and\n"	279	usage += "If the -f parameter is not utilized, the script will read and\n"
215	usage += "write to stdin and stdout respectively. The use of quotes on\n"	280	usage += "write to stdin and stdout respectively. The use of quotes on\n"
216	usage += "parameters will break the script.\n"	281	usage += "parameters will break the script.\n"
		282	usage += "<path> -'xy[@tz=\"main\"/qr` will match all 'qr' elements of all 'xy' elements which 'tz' attributes value is main\n"
		283	usage += "Matching:\n"
		284	usage += "-c with -v only : only existing attribute defined by match-path will be changed\n"
		285	usage += "-c with -v and -a : the given attribute is added/changed to the element defined by match-path\n"
		286	usage += "-c with -e only : the given element is added to the element defined by match-path\n"
217		287
218		288
219	def error(message):	289	def error(message):
220	print "ERROR: " + message	290	print "ERROR: " + message
221	sys.exit(1)	291	sys.exit(1)
222		292
223
224	# if len(sys.argv) == 1:	293	# if len(sys.argv) == 1:
225	# usage(True)	294	# usage(True)
226		295
Lines 232-249 Link Here
232	make_option ("-e", "--element", action="append", dest="elements", help="Tag of the element of which the attributes to be changed. These can be chained for multiple elements."),	301	make_option ("-e", "--element", action="append", dest="elements", help="Tag of the element of which the attributes to be changed. These can be chained for multiple elements."),
233	make_option ("-a", "--attribute", action="append", dest="attributes", help="Attribute of the matching elements to change. These can be chained for multiple value-attribute pairs"),	302	make_option ("-a", "--attribute", action="append", dest="attributes", help="Attribute of the matching elements to change. These can be chained for multiple value-attribute pairs"),
234	make_option ("-v", "--value", action="append", dest="values", help="Value to set the attribute to."),	303	make_option ("-v", "--value", action="append", dest="values", help="Value to set the attribute to."),
235	make_option ("-r", "--source-element", action="append", dest="source_elements", help="Tag of the element of which the attributes to be changed just in source scope. These can be chained for multiple elements."),	304	# make_option ("-r", "--source-element", action="append", dest="source_elements", help="Tag of the element of which the attributes to be changed just in source scope. These can be chained for multiple elements."),
236	make_option ("-t","--source-attribute", action="append", dest="source_attributes", help="Attribute of the matching elements to change. These can be chained for multiple value-attribute pairs (for source only)"),	305	# make_option ("-t", "--source-attribute", action="append", dest="source_attributes", help="Attribute of the matching elements to change. These can be chained for multiple value-attribute pairs (for source only)"),
237	make_option ("-y", "--source-value", action="append", dest="source_values", help="Value to set the attribute to. (sourceonly)"),	306	# make_option ("-y", "--source-value", action="append", dest="source_values", help="Value to set the attribute to. (sourceonly)"),
238	make_option ("-j", "--target-element", action="append", dest="target_elements", help="Tag of the element of which the attributes to be changed just in target scope. These can be chained for multiple elements."),	307	# make_option ("-j", "--target-element", action="append", dest="target_elements", help="Tag of the element of which the attributes to be changed just in target scope. These can be chained for multiple elements."),
239	make_option ("-k", "--target-attribute", action="append", dest="target_attributes", help="Attribute of the matching elements to change. These can be chained for multiple value-attribute pairs (for targetonly)"),	308	# make_option ("-k", "--target-attribute", action="append", dest="target_attributes", help="Attribute of the matching elements to change. These can be chained for multiple value-attribute pairs (for targetonly)"),
240	make_option ("-l", "--target-value", action="append", dest="target_values", help="Value to set the attribute to (targeronly)."),	309	# make_option ("-l", "--target-value", action="append", dest="target_values", help="Value to set the attribute to (targeronly)."),
241	make_option ("-i", "--index", type="int", dest="index", help="Index of the match. If none is specified, the changes will be applied to all matches within the document. Starts from zero.")	310	make_option ("-r", "--source-element", action="append", dest="elements", help="Deprecated. please use -e or --element"),
		311	make_option ("-t", "--source-attribute", action="append", dest="attributes", help="Deprecated. please use -a or --attribute"),
		312	make_option ("-y", "--source-value", action="append", dest="values", help="Deprecated. please use -v or --value"),
		313	make_option ("-j", "--target-element", action="append", dest="elements", help="Deprecated. please use -e or --element"),
		314	make_option ("-k", "--target-attribute", action="append", dest="attributes", help="Deprecated. please use -a or --attribute"),
		315	make_option ("-l", "--target-value", action="append", dest="values", help="Deprecated. please use -v or --value"),
		316	make_option ("-i", "--index", type="int", dest="index", help="Index of the match. If none is specified, the changes will be applied to all matches within the document. Starts from zero."),
		317	make_option ("-m", "--match", action="append", dest="matches", help="match a node by the given path. If -c is set, -e and/or -a with -v are set/appended to path. If -d is set nodes matching path are deleted"),
		318	make_option ("-q", "--qualify", type="string", action="append", dest="qualifier", help="qualifies any node if given path matches to -m matched node as base. All other -m matching nodes are ignored")
242	]	319	]
243		320
244	parser = OptionParser(usage, options_list)	321	parser = OptionParser(usage, options_list)
245	(options, args) = parser.parse_args()	322	(options, args) = parser.parse_args()
246		323
		324	# TODO -r -j can be appended to elements since nothing would change
		325	# TODO -t -k can be appended to attributes since nothing would change
		326	# TODO -y -l can be appended to values since nothing would change
247		327
248	# Invalid Arguments Must be smited!	328	# Invalid Arguments Must be smited!
249	if not options.doAdd and not options.doDelete and not options.gentoo_classpath:	329	if not options.doAdd and not options.doDelete and not options.gentoo_classpath:
Lines 255-287 Link Here
255	if options.doAdd and options.doDelete:	335	if options.doAdd and options.doDelete:
256	error("Unable to perform multiple actions simultaneously.")	336	error("Unable to perform multiple actions simultaneously.")
257		337
258	if not options.elements and not options.target_elements and not options.source_elements:	338	if not options.elements and not options.matches:
259	error("At least one element (global, source only or target only) and attribute must be specified.")	339	error("At least one element must be specified or given by match")
260		340
261	for elem in ( options.source_attributes or [] ):	341	if options.doAdd and not options.matches and len(options.values or []) != len(options.attributes or []):
262	if elem in ( options.attributes or [] ):
263	error("You can't set an attribute in global and source scope at the same time")
264
265	for elem in ( options.target_attributes or [] ):
266	if elem in ( options.attributes or [] ):
267	error("You can't set an attribute in global and target scope at the same time")
268
269	if options.doAdd and (len(options.values or []) != len(options.attributes or [])
270	or len(options.source_values or [] ) != len(options.source_attributes or [])
271	or len(options.target_values or [] ) != len(options.target_attributes or [])):
272	error("You must give attribute(s)/value(s) for every element you are changing.")	342	error("You must give attribute(s)/value(s) for every element you are changing.")
273		343
		344	if options.doDelete and options.matches and (options.elements or options.attributes or options.values):
		345	error("You can't set -e, -a or -v while using a match to delete")
		346
		347	if options.matches and options.elements and (len(options.matches or []) != len(options.elements or [])):
		348	error("You must give a element for every match you define")
		349
		350	if options.matches and options.values and options.elements and not options.attributes:
		351	error("you can't define new Element and value without an attribute")
		352
		353	if options.matches and options.values and (len(options.matches or []) != len(options.values or [])):
		354	error("You must give a value for every match you define")
		355
		356	if options.matches and options.qualifier and (len(options.matches or []) != len(options.qualifier or [])):
		357	error("You must give a qualifier for every match you define")
		358
		359	if options.matches and options.attributes and (len(options.attributes or []) != len(options.values or [])):
		360	error("You must give a value for each attribute you give")
		361
		362
274	# End Invalid Arguments Check	363	# End Invalid Arguments Check
275		364
276	def get_rewriter(options):	365	def get_rewriter(options):
277	if options.index or options.doDelete or options.gentoo_classpath:	366	if options.index or options.doDelete or options.gentoo_classpath or options.matches:
278	# java-ant-2.eclass does not use these options so we can optimize the ExpatWriter	367	# java-ant-2.eclass does not use these options so we can optimize the ExpatWriter
279	# and let the DomRewriter do these. Also keeps the index option compatible for sure.	368	# and let the DomRewriter do these. Also keeps the index option compatible for sure.
280	rewriter = DomRewriter(options.elements, options.attributes, options.values, options.index)	369	rewriter = DomRewriter(options)
281	else:	370	else:
282	rewriter = SaxRewriter(options.elements, options.attributes, options.values, options.index,	371	rewriter = SaxRewriter(options.elements, options.attributes, options.values, options.index)
283	options.source_elements, options.source_attributes, options.source_values,
284	options.target_elements, options.target_attributes, options.target_values)
285		372
286	return rewriter	373	return rewriter
287		374

Return to bug 405865