1
# Copyright (C) 2005, 2006 Canonical Ltd
3
# This program is free software; you can redistribute it and/or modify
4
# it under the terms of the GNU General Public License as published by
5
# the Free Software Foundation; either version 2 of the License, or
6
# (at your option) any later version.
8
# This program is distributed in the hope that it will be useful,
9
# but WITHOUT ANY WARRANTY; without even the implied warranty of
10
# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
11
# GNU General Public License for more details.
13
# You should have received a copy of the GNU General Public License
14
# along with this program; if not, write to the Free Software
15
# Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
24
from bzrlib.xml_serializer import SubElement, Element, Serializer
25
from bzrlib.inventory import ROOT_ID, Inventory, InventoryEntry
26
from bzrlib.revision import Revision
27
from bzrlib.errors import BzrError
34
"'":"'", # FIXME: overkill
41
def _ensure_utf8_re():
42
"""Make sure the _utf8_re and _unicode_re regexes have been compiled."""
43
global _utf8_re, _unicode_re
45
_utf8_re = re.compile('[&<>\'\"]|[\x80-\xff]+')
46
if _unicode_re is None:
47
_unicode_re = re.compile(u'[&<>\'\"\u0080-\uffff]')
50
def _unicode_escape_replace(match, _map=_xml_escape_map):
51
"""Replace a string of non-ascii, non XML safe characters with their escape
53
This will escape both Standard XML escapes, like <>"', etc.
54
As well as escaping non ascii characters, because ElementTree did.
55
This helps us remain compatible to older versions of bzr. We may change
56
our policy in the future, though.
58
# jam 20060816 Benchmarks show that try/KeyError is faster if you
59
# expect the entity to rarely miss. There is about a 10% difference
60
# in overall time. But if you miss frequently, then if None is much
61
# faster. For our use case, we *rarely* have a revision id, file id
62
# or path name that is unicode. So use try/KeyError.
64
return _map[match.group()]
66
return "&#%d;" % ord(match.group())
69
def _utf8_escape_replace(match, _map=_xml_escape_map):
70
"""Escape utf8 characters into XML safe ones.
72
This uses 2 tricks. It is either escaping "standard" characters, like "&<>,
73
or it is handling characters with the high-bit set. For ascii characters,
74
we just lookup the replacement in the dictionary. For everything else, we
75
decode back into Unicode, and then use the XML escape code.
78
return _map[match.group()]
80
return ''.join('&#%d;' % ord(uni_chr)
81
for uni_chr in match.group().decode('utf8'))
86
def _encode_and_escape(unicode_or_utf8_str, _map=_to_escaped_map):
87
"""Encode the string into utf8, and escape invalid XML characters"""
88
# We frequently get entities we have not seen before, so it is better
89
# to check if None, rather than try/KeyError
90
text = _map.get(unicode_or_utf8_str)
92
if unicode_or_utf8_str.__class__ == unicode:
93
# The alternative policy is to do a regular UTF8 encoding
94
# and then escape only XML meta characters.
95
# Performance is equivalent once you use cache_utf8. *However*
96
# this makes the serialized texts incompatible with old versions
97
# of bzr. So no net gain. (Perhaps the read code would handle utf8
98
# better than entity escapes, but cElementTree seems to do just fine
100
text = str(_unicode_re.sub(_unicode_escape_replace,
101
unicode_or_utf8_str)) + '"'
103
# Plain strings are considered to already be in utf-8 so we do a
104
# slightly different method for escaping.
105
text = _utf8_re.sub(_utf8_escape_replace,
106
unicode_or_utf8_str) + '"'
107
_map[unicode_or_utf8_str] = text
111
def _get_utf8_or_ascii(a_str,
112
_encode_utf8=cache_utf8.encode,
113
_get_cached_ascii=cache_utf8.get_cached_ascii):
114
"""Return a cached version of the string.
116
cElementTree will return a plain string if the XML is plain ascii. It only
117
returns Unicode when it needs to. We want to work in utf-8 strings. So if
118
cElementTree returns a plain string, we can just return the cached version.
119
If it is Unicode, then we need to encode it.
121
:param a_str: An 8-bit string or Unicode as returned by
122
cElementTree.Element.get()
123
:return: A utf-8 encoded 8-bit string.
125
# This is fairly optimized because we know what cElementTree does, this is
126
# not meant as a generic function for all cases. Because it is possible for
127
# an 8-bit string to not be ascii or valid utf8.
128
if a_str.__class__ == unicode:
129
return _encode_utf8(a_str)
131
return _get_cached_ascii(a_str)
135
"""Clean out the unicode => escaped map"""
136
_to_escaped_map.clear()
139
class Serializer_v5(Serializer):
140
"""Version 5 serializer
142
Packs objects into XML and vice versa.
147
support_altered_by_hack = True
148
# This format supports the altered-by hack that reads file ids directly out
149
# of the versionedfile, without doing XML parsing.
151
def write_inventory_to_string(self, inv):
152
"""Just call write_inventory with a StringIO and return the value"""
153
sio = cStringIO.StringIO()
154
self.write_inventory(inv, sio)
155
return sio.getvalue()
157
def write_inventory(self, inv, f):
158
"""Write inventory to a file.
160
:param inv: the inventory to write.
161
:param f: the file to write.
165
append = output.append
166
self._append_inventory_root(append, inv)
167
entries = inv.iter_entries()
169
root_path, root_ie = entries.next()
170
for path, ie in entries:
171
self._append_entry(append, ie)
172
append('</inventory>\n')
174
# Just to keep the cache from growing without bounds
175
# but we may actually not want to do clear the cache
178
def _append_inventory_root(self, append, inv):
179
"""Append the inventory root to output."""
181
if inv.root.file_id not in (None, ROOT_ID):
183
append(_encode_and_escape(inv.root.file_id))
184
append(' format="5"')
185
if inv.revision_id is not None:
186
append(' revision_id="')
187
append(_encode_and_escape(inv.revision_id))
190
def _append_entry(self, append, ie):
191
"""Convert InventoryEntry to XML element and append to output."""
192
# TODO: should just be a plain assertion
193
assert InventoryEntry.versionable_kind(ie.kind), \
194
'unsupported entry kind %s' % ie.kind
199
append(' executable="yes"')
201
append(_encode_and_escape(ie.file_id))
203
append(_encode_and_escape(ie.name))
204
if self._parent_condition(ie):
205
assert isinstance(ie.parent_id, basestring)
206
append(' parent_id="')
207
append(_encode_and_escape(ie.parent_id))
208
if ie.revision is not None:
209
append(' revision="')
210
append(_encode_and_escape(ie.revision))
211
if ie.symlink_target is not None:
212
append(' symlink_target="')
213
append(_encode_and_escape(ie.symlink_target))
214
if ie.text_sha1 is not None:
215
append(' text_sha1="')
218
if ie.text_size is not None:
219
append(' text_size="%d"' % ie.text_size)
223
def _parent_condition(self, ie):
224
return ie.parent_id != ROOT_ID
226
def _pack_revision(self, rev):
227
"""Revision object -> xml tree"""
228
# For the XML format, we need to write them as Unicode rather than as
229
# utf-8 strings. So that cElementTree can handle properly escaping
231
decode_utf8 = cache_utf8.decode
232
revision_id = rev.revision_id
233
if isinstance(revision_id, str):
234
revision_id = decode_utf8(revision_id)
235
root = Element('revision',
236
committer = rev.committer,
237
timestamp = '%.3f' % rev.timestamp,
238
revision_id = revision_id,
239
inventory_sha1 = rev.inventory_sha1,
242
if rev.timezone is not None:
243
root.set('timezone', str(rev.timezone))
245
msg = SubElement(root, 'message')
246
msg.text = rev.message
249
pelts = SubElement(root, 'parents')
250
pelts.tail = pelts.text = '\n'
251
for parent_id in rev.parent_ids:
252
assert isinstance(parent_id, basestring)
253
p = SubElement(pelts, 'revision_ref')
255
if isinstance(parent_id, str):
256
parent_id = decode_utf8(parent_id)
257
p.set('revision_id', parent_id)
259
self._pack_revision_properties(rev, root)
262
def _pack_revision_properties(self, rev, under_element):
263
top_elt = SubElement(under_element, 'properties')
264
for prop_name, prop_value in sorted(rev.properties.items()):
265
assert isinstance(prop_name, basestring)
266
assert isinstance(prop_value, basestring)
267
prop_elt = SubElement(top_elt, 'property')
268
prop_elt.set('name', prop_name)
269
prop_elt.text = prop_value
273
def _unpack_inventory(self, elt):
274
"""Construct from XML Element
276
assert elt.tag == 'inventory'
277
root_id = elt.get('file_id') or ROOT_ID
278
format = elt.get('format')
279
if format is not None:
281
raise BzrError("invalid format version %r on inventory"
283
revision_id = elt.get('revision_id')
284
if revision_id is not None:
285
revision_id = cache_utf8.encode(revision_id)
286
inv = Inventory(root_id, revision_id=revision_id)
288
ie = self._unpack_entry(e)
289
if ie.parent_id is None:
290
ie.parent_id = root_id
294
def _unpack_entry(self, elt):
296
if not InventoryEntry.versionable_kind(kind):
297
raise AssertionError('unsupported entry kind %s' % kind)
299
get_cached = _get_utf8_or_ascii
301
parent_id = elt.get('parent_id')
302
# TODO: jam 20060817 At present, caching file ids costs us too
303
# much time. It slows down overall read performances from
304
# approx 500ms to 700ms. And doesn't improve future reads.
305
# it might be because revision ids and file ids are mixing.
306
# Consider caching *just* the file ids, for a limited period
308
#parent_id = get_cached(parent_id)
309
#file_id = get_cached(elt.get('file_id'))
310
file_id = elt.get('file_id')
312
if kind == 'directory':
313
ie = inventory.InventoryDirectory(file_id,
317
ie = inventory.InventoryFile(file_id,
320
ie.text_sha1 = elt.get('text_sha1')
321
if elt.get('executable') == 'yes':
323
v = elt.get('text_size')
324
ie.text_size = v and int(v)
325
elif kind == 'symlink':
326
ie = inventory.InventoryLink(file_id,
329
ie.symlink_target = elt.get('symlink_target')
331
raise BzrError("unknown kind %r" % kind)
332
revision = elt.get('revision')
333
if revision is not None:
334
revision = get_cached(revision)
335
ie.revision = revision
339
def _unpack_revision(self, elt):
340
"""XML Element -> Revision object"""
341
assert elt.tag == 'revision'
342
format = elt.get('format')
343
if format is not None:
345
raise BzrError("invalid format version %r on inventory"
347
get_cached = _get_utf8_or_ascii
348
rev = Revision(committer = elt.get('committer'),
349
timestamp = float(elt.get('timestamp')),
350
revision_id = get_cached(elt.get('revision_id')),
351
inventory_sha1 = elt.get('inventory_sha1')
353
parents = elt.find('parents') or []
355
assert p.tag == 'revision_ref', \
356
"bad parent node tag %r" % p.tag
357
rev.parent_ids.append(get_cached(p.get('revision_id')))
358
self._unpack_revision_properties(elt, rev)
359
v = elt.get('timezone')
363
rev.timezone = int(v)
364
rev.message = elt.findtext('message') # text of <message>
367
def _unpack_revision_properties(self, elt, rev):
368
"""Unpack properties onto a revision."""
369
props_elt = elt.find('properties')
370
assert len(rev.properties) == 0
373
for prop_elt in props_elt:
374
assert prop_elt.tag == 'property', \
375
"bad tag under properties list: %r" % prop_elt.tag
376
name = prop_elt.get('name')
377
value = prop_elt.text
378
# If a property had an empty value ('') cElementTree reads
379
# that back as None, convert it back to '', so that all
380
# properties have string values
383
assert name not in rev.properties, \
384
"repeated property %r" % name
385
rev.properties[name] = value
388
serializer_v5 = Serializer_v5()
added
removed
bzrlib/xml5.py
