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
25
from bzrlib.xml_serializer import SubElement, Element, Serializer
26
from bzrlib.inventory import ROOT_ID, Inventory, InventoryEntry
27
from bzrlib.revision import Revision
28
from bzrlib.errors import BzrError
35
"'":"'", # FIXME: overkill
42
def _ensure_utf8_re():
43
"""Make sure the _utf8_re and _unicode_re regexes have been compiled."""
44
global _utf8_re, _unicode_re
46
_utf8_re = re.compile('[&<>\'\"]|[\x80-\xff]+')
47
if _unicode_re is None:
48
_unicode_re = re.compile(u'[&<>\'\"\u0080-\uffff]')
51
def _unicode_escape_replace(match, _map=_xml_escape_map):
52
"""Replace a string of non-ascii, non XML safe characters with their escape
54
This will escape both Standard XML escapes, like <>"', etc.
55
As well as escaping non ascii characters, because ElementTree did.
56
This helps us remain compatible to older versions of bzr. We may change
57
our policy in the future, though.
59
# jam 20060816 Benchmarks show that try/KeyError is faster if you
60
# expect the entity to rarely miss. There is about a 10% difference
61
# in overall time. But if you miss frequently, then if None is much
62
# faster. For our use case, we *rarely* have a revision id, file id
63
# or path name that is unicode. So use try/KeyError.
65
return _map[match.group()]
67
return "&#%d;" % ord(match.group())
70
def _utf8_escape_replace(match, _map=_xml_escape_map):
71
"""Escape utf8 characters into XML safe ones.
73
This uses 2 tricks. It is either escaping "standard" characters, like "&<>,
74
or it is handling characters with the high-bit set. For ascii characters,
75
we just lookup the replacement in the dictionary. For everything else, we
76
decode back into Unicode, and then use the XML escape code.
79
return _map[match.group()]
81
return ''.join('&#%d;' % ord(uni_chr)
82
for uni_chr in match.group().decode('utf8'))
87
def _encode_and_escape(unicode_or_utf8_str, _map=_to_escaped_map):
88
"""Encode the string into utf8, and escape invalid XML characters"""
89
# We frequently get entities we have not seen before, so it is better
90
# to check if None, rather than try/KeyError
91
text = _map.get(unicode_or_utf8_str)
93
if unicode_or_utf8_str.__class__ == unicode:
94
# The alternative policy is to do a regular UTF8 encoding
95
# and then escape only XML meta characters.
96
# Performance is equivalent once you use cache_utf8. *However*
97
# this makes the serialized texts incompatible with old versions
98
# of bzr. So no net gain. (Perhaps the read code would handle utf8
99
# better than entity escapes, but cElementTree seems to do just fine
101
text = str(_unicode_re.sub(_unicode_escape_replace,
102
unicode_or_utf8_str)) + '"'
104
# Plain strings are considered to already be in utf-8 so we do a
105
# slightly different method for escaping.
106
text = _utf8_re.sub(_utf8_escape_replace,
107
unicode_or_utf8_str) + '"'
108
_map[unicode_or_utf8_str] = text
112
def _get_utf8_or_ascii(a_str,
113
_encode_utf8=cache_utf8.encode,
114
_get_cached_ascii=cache_utf8.get_cached_ascii):
115
"""Return a cached version of the string.
117
cElementTree will return a plain string if the XML is plain ascii. It only
118
returns Unicode when it needs to. We want to work in utf-8 strings. So if
119
cElementTree returns a plain string, we can just return the cached version.
120
If it is Unicode, then we need to encode it.
122
:param a_str: An 8-bit string or Unicode as returned by
123
cElementTree.Element.get()
124
:return: A utf-8 encoded 8-bit string.
126
# This is fairly optimized because we know what cElementTree does, this is
127
# not meant as a generic function for all cases. Because it is possible for
128
# an 8-bit string to not be ascii or valid utf8.
129
if a_str.__class__ == unicode:
130
return _encode_utf8(a_str)
132
return _get_cached_ascii(a_str)
136
"""Clean out the unicode => escaped map"""
137
_to_escaped_map.clear()
140
class Serializer_v5(Serializer):
141
"""Version 5 serializer
143
Packs objects into XML and vice versa.
148
support_altered_by_hack = True
149
# This format supports the altered-by hack that reads file ids directly out
150
# of the versionedfile, without doing XML parsing.
152
supported_kinds = set(['file', 'directory', 'symlink'])
154
def write_inventory_to_string(self, inv):
155
"""Just call write_inventory with a StringIO and return the value"""
156
sio = cStringIO.StringIO()
157
self.write_inventory(inv, sio)
158
return sio.getvalue()
160
def write_inventory(self, inv, f):
161
"""Write inventory to a file.
163
:param inv: the inventory to write.
164
:param f: the file to write.
168
append = output.append
169
self._append_inventory_root(append, inv)
170
entries = inv.iter_entries()
172
root_path, root_ie = entries.next()
173
for path, ie in entries:
174
self._append_entry(append, ie)
175
append('</inventory>\n')
177
# Just to keep the cache from growing without bounds
178
# but we may actually not want to do clear the cache
181
def _append_inventory_root(self, append, inv):
182
"""Append the inventory root to output."""
184
if inv.root.file_id not in (None, ROOT_ID):
186
append(_encode_and_escape(inv.root.file_id))
187
append(' format="5"')
188
if inv.revision_id is not None:
189
append(' revision_id="')
190
append(_encode_and_escape(inv.revision_id))
193
def _append_entry(self, append, ie):
194
"""Convert InventoryEntry to XML element and append to output."""
195
# TODO: should just be a plain assertion
196
if ie.kind not in self.supported_kinds:
197
raise errors.UnsupportedInventoryKind(ie.kind)
202
append(' executable="yes"')
204
append(_encode_and_escape(ie.file_id))
206
append(_encode_and_escape(ie.name))
207
if self._parent_condition(ie):
208
assert isinstance(ie.parent_id, basestring)
209
append(' parent_id="')
210
append(_encode_and_escape(ie.parent_id))
211
if ie.revision is not None:
212
append(' revision="')
213
append(_encode_and_escape(ie.revision))
214
if ie.symlink_target is not None:
215
append(' symlink_target="')
216
append(_encode_and_escape(ie.symlink_target))
217
if ie.text_sha1 is not None:
218
append(' text_sha1="')
221
if ie.text_size is not None:
222
append(' text_size="%d"' % ie.text_size)
223
if getattr(ie, 'reference_revision', None) is not None:
224
append(' reference_revision="')
225
append(_encode_and_escape(ie.reference_revision))
229
def _parent_condition(self, ie):
230
return ie.parent_id != ROOT_ID
232
def _pack_revision(self, rev):
233
"""Revision object -> xml tree"""
234
# For the XML format, we need to write them as Unicode rather than as
235
# utf-8 strings. So that cElementTree can handle properly escaping
237
decode_utf8 = cache_utf8.decode
238
revision_id = rev.revision_id
239
if isinstance(revision_id, str):
240
revision_id = decode_utf8(revision_id)
241
root = Element('revision',
242
committer = rev.committer,
243
timestamp = '%.3f' % rev.timestamp,
244
revision_id = revision_id,
245
inventory_sha1 = rev.inventory_sha1,
248
if rev.timezone is not None:
249
root.set('timezone', str(rev.timezone))
251
msg = SubElement(root, 'message')
252
msg.text = rev.message
255
pelts = SubElement(root, 'parents')
256
pelts.tail = pelts.text = '\n'
257
for parent_id in rev.parent_ids:
258
assert isinstance(parent_id, basestring)
259
p = SubElement(pelts, 'revision_ref')
261
if isinstance(parent_id, str):
262
parent_id = decode_utf8(parent_id)
263
p.set('revision_id', parent_id)
265
self._pack_revision_properties(rev, root)
268
def _pack_revision_properties(self, rev, under_element):
269
top_elt = SubElement(under_element, 'properties')
270
for prop_name, prop_value in sorted(rev.properties.items()):
271
assert isinstance(prop_name, basestring)
272
assert isinstance(prop_value, basestring)
273
prop_elt = SubElement(top_elt, 'property')
274
prop_elt.set('name', prop_name)
275
prop_elt.text = prop_value
279
def _unpack_inventory(self, elt):
280
"""Construct from XML Element
282
assert elt.tag == 'inventory'
283
root_id = elt.get('file_id') or ROOT_ID
284
format = elt.get('format')
285
if format is not None:
287
raise BzrError("invalid format version %r on inventory"
289
revision_id = elt.get('revision_id')
290
if revision_id is not None:
291
revision_id = cache_utf8.encode(revision_id)
292
inv = Inventory(root_id, revision_id=revision_id)
294
ie = self._unpack_entry(e)
295
if ie.parent_id == ROOT_ID:
296
ie.parent_id = root_id
300
def _unpack_entry(self, elt, none_parents=False):
302
if not InventoryEntry.versionable_kind(kind):
303
raise AssertionError('unsupported entry kind %s' % kind)
305
get_cached = _get_utf8_or_ascii
307
parent_id = elt.get('parent_id')
308
if parent_id is None and not none_parents:
310
# TODO: jam 20060817 At present, caching file ids costs us too
311
# much time. It slows down overall read performances from
312
# approx 500ms to 700ms. And doesn't improve future reads.
313
# it might be because revision ids and file ids are mixing.
314
# Consider caching *just* the file ids, for a limited period
316
#parent_id = get_cached(parent_id)
317
#file_id = get_cached(elt.get('file_id'))
318
file_id = elt.get('file_id')
320
if kind == 'directory':
321
ie = inventory.InventoryDirectory(file_id,
325
ie = inventory.InventoryFile(file_id,
328
ie.text_sha1 = elt.get('text_sha1')
329
if elt.get('executable') == 'yes':
331
v = elt.get('text_size')
332
ie.text_size = v and int(v)
333
elif kind == 'symlink':
334
ie = inventory.InventoryLink(file_id,
337
ie.symlink_target = elt.get('symlink_target')
339
raise errors.UnsupportedInventoryKind(kind)
340
revision = elt.get('revision')
341
if revision is not None:
342
revision = get_cached(revision)
343
ie.revision = revision
347
def _unpack_revision(self, elt):
348
"""XML Element -> Revision object"""
349
assert elt.tag == 'revision'
350
format = elt.get('format')
351
if format is not None:
353
raise BzrError("invalid format version %r on inventory"
355
get_cached = _get_utf8_or_ascii
356
rev = Revision(committer = elt.get('committer'),
357
timestamp = float(elt.get('timestamp')),
358
revision_id = get_cached(elt.get('revision_id')),
359
inventory_sha1 = elt.get('inventory_sha1')
361
parents = elt.find('parents') or []
363
assert p.tag == 'revision_ref', \
364
"bad parent node tag %r" % p.tag
365
rev.parent_ids.append(get_cached(p.get('revision_id')))
366
self._unpack_revision_properties(elt, rev)
367
v = elt.get('timezone')
371
rev.timezone = int(v)
372
rev.message = elt.findtext('message') # text of <message>
375
def _unpack_revision_properties(self, elt, rev):
376
"""Unpack properties onto a revision."""
377
props_elt = elt.find('properties')
378
assert len(rev.properties) == 0
381
for prop_elt in props_elt:
382
assert prop_elt.tag == 'property', \
383
"bad tag under properties list: %r" % prop_elt.tag
384
name = prop_elt.get('name')
385
value = prop_elt.text
386
# If a property had an empty value ('') cElementTree reads
387
# that back as None, convert it back to '', so that all
388
# properties have string values
391
assert name not in rev.properties, \
392
"repeated property %r" % name
393
rev.properties[name] = value
396
serializer_v5 = Serializer_v5()
added
removed
bzrlib/xml5.py
