22
22
# importing this module is fairly slow because it has to load several
25
from bzrlib.serializer import Serializer
26
from bzrlib.trace import mutter
30
# it's in this package in python2.5
31
from xml.etree.cElementTree import (ElementTree, SubElement, Element,
32
XMLTreeBuilder, fromstring, tostring)
33
import xml.etree as elementtree
35
from cElementTree import (ElementTree, SubElement, Element,
36
XMLTreeBuilder, fromstring, tostring)
37
import elementtree.ElementTree
38
ParseError = SyntaxError
26
from util.cElementTree import (ElementTree, SubElement, Element,
39
28
except ImportError:
40
mutter('WARNING: using slower ElementTree; consider installing cElementTree'
41
" and make sure it's on your PYTHONPATH")
42
# this copy is shipped with bzr
43
29
from util.elementtree.ElementTree import (ElementTree, SubElement,
44
Element, XMLTreeBuilder,
46
import util.elementtree as elementtree
47
from xml.parsers.expat import ExpatError as ParseError
49
from bzrlib import errors
52
class XMLSerializer(Serializer):
53
"""Abstract XML object serialize/deserialize"""
55
squashes_xml_invalid_characters = True
57
def read_inventory_from_string(self, xml_string, revision_id=None,
58
entry_cache=None, return_from_cache=False):
59
"""Read xml_string into an inventory object.
61
:param xml_string: The xml to read.
62
:param revision_id: If not-None, the expected revision id of the
63
inventory. Some serialisers use this to set the results' root
64
revision. This should be supplied for deserialising all
65
from-repository inventories so that xml5 inventories that were
66
serialised without a revision identifier can be given the right
67
revision id (but not for working tree inventories where users can
68
edit the data without triggering checksum errors or anything).
69
:param entry_cache: An optional cache of InventoryEntry objects. If
70
supplied we will look up entries via (file_id, revision_id) which
71
should map to a valid InventoryEntry (File/Directory/etc) object.
72
:param return_from_cache: Return entries directly from the cache,
73
rather than copying them first. This is only safe if the caller
74
promises not to mutate the returned inventory entries, but it can
75
make some operations significantly faster.
78
return self._unpack_inventory(fromstring(xml_string), revision_id,
79
entry_cache=entry_cache,
80
return_from_cache=return_from_cache)
82
raise errors.UnexpectedInventoryFormat(e)
84
def read_inventory(self, f, revision_id=None):
86
return self._unpack_inventory(self._read_element(f),
89
raise errors.UnexpectedInventoryFormat(e)
30
Element, XMLTreeBuilder)
32
from bzrlib.inventory import ROOT_ID, Inventory, InventoryEntry
33
from bzrlib.revision import Revision, RevisionReference
34
from bzrlib.errors import BzrError
37
class Serializer(object):
38
"""Abstract object serialize/deserialize"""
39
def write_inventory(self, inv, f):
40
"""Write inventory to a file"""
41
elt = self._pack_inventory(inv)
42
self._write_element(elt, f)
44
def read_inventory_from_string(self, xml_string):
45
return self._unpack_inventory(self._parse_string(xml_string))
47
def read_inventory(self, f):
48
return self._unpack_inventory(self._read_element(f))
91
50
def write_revision(self, rev, f):
92
51
self._write_element(self._pack_revision(rev), f)
94
def write_revision_to_string(self, rev):
95
return tostring(self._pack_revision(rev)) + '\n'
97
53
def read_revision(self, f):
98
54
return self._unpack_revision(self._read_element(f))
100
56
def read_revision_from_string(self, xml_string):
101
return self._unpack_revision(fromstring(xml_string))
57
return self._unpack_revision(self._parse_string(xml_string))
103
59
def _write_element(self, elt, f):
104
60
ElementTree(elt).write(f, 'utf-8')
107
63
def _read_element(self, f):
108
64
return ElementTree().parse(f)
111
# performance tuning for elementree's serialiser. This should be
112
# sent upstream - RBC 20060523.
113
# the functions here are patched into elementtree at runtime.
115
escape_re = re.compile("[&'\"<>]")
118
"'":"'", # FIXME: overkill
123
def _escape_replace(match, map=escape_map):
124
return map[match.group()]
126
def _escape_attrib(text, encoding=None, replace=None):
127
# escape attribute value
131
text = elementtree.ElementTree._encode(text, encoding)
133
return elementtree.ElementTree._encode_entity(text)
135
return escape_re.sub(_escape_replace, text)
137
text = replace(text, "&", "&")
138
text = replace(text, "'", "'") # FIXME: overkill
139
text = replace(text, "\"", """)
140
text = replace(text, "<", "<")
141
text = replace(text, ">", ">")
143
except (TypeError, AttributeError):
144
elementtree.ElementTree._raise_serialization_error(text)
146
elementtree.ElementTree._escape_attrib = _escape_attrib
148
escape_cdata_re = re.compile("[&<>]")
154
def _escape_cdata_replace(match, map=escape_cdata_map):
155
return map[match.group()]
157
def _escape_cdata(text, encoding=None, replace=None):
158
# escape character data
162
text = elementtree.ElementTree._encode(text, encoding)
164
return elementtree.ElementTree._encode_entity(text)
166
return escape_cdata_re.sub(_escape_cdata_replace, text)
168
text = replace(text, "&", "&")
169
text = replace(text, "<", "<")
170
text = replace(text, ">", ">")
172
except (TypeError, AttributeError):
173
elementtree.ElementTree._raise_serialization_error(text)
175
elementtree.ElementTree._escape_cdata = _escape_cdata
178
def escape_invalid_chars(message):
179
"""Escape the XML-invalid characters in a commit message.
181
:param message: Commit message to escape
182
:return: tuple with escaped message and number of characters escaped
186
# Python strings can include characters that can't be
187
# represented in well-formed XML; escape characters that
188
# aren't listed in the XML specification
189
# (http://www.w3.org/TR/REC-xml/#NT-Char).
190
return re.subn(u'[^\x09\x0A\x0D\u0020-\uD7FF\uE000-\uFFFD]+',
191
lambda match: match.group(0).encode('unicode_escape'),
66
def _parse_string(self, xml_string):
67
parser = XMLTreeBuilder()
68
parser.feed(xml_string)
73
class _Serializer_v4(Serializer):
74
"""Version 0.0.4 serializer
76
You should use the serialzer_v4 singleton."""
80
def _pack_inventory(self, inv):
81
"""Convert to XML Element"""
82
e = Element('inventory')
84
if inv.root.file_id not in (None, ROOT_ID):
85
e.set('file_id', inv.root.file_id)
86
for path, ie in inv.iter_entries():
87
e.append(self._pack_entry(ie))
91
def _pack_entry(self, ie):
92
"""Convert InventoryEntry to XML element"""
94
e.set('name', ie.name)
95
e.set('file_id', ie.file_id)
96
e.set('kind', ie.kind)
98
if ie.text_size != None:
99
e.set('text_size', '%d' % ie.text_size)
101
for f in ['text_id', 'text_sha1']:
106
# to be conservative, we don't externalize the root pointers
107
# for now, leaving them as null in the xml form. in a future
108
# version it will be implied by nested elements.
109
if ie.parent_id != ROOT_ID:
110
assert isinstance(ie.parent_id, basestring)
111
e.set('parent_id', ie.parent_id)
118
def _unpack_inventory(self, elt):
119
"""Construct from XML Element
121
assert elt.tag == 'inventory'
122
root_id = elt.get('file_id') or ROOT_ID
123
inv = Inventory(root_id)
125
ie = self._unpack_entry(e)
126
if ie.parent_id == ROOT_ID:
127
ie.parent_id = root_id
132
def _unpack_entry(self, elt):
133
assert elt.tag == 'entry'
135
## original format inventories don't have a parent_id for
136
## nodes in the root directory, but it's cleaner to use one
138
parent_id = elt.get('parent_id')
139
if parent_id == None:
142
ie = InventoryEntry(elt.get('file_id'),
146
ie.text_id = elt.get('text_id')
147
ie.text_sha1 = elt.get('text_sha1')
149
## mutter("read inventoryentry: %r" % (elt.attrib))
151
v = elt.get('text_size')
152
ie.text_size = v and int(v)
157
def _pack_revision(self, rev):
158
"""Revision object -> xml tree"""
159
root = Element('revision',
160
committer = rev.committer,
161
timestamp = '%.9f' % rev.timestamp,
162
revision_id = rev.revision_id,
163
inventory_id = rev.inventory_id,
164
inventory_sha1 = rev.inventory_sha1,
167
root.set('timezone', str(rev.timezone))
170
msg = SubElement(root, 'message')
171
msg.text = rev.message
175
pelts = SubElement(root, 'parents')
176
pelts.tail = pelts.text = '\n'
177
for rr in rev.parents:
178
assert isinstance(rr, RevisionReference)
179
p = SubElement(pelts, 'revision_ref')
181
assert rr.revision_id
182
p.set('revision_id', rr.revision_id)
184
p.set('revision_sha1', rr.revision_sha1)
189
def _unpack_revision(self, elt):
190
"""XML Element -> Revision object"""
192
# <changeset> is deprecated...
193
if elt.tag not in ('revision', 'changeset'):
194
raise BzrError("unexpected tag in revision file: %r" % elt)
196
rev = Revision(committer = elt.get('committer'),
197
timestamp = float(elt.get('timestamp')),
198
revision_id = elt.get('revision_id'),
199
inventory_id = elt.get('inventory_id'),
200
inventory_sha1 = elt.get('inventory_sha1')
203
precursor = elt.get('precursor')
204
precursor_sha1 = elt.get('precursor_sha1')
206
pelts = elt.find('parents')
210
assert p.tag == 'revision_ref', \
211
"bad parent node tag %r" % p.tag
212
rev_ref = RevisionReference(p.get('revision_id'),
213
p.get('revision_sha1'))
214
rev.parents.append(rev_ref)
218
prec_parent = rev.parents[0].revision_id
219
assert prec_parent == precursor
221
# revisions written prior to 0.0.5 have a single precursor
222
# give as an attribute
223
rev_ref = RevisionReference(precursor, precursor_sha1)
224
rev.parents.append(rev_ref)
226
v = elt.get('timezone')
227
rev.timezone = v and int(v)
229
rev.message = elt.findtext('message') # text of <message>
235
"""singleton instance"""
236
serializer_v4 = _Serializer_v4()
added
removed
bzrlib/xml.py
