22
22
# importing this module is fairly slow because it has to load several
25
from bzrlib.trace import mutter, warning
25
from bzrlib.serializer import Serializer
26
from bzrlib.trace import mutter
28
from cElementTree import (ElementTree, SubElement, Element,
29
XMLTreeBuilder, fromstring, tostring)
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
30
39
except ImportError:
31
from warnings import warn
32
warn('using slower ElementTree; consider installing cElementTree')
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
33
43
from util.elementtree.ElementTree import (ElementTree, SubElement,
34
44
Element, XMLTreeBuilder,
35
45
fromstring, tostring)
37
from bzrlib.inventory import ROOT_ID, Inventory, InventoryEntry
38
from bzrlib.revision import Revision, RevisionReference
39
from bzrlib.errors import BzrError
42
class Serializer(object):
43
"""Abstract object serialize/deserialize"""
44
def write_inventory(self, inv, f):
45
"""Write inventory to a file"""
46
elt = self._pack_inventory(inv)
47
self._write_element(elt, f)
49
def write_inventory_to_string(self, inv):
50
return tostring(self._pack_inventory(inv))
52
def read_inventory_from_string(self, xml_string):
53
return self._unpack_inventory(fromstring(xml_string))
55
def read_inventory(self, f):
56
return self._unpack_inventory(self._read_element(f))
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)
58
91
def write_revision(self, rev, f):
59
92
self._write_element(self._pack_revision(rev), f)
61
94
def write_revision_to_string(self, rev):
62
return tostring(self._pack_revision(rev), f)
95
return tostring(self._pack_revision(rev)) + '\n'
64
97
def read_revision(self, f):
65
98
return self._unpack_revision(self._read_element(f))
75
108
return ElementTree().parse(f)
78
class _Serializer_v4(Serializer):
79
"""Version 0.0.4 serializer
81
You should use the serialzer_v4 singleton."""
85
def _pack_inventory(self, inv):
86
"""Convert to XML Element"""
87
e = Element('inventory')
89
if inv.root.file_id not in (None, ROOT_ID):
90
e.set('file_id', inv.root.file_id)
91
for path, ie in inv.iter_entries():
92
e.append(self._pack_entry(ie))
96
def _pack_entry(self, ie):
97
"""Convert InventoryEntry to XML element"""
99
e.set('name', ie.name)
100
e.set('file_id', ie.file_id)
101
e.set('kind', ie.kind)
103
if ie.text_size != None:
104
e.set('text_size', '%d' % ie.text_size)
106
for f in ['text_id', 'text_sha1']:
111
# to be conservative, we don't externalize the root pointers
112
# for now, leaving them as null in the xml form. in a future
113
# version it will be implied by nested elements.
114
if ie.parent_id != ROOT_ID:
115
assert isinstance(ie.parent_id, basestring)
116
e.set('parent_id', ie.parent_id)
123
def _unpack_inventory(self, elt):
124
"""Construct from XML Element
126
assert elt.tag == 'inventory'
127
root_id = elt.get('file_id') or ROOT_ID
128
inv = Inventory(root_id)
130
ie = self._unpack_entry(e)
131
if ie.parent_id == ROOT_ID:
132
ie.parent_id = root_id
137
def _unpack_entry(self, elt):
138
assert elt.tag == 'entry'
140
## original format inventories don't have a parent_id for
141
## nodes in the root directory, but it's cleaner to use one
143
parent_id = elt.get('parent_id')
144
if parent_id == None:
147
ie = InventoryEntry(elt.get('file_id'),
151
ie.text_id = elt.get('text_id')
152
ie.text_sha1 = elt.get('text_sha1')
154
## mutter("read inventoryentry: %r" % (elt.attrib))
156
v = elt.get('text_size')
157
ie.text_size = v and int(v)
162
def _pack_revision(self, rev):
163
"""Revision object -> xml tree"""
164
root = Element('revision',
165
committer = rev.committer,
166
timestamp = '%.9f' % rev.timestamp,
167
revision_id = rev.revision_id,
168
inventory_id = rev.inventory_id,
169
inventory_sha1 = rev.inventory_sha1,
172
root.set('timezone', str(rev.timezone))
175
msg = SubElement(root, 'message')
176
msg.text = rev.message
180
pelts = SubElement(root, 'parents')
181
pelts.tail = pelts.text = '\n'
182
for rr in rev.parents:
183
assert isinstance(rr, RevisionReference)
184
p = SubElement(pelts, 'revision_ref')
186
assert rr.revision_id
187
p.set('revision_id', rr.revision_id)
189
p.set('revision_sha1', rr.revision_sha1)
194
def _unpack_revision(self, elt):
195
"""XML Element -> Revision object"""
197
# <changeset> is deprecated...
198
if elt.tag not in ('revision', 'changeset'):
199
raise BzrError("unexpected tag in revision file: %r" % elt)
201
rev = Revision(committer = elt.get('committer'),
202
timestamp = float(elt.get('timestamp')),
203
revision_id = elt.get('revision_id'),
204
inventory_id = elt.get('inventory_id'),
205
inventory_sha1 = elt.get('inventory_sha1')
208
precursor = elt.get('precursor')
209
precursor_sha1 = elt.get('precursor_sha1')
211
pelts = elt.find('parents')
215
assert p.tag == 'revision_ref', \
216
"bad parent node tag %r" % p.tag
217
rev_ref = RevisionReference(p.get('revision_id'),
218
p.get('revision_sha1'))
219
rev.parents.append(rev_ref)
223
prec_parent = rev.parents[0].revision_id
224
assert prec_parent == precursor
226
# revisions written prior to 0.0.5 have a single precursor
227
# give as an attribute
228
rev_ref = RevisionReference(precursor, precursor_sha1)
229
rev.parents.append(rev_ref)
231
v = elt.get('timezone')
232
rev.timezone = v and int(v)
234
rev.message = elt.findtext('message') # text of <message>
240
"""singleton instance"""
241
serializer_v4 = _Serializer_v4()
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'),
added
removed
bzrlib/xml_serializer.py
