← Back to branch summary

~bzr-pqm/bzr/bzr.dev

« back to all changes in this revision

Viewing changes to bzrlib/xml_serializer.py

  • Committer: Canonical.com Patch Queue Manager
  • Date: 2010-09-01 08:02:42 UTC
  • mfrom: (5390.3.3 faster-revert-593560)
  • Revision ID: pqm@pqm.ubuntu.com-20100901080242-esg62ody4frwmy66
(spiv) Avoid repeatedly calling self.target.all_file_ids() in
 InterTree.iter_changes. (Andrew Bennetts)

Show diffs side-by-side

added added

removed removed

Lines of Context:
bzrlib/xml_serializer.py
1
 
# Copyright (C) 2005, 2006 Canonical Ltd
 
1
# Copyright (C) 2005-2010 Canonical Ltd
2
2
#
3
3
# This program is free software; you can redistribute it and/or modify
4
4
# it under the terms of the GNU General Public License as published by
22
22
# importing this module is fairly slow because it has to load several
23
23
# ElementTree bits
24
24
 
25
 
from bzrlib import registry
26
 
from bzrlib.trace import mutter, warning
 
25
from bzrlib.serializer import Serializer
 
26
from bzrlib.trace import mutter
27
27
 
28
28
try:
29
29
    try:
49
49
from bzrlib import errors
50
50
 
51
51
 
52
 
class Serializer(object):
53
 
    """Abstract object serialize/deserialize"""
54
 
 
55
 
    def write_inventory(self, inv, f):
56
 
        """Write inventory to a file"""
57
 
        raise NotImplementedError(self.write_inventory)
58
 
 
59
 
    def write_inventory_to_string(self, inv):
60
 
        raise NotImplementedError(self.write_inventory_to_string)
 
52
class XMLSerializer(Serializer):
 
53
    """Abstract XML object serialize/deserialize"""
 
54
 
 
55
    squashes_xml_invalid_characters = True
61
56
 
62
57
    def read_inventory_from_string(self, xml_string, revision_id=None,
63
 
                                   entry_cache=None):
 
58
                                   entry_cache=None, return_from_cache=False):
64
59
        """Read xml_string into an inventory object.
65
60
 
66
61
        :param xml_string: The xml to read.
74
69
        :param entry_cache: An optional cache of InventoryEntry objects. If
75
70
            supplied we will look up entries via (file_id, revision_id) which
76
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.
77
76
        """
78
77
        try:
79
78
            return self._unpack_inventory(fromstring(xml_string), revision_id,
80
 
                                          entry_cache=entry_cache)
 
79
                                          entry_cache=entry_cache,
 
80
                                          return_from_cache=return_from_cache)
81
81
        except ParseError, e:
82
82
            raise errors.UnexpectedInventoryFormat(e)
83
83
 
84
84
    def read_inventory(self, f, revision_id=None):
85
85
        try:
86
 
            return self._unpack_inventory(self._read_element(f),
87
 
                revision_id=None)
 
86
            try:
 
87
                return self._unpack_inventory(self._read_element(f),
 
88
                    revision_id=None)
 
89
            finally:
 
90
                f.close()
88
91
        except ParseError, e:
89
92
            raise errors.UnexpectedInventoryFormat(e)
90
93
 
175
178
elementtree.ElementTree._escape_cdata = _escape_cdata
176
179
 
177
180
 
178
 
class SerializerRegistry(registry.Registry):
179
 
    """Registry for serializer objects"""
180
 
 
181
 
 
182
 
format_registry = SerializerRegistry()
183
 
format_registry.register_lazy('4', 'bzrlib.xml4', 'serializer_v4')
184
 
format_registry.register_lazy('5', 'bzrlib.xml5', 'serializer_v5')
185
 
format_registry.register_lazy('6', 'bzrlib.xml6', 'serializer_v6')
186
 
format_registry.register_lazy('7', 'bzrlib.xml7', 'serializer_v7')
187
 
format_registry.register_lazy('8', 'bzrlib.xml8', 'serializer_v8')
 
181
def escape_invalid_chars(message):
 
182
    """Escape the XML-invalid characters in a commit message.
 
183
 
 
184
    :param message: Commit message to escape
 
185
    :return: tuple with escaped message and number of characters escaped
 
186
    """
 
187
    if message is None:
 
188
        return None, 0
 
189
    # Python strings can include characters that can't be
 
190
    # represented in well-formed XML; escape characters that
 
191
    # aren't listed in the XML specification
 
192
    # (http://www.w3.org/TR/REC-xml/#NT-Char).
 
193
    return re.subn(u'[^\x09\x0A\x0D\u0020-\uD7FF\uE000-\uFFFD]+',
 
194
            lambda match: match.group(0).encode('unicode_escape'),
 
195
            message)