← Back to branch summary

~bzr-pqm/bzr/bzr.dev

« back to all changes in this revision

Viewing changes to bzrlib/xml_serializer.py

  • Committer: Andrew Bennetts
  • Date: 2009-04-02 05:53:12 UTC
  • mto: This revision was merged to the branch mainline in revision 4242.
  • Revision ID: andrew.bennetts@canonical.com-20090402055312-h7mvgumvm7e620mj
Fix nits in spelling and naming.

Show diffs side-by-side

added added

removed removed

Lines of Context:
bzrlib/xml_serializer.py
1
 
# Copyright (C) 2005-2010 Canonical Ltd
 
1
# Copyright (C) 2005, 2006 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
 
import re
26
 
 
27
 
from bzrlib.serializer import Serializer
28
 
from bzrlib.trace import mutter
 
25
from bzrlib import registry
 
26
from bzrlib.trace import mutter, warning
29
27
 
30
28
try:
31
29
    try:
33
31
        from xml.etree.cElementTree import (ElementTree, SubElement, Element,
34
32
            XMLTreeBuilder, fromstring, tostring)
35
33
        import xml.etree as elementtree
36
 
        # Also import ElementTree module so monkey-patching below always works
37
 
        import xml.etree.ElementTree
38
34
    except ImportError:
39
35
        from cElementTree import (ElementTree, SubElement, Element,
40
36
                                  XMLTreeBuilder, fromstring, tostring)
53
49
from bzrlib import errors
54
50
 
55
51
 
56
 
class XMLSerializer(Serializer):
57
 
    """Abstract XML object serialize/deserialize"""
58
 
 
59
 
    squashes_xml_invalid_characters = True
 
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)
60
61
 
61
62
    def read_inventory_from_string(self, xml_string, revision_id=None,
62
 
                                   entry_cache=None, return_from_cache=False):
 
63
                                   entry_cache=None):
63
64
        """Read xml_string into an inventory object.
64
65
 
65
66
        :param xml_string: The xml to read.
73
74
        :param entry_cache: An optional cache of InventoryEntry objects. If
74
75
            supplied we will look up entries via (file_id, revision_id) which
75
76
            should map to a valid InventoryEntry (File/Directory/etc) object.
76
 
        :param return_from_cache: Return entries directly from the cache,
77
 
            rather than copying them first. This is only safe if the caller
78
 
            promises not to mutate the returned inventory entries, but it can
79
 
            make some operations significantly faster.
80
77
        """
81
78
        try:
82
79
            return self._unpack_inventory(fromstring(xml_string), revision_id,
83
 
                                          entry_cache=entry_cache,
84
 
                                          return_from_cache=return_from_cache)
 
80
                                          entry_cache=entry_cache)
85
81
        except ParseError, e:
86
82
            raise errors.UnexpectedInventoryFormat(e)
87
83
 
88
84
    def read_inventory(self, f, revision_id=None):
89
85
        try:
90
 
            try:
91
 
                return self._unpack_inventory(self._read_element(f),
92
 
                    revision_id=None)
93
 
            finally:
94
 
                f.close()
 
86
            return self._unpack_inventory(self._read_element(f),
 
87
                revision_id=None)
95
88
        except ParseError, e:
96
89
            raise errors.UnexpectedInventoryFormat(e)
97
90
 
115
108
        return ElementTree().parse(f)
116
109
 
117
110
 
118
 
def escape_invalid_chars(message):
119
 
    """Escape the XML-invalid characters in a commit message.
120
 
 
121
 
    :param message: Commit message to escape
122
 
    :return: tuple with escaped message and number of characters escaped
123
 
    """
124
 
    if message is None:
125
 
        return None, 0
126
 
    # Python strings can include characters that can't be
127
 
    # represented in well-formed XML; escape characters that
128
 
    # aren't listed in the XML specification
129
 
    # (http://www.w3.org/TR/REC-xml/#NT-Char).
130
 
    return re.subn(u'[^\x09\x0A\x0D\u0020-\uD7FF\uE000-\uFFFD]+',
131
 
            lambda match: match.group(0).encode('unicode_escape'),
132
 
            message)
 
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.
 
114
import re
 
115
escape_re = re.compile("[&'\"<>]")
 
116
escape_map = {
 
117
    "&":'&amp;',
 
118
    "'":"&apos;", # FIXME: overkill
 
119
    "\"":"&quot;",
 
120
    "<":"&lt;",
 
121
    ">":"&gt;",
 
122
    }
 
123
def _escape_replace(match, map=escape_map):
 
124
    return map[match.group()]
 
125
 
 
126
def _escape_attrib(text, encoding=None, replace=None):
 
127
    # escape attribute value
 
128
    try:
 
129
        if encoding:
 
130
            try:
 
131
                text = elementtree.ElementTree._encode(text, encoding)
 
132
            except UnicodeError:
 
133
                return elementtree.ElementTree._encode_entity(text)
 
134
        if replace is None:
 
135
            return escape_re.sub(_escape_replace, text)
 
136
        else:
 
137
            text = replace(text, "&", "&amp;")
 
138
            text = replace(text, "'", "&apos;") # FIXME: overkill
 
139
            text = replace(text, "\"", "&quot;")
 
140
            text = replace(text, "<", "&lt;")
 
141
            text = replace(text, ">", "&gt;")
 
142
            return text
 
143
    except (TypeError, AttributeError):
 
144
        elementtree.ElementTree._raise_serialization_error(text)
 
145
 
 
146
elementtree.ElementTree._escape_attrib = _escape_attrib
 
147
 
 
148
escape_cdata_re = re.compile("[&<>]")
 
149
escape_cdata_map = {
 
150
    "&":'&amp;',
 
151
    "<":"&lt;",
 
152
    ">":"&gt;",
 
153
    }
 
154
def _escape_cdata_replace(match, map=escape_cdata_map):
 
155
    return map[match.group()]
 
156
 
 
157
def _escape_cdata(text, encoding=None, replace=None):
 
158
    # escape character data
 
159
    try:
 
160
        if encoding:
 
161
            try:
 
162
                text = elementtree.ElementTree._encode(text, encoding)
 
163
            except UnicodeError:
 
164
                return elementtree.ElementTree._encode_entity(text)
 
165
        if replace is None:
 
166
            return escape_cdata_re.sub(_escape_cdata_replace, text)
 
167
        else:
 
168
            text = replace(text, "&", "&amp;")
 
169
            text = replace(text, "<", "&lt;")
 
170
            text = replace(text, ">", "&gt;")
 
171
            return text
 
172
    except (TypeError, AttributeError):
 
173
        elementtree.ElementTree._raise_serialization_error(text)
 
174
 
 
175
elementtree.ElementTree._escape_cdata = _escape_cdata
 
176
 
 
177
 
 
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')