← 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: 2007-06-26 19:10:00 UTC
  • mfrom: (2552.1.1 jam-integration)
  • Revision ID: pqm@pqm.ubuntu.com-20070626191000-flte155pupv54bcs
(Adeodato Simó) fix bug #120591 by using exact encoding for merge directives.

Show diffs side-by-side

added added

removed removed

Lines of Context:
bzrlib/xml_serializer.py
1
 
#! /usr/bin/env python
2
 
# -*- coding: UTF-8 -*-
3
 
 
 
1
# Copyright (C) 2005, 2006 Canonical Ltd
 
2
#
4
3
# This program is free software; you can redistribute it and/or modify
5
4
# it under the terms of the GNU General Public License as published by
6
5
# the Free Software Foundation; either version 2 of the License, or
7
6
# (at your option) any later version.
8
 
 
 
7
#
9
8
# This program is distributed in the hope that it will be useful,
10
9
# but WITHOUT ANY WARRANTY; without even the implied warranty of
11
10
# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
12
11
# GNU General Public License for more details.
13
 
 
 
12
#
14
13
# You should have received a copy of the GNU General Public License
15
14
# along with this program; if not, write to the Free Software
16
15
# Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307  USA
26
25
from bzrlib.trace import mutter, warning
27
26
 
28
27
try:
29
 
    from cElementTree import (ElementTree, SubElement, Element,
30
 
                              XMLTreeBuilder, fromstring, tostring)
 
28
    try:
 
29
        # it's in this package in python2.5
 
30
        from xml.etree.cElementTree import (ElementTree, SubElement, Element,
 
31
            XMLTreeBuilder, fromstring, tostring)
 
32
        import xml.etree as elementtree
 
33
    except ImportError:
 
34
        from cElementTree import (ElementTree, SubElement, Element,
 
35
                                  XMLTreeBuilder, fromstring, tostring)
 
36
        import elementtree
 
37
    ParseError = SyntaxError
31
38
except ImportError:
32
 
    ## from warnings import warn
33
 
    ## warn('using slower ElementTree; consider installing cElementTree')
 
39
    mutter('WARNING: using slower ElementTree; consider installing cElementTree'
 
40
           " and make sure it's on your PYTHONPATH")
 
41
    # this copy is shipped with bzr
34
42
    from util.elementtree.ElementTree import (ElementTree, SubElement,
35
43
                                              Element, XMLTreeBuilder,
36
44
                                              fromstring, tostring)
 
45
    import util.elementtree as elementtree
 
46
    from xml.parsers.expat import ExpatError as ParseError
37
47
 
38
 
from bzrlib.errors import BzrError
 
48
from bzrlib import errors
39
49
 
40
50
 
41
51
class Serializer(object):
46
56
        self._write_element(elt, f)
47
57
 
48
58
    def write_inventory_to_string(self, inv):
49
 
        return tostring(self._pack_inventory(inv))
 
59
        return tostring(self._pack_inventory(inv)) + '\n'
50
60
 
51
61
    def read_inventory_from_string(self, xml_string):
52
 
        return self._unpack_inventory(fromstring(xml_string))
 
62
        try:
 
63
            return self._unpack_inventory(fromstring(xml_string))
 
64
        except ParseError, e:
 
65
            raise errors.UnexpectedInventoryFormat(e)
53
66
 
54
67
    def read_inventory(self, f):
55
 
        return self._unpack_inventory(self._read_element(f))
 
68
        try:
 
69
            return self._unpack_inventory(self._read_element(f))
 
70
        except ParseError, e:
 
71
            raise errors.UnexpectedInventoryFormat(e)
56
72
 
57
73
    def write_revision(self, rev, f):
58
74
        self._write_element(self._pack_revision(rev), f)
59
75
 
60
76
    def write_revision_to_string(self, rev):
61
 
        return tostring(self._pack_revision(rev), f)
 
77
        return tostring(self._pack_revision(rev)) + '\n'
62
78
 
63
79
    def read_revision(self, f):
64
80
        return self._unpack_revision(self._read_element(f))
72
88
 
73
89
    def _read_element(self, f):
74
90
        return ElementTree().parse(f)
 
91
 
 
92
 
 
93
# performance tuning for elementree's serialiser. This should be
 
94
# sent upstream - RBC 20060523.
 
95
# the functions here are patched into elementtree at runtime.
 
96
import re
 
97
escape_re = re.compile("[&'\"<>]")
 
98
escape_map = {
 
99
    "&":'&amp;',
 
100
    "'":"&apos;", # FIXME: overkill
 
101
    "\"":"&quot;",
 
102
    "<":"&lt;",
 
103
    ">":"&gt;",
 
104
    }
 
105
def _escape_replace(match, map=escape_map):
 
106
    return map[match.group()]
 
107
 
 
108
def _escape_attrib(text, encoding=None, replace=None):
 
109
    # escape attribute value
 
110
    try:
 
111
        if encoding:
 
112
            try:
 
113
                text = elementtree.ElementTree._encode(text, encoding)
 
114
            except UnicodeError:
 
115
                return elementtree.ElementTree._encode_entity(text)
 
116
        if replace is None:
 
117
            return escape_re.sub(_escape_replace, text)
 
118
        else:
 
119
            text = replace(text, "&", "&amp;")
 
120
            text = replace(text, "'", "&apos;") # FIXME: overkill
 
121
            text = replace(text, "\"", "&quot;")
 
122
            text = replace(text, "<", "&lt;")
 
123
            text = replace(text, ">", "&gt;")
 
124
            return text
 
125
    except (TypeError, AttributeError):
 
126
        elementtree.ElementTree._raise_serialization_error(text)
 
127
 
 
128
elementtree.ElementTree._escape_attrib = _escape_attrib
 
129
 
 
130
escape_cdata_re = re.compile("[&<>]")
 
131
escape_cdata_map = {
 
132
    "&":'&amp;',
 
133
    "<":"&lt;",
 
134
    ">":"&gt;",
 
135
    }
 
136
def _escape_cdata_replace(match, map=escape_cdata_map):
 
137
    return map[match.group()]
 
138
 
 
139
def _escape_cdata(text, encoding=None, replace=None):
 
140
    # escape character data
 
141
    try:
 
142
        if encoding:
 
143
            try:
 
144
                text = elementtree.ElementTree._encode(text, encoding)
 
145
            except UnicodeError:
 
146
                return elementtree.ElementTree._encode_entity(text)
 
147
        if replace is None:
 
148
            return escape_cdata_re.sub(_escape_cdata_replace, text)
 
149
        else:
 
150
            text = replace(text, "&", "&amp;")
 
151
            text = replace(text, "<", "&lt;")
 
152
            text = replace(text, ">", "&gt;")
 
153
            return text
 
154
    except (TypeError, AttributeError):
 
155
        elementtree.ElementTree._raise_serialization_error(text)
 
156
 
 
157
elementtree.ElementTree._escape_cdata = _escape_cdata