« back to all changes in this revision
Viewing changes to bzrlib/pack.py
- browse files at revision 2506.6.1
- compare with another revision
- compare with revision 6020
- download diff
- download tarball
- view history from revision 2506.6.1
- Committer: Andrew Bennetts
- Date: 2007-06-14 11:23:38 UTC
- mto: (2506.2.7 container-format)
- mto: This revision was merged to the branch mainline in revision 2572.
- Revision ID: andrew.bennetts@canonical.com-20070614112338-6u3900u6nkag66u8
Return a callable instead of a str from read, and add more validation.
added
removed
bzrlib/pack.py
37
37
name.
38
38
39
39
:raises InvalidRecordError: if name is not valid.
40
:seealso: _check_name_encoding
40
41
"""
41
# XXX: consider checking that name is a str of valid UTF-8 too?
42
42
if _whitespace_re.search(name) is not None:
43
43
raise errors.InvalidRecordError("%r is not a valid name." % (name,))
44
44
45
45
46
def _check_name_encoding(name):
47
"""Check that 'name' is valid UTF-8.
48
49
This is separate from _check_name because UTF-8 decoding is relatively
50
expensive, and we usually want to avoid it.
51
52
:raises InvalidRecordError: if name is not valid UTF-8.
53
"""
54
try:
55
name.decode('utf-8')
56
except UnicodeDecodeError, e:
57
raise errors.InvalidRecordError(str(e))
58
59
46
60
class ContainerWriter(object):
47
61
"""A class for writing containers."""
48
62
122
136
Each yielded record will be a 2-tuple of (names, bytes), where names is
123
137
a ``list`` and bytes is a ``str``.
124
138
125
:raises UnknownContainerFormatError: if the format of the container is
126
unrecognised.
139
:raises ContainerError: if any sort of containter corruption is
140
detected, e.g. UnknownContainerFormatError is the format of the
141
container is unrecognised.
127
142
"""
128
format = self._read_line()
129
if format != FORMAT_ONE:
130
raise errors.UnknownContainerFormatError(format)
143
self._read_format()
131
144
return self._iter_records()
132
145
146
def iter_record_objects(self):
147
"""Iterate over the container, yielding each record as it is read.
148
149
Each yielded record will be an object with ``read`` and ``validate``
150
methods.
151
152
:raises ContainerError: if any sort of containter corruption is
153
detected, e.g. UnknownContainerFormatError is the format of the
154
container is unrecognised.
155
"""
156
self._read_format()
157
return self._iter_record_objects()
158
133
159
def _iter_records(self):
160
for record in self._iter_record_objects():
161
yield record.read()
162
163
def _iter_record_objects(self):
134
164
while True:
135
165
record_kind = self.reader_func(1)
136
166
if record_kind == 'B':
137
167
# Bytes record.
138
168
reader = BytesRecordReader(self.reader_func)
139
yield reader.read()
169
yield reader
140
170
elif record_kind == 'E':
141
171
# End marker. There are no more records.
142
172
return
148
178
# Unknown record type.
149
179
raise errors.UnknownRecordTypeError(record_kind)
150
180
181
def _read_format(self):
182
format = self._read_line()
183
if format != FORMAT_ONE:
184
raise errors.UnknownContainerFormatError(format)
185
151
186
def validate(self):
152
187
"""Validate this container and its records.
153
188
155
190
156
191
:raises ContainerError: if something is invalid.
157
192
"""
158
for names, bytes in self.iter_records():
159
# XXX: bytes is str, should be callable to get bytes.
160
# callable()
161
pass
193
all_names = set()
194
for record_names, read_bytes in self.iter_records():
195
read_bytes(None)
196
for name in record_names:
197
_check_name_encoding(name)
198
# Check that the name is unique. Note that Python will refuse
199
# to decode non-shortest forms of UTF-8 encoding, so there is no
200
# risk that the same unicode string has been encoded two
201
# different ways.
202
if name in all_names:
203
raise errors.DuplicateRecordNameError(name)
204
all_names.add(name)
162
205
excess_bytes = self.reader_func(1)
163
206
if excess_bytes != '':
164
207
raise errors.ContainerHasExcessDataError(excess_bytes)
171
214
172
215
You can either validate or read, you can't do both.
173
216
174
:returns: (names, bytes)
217
:returns: A tuple of (names, callable). The callable can be called
218
repeatedly to obtain the bytes for the record, with a max_length
219
argument. If max_length is None, returns all the bytes. Because
220
records can be arbitrarily large, using None is not recommended
221
unless you have reason to believe the content will fit in memory.
175
222
"""
176
223
# Read the content length.
177
224
length_line = self._read_line()
189
236
break
190
237
_check_name(name)
191
238
names.append(name)
192
bytes = self.reader_func(length)
193
if len(bytes) != length:
239
240
self._remaining_length = length
241
return names, self._content_reader
242
243
def _content_reader(self, max_length):
244
if max_length is None:
245
length_to_read = self._remaining_length
246
else:
247
length_to_read = min(max_length, self._remaining_length)
248
self._remaining_length -= length_to_read
249
bytes = self.reader_func(length_to_read)
250
if len(bytes) != length_to_read:
194
251
raise errors.UnexpectedEndOfContainerError()
195
return names, bytes
252
return bytes
196
253
197
254
def validate(self):
198
255
"""Validate this record.
201
258
202
259
:raises ContainerError: if this record is invalid.
203
260
"""
204
self.read()
261
names, read_bytes = self.read()
262
for name in names:
263
_check_name_encoding(name)
264
read_bytes(None)
265
Loggerhead is a web-based interface for Breezy
Version: 2.0.1