13
13
# You should have received a copy of the GNU General Public License
14
14
# along with this program; if not, write to the Free Software
15
# Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
15
# Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
17
17
"""Container format for Bazaar data.
19
"Containers" and "records" are described in doc/developers/container-format.txt.
19
"Containers" and "records" are described in
20
doc/developers/container-format.txt.
22
23
from cStringIO import StringIO
58
59
raise errors.InvalidRecordError(str(e))
62
class ContainerSerialiser(object):
63
"""A helper class for serialising containers.
65
It simply returns bytes from method calls to 'begin', 'end' and
66
'bytes_record'. You may find ContainerWriter to be a more convenient
71
"""Return the bytes to begin a container."""
72
return FORMAT_ONE + "\n"
75
"""Return the bytes to finish a container."""
78
def bytes_record(self, bytes, names):
79
"""Return the bytes for a Bytes record with the given name and
85
byte_sections.append(str(len(bytes)) + "\n")
87
for name_tuple in names:
88
# Make sure we're writing valid names. Note that we will leave a
89
# half-written record if a name is bad!
90
for name in name_tuple:
92
byte_sections.append('\x00'.join(name_tuple) + "\n")
94
byte_sections.append("\n")
95
# Finally, the contents.
96
byte_sections.append(bytes)
97
# XXX: This causes a memory copy of bytes in size, but is usually
98
# faster than two write calls (12 vs 13 seconds to output a gig of
99
# 1k records.) - results may differ on significantly larger records
100
# like .iso's but as they should be rare in any case and thus not
101
# likely to be the common case. The biggest issue is causing extreme
102
# memory pressure in that case. One possibly improvement here is to
103
# check the size of the content before deciding to join here vs call
105
return ''.join(byte_sections)
61
108
class ContainerWriter(object):
62
"""A class for writing containers.
109
"""A class for writing containers to a file.
64
111
:attribute records_written: The number of user records added to the
65
112
container. This does not count the prelude or suffix of the container
103
151
and thus are only suitable for use by a ContainerReader.
105
153
current_offset = self.current_offset
109
self.write_func(str(len(bytes)) + "\n")
111
for name_tuple in names:
112
# Make sure we're writing valid names. Note that we will leave a
113
# half-written record if a name is bad!
114
for name in name_tuple:
116
self.write_func('\x00'.join(name_tuple) + "\n")
118
self.write_func("\n")
119
# Finally, the contents.
120
self.write_func(bytes)
154
serialised_record = self._serialiser.bytes_record(bytes, names)
155
self.write_func(serialised_record)
121
156
self.records_written += 1
122
157
# return a memo of where we wrote data to allow random access.
123
158
return current_offset, self.current_offset - current_offset
126
161
class ReadVFile(object):
127
"""Adapt a readv result iterator to a file like protocol."""
162
"""Adapt a readv result iterator to a file like protocol.
164
The readv result must support the iterator protocol returning (offset,
168
# XXX: This could be a generic transport class, as other code may want to
169
# gradually consume the readv result.
129
171
def __init__(self, readv_result):
172
"""Construct a new ReadVFile wrapper.
174
:seealso: make_readv_reader
176
:param readv_result: the most recent readv result - list or generator
178
# readv can return a sequence or an iterator, but we require an
179
# iterator to know how much has been consumed.
180
readv_result = iter(readv_result)
130
181
self.readv_result = readv_result
131
# the most recent readv result block
132
182
self._string = None
135
185
if (self._string is None or
136
186
self._string.tell() == self._string_length):
137
length, data = self.readv_result.next()
187
offset, data = self.readv_result.next()
138
188
self._string_length = len(data)
139
189
self._string = StringIO(data)
199
252
is a ``list`` and bytes is a function that takes one argument,
202
You **must not** call the callable after advancing the interator to the
255
You **must not** call the callable after advancing the iterator to the
203
256
next record. That is, this code is invalid::
205
258
record_iter = container.iter_records()
206
259
names1, callable1 = record_iter.next()
207
260
names2, callable2 = record_iter.next()
208
261
bytes1 = callable1(None)
210
263
As it will give incorrect results and invalidate the state of the
213
:raises ContainerError: if any sort of containter corruption is
266
:raises ContainerError: if any sort of container corruption is
214
267
detected, e.g. UnknownContainerFormatError is the format of the
215
268
container is unrecognised.
216
269
:seealso: ContainerReader.read
218
271
self._read_format()
219
272
return self._iter_records()
221
274
def iter_record_objects(self):
222
275
"""Iterate over the container, yielding each record as it is read.
346
399
_check_name_encoding(name)
403
class ContainerPushParser(object):
404
"""A "push" parser for container format 1.
406
It accepts bytes via the ``accept_bytes`` method, and parses them into
407
records which can be retrieved via the ``read_pending_records`` method.
412
self._state_handler = self._state_expecting_format_line
413
self._parsed_records = []
414
self._reset_current_record()
415
self.finished = False
417
def _reset_current_record(self):
418
self._current_record_length = None
419
self._current_record_names = []
421
def accept_bytes(self, bytes):
422
self._buffer += bytes
423
# Keep iterating the state machine until it stops consuming bytes from
425
last_buffer_length = None
426
cur_buffer_length = len(self._buffer)
427
last_state_handler = None
428
while (cur_buffer_length != last_buffer_length
429
or last_state_handler != self._state_handler):
430
last_buffer_length = cur_buffer_length
431
last_state_handler = self._state_handler
432
self._state_handler()
433
cur_buffer_length = len(self._buffer)
435
def read_pending_records(self, max=None):
437
records = self._parsed_records[:max]
438
del self._parsed_records[:max]
441
records = self._parsed_records
442
self._parsed_records = []
445
def _consume_line(self):
446
"""Take a line out of the buffer, and return the line.
448
If a newline byte is not found in the buffer, the buffer is
449
unchanged and this returns None instead.
451
newline_pos = self._buffer.find('\n')
452
if newline_pos != -1:
453
line = self._buffer[:newline_pos]
454
self._buffer = self._buffer[newline_pos+1:]
459
def _state_expecting_format_line(self):
460
line = self._consume_line()
462
if line != FORMAT_ONE:
463
raise errors.UnknownContainerFormatError(line)
464
self._state_handler = self._state_expecting_record_type
466
def _state_expecting_record_type(self):
467
if len(self._buffer) >= 1:
468
record_type = self._buffer[0]
469
self._buffer = self._buffer[1:]
470
if record_type == 'B':
471
self._state_handler = self._state_expecting_length
472
elif record_type == 'E':
474
self._state_handler = self._state_expecting_nothing
476
raise errors.UnknownRecordTypeError(record_type)
478
def _state_expecting_length(self):
479
line = self._consume_line()
482
self._current_record_length = int(line)
484
raise errors.InvalidRecordError(
485
"%r is not a valid length." % (line,))
486
self._state_handler = self._state_expecting_name
488
def _state_expecting_name(self):
489
encoded_name_parts = self._consume_line()
490
if encoded_name_parts == '':
491
self._state_handler = self._state_expecting_body
492
elif encoded_name_parts:
493
name_parts = tuple(encoded_name_parts.split('\x00'))
494
for name_part in name_parts:
495
_check_name(name_part)
496
self._current_record_names.append(name_parts)
498
def _state_expecting_body(self):
499
if len(self._buffer) >= self._current_record_length:
500
body_bytes = self._buffer[:self._current_record_length]
501
self._buffer = self._buffer[self._current_record_length:]
502
record = (self._current_record_names, body_bytes)
503
self._parsed_records.append(record)
504
self._reset_current_record()
505
self._state_handler = self._state_expecting_record_type
507
def _state_expecting_nothing(self):
510
def read_size_hint(self):
512
if self._state_handler == self._state_expecting_body:
513
remaining = self._current_record_length - len(self._buffer)
516
return max(hint, remaining)
520
def iter_records_from_file(source_file):
521
parser = ContainerPushParser()
523
bytes = source_file.read(parser.read_size_hint())
524
parser.accept_bytes(bytes)
525
for record in parser.read_pending_records():
added
removed
bzrlib/pack.py
