~bzr-pqm/bzr/bzr.dev

« back to all changes in this revision

Viewing changes to bzrlib/bundle/serializer/v08.py

  • Committer: Patch Queue Manager
  • Date: 2011-09-22 14:12:18 UTC
  • mfrom: (6155.3.1 jam)
  • Revision ID: pqm@pqm.ubuntu.com-20110922141218-86s4uu6nqvourw4f
(jameinel) Cleanup comments bzrlib/smart/__init__.py (John A Meinel)

Show diffs side-by-side

added added

removed removed

Lines of Context:
1
 
# (C) 2005 Canonical Development Ltd
2
 
 
 
1
# Copyright (C) 2005, 2006, 2009 Canonical Ltd
 
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
5
5
# the Free Software Foundation; either version 2 of the License, or
6
6
# (at your option) any later version.
7
 
 
 
7
#
8
8
# This program is distributed in the hope that it will be useful,
9
9
# but WITHOUT ANY WARRANTY; without even the implied warranty of
10
10
# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
11
11
# GNU General Public License for more details.
12
 
 
 
12
#
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
16
16
 
17
17
"""Serializer factory for reading and writing bundles.
18
18
"""
19
19
 
20
20
import os
21
21
 
22
 
from bzrlib.bundle.serializer import (BundleSerializer, 
23
 
                                      BUNDLE_HEADER, 
24
 
                                      format_highres_date,
25
 
                                      unpack_highres_date,
 
22
from bzrlib import (
 
23
    errors,
 
24
    ui,
 
25
    )
 
26
from bzrlib.bundle.serializer import (BundleSerializer,
 
27
                                      _get_bundle_header,
26
28
                                     )
27
29
from bzrlib.bundle.serializer import binary_diff
28
 
from bzrlib.delta import compare_trees
 
30
from bzrlib.bundle.bundle_data import (RevisionInfo, BundleInfo, BundleTree)
29
31
from bzrlib.diff import internal_diff
30
 
import bzrlib.errors as errors
31
32
from bzrlib.osutils import pathjoin
32
 
from bzrlib.progress import DummyProgress
33
33
from bzrlib.revision import NULL_REVISION
34
 
from bzrlib.rio import RioWriter, read_stanzas
35
 
import bzrlib.ui
36
34
from bzrlib.testament import StrictTestament
 
35
from bzrlib.timestamp import (
 
36
    format_highres_date,
 
37
    unpack_highres_date,
 
38
)
37
39
from bzrlib.textfile import text_file
 
40
from bzrlib.trace import mutter
38
41
 
39
42
bool_text = {True: 'yes', False: 'no'}
40
43
 
53
56
        else:
54
57
            self.properties = properties
55
58
 
 
59
    def add_utf8_property(self, name, value):
 
60
        """Add a property whose value is currently utf8 to the action."""
 
61
        self.properties.append((name, value.decode('utf8')))
 
62
 
56
63
    def add_property(self, name, value):
57
64
        """Add a property to the action"""
58
65
        self.properties.append((name, value))
84
91
        to_file.write(text_line+'\n')
85
92
 
86
93
 
87
 
class BundleSerializerV07(BundleSerializer):
 
94
class BundleSerializerV08(BundleSerializer):
88
95
    def read(self, f):
89
96
        """Read the rest of the bundles from the supplied file.
90
97
 
91
98
        :param f: The file to read from
92
99
        :return: A list of bundles
93
100
        """
94
 
        assert self.version == '0.7'
95
 
        # The first line of the header should have been read
96
 
        raise NotImplementedError
 
101
        return BundleReader(f).info
 
102
 
 
103
    def check_compatible(self):
 
104
        if self.source.supports_rich_root():
 
105
            raise errors.IncompatibleBundleFormat('0.8', repr(self.source))
97
106
 
98
107
    def write(self, source, revision_ids, forced_bases, f):
99
108
        """Write the bundless to the supplied files.
107
116
        self.revision_ids = revision_ids
108
117
        self.forced_bases = forced_bases
109
118
        self.to_file = f
 
119
        self.check_compatible()
110
120
        source.lock_read()
111
121
        try:
112
122
            self._write_main_header()
113
 
            pb = DummyProgress()
 
123
            pb = ui.ui_factory.nested_progress_bar()
114
124
            try:
115
125
                self._write_revisions(pb)
116
126
            finally:
117
 
                pass
118
 
                #pb.finished()
 
127
                pb.finished()
119
128
        finally:
120
129
            source.unlock()
121
130
 
 
131
    def write_bundle(self, repository, target, base, fileobj):
 
132
        return self._write_bundle(repository, target, base, fileobj)
 
133
 
122
134
    def _write_main_header(self):
123
135
        """Write the header for the changes"""
124
136
        f = self.to_file
125
 
        f.write(BUNDLE_HEADER)
126
 
        f.write('0.7\n')
 
137
        f.write(_get_bundle_header('0.8'))
127
138
        f.write('#\n')
128
139
 
129
 
    def _write(self, key, value, indent=1):
130
 
        """Write out meta information, with proper indenting, etc"""
131
 
        assert indent > 0, 'indentation must be greater than 0'
 
140
    def _write(self, key, value, indent=1, trailing_space_when_empty=False):
 
141
        """Write out meta information, with proper indenting, etc.
 
142
 
 
143
        :param trailing_space_when_empty: To work around a bug in earlier
 
144
            bundle readers, when writing an empty property, we use "prop: \n"
 
145
            rather than writing "prop:\n".
 
146
            If this parameter is True, and value is the empty string, we will
 
147
            write an extra space.
 
148
        """
 
149
        if indent < 1:
 
150
            raise ValueError('indentation must be greater than 0')
132
151
        f = self.to_file
133
152
        f.write('#' + (' ' * indent))
134
153
        f.write(key.encode('utf-8'))
135
154
        if not value:
136
 
            f.write(':\n')
137
 
        elif isinstance(value, basestring):
 
155
            if trailing_space_when_empty and value == '':
 
156
                f.write(': \n')
 
157
            else:
 
158
                f.write(':\n')
 
159
        elif isinstance(value, str):
 
160
            f.write(': ')
 
161
            f.write(value)
 
162
            f.write('\n')
 
163
        elif isinstance(value, unicode):
138
164
            f.write(': ')
139
165
            f.write(value.encode('utf-8'))
140
166
            f.write('\n')
142
168
            f.write(':\n')
143
169
            for entry in value:
144
170
                f.write('#' + (' ' * (indent+2)))
145
 
                f.write(entry.encode('utf-8'))
 
171
                if isinstance(entry, str):
 
172
                    f.write(entry)
 
173
                else:
 
174
                    f.write(entry.encode('utf-8'))
146
175
                f.write('\n')
147
176
 
148
177
    def _write_revisions(self, pb):
152
181
        last_rev_id = None
153
182
        last_rev_tree = None
154
183
 
155
 
        i_max = len(self.revision_ids) 
 
184
        i_max = len(self.revision_ids)
156
185
        for i, rev_id in enumerate(self.revision_ids):
157
 
            pb.update("Generating revsion data", i, i_max)
 
186
            pb.update("Generating revision data", i, i_max)
158
187
            rev = self.source.get_revision(rev_id)
159
188
            if rev_id == last_rev_id:
160
189
                rev_tree = last_rev_tree
161
190
            else:
162
 
                base_tree = self.source.revision_tree(rev_id)
163
 
            rev_tree = self.source.revision_tree(rev_id)
 
191
                rev_tree = self.source.revision_tree(rev_id)
164
192
            if rev_id in self.forced_bases:
165
193
                explicit_base = True
166
194
                base_id = self.forced_bases[rev_id]
177
205
                base_tree = last_rev_tree
178
206
            else:
179
207
                base_tree = self.source.revision_tree(base_id)
180
 
 
181
 
            self._write_revision(rev, rev_tree, base_id, base_tree, 
182
 
                                 explicit_base)
 
208
            force_binary = (i != 0)
 
209
            self._write_revision(rev, rev_tree, base_id, base_tree,
 
210
                                 explicit_base, force_binary)
183
211
 
184
212
            last_rev_id = base_id
185
213
            last_rev_tree = base_tree
186
214
 
187
 
    def _write_revision(self, rev, rev_tree, base_rev, base_tree, 
188
 
                        explicit_base):
 
215
    def _testament_sha1(self, revision_id):
 
216
        return StrictTestament.from_revision(self.source,
 
217
                                             revision_id).as_sha1()
 
218
 
 
219
    def _write_revision(self, rev, rev_tree, base_rev, base_tree,
 
220
                        explicit_base, force_binary):
189
221
        """Write out the information for a revision."""
190
222
        def w(key, value):
191
223
            self._write(key, value, indent=1)
195
227
        w('date', format_highres_date(rev.timestamp, rev.timezone))
196
228
        self.to_file.write('\n')
197
229
 
198
 
        self._write_delta(rev_tree, base_tree, rev.revision_id)
 
230
        self._write_delta(rev_tree, base_tree, rev.revision_id, force_binary)
199
231
 
200
232
        w('revision id', rev.revision_id)
201
 
        w('sha1', StrictTestament.from_revision(self.source, 
202
 
                                                rev.revision_id).as_sha1())
 
233
        w('sha1', self._testament_sha1(rev.revision_id))
203
234
        w('inventory sha1', rev.inventory_sha1)
204
235
        if rev.parent_ids:
205
236
            w('parent ids', rev.parent_ids)
207
238
            w('base id', base_rev)
208
239
        if rev.properties:
209
240
            self._write('properties', None, indent=1)
210
 
            for name, value in rev.properties.items():
211
 
                self._write(name, value, indent=3)
212
 
        
 
241
            for name, value in sorted(rev.properties.items()):
 
242
                self._write(name, value, indent=3,
 
243
                            trailing_space_when_empty=True)
 
244
 
213
245
        # Add an extra blank space at the end
214
246
        self.to_file.write('\n')
215
247
 
222
254
        self.to_file.write(' // '.join(p_texts).encode('utf-8'))
223
255
        self.to_file.write('\n')
224
256
 
225
 
    def _write_delta(self, new_tree, old_tree, default_revision_id):
 
257
    def _write_delta(self, new_tree, old_tree, default_revision_id,
 
258
                     force_binary):
226
259
        """Write out the changes between the trees."""
227
260
        DEVNULL = '/dev/null'
228
261
        old_label = ''
229
262
        new_label = ''
230
263
 
231
 
        def do_diff(file_id, old_path, new_path, action):
 
264
        def do_diff(file_id, old_path, new_path, action, force_binary):
232
265
            def tree_lines(tree, require_text=False):
233
 
                if file_id in tree:
 
266
                if tree.has_id(file_id):
234
267
                    tree_file = tree.get_file(file_id)
235
268
                    if require_text is True:
236
269
                        tree_file = text_file(tree_file)
239
272
                    return []
240
273
 
241
274
            try:
 
275
                if force_binary:
 
276
                    raise errors.BinaryFile()
242
277
                old_lines = tree_lines(old_tree, require_text=True)
243
278
                new_lines = tree_lines(new_tree, require_text=True)
244
279
                action.write(self.to_file)
245
 
                internal_diff(old_path, old_lines, new_path, new_lines, 
 
280
                internal_diff(old_path, old_lines, new_path, new_lines,
246
281
                              self.to_file)
247
282
            except errors.BinaryFile:
248
283
                old_lines = tree_lines(old_tree, require_text=False)
249
284
                new_lines = tree_lines(new_tree, require_text=False)
250
285
                action.add_property('encoding', 'base64')
251
286
                action.write(self.to_file)
252
 
                binary_diff(old_path, old_lines, new_path, new_lines, 
 
287
                binary_diff(old_path, old_lines, new_path, new_lines,
253
288
                            self.to_file)
254
289
 
255
290
        def finish_action(action, file_id, kind, meta_modified, text_modified,
256
291
                          old_path, new_path):
257
292
            entry = new_tree.inventory[file_id]
258
293
            if entry.revision != default_revision_id:
259
 
                action.add_property('last-changed', entry.revision)
 
294
                action.add_utf8_property('last-changed', entry.revision)
260
295
            if meta_modified:
261
296
                action.add_bool_property('executable', entry.executable)
262
297
            if text_modified and kind == "symlink":
263
298
                action.add_property('target', entry.symlink_target)
264
299
            if text_modified and kind == "file":
265
 
                do_diff(file_id, old_path, new_path, action)
 
300
                do_diff(file_id, old_path, new_path, action, force_binary)
266
301
            else:
267
302
                action.write(self.to_file)
268
303
 
269
 
        delta = compare_trees(old_tree, new_tree, want_unchanged=True)
 
304
        delta = new_tree.changes_from(old_tree, want_unchanged=True,
 
305
                                      include_root=True)
270
306
        for path, file_id, kind in delta.removed:
271
307
            action = Action('removed', [kind, path]).write(self.to_file)
272
308
 
273
309
        for path, file_id, kind in delta.added:
274
310
            action = Action('added', [kind, path], [('file-id', file_id)])
275
 
            meta_modified = (kind=='file' and 
 
311
            meta_modified = (kind=='file' and
276
312
                             new_tree.is_executable(file_id))
277
313
            finish_action(action, file_id, kind, meta_modified, True,
278
314
                          DEVNULL, path)
296
332
                continue
297
333
            old_rev = getattr(old_tree.inventory[ie.file_id], 'revision', None)
298
334
            if new_rev != old_rev:
299
 
                action = Action('modified', [ie.kind, 
 
335
                action = Action('modified', [ie.kind,
300
336
                                             new_tree.id2path(ie.file_id)])
301
 
                action.add_property('last-changed', ie.revision)
 
337
                action.add_utf8_property('last-changed', ie.revision)
302
338
                action.write(self.to_file)
 
339
 
 
340
 
 
341
class BundleReader(object):
 
342
    """This class reads in a bundle from a file, and returns
 
343
    a Bundle object, which can then be applied against a tree.
 
344
    """
 
345
    def __init__(self, from_file):
 
346
        """Read in the bundle from the file.
 
347
 
 
348
        :param from_file: A file-like object (must have iterator support).
 
349
        """
 
350
        object.__init__(self)
 
351
        self.from_file = iter(from_file)
 
352
        self._next_line = None
 
353
 
 
354
        self.info = self._get_info()
 
355
        # We put the actual inventory ids in the footer, so that the patch
 
356
        # is easier to read for humans.
 
357
        # Unfortunately, that means we need to read everything before we
 
358
        # can create a proper bundle.
 
359
        self._read()
 
360
        self._validate()
 
361
 
 
362
    def _get_info(self):
 
363
        return BundleInfo08()
 
364
 
 
365
    def _read(self):
 
366
        self._next().next()
 
367
        while self._next_line is not None:
 
368
            if not self._read_revision_header():
 
369
                break
 
370
            if self._next_line is None:
 
371
                break
 
372
            self._read_patches()
 
373
            self._read_footer()
 
374
 
 
375
    def _validate(self):
 
376
        """Make sure that the information read in makes sense
 
377
        and passes appropriate checksums.
 
378
        """
 
379
        # Fill in all the missing blanks for the revisions
 
380
        # and generate the real_revisions list.
 
381
        self.info.complete_info()
 
382
 
 
383
    def _next(self):
 
384
        """yield the next line, but secretly
 
385
        keep 1 extra line for peeking.
 
386
        """
 
387
        for line in self.from_file:
 
388
            last = self._next_line
 
389
            self._next_line = line
 
390
            if last is not None:
 
391
                #mutter('yielding line: %r' % last)
 
392
                yield last
 
393
        last = self._next_line
 
394
        self._next_line = None
 
395
        #mutter('yielding line: %r' % last)
 
396
        yield last
 
397
 
 
398
    def _read_revision_header(self):
 
399
        found_something = False
 
400
        self.info.revisions.append(RevisionInfo(None))
 
401
        for line in self._next():
 
402
            # The bzr header is terminated with a blank line
 
403
            # which does not start with '#'
 
404
            if line is None or line == '\n':
 
405
                break
 
406
            if not line.startswith('#'):
 
407
                continue
 
408
            found_something = True
 
409
            self._handle_next(line)
 
410
        if not found_something:
 
411
            # Nothing was there, so remove the added revision
 
412
            self.info.revisions.pop()
 
413
        return found_something
 
414
 
 
415
    def _read_next_entry(self, line, indent=1):
 
416
        """Read in a key-value pair
 
417
        """
 
418
        if not line.startswith('#'):
 
419
            raise errors.MalformedHeader('Bzr header did not start with #')
 
420
        line = line[1:-1].decode('utf-8') # Remove the '#' and '\n'
 
421
        if line[:indent] == ' '*indent:
 
422
            line = line[indent:]
 
423
        if not line:
 
424
            return None, None# Ignore blank lines
 
425
 
 
426
        loc = line.find(': ')
 
427
        if loc != -1:
 
428
            key = line[:loc]
 
429
            value = line[loc+2:]
 
430
            if not value:
 
431
                value = self._read_many(indent=indent+2)
 
432
        elif line[-1:] == ':':
 
433
            key = line[:-1]
 
434
            value = self._read_many(indent=indent+2)
 
435
        else:
 
436
            raise errors.MalformedHeader('While looking for key: value pairs,'
 
437
                    ' did not find the colon %r' % (line))
 
438
 
 
439
        key = key.replace(' ', '_')
 
440
        #mutter('found %s: %s' % (key, value))
 
441
        return key, value
 
442
 
 
443
    def _handle_next(self, line):
 
444
        if line is None:
 
445
            return
 
446
        key, value = self._read_next_entry(line, indent=1)
 
447
        mutter('_handle_next %r => %r' % (key, value))
 
448
        if key is None:
 
449
            return
 
450
 
 
451
        revision_info = self.info.revisions[-1]
 
452
        if key in revision_info.__dict__:
 
453
            if getattr(revision_info, key) is None:
 
454
                if key in ('file_id', 'revision_id', 'base_id'):
 
455
                    value = value.encode('utf8')
 
456
                elif key in ('parent_ids'):
 
457
                    value = [v.encode('utf8') for v in value]
 
458
                setattr(revision_info, key, value)
 
459
            else:
 
460
                raise errors.MalformedHeader('Duplicated Key: %s' % key)
 
461
        else:
 
462
            # What do we do with a key we don't recognize
 
463
            raise errors.MalformedHeader('Unknown Key: "%s"' % key)
 
464
 
 
465
    def _read_many(self, indent):
 
466
        """If a line ends with no entry, that means that it should be
 
467
        followed with multiple lines of values.
 
468
 
 
469
        This detects the end of the list, because it will be a line that
 
470
        does not start properly indented.
 
471
        """
 
472
        values = []
 
473
        start = '#' + (' '*indent)
 
474
 
 
475
        if self._next_line is None or self._next_line[:len(start)] != start:
 
476
            return values
 
477
 
 
478
        for line in self._next():
 
479
            values.append(line[len(start):-1].decode('utf-8'))
 
480
            if self._next_line is None or self._next_line[:len(start)] != start:
 
481
                break
 
482
        return values
 
483
 
 
484
    def _read_one_patch(self):
 
485
        """Read in one patch, return the complete patch, along with
 
486
        the next line.
 
487
 
 
488
        :return: action, lines, do_continue
 
489
        """
 
490
        #mutter('_read_one_patch: %r' % self._next_line)
 
491
        # Peek and see if there are no patches
 
492
        if self._next_line is None or self._next_line.startswith('#'):
 
493
            return None, [], False
 
494
 
 
495
        first = True
 
496
        lines = []
 
497
        for line in self._next():
 
498
            if first:
 
499
                if not line.startswith('==='):
 
500
                    raise errors.MalformedPatches('The first line of all patches'
 
501
                        ' should be a bzr meta line "==="'
 
502
                        ': %r' % line)
 
503
                action = line[4:-1].decode('utf-8')
 
504
            elif line.startswith('... '):
 
505
                action += line[len('... '):-1].decode('utf-8')
 
506
 
 
507
            if (self._next_line is not None and
 
508
                self._next_line.startswith('===')):
 
509
                return action, lines, True
 
510
            elif self._next_line is None or self._next_line.startswith('#'):
 
511
                return action, lines, False
 
512
 
 
513
            if first:
 
514
                first = False
 
515
            elif not line.startswith('... '):
 
516
                lines.append(line)
 
517
 
 
518
        return action, lines, False
 
519
 
 
520
    def _read_patches(self):
 
521
        do_continue = True
 
522
        revision_actions = []
 
523
        while do_continue:
 
524
            action, lines, do_continue = self._read_one_patch()
 
525
            if action is not None:
 
526
                revision_actions.append((action, lines))
 
527
        if self.info.revisions[-1].tree_actions is not None:
 
528
            raise AssertionError()
 
529
        self.info.revisions[-1].tree_actions = revision_actions
 
530
 
 
531
    def _read_footer(self):
 
532
        """Read the rest of the meta information.
 
533
 
 
534
        :param first_line:  The previous step iterates past what it
 
535
                            can handle. That extra line is given here.
 
536
        """
 
537
        for line in self._next():
 
538
            self._handle_next(line)
 
539
            if self._next_line is None:
 
540
                break
 
541
            if not self._next_line.startswith('#'):
 
542
                # Consume the trailing \n and stop processing
 
543
                self._next().next()
 
544
                break
 
545
 
 
546
class BundleInfo08(BundleInfo):
 
547
 
 
548
    def _update_tree(self, bundle_tree, revision_id):
 
549
        bundle_tree.note_last_changed('', revision_id)
 
550
        BundleInfo._update_tree(self, bundle_tree, revision_id)
 
551
 
 
552
    def _testament_sha1_from_revision(self, repository, revision_id):
 
553
        testament = StrictTestament.from_revision(repository, revision_id)
 
554
        return testament.as_sha1()
 
555
 
 
556
    def _testament_sha1(self, revision, tree):
 
557
        return StrictTestament(revision, tree).as_sha1()