58
72
from binascii import hexlify
59
73
from cStringIO import StringIO
61
from bzrlib.osutils import (local_time_offset, username,
62
rand_bytes, compact_date, user_email,
75
from bzrlib.atomicfile import AtomicFile
76
from bzrlib.osutils import (local_time_offset,
77
rand_bytes, compact_date,
63
78
kind_marker, is_inside_any, quotefn,
64
sha_string, sha_strings, sha_file, isdir, isfile,
79
sha_file, isdir, isfile,
66
from bzrlib.branch import gen_file_id, INVENTORY_FILEID, ANCESTRY_FILEID
82
import bzrlib.errors as errors
67
83
from bzrlib.errors import (BzrError, PointlessCommit,
70
from bzrlib.revision import Revision, RevisionReference
88
from bzrlib.revision import Revision
89
from bzrlib.testament import Testament
71
90
from bzrlib.trace import mutter, note, warning
72
91
from bzrlib.xml5 import serializer_v5
73
from bzrlib.inventory import Inventory
74
from bzrlib.weave import Weave
75
from bzrlib.weavefile import read_weave, write_weave_v5
76
from bzrlib.atomicfile import AtomicFile
92
from bzrlib.inventory import Inventory, ROOT_ID, InventoryEntry
93
from bzrlib.symbol_versioning import *
94
from bzrlib.workingtree import WorkingTree
97
@deprecated_function(zero_seven)
79
98
def commit(*args, **kwargs):
80
99
"""Commit a new revision to a branch.
90
109
class NullCommitReporter(object):
91
110
"""I report on progress of a commit."""
92
def added(self, path):
95
def removed(self, path):
98
def renamed(self, old_path, new_path):
112
def snapshot_change(self, change, path):
115
def completed(self, revno, rev_id):
118
def deleted(self, file_id):
121
def escaped(self, escape_count, message):
124
def missing(self, path):
127
def renamed(self, change, old_path, new_path):
102
131
class ReportCommitToLog(NullCommitReporter):
103
def added(self, path):
104
note('added %s', path)
106
def removed(self, path):
107
note('removed %s', path)
109
def renamed(self, old_path, new_path):
110
note('renamed %s => %s', old_path, new_path)
133
# this may be more useful if 'note' was replaced by an overridable
134
# method on self, which would allow more trivial subclassing.
135
# alternative, a callable could be passed in, allowing really trivial
136
# reuse for some uis. RBC 20060511
138
def snapshot_change(self, change, path):
139
if change == 'unchanged':
141
note("%s %s", change, path)
143
def completed(self, revno, rev_id):
144
note('Committed revision %d.', revno)
146
def deleted(self, file_id):
147
note('deleted %s', file_id)
149
def escaped(self, escape_count, message):
150
note("replaced %d control characters in message", escape_count)
152
def missing(self, path):
153
note('missing %s', path)
155
def renamed(self, change, old_path, new_path):
156
note('%s %s => %s', change, old_path, new_path)
113
159
class Commit(object):
155
216
allow_pointless -- If true (default), commit even if nothing
156
217
has changed and no merges are recorded.
219
strict -- If true, don't allow a commit if the working tree
220
contains unknown files.
222
revprops -- Properties for new revision
223
:param local: Perform a local only commit.
158
225
mutter('preparing to commit')
161
self.weave_store = branch.weave_store
227
if deprecated_passed(branch):
228
warn("Commit.commit (branch, ...): The branch parameter is "
229
"deprecated as of bzr 0.8. Please use working_tree= instead.",
230
DeprecationWarning, stacklevel=2)
232
self.work_tree = self.branch.bzrdir.open_workingtree()
233
elif working_tree is None:
234
raise BzrError("One of branch and working_tree must be passed into commit().")
236
self.work_tree = working_tree
237
self.branch = self.work_tree.branch
239
raise BzrError("The message keyword parameter is required for commit().")
241
self.weave_store = self.branch.repository.weave_store
242
self.bound_branch = None
244
self.master_branch = None
245
self.master_locked = False
162
246
self.rev_id = rev_id
163
247
self.specific_files = specific_files
164
248
self.allow_pointless = allow_pointless
166
if timestamp is None:
167
self.timestamp = time.time()
169
self.timestamp = long(timestamp)
172
self.rev_id = _gen_revision_id(self.branch, self.timestamp)
176
if committer is None:
177
self.committer = username(self.branch)
179
assert isinstance(committer, basestring), type(committer)
180
self.committer = committer
183
self.timezone = local_time_offset()
185
self.timezone = int(timezone)
187
assert isinstance(message, basestring), type(message)
188
self.message = message
190
self.branch.lock_write()
250
if revprops is not None:
251
self.revprops.update(revprops)
253
if reporter is None and self.reporter is None:
254
self.reporter = NullCommitReporter()
255
elif reporter is not None:
256
self.reporter = reporter
258
self.work_tree.lock_write()
259
self.pb = bzrlib.ui.ui_factory.nested_progress_bar()
192
self.work_tree = self.branch.working_tree()
261
# Cannot commit with conflicts present.
262
if len(self.work_tree.conflicts())>0:
263
raise ConflictsInTree
265
# setup the bound branch variables as needed.
266
self._check_bound_branch()
268
# check for out of date working trees
269
# if we are bound, then self.branch is the master branch and this
270
# test is thus all we need.
271
if self.work_tree.last_revision() != self.master_branch.last_revision():
272
raise errors.OutOfDateTree(self.work_tree)
275
# raise an exception as soon as we find a single unknown.
276
for unknown in self.work_tree.unknowns():
277
raise StrictCommitFailed()
279
if timestamp is None:
280
self.timestamp = time.time()
282
self.timestamp = long(timestamp)
284
if self.config is None:
285
self.config = bzrlib.config.BranchConfig(self.branch)
288
self.rev_id = _gen_revision_id(self.config, self.timestamp)
292
if committer is None:
293
self.committer = self.config.username()
295
assert isinstance(committer, basestring), type(committer)
296
self.committer = committer
299
self.timezone = local_time_offset()
301
self.timezone = int(timezone)
303
if isinstance(message, str):
304
message = message.decode(bzrlib.user_encoding)
305
assert isinstance(message, unicode), type(message)
306
self.message = message
307
self._escape_commit_message()
193
309
self.work_inv = self.work_tree.inventory
194
self.basis_tree = self.branch.basis_tree()
310
self.basis_tree = self.work_tree.basis_tree()
195
311
self.basis_inv = self.basis_tree.inventory
312
# one to finish, one for rev and inventory, and one for each
313
# inventory entry, and the same for the new inventory.
314
# note that this estimate is too long when we do a partial tree
315
# commit which excludes some new files from being considered.
316
# The estimate is corrected when we populate the new inv.
317
self.pb_total = len(self.basis_inv) + len(self.work_inv) + 3 - 1
197
320
self._gather_parents()
321
if len(self.parents) > 1 and self.specific_files:
322
raise NotImplementedError('selected-file commit of merges is not supported yet: files %r',
198
324
self._check_parents_present()
200
326
self._remove_deleted()
201
self.new_inv = Inventory()
327
self._populate_new_inv()
328
self._store_snapshot()
203
329
self._report_deletes()
205
331
if not (self.allow_pointless
207
333
or self.new_inv != self.basis_inv):
208
334
raise PointlessCommit()
210
self._record_inventory()
211
self._record_ancestry()
336
self._emit_progress_update()
337
self.inv_sha1 = self.branch.repository.add_inventory(
342
self._emit_progress_update()
212
343
self._make_revision()
213
note('committed r%d {%s}', (self.branch.revno() + 1),
344
# revision data is in the local branch now.
346
# upload revision data to the master.
347
# this will propogate merged revisions too if needed.
348
if self.bound_branch:
349
self.master_branch.repository.fetch(self.branch.repository,
350
revision_id=self.rev_id)
351
# now the master has the revision data
352
# 'commit' to the master first so a timeout here causes the local
353
# branch to be out of date
354
self.master_branch.append_revision(self.rev_id)
356
# and now do the commit locally.
215
357
self.branch.append_revision(self.rev_id)
216
self.branch.set_pending_merges([])
359
self.work_tree.set_pending_merges([])
360
self.work_tree.set_last_revision(self.rev_id)
361
# now the work tree is up to date with the branch
363
self.reporter.completed(self.branch.revno(), self.rev_id)
364
if self.config.post_commit() is not None:
365
hooks = self.config.post_commit().split(' ')
366
# this would be nicer with twisted.python.reflect.namedAny
368
result = eval(hook + '(branch, rev_id)',
369
{'branch':self.branch,
371
'rev_id':self.rev_id})
372
self._emit_progress_update()
222
def _record_inventory(self):
223
"""Store the inventory for the new revision."""
224
inv_text = serializer_v5.write_inventory_to_string(self.new_inv)
225
self.inv_sha1 = sha_string(inv_text)
226
self.weave_store.add_text(INVENTORY_FILEID, self.rev_id,
227
split_lines(inv_text), self.parents)
230
def _record_ancestry(self):
231
"""Append merged revision ancestry to the ancestry file.
233
This should be the merged ancestry of all parents, plus the
235
w = self.weave_store.get_weave_or_empty(ANCESTRY_FILEID)
236
lines = self._merge_ancestry_lines(w)
237
w.add(self.rev_id, self.parents, lines)
238
self.weave_store.put_weave(ANCESTRY_FILEID, w)
241
def _merge_ancestry_lines(self, ancestry_weave):
242
"""Return merged ancestry lines.
244
The lines are revision-ids followed by newlines."""
247
for parent_id in self.parents:
248
for line in ancestry_weave.get(parent_id):
249
assert line[-1] == '\n'
253
r = self.rev_id + '\n'
256
mutter('merged ancestry of {%s}:\n%s', self.rev_id, ''.join(ancs))
376
def _check_bound_branch(self):
377
"""Check to see if the local branch is bound.
379
If it is bound, then most of the commit will actually be
380
done using the remote branch as the target branch.
381
Only at the end will the local branch be updated.
383
if self.local and not self.branch.get_bound_location():
384
raise errors.LocalRequiresBoundBranch()
387
self.master_branch = self.branch.get_master_branch()
389
if not self.master_branch:
390
# make this branch the reference branch for out of date checks.
391
self.master_branch = self.branch
394
# If the master branch is bound, we must fail
395
master_bound_location = self.master_branch.get_bound_location()
396
if master_bound_location:
397
raise errors.CommitToDoubleBoundBranch(self.branch,
398
self.master_branch, master_bound_location)
400
# TODO: jam 20051230 We could automatically push local
401
# commits to the remote branch if they would fit.
402
# But for now, just require remote to be identical
405
# Make sure the local branch is identical to the master
406
master_rh = self.master_branch.revision_history()
407
local_rh = self.branch.revision_history()
408
if local_rh != master_rh:
409
raise errors.BoundBranchOutOfDate(self.branch,
412
# Now things are ready to change the master branch
414
self.bound_branch = self.branch
415
self.master_branch.lock_write()
416
self.master_locked = True
418
#### # Check to see if we have any pending merges. If we do
419
#### # those need to be pushed into the master branch
420
#### pending_merges = self.work_tree.pending_merges()
421
#### if pending_merges:
422
#### for revision_id in pending_merges:
423
#### self.master_branch.repository.fetch(self.bound_branch.repository,
424
#### revision_id=revision_id)
427
"""Cleanup any open locks, progress bars etc."""
428
cleanups = [self._cleanup_bound_branch,
429
self.work_tree.unlock,
431
found_exception = None
432
for cleanup in cleanups:
435
# we want every cleanup to run no matter what.
436
# so we have a catchall here, but we will raise the
437
# last encountered exception up the stack: and
438
# typically this will be useful enough.
441
if found_exception is not None:
442
# dont do a plan raise, because the last exception may have been
443
# trashed, e is our sure-to-work exception even though it loses the
444
# full traceback. XXX: RBC 20060421 perhaps we could check the
445
# exc_info and if its the same one do a plain raise otherwise
446
# 'raise e' as we do now.
449
def _cleanup_bound_branch(self):
450
"""Executed at the end of a try/finally to cleanup a bound branch.
452
If the branch wasn't bound, this is a no-op.
453
If it was, it resents self.branch to the local branch, instead
456
if not self.bound_branch:
458
if self.master_locked:
459
self.master_branch.unlock()
461
def _escape_commit_message(self):
462
"""Replace xml-incompatible control characters."""
463
# FIXME: RBC 20060419 this should be done by the revision
464
# serialiser not by commit. Then we can also add an unescaper
465
# in the deserializer and start roundtripping revision messages
466
# precisely. See repository_implementations/test_repository.py
468
# Python strings can include characters that can't be
469
# represented in well-formed XML; escape characters that
470
# aren't listed in the XML specification
471
# (http://www.w3.org/TR/REC-xml/#NT-Char).
472
self.message, escape_count = re.subn(
473
u'[^\x09\x0A\x0D\u0020-\uD7FF\uE000-\uFFFD]+',
474
lambda match: match.group(0).encode('unicode_escape'),
477
self.reporter.escaped(escape_count, self.message)
260
479
def _gather_parents(self):
261
pending_merges = self.branch.pending_merges()
480
"""Record the parents of a merge for merge detection."""
481
pending_merges = self.work_tree.pending_merges()
262
482
self.parents = []
263
self.parent_trees = []
483
self.parent_invs = []
484
self.present_parents = []
264
485
precursor_id = self.branch.last_revision()
266
487
self.parents.append(precursor_id)
267
self.parent_trees.append(self.basis_tree)
268
488
self.parents += pending_merges
269
self.parent_trees.extend(map(self.branch.revision_tree, pending_merges))
489
for revision in self.parents:
490
if self.branch.repository.has_revision(revision):
491
inventory = self.branch.repository.get_inventory(revision)
492
self.parent_invs.append(inventory)
493
self.present_parents.append(revision)
272
495
def _check_parents_present(self):
273
496
for parent_id in self.parents:
274
497
mutter('commit parent revision {%s}', parent_id)
275
if not self.branch.has_revision(parent_id):
276
warning("can't commit a merge from an absent parent")
277
raise HistoryMissing(self.branch, 'revision', parent_id)
498
if not self.branch.repository.has_revision(parent_id):
499
if parent_id == self.branch.last_revision():
500
warning("parent is missing %r", parent_id)
501
raise HistoryMissing(self.branch, 'revision', parent_id)
503
mutter("commit will ghost revision %r", parent_id)
280
505
def _make_revision(self):
281
506
"""Record a new revision object for this commit."""
282
self.rev = Revision(timestamp=self.timestamp,
283
timezone=self.timezone,
284
committer=self.committer,
285
message=self.message,
286
inventory_sha1=self.inv_sha1,
287
revision_id=self.rev_id)
288
self.rev.parents = map(RevisionReference, self.parents)
290
serializer_v5.write_revision(self.rev, rev_tmp)
292
self.branch.revision_store.add(rev_tmp, self.rev_id)
293
mutter('new revision_id is {%s}', self.rev_id)
507
rev = Revision(timestamp=self.timestamp,
508
timezone=self.timezone,
509
committer=self.committer,
510
message=self.message,
511
inventory_sha1=self.inv_sha1,
512
revision_id=self.rev_id,
513
properties=self.revprops)
514
rev.parent_ids = self.parents
515
self.branch.repository.add_revision(self.rev_id, rev, self.new_inv, self.config)
296
517
def _remove_deleted(self):
297
518
"""Remove deleted files from the working inventories.
310
531
if specific and not is_inside_any(specific, path):
312
533
if not self.work_tree.has_filename(path):
313
note('missing %s', path)
314
deleted_ids.append(ie.file_id)
534
self.reporter.missing(path)
535
deleted_ids.append((path, ie.file_id))
316
for file_id in deleted_ids:
537
deleted_ids.sort(reverse=True)
538
for path, file_id in deleted_ids:
317
539
del self.work_inv[file_id]
318
self.branch._write_inventory(self.work_inv)
321
def _find_file_parents(self, file_id):
322
"""Return the text versions and hashes for all file parents.
324
Returned as a map from text version to inventory entry.
326
This is a set containing the file versions in all parents
327
revisions containing the file. If the file is new, the set
330
for tree in self.parent_trees:
331
if file_id in tree.inventory:
332
ie = tree.inventory[file_id]
333
assert ie.kind == 'file'
334
assert ie.file_id == file_id
335
if ie.text_version in r:
336
assert r[ie.text_version] == ie
338
r[ie.text_version] = ie
342
def _store_files(self):
343
"""Store new texts of modified/added files.
345
This is called with new_inv set to a copy of the working
346
inventory, with deleted/removed files already cut out. So
347
this code only needs to deal with setting text versions, and
348
possibly recording new file texts."""
540
self.work_tree._write_inventory(self.work_inv)
542
def _store_snapshot(self):
543
"""Pass over inventory and record a snapshot.
545
Entries get a new revision when they are modified in
546
any way, which includes a merge with a new set of
547
parents that have the same entry.
549
# XXX: Need to think more here about when the user has
550
# made a specific decision on a particular value -- c.f.
553
# iter_entries does not visit the ROOT_ID node so we need to call
554
# self._emit_progress_update once by hand.
555
self._emit_progress_update()
556
for path, ie in self.new_inv.iter_entries():
557
self._emit_progress_update()
558
previous_entries = ie.find_previous_heads(
561
self.branch.repository.get_transaction())
562
if ie.revision is None:
563
# we are creating a new revision for ie in the history store
565
ie.snapshot(self.rev_id, path, previous_entries,
566
self.work_tree, self.weave_store,
567
self.branch.repository.get_transaction())
568
# describe the nature of the change that has occured relative to
569
# the basis inventory.
570
if (self.basis_inv.has_id(ie.file_id)):
571
basis_ie = self.basis_inv[ie.file_id]
574
change = ie.describe_change(basis_ie, ie)
575
if change in (InventoryEntry.RENAMED,
576
InventoryEntry.MODIFIED_AND_RENAMED):
577
old_path = self.basis_inv.id2path(ie.file_id)
578
self.reporter.renamed(change, old_path, path)
580
self.reporter.snapshot_change(change, path)
582
def _populate_new_inv(self):
583
"""Build revision inventory.
585
This creates a new empty inventory. Depending on
586
which files are selected for commit, and what is present in the
587
current tree, the new inventory is populated. inventory entries
588
which are candidates for modification have their revision set to
589
None; inventory entries that are carried over untouched have their
590
revision set to their prior value.
592
mutter("Selecting files for commit with filter %s", self.specific_files)
593
self.new_inv = Inventory(revision_id=self.rev_id)
594
# iter_entries does not visit the ROOT_ID node so we need to call
595
# self._emit_progress_update once by hand.
596
self._emit_progress_update()
349
597
for path, new_ie in self.work_inv.iter_entries():
598
self._emit_progress_update()
350
599
file_id = new_ie.file_id
351
600
mutter('check %s {%s}', path, new_ie.file_id)
352
601
if self.specific_files:
353
602
if not is_inside_any(self.specific_files, path):
354
603
mutter('%s not selected for commit', path)
355
self._carry_file(file_id)
357
if new_ie.kind != 'file':
358
self._commit_nonfile(file_id)
361
file_parents = self._find_file_parents(file_id)
362
if len(file_parents) == 1:
363
parent_ie = file_parents.values()[0]
364
wc_sha1 = self.work_tree.get_file_sha1(file_id)
365
if parent_ie.text_sha1 == wc_sha1:
366
# text not changed or merged
367
self._commit_old_text(file_id, parent_ie)
370
mutter('parents of %s are %r', path, file_parents)
372
# file is either new, or a file merge; need to record
374
if len(file_parents) > 1:
375
note('merged %s', path)
376
elif len(file_parents) == 0:
377
note('added %s', path)
379
note('modified %s', path)
380
self._commit_file(new_ie, file_id, file_parents)
383
def _commit_nonfile(self, file_id):
384
self.new_inv.add(self.work_inv[file_id].copy())
387
def _carry_file(self, file_id):
604
self._carry_entry(file_id)
607
# this is selected, ensure its parents are too.
608
parent_id = new_ie.parent_id
609
while parent_id != ROOT_ID:
610
if not self.new_inv.has_id(parent_id):
611
ie = self._select_entry(self.work_inv[parent_id])
612
mutter('%s selected for commit because of %s',
613
self.new_inv.id2path(parent_id), path)
615
ie = self.new_inv[parent_id]
616
if ie.revision is not None:
618
mutter('%s selected for commit because of %s',
619
self.new_inv.id2path(parent_id), path)
620
parent_id = ie.parent_id
621
mutter('%s selected for commit', path)
622
self._select_entry(new_ie)
624
def _emit_progress_update(self):
625
"""Emit an update to the progress bar."""
626
self.pb.update("Committing", self.pb_count, self.pb_total)
629
def _select_entry(self, new_ie):
630
"""Make new_ie be considered for committing."""
636
def _carry_entry(self, file_id):
388
637
"""Carry the file unchanged from the basis revision."""
389
638
if self.basis_inv.has_id(file_id):
390
639
self.new_inv.add(self.basis_inv[file_id].copy())
393
def _commit_old_text(self, file_id, parent_ie):
394
"""Keep the same text as last time, but possibly a different name."""
395
ie = self.work_inv[file_id].copy()
396
ie.text_version = parent_ie.text_version
397
ie.text_size = parent_ie.text_size
398
ie.text_sha1 = parent_ie.text_sha1
641
# this entry is new and not being committed
402
644
def _report_deletes(self):
403
for file_id in self.basis_inv:
404
if file_id not in self.new_inv:
405
note('deleted %s', self.basis_inv.id2path(file_id))
408
def _commit_file(self, new_ie, file_id, file_parents):
409
mutter('store new text for {%s} in revision {%s}',
410
file_id, self.rev_id)
411
new_lines = self.work_tree.get_file(file_id).readlines()
412
self._add_text_to_weave(file_id, new_lines, file_parents)
413
new_ie.text_version = self.rev_id
414
new_ie.text_sha1 = sha_strings(new_lines)
415
new_ie.text_size = sum(map(len, new_lines))
416
self.new_inv.add(new_ie)
419
def _add_text_to_weave(self, file_id, new_lines, parents):
420
if file_id.startswith('__'):
421
raise ValueError('illegal file-id %r for text file' % file_id)
422
self.weave_store.add_text(file_id, self.rev_id, new_lines, parents)
425
def _gen_revision_id(branch, when):
645
for path, ie in self.basis_inv.iter_entries():
646
if ie.file_id not in self.new_inv:
647
self.reporter.deleted(path)
649
def _gen_revision_id(config, when):
426
650
"""Return new revision-id."""
427
s = '%s-%s-' % (user_email(branch), compact_date(when))
651
s = '%s-%s-' % (config.user_email(), compact_date(when))
428
652
s += hexlify(rand_bytes(8))
added
removed
bzrlib/commit.py
