55
61
# merges from, then it should still be reported as newly added
56
62
# relative to the basis revision.
58
# TODO: Do checks that the tree can be committed *before* running the
59
# editor; this should include checks for a pointless commit and for
60
# unknown or missing files.
62
# TODO: If commit fails, leave the message in a file somewhere.
71
from binascii import hexlify
70
72
from cStringIO import StringIO
73
import bzrlib.errors as errors
74
from bzrlib.osutils import (local_time_offset, username,
75
rand_bytes, compact_date, user_email,
76
kind_marker, is_inside_any, quotefn,
77
sha_string, sha_strings, sha_file, isdir, isfile,
79
from bzrlib.branch import gen_file_id
74
80
from bzrlib.errors import (BzrError, PointlessCommit,
78
from bzrlib.osutils import (kind_marker, isdir,isfile, is_inside_any,
79
is_inside_or_parent_of_any,
80
quotefn, sha_file, split_lines)
81
from bzrlib.testament import Testament
83
from bzrlib.revision import Revision
82
84
from bzrlib.trace import mutter, note, warning
83
85
from bzrlib.xml5 import serializer_v5
84
from bzrlib.inventory import Inventory, ROOT_ID, InventoryEntry
85
from bzrlib import symbol_versioning
86
from bzrlib.symbol_versioning import (deprecated_passed,
89
from bzrlib.workingtree import WorkingTree
86
from bzrlib.inventory import Inventory
87
from bzrlib.weave import Weave
88
from bzrlib.weavefile import read_weave, write_weave_v5
89
from bzrlib.atomicfile import AtomicFile
92
def commit(*args, **kwargs):
93
"""Commit a new revision to a branch.
95
Function-style interface for convenience of old callers.
97
New code should use the Commit class instead.
99
## XXX: Remove this in favor of Branch.commit?
100
Commit().commit(*args, **kwargs)
92
103
class NullCommitReporter(object):
93
104
"""I report on progress of a commit."""
95
def snapshot_change(self, change, path):
98
def completed(self, revno, rev_id):
101
def deleted(self, file_id):
104
def escaped(self, escape_count, message):
107
def missing(self, path):
110
def renamed(self, change, old_path, new_path):
105
def added(self, path):
108
def removed(self, path):
111
def renamed(self, old_path, new_path):
114
115
class ReportCommitToLog(NullCommitReporter):
116
# this may be more useful if 'note' was replaced by an overridable
117
# method on self, which would allow more trivial subclassing.
118
# alternative, a callable could be passed in, allowing really trivial
119
# reuse for some uis. RBC 20060511
121
def snapshot_change(self, change, path):
122
if change == 'unchanged':
124
note("%s %s", change, path)
126
def completed(self, revno, rev_id):
127
note('Committed revision %d.', revno)
129
def deleted(self, file_id):
130
note('deleted %s', file_id)
132
def escaped(self, escape_count, message):
133
note("replaced %d control characters in message", escape_count)
135
def missing(self, path):
136
note('missing %s', path)
138
def renamed(self, change, old_path, new_path):
139
note('%s %s => %s', change, old_path, new_path)
116
def added(self, path):
117
note('added %s', path)
119
def removed(self, path):
120
note('removed %s', path)
122
def renamed(self, old_path, new_path):
123
note('renamed %s => %s', old_path, new_path)
142
126
class Commit(object):
199
168
allow_pointless -- If true (default), commit even if nothing
200
169
has changed and no merges are recorded.
202
strict -- If true, don't allow a commit if the working tree
203
contains unknown files.
205
revprops -- Properties for new revision
206
:param local: Perform a local only commit.
208
171
mutter('preparing to commit')
210
if deprecated_passed(branch):
211
symbol_versioning.warn("Commit.commit (branch, ...): The branch parameter is "
212
"deprecated as of bzr 0.8. Please use working_tree= instead.",
213
DeprecationWarning, stacklevel=2)
215
self.work_tree = self.branch.bzrdir.open_workingtree()
216
elif working_tree is None:
217
raise BzrError("One of branch and working_tree must be passed into commit().")
219
self.work_tree = working_tree
220
self.branch = self.work_tree.branch
222
raise BzrError("The message keyword parameter is required for commit().")
224
self.bound_branch = None
226
self.master_branch = None
227
self.master_locked = False
174
self.weave_store = branch.weave_store
229
176
self.specific_files = specific_files
230
177
self.allow_pointless = allow_pointless
232
if reporter is None and self.reporter is None:
233
self.reporter = NullCommitReporter()
234
elif reporter is not None:
235
self.reporter = reporter
237
self.work_tree.lock_write()
238
self.pb = bzrlib.ui.ui_factory.nested_progress_bar()
179
if timestamp is None:
180
self.timestamp = time.time()
182
self.timestamp = long(timestamp)
185
self.rev_id = _gen_revision_id(self.branch, self.timestamp)
189
if committer is None:
190
self.committer = username(self.branch)
192
assert isinstance(committer, basestring), type(committer)
193
self.committer = committer
196
self.timezone = local_time_offset()
198
self.timezone = int(timezone)
200
assert isinstance(message, basestring), type(message)
201
self.message = message
202
self._escape_commit_message()
204
self.branch.lock_write()
240
# Cannot commit with conflicts present.
241
if len(self.work_tree.conflicts())>0:
242
raise ConflictsInTree
244
# setup the bound branch variables as needed.
245
self._check_bound_branch()
247
# check for out of date working trees
248
# if we are bound, then self.branch is the master branch and this
249
# test is thus all we need.
250
if self.work_tree.last_revision() != self.master_branch.last_revision():
251
raise errors.OutOfDateTree(self.work_tree)
254
# raise an exception as soon as we find a single unknown.
255
for unknown in self.work_tree.unknowns():
256
raise StrictCommitFailed()
258
if self.config is None:
259
self.config = self.branch.get_config()
261
if isinstance(message, str):
262
message = message.decode(bzrlib.user_encoding)
263
assert isinstance(message, unicode), type(message)
264
self.message = message
265
self._escape_commit_message()
206
self.work_tree = self.branch.working_tree()
267
207
self.work_inv = self.work_tree.inventory
268
self.basis_tree = self.work_tree.basis_tree()
208
self.basis_tree = self.branch.basis_tree()
269
209
self.basis_inv = self.basis_tree.inventory
270
# one to finish, one for rev and inventory, and one for each
271
# inventory entry, and the same for the new inventory.
272
# note that this estimate is too long when we do a partial tree
273
# commit which excludes some new files from being considered.
274
# The estimate is corrected when we populate the new inv.
275
self.pb_total = len(self.work_inv) + 5
278
211
self._gather_parents()
279
212
if len(self.parents) > 1 and self.specific_files:
280
raise NotImplementedError('selected-file commit of merges is not supported yet: files %r',
213
raise NotImplementedError('selected-file commit of merges is not supported yet')
282
214
self._check_parents_present()
283
self.builder = self.branch.get_commit_builder(self.parents,
284
self.config, timestamp, timezone, committer, revprops, rev_id)
286
216
self._remove_deleted()
287
217
self._populate_new_inv()
218
self._store_snapshot()
288
219
self._report_deletes()
290
221
if not (self.allow_pointless
291
222
or len(self.parents) > 1
292
or self.builder.new_inventory != self.basis_inv):
223
or self.new_inv != self.basis_inv):
293
224
raise PointlessCommit()
295
self._emit_progress_update()
296
# TODO: Now the new inventory is known, check for conflicts and
297
# prompt the user for a commit message.
298
# ADHB 2006-08-08: If this is done, populate_new_inv should not add
299
# weave lines, because nothing should be recorded until it is known
300
# that commit will succeed.
301
self.builder.finish_inventory()
302
self._emit_progress_update()
303
self.rev_id = self.builder.commit(self.message)
304
self._emit_progress_update()
305
# revision data is in the local branch now.
307
# upload revision data to the master.
308
# this will propagate merged revisions too if needed.
309
if self.bound_branch:
310
self.master_branch.repository.fetch(self.branch.repository,
311
revision_id=self.rev_id)
312
# now the master has the revision data
313
# 'commit' to the master first so a timeout here causes the local
314
# branch to be out of date
315
self.master_branch.append_revision(self.rev_id)
317
# and now do the commit locally.
226
self._record_inventory()
227
self._record_ancestry()
228
self._make_revision()
229
note('committed r%d {%s}', (self.branch.revno() + 1),
318
231
self.branch.append_revision(self.rev_id)
320
self.work_tree.set_pending_merges([])
321
self.work_tree.set_last_revision(self.rev_id)
322
# now the work tree is up to date with the branch
324
self.reporter.completed(self.branch.revno(), self.rev_id)
325
if self.config.post_commit() is not None:
326
hooks = self.config.post_commit().split(' ')
327
# this would be nicer with twisted.python.reflect.namedAny
329
result = eval(hook + '(branch, rev_id)',
330
{'branch':self.branch,
332
'rev_id':self.rev_id})
333
self._emit_progress_update()
232
self.branch.set_pending_merges([])
338
def _check_bound_branch(self):
339
"""Check to see if the local branch is bound.
341
If it is bound, then most of the commit will actually be
342
done using the remote branch as the target branch.
343
Only at the end will the local branch be updated.
345
if self.local and not self.branch.get_bound_location():
346
raise errors.LocalRequiresBoundBranch()
349
self.master_branch = self.branch.get_master_branch()
351
if not self.master_branch:
352
# make this branch the reference branch for out of date checks.
353
self.master_branch = self.branch
356
# If the master branch is bound, we must fail
357
master_bound_location = self.master_branch.get_bound_location()
358
if master_bound_location:
359
raise errors.CommitToDoubleBoundBranch(self.branch,
360
self.master_branch, master_bound_location)
362
# TODO: jam 20051230 We could automatically push local
363
# commits to the remote branch if they would fit.
364
# But for now, just require remote to be identical
367
# Make sure the local branch is identical to the master
368
master_rh = self.master_branch.revision_history()
369
local_rh = self.branch.revision_history()
370
if local_rh != master_rh:
371
raise errors.BoundBranchOutOfDate(self.branch,
374
# Now things are ready to change the master branch
376
self.bound_branch = self.branch
377
self.master_branch.lock_write()
378
self.master_locked = True
381
"""Cleanup any open locks, progress bars etc."""
382
cleanups = [self._cleanup_bound_branch,
383
self.work_tree.unlock,
385
found_exception = None
386
for cleanup in cleanups:
389
# we want every cleanup to run no matter what.
390
# so we have a catchall here, but we will raise the
391
# last encountered exception up the stack: and
392
# typically this will be useful enough.
395
if found_exception is not None:
396
# don't do a plan raise, because the last exception may have been
397
# trashed, e is our sure-to-work exception even though it loses the
398
# full traceback. XXX: RBC 20060421 perhaps we could check the
399
# exc_info and if its the same one do a plain raise otherwise
400
# 'raise e' as we do now.
403
def _cleanup_bound_branch(self):
404
"""Executed at the end of a try/finally to cleanup a bound branch.
406
If the branch wasn't bound, this is a no-op.
407
If it was, it resents self.branch to the local branch, instead
410
if not self.bound_branch:
412
if self.master_locked:
413
self.master_branch.unlock()
236
def _record_inventory(self):
237
"""Store the inventory for the new revision."""
238
inv_text = serializer_v5.write_inventory_to_string(self.new_inv)
239
self.inv_sha1 = sha_string(inv_text)
240
s = self.branch.control_weaves
241
s.add_text('inventory', self.rev_id,
242
split_lines(inv_text), self.present_parents)
415
244
def _escape_commit_message(self):
416
245
"""Replace xml-incompatible control characters."""
417
# FIXME: RBC 20060419 this should be done by the revision
418
# serialiser not by commit. Then we can also add an unescaper
419
# in the deserializer and start roundtripping revision messages
420
# precisely. See repository_implementations/test_repository.py
422
246
# Python strings can include characters that can't be
423
247
# represented in well-formed XML; escape characters that
424
248
# aren't listed in the XML specification
425
249
# (http://www.w3.org/TR/REC-xml/#NT-Char).
250
if isinstance(self.message, unicode):
251
char_pattern = u'[^\x09\x0A\x0D\u0020-\uD7FF\uE000-\uFFFD]'
253
# Use a regular 'str' as pattern to avoid having re.subn
254
# return 'unicode' results.
255
char_pattern = '[^x09\x0A\x0D\x20-\xFF]'
426
256
self.message, escape_count = re.subn(
427
u'[^\x09\x0A\x0D\u0020-\uD7FF\uE000-\uFFFD]+',
428
258
lambda match: match.group(0).encode('unicode_escape'),
431
self.reporter.escaped(escape_count, self.message)
261
note("replaced %d control characters in message", escape_count)
263
def _record_ancestry(self):
264
"""Append merged revision ancestry to the ancestry file.
266
This should be the merged ancestry of all parents, plus the
268
s = self.branch.control_weaves
269
w = s.get_weave_or_empty('ancestry')
270
lines = self._make_ancestry(w)
271
w.add(self.rev_id, self.present_parents, lines)
272
s.put_weave('ancestry', w)
274
def _make_ancestry(self, ancestry_weave):
275
"""Return merged ancestry lines.
277
The lines are revision-ids followed by newlines."""
278
parent_ancestries = [ancestry_weave.get(p) for p in self.present_parents]
279
new_lines = merge_ancestry_lines(self.rev_id, parent_ancestries)
280
mutter('merged ancestry of {%s}:\n%s', self.rev_id, ''.join(new_lines))
433
283
def _gather_parents(self):
434
284
"""Record the parents of a merge for merge detection."""
435
# TODO: Make sure that this list doesn't contain duplicate
436
# entries and the order is preserved when doing this.
437
self.parents = self.work_tree.get_parent_ids()
438
self.parent_invs = []
285
pending_merges = self.branch.pending_merges()
287
self.parent_trees = []
288
self.present_parents = []
289
precursor_id = self.branch.last_revision()
291
self.parents.append(precursor_id)
292
self.parents += pending_merges
439
293
for revision in self.parents:
440
if self.branch.repository.has_revision(revision):
441
inventory = self.branch.repository.get_inventory(revision)
442
self.parent_invs.append(inventory)
294
if self.branch.has_revision(revision):
295
self.parent_trees.append(self.branch.revision_tree(revision))
296
self.present_parents.append(revision)
444
298
def _check_parents_present(self):
445
299
for parent_id in self.parents:
446
300
mutter('commit parent revision {%s}', parent_id)
447
if not self.branch.repository.has_revision(parent_id):
301
if not self.branch.has_revision(parent_id):
448
302
if parent_id == self.branch.last_revision():
449
warning("parent is missing %r", parent_id)
450
raise BzrCheckError("branch %s is missing revision {%s}"
451
% (self.branch, parent_id))
303
warning("parent is pissing %r", parent_id)
304
raise HistoryMissing(self.branch, 'revision', parent_id)
306
mutter("commit will ghost revision %r", parent_id)
308
def _make_revision(self):
309
"""Record a new revision object for this commit."""
310
self.rev = Revision(timestamp=self.timestamp,
311
timezone=self.timezone,
312
committer=self.committer,
313
message=self.message,
314
inventory_sha1=self.inv_sha1,
315
revision_id=self.rev_id)
316
self.rev.parent_ids = self.parents
318
serializer_v5.write_revision(self.rev, rev_tmp)
320
self.branch.revision_store.add(rev_tmp, self.rev_id)
321
mutter('new revision_id is {%s}', self.rev_id)
453
324
def _remove_deleted(self):
454
325
"""Remove deleted files from the working inventories.
467
338
if specific and not is_inside_any(specific, path):
469
340
if not self.work_tree.has_filename(path):
470
self.reporter.missing(path)
471
deleted_ids.append((path, ie.file_id))
341
note('missing %s', path)
342
deleted_ids.append(ie.file_id)
473
deleted_ids.sort(reverse=True)
474
for path, file_id in deleted_ids:
475
del self.work_inv[file_id]
476
self.work_tree._write_inventory(self.work_inv)
344
for file_id in deleted_ids:
345
if file_id in self.work_inv:
346
del self.work_inv[file_id]
347
self.branch._write_inventory(self.work_inv)
350
def _find_entry_parents(self, file_id):
351
"""Return the text versions and hashes for all file parents.
353
Returned as a map from text version to inventory entry.
355
This is a set containing the file versions in all parents
356
revisions containing the file. If the file is new, the set
359
for tree in self.parent_trees:
360
if file_id in tree.inventory:
361
ie = tree.inventory[file_id]
362
assert ie.file_id == file_id
364
assert r[ie.revision] == ie
369
def _store_snapshot(self):
370
"""Pass over inventory and record a snapshot.
372
Entries get a new revision when they are modified in
373
any way, which includes a merge with a new set of
374
parents that have the same entry. Currently we do not
375
check for that set being ancestors of each other - and
376
we should - only parallel children should count for this
377
test see find_entry_parents to correct this. FIXME <---
378
I.e. if we are merging in revision FOO, and our
379
copy of file id BAR is identical to FOO.BAR, we should
380
generate a new revision of BAR IF and only IF FOO is
381
neither a child of our current tip, nor an ancestor of
382
our tip. The presence of FOO in our store should not
383
affect this logic UNLESS we are doing a merge of FOO,
386
# XXX: Need to think more here about when the user has
387
# made a specific decision on a particular value -- c.f.
389
for path, ie in self.new_inv.iter_entries():
390
previous_entries = self._find_entry_parents(ie. file_id)
391
if ie.revision is None:
392
change = ie.snapshot(self.rev_id, path, previous_entries,
393
self.work_tree, self.weave_store)
396
note("%s %s", change, path)
478
398
def _populate_new_inv(self):
479
399
"""Build revision inventory.
485
405
None; inventory entries that are carried over untouched have their
486
406
revision set to their prior value.
488
# ESEPARATIONOFCONCERNS: this function is diffing and using the diff
489
# results to create a new inventory at the same time, which results
490
# in bugs like #46635. Any reason not to use/enhance Tree.changes_from?
492
408
mutter("Selecting files for commit with filter %s", self.specific_files)
493
entries = self.work_inv.iter_entries()
494
if not self.builder.record_root_entry:
495
symbol_versioning.warn('CommitBuilders should support recording'
496
' the root entry as of bzr 0.10.', DeprecationWarning,
498
self.builder.new_inventory.add(self.basis_inv.root.copy())
500
self._emit_progress_update()
501
for path, new_ie in entries:
502
self._emit_progress_update()
409
self.new_inv = Inventory()
410
for path, new_ie in self.work_inv.iter_entries():
503
411
file_id = new_ie.file_id
504
# mutter('check %s {%s}', path, file_id)
505
if (not self.specific_files or
506
is_inside_or_parent_of_any(self.specific_files, path)):
507
# mutter('%s selected for commit', path)
511
# mutter('%s not selected for commit', path)
512
if self.basis_inv.has_id(file_id):
513
ie = self.basis_inv[file_id].copy()
515
# this entry is new and not being committed
412
mutter('check %s {%s}', path, new_ie.file_id)
413
if self.specific_files:
414
if not is_inside_any(self.specific_files, path):
415
mutter('%s not selected for commit', path)
416
self._carry_file(file_id)
518
self.builder.record_entry_contents(ie, self.parent_invs,
519
path, self.work_tree)
520
# describe the nature of the change that has occurred relative to
521
# the basis inventory.
522
if (self.basis_inv.has_id(ie.file_id)):
523
basis_ie = self.basis_inv[ie.file_id]
526
change = ie.describe_change(basis_ie, ie)
527
if change in (InventoryEntry.RENAMED,
528
InventoryEntry.MODIFIED_AND_RENAMED):
529
old_path = self.basis_inv.id2path(ie.file_id)
530
self.reporter.renamed(change, old_path, path)
532
self.reporter.snapshot_change(change, path)
534
if not self.specific_files:
537
# ignore removals that don't match filespec
538
for path, new_ie in self.basis_inv.iter_entries():
539
if new_ie.file_id in self.work_inv:
541
if is_inside_any(self.specific_files, path):
418
mutter('%s selected for commit', path)
543
419
ie = new_ie.copy()
544
420
ie.revision = None
545
self.builder.record_entry_contents(ie, self.parent_invs, path,
548
def _emit_progress_update(self):
549
"""Emit an update to the progress bar."""
550
self.pb.update("Committing", self.pb_count, self.pb_total)
423
def _carry_file(self, file_id):
424
"""Carry the file unchanged from the basis revision."""
425
if self.basis_inv.has_id(file_id):
426
self.new_inv.add(self.basis_inv[file_id].copy())
553
428
def _report_deletes(self):
554
for path, ie in self.basis_inv.iter_entries():
555
if ie.file_id not in self.builder.new_inventory:
556
self.reporter.deleted(path)
429
for file_id in self.basis_inv:
430
if file_id not in self.new_inv:
431
note('deleted %s', self.basis_inv.id2path(file_id))
435
def _gen_revision_id(branch, when):
436
"""Return new revision-id."""
437
s = '%s-%s-' % (user_email(branch), compact_date(when))
438
s += hexlify(rand_bytes(8))
444
def merge_ancestry_lines(rev_id, ancestries):
445
"""Return merged ancestry lines.
447
rev_id -- id of the new revision
449
ancestries -- a sequence of ancestries for parent revisions,
450
as newline-terminated line lists.
452
if len(ancestries) == 0:
453
return [rev_id + '\n']
454
seen = set(ancestries[0])
455
ancs = ancestries[0][:]
456
for parent_ancestry in ancestries[1:]:
457
for line in parent_ancestry:
458
assert line[-1] == '\n'
added
removed
bzrlib/commit.py
