1
1
# Copyright (C) 2005 Canonical Ltd
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.
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.
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
15
# Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
18
## XXX: Can we do any better about making interrupted commits change
21
## XXX: If we merged two versions of a file then we still need to
22
## create a new version representing that merge, even if it didn't
23
## change from the parent.
25
## TODO: Read back the just-generated changeset, and make sure it
26
## applies and recreates the right state.
18
# XXX: Can we do any better about making interrupted commits change
19
# nothing? Perhaps the best approach is to integrate commit of
20
# AtomicFiles with releasing the lock on the Branch.
22
# TODO: Separate 'prepare' phase where we find a list of potentially
23
# committed files. We then can then pause the commit to prompt for a
24
# commit message, knowing the summary will be the same as what's
25
# actually used for the commit. (But perhaps simpler to simply get
26
# the tree status, then use that for a selective commit?)
28
# The newly committed revision is going to have a shape corresponding
29
# to that of the working inventory. Files that are not in the
30
# working tree and that were in the predecessor are reported as
31
# removed --- this can include files that were either removed from the
32
# inventory or deleted in the working tree. If they were only
33
# deleted from disk, they are removed from the working inventory.
35
# We then consider the remaining entries, which will be in the new
36
# version. Directory entries are simply copied across. File entries
37
# must be checked to see if a new version of the file should be
38
# recorded. For each parent revision inventory, we check to see what
39
# version of the file was present. If the file was present in at
40
# least one tree, and if it was the same version in all the trees,
41
# then we can just refer to that version. Otherwise, a new version
42
# representing the merger of the file versions must be added.
44
# TODO: Update hashcache before and after - or does the WorkingTree
47
# TODO: Rather than mashing together the ancestry and storing it back,
48
# perhaps the weave should have single method which does it all in one
49
# go, avoiding a lot of redundant work.
51
# TODO: Perhaps give a warning if one of the revisions marked as
52
# merged is already in the ancestry, and then don't record it as a
55
# TODO: If the file is newly merged but unchanged from the version it
56
# merges from, then it should still be reported as newly added
57
# relative to the basis revision.
59
# TODO: Do checks that the tree can be committed *before* running the
60
# editor; this should include checks for a pointless commit and for
61
# unknown or missing files.
63
# TODO: If commit fails, leave the message in a file somewhere.
37
72
from binascii import hexlify
38
73
from cStringIO import StringIO
40
from bzrlib.osutils import (local_time_offset, username,
41
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,
42
78
kind_marker, is_inside_any, quotefn,
43
sha_string, sha_file, isdir, isfile)
44
from bzrlib.branch import gen_file_id
45
from bzrlib.errors import BzrError, PointlessCommit
46
from bzrlib.revision import Revision, RevisionReference
47
from bzrlib.trace import mutter, note
79
sha_string, sha_strings, sha_file, isdir, isfile,
82
from bzrlib.errors import (BzrError, PointlessCommit,
87
import bzrlib.gpg as gpg
88
from bzrlib.revision import Revision
89
from bzrlib.testament import Testament
90
from bzrlib.trace import mutter, note, warning
48
91
from bzrlib.xml5 import serializer_v5
49
from bzrlib.inventory import Inventory
50
from bzrlib.delta import compare_trees
92
from bzrlib.inventory import Inventory, ROOT_ID
93
from bzrlib.symbol_versioning import *
51
94
from bzrlib.weave import Weave
52
95
from bzrlib.weavefile import read_weave, write_weave_v5
53
from bzrlib.atomicfile import AtomicFile
96
from bzrlib.workingtree import WorkingTree
99
@deprecated_function(zero_seven)
56
100
def commit(*args, **kwargs):
57
101
"""Commit a new revision to a branch.
61
105
New code should use the Commit class instead.
107
## XXX: Remove this in favor of Branch.commit?
63
108
Commit().commit(*args, **kwargs)
66
111
class NullCommitReporter(object):
67
112
"""I report on progress of a commit."""
68
def added(self, path):
71
def removed(self, path):
74
def renamed(self, old_path, new_path):
114
def snapshot_change(self, change, path):
117
def completed(self, revno, rev_id):
120
def deleted(self, file_id):
123
def escaped(self, escape_count, message):
126
def missing(self, path):
78
130
class ReportCommitToLog(NullCommitReporter):
79
def added(self, path):
80
note('added %s', path)
82
def removed(self, path):
83
note('removed %s', path)
85
def renamed(self, old_path, new_path):
86
note('renamed %s => %s', old_path, new_path)
132
def snapshot_change(self, change, path):
133
note("%s %s", change, path)
135
def completed(self, revno, rev_id):
136
note('committed r%d {%s}', revno, rev_id)
138
def deleted(self, file_id):
139
note('deleted %s', file_id)
141
def escaped(self, escape_count, message):
142
note("replaced %d control characters in message", escape_count)
144
def missing(self, path):
145
note('missing %s', path)
89
148
class Commit(object):
99
158
working inventory.
101
160
def __init__(self,
103
163
if reporter is not None:
104
164
self.reporter = reporter
106
166
self.reporter = NullCommitReporter()
167
if config is not None:
173
branch=DEPRECATED_PARAMETER, message=None,
114
177
specific_files=None,
116
allow_pointless=True):
179
allow_pointless=True,
117
184
"""Commit working copy as a new revision.
119
The basic approach is to add all the file texts into the
120
store, then the inventory, then make a new revision pointing
121
to that inventory and store that.
123
This is not quite safe if the working copy changes during the
124
commit; for the moment that is simply not allowed. A better
125
approach is to make a temporary copy of the files before
126
computing their hashes, and then add those hashes in turn to
127
the inventory. This should mean at least that there are no
128
broken hash pointers. There is no way we can get a snapshot
129
of the whole directory at an instant. This would also have to
130
be robust against files disappearing, moving, etc. So the
131
whole thing is a bit hard.
133
This raises PointlessCommit if there are no changes, no new merges,
134
and allow_pointless is false.
186
branch -- the deprecated branch to commit to. New callers should pass in
189
message -- the commit message, a mandatory parameter
136
191
timestamp -- if not None, seconds-since-epoch for a
137
192
postdated/predated commit.
140
If true, commit only those files.
194
specific_files -- If true, commit only those files.
143
If set, use this as the new revision id.
196
rev_id -- If set, use this as the new revision id.
144
197
Useful for test or import commands that need to tightly
145
198
control what revisions are assigned. If you duplicate
146
199
a revision id that exists elsewhere it is your own fault.
147
200
If null (default), a time/random revision id is generated.
202
allow_pointless -- If true (default), commit even if nothing
203
has changed and no merges are recorded.
205
strict -- If true, don't allow a commit if the working tree
206
contains unknown files.
208
revprops -- Properties for new revision
151
self.branch.lock_write()
210
mutter('preparing to commit')
212
if deprecated_passed(branch):
213
warn("Commit.commit (branch, ...): The branch parameter is "
214
"deprecated as of bzr 0.8. Please use working_tree= instead.",
215
DeprecationWarning, stacklevel=2)
217
self.work_tree = WorkingTree(branch.base, branch)
218
elif working_tree is None:
219
raise BzrError("One of branch and working_tree must be passed into commit().")
221
self.work_tree = working_tree
222
self.branch = self.work_tree.branch
224
raise BzrError("The message keyword parameter is required for commit().")
226
self.weave_store = self.branch.repository.weave_store
152
227
self.rev_id = rev_id
153
228
self.specific_files = specific_files
154
229
self.allow_pointless = allow_pointless
230
self.revprops = {'branch-nick': self.branch.nick}
232
self.revprops.update(revprops)
235
# raise an exception as soon as we find a single unknown.
236
for unknown in self.work_tree.unknowns():
237
raise StrictCommitFailed()
156
239
if timestamp is None:
157
240
self.timestamp = time.time()
159
242
self.timestamp = long(timestamp)
244
if self.config is None:
245
self.config = bzrlib.config.BranchConfig(self.branch)
248
self.rev_id = _gen_revision_id(self.config, self.timestamp)
161
252
if committer is None:
162
self.committer = username(self.branch)
253
self.committer = self.config.username()
164
255
assert isinstance(committer, basestring), type(committer)
165
256
self.committer = committer
170
261
self.timezone = int(timezone)
172
assert isinstance(message, basestring), type(message)
263
if isinstance(message, str):
264
message = message.decode(bzrlib.user_encoding)
265
assert isinstance(message, unicode), type(message)
173
266
self.message = message
267
self._escape_commit_message()
269
self.branch.lock_write()
176
# First walk over the working inventory; and both update that
177
# and also build a new revision inventory. The revision
178
# inventory needs to hold the text-id, sha1 and size of the
179
# actual file versions committed in the revision. (These are
180
# not present in the working inventory.) We also need to
181
# detect missing/deleted files, and remove them from the
184
self.work_tree = self.branch.working_tree()
185
271
self.work_inv = self.work_tree.inventory
186
272
self.basis_tree = self.branch.basis_tree()
187
273
self.basis_inv = self.basis_tree.inventory
189
self.pending_merges = self.branch.pending_merges()
190
if self.pending_merges:
191
raise NotImplementedError("sorry, can't commit merges to the weave format yet")
275
self._gather_parents()
276
if len(self.parents) > 1 and self.specific_files:
277
raise NotImplementedError('selected-file commit of merges is not supported yet')
278
self._check_parents_present()
193
if self.rev_id is None:
194
self.rev_id = _gen_revision_id(self.branch, time.time())
196
# todo: update hashcache
197
self.delta = compare_trees(self.basis_tree, self.work_tree,
198
specific_files=self.specific_files)
200
if not (self.delta.has_changed()
201
or self.allow_pointless
202
or self.pending_merges):
280
self._remove_deleted()
281
self._populate_new_inv()
282
self._store_snapshot()
283
self._report_deletes()
285
if not (self.allow_pointless
286
or len(self.parents) > 1
287
or self.new_inv != self.basis_inv):
203
288
raise PointlessCommit()
205
self.new_inv = self.basis_inv.copy()
207
self.delta.show(sys.stdout)
209
self._remove_deleted()
212
self.branch._write_inventory(self.work_inv)
290
if len(list(self.work_tree.iter_conflicts()))>0:
291
raise ConflictsInTree
213
293
self._record_inventory()
215
294
self._make_revision()
216
note('committted r%d {%s}', (self.branch.revno() + 1),
295
self.work_tree.set_pending_merges([])
218
296
self.branch.append_revision(self.rev_id)
219
self.branch.set_pending_merges([])
297
if len(self.parents):
298
precursor = self.parents[0]
301
self.work_tree.set_last_revision(self.rev_id, precursor)
302
self.reporter.completed(self.branch.revno()+1, self.rev_id)
303
if self.config.post_commit() is not None:
304
hooks = self.config.post_commit().split(' ')
305
# this would be nicer with twisted.python.reflect.namedAny
307
result = eval(hook + '(branch, rev_id)',
308
{'branch':self.branch,
310
'rev_id':self.rev_id})
221
312
self.branch.unlock()
224
314
def _record_inventory(self):
226
serializer_v5.write_inventory(self.new_inv, inv_tmp)
227
self.inv_sha1 = sha_string(inv_tmp.getvalue())
229
self.branch.inventory_store.add(inv_tmp, self.rev_id)
315
"""Store the inventory for the new revision."""
316
inv_text = serializer_v5.write_inventory_to_string(self.new_inv)
317
self.inv_sha1 = sha_string(inv_text)
318
s = self.branch.repository.control_weaves
319
s.add_text('inventory', self.rev_id,
320
split_lines(inv_text), self.present_parents,
321
self.branch.get_transaction())
323
def _escape_commit_message(self):
324
"""Replace xml-incompatible control characters."""
325
# Python strings can include characters that can't be
326
# represented in well-formed XML; escape characters that
327
# aren't listed in the XML specification
328
# (http://www.w3.org/TR/REC-xml/#NT-Char).
329
self.message, escape_count = re.subn(
330
u'[^\x09\x0A\x0D\u0020-\uD7FF\uE000-\uFFFD]+',
331
lambda match: match.group(0).encode('unicode_escape'),
334
self.reporter.escaped(escape_count, self.message)
336
def _gather_parents(self):
337
"""Record the parents of a merge for merge detection."""
338
pending_merges = self.work_tree.pending_merges()
340
self.parent_invs = []
341
self.present_parents = []
342
precursor_id = self.branch.last_revision()
344
self.parents.append(precursor_id)
345
self.parents += pending_merges
346
for revision in self.parents:
347
if self.branch.repository.has_revision(revision):
348
inventory = self.branch.repository.get_inventory(revision)
349
self.parent_invs.append(inventory)
350
self.present_parents.append(revision)
352
def _check_parents_present(self):
353
for parent_id in self.parents:
354
mutter('commit parent revision {%s}', parent_id)
355
if not self.branch.repository.has_revision(parent_id):
356
if parent_id == self.branch.last_revision():
357
warning("parent is missing %r", parent_id)
358
raise HistoryMissing(self.branch, 'revision', parent_id)
360
mutter("commit will ghost revision %r", parent_id)
232
362
def _make_revision(self):
233
363
"""Record a new revision object for this commit."""
234
364
self.rev = Revision(timestamp=self.timestamp,
236
366
committer=self.committer,
237
367
message=self.message,
238
368
inventory_sha1=self.inv_sha1,
239
revision_id=self.rev_id)
241
self.rev.parents = []
242
precursor_id = self.branch.last_patch()
244
self.rev.parents.append(RevisionReference(precursor_id))
245
for merge_rev in self.pending_merges:
246
rev.parents.append(RevisionReference(merge_rev))
248
rev_tmp = tempfile.TemporaryFile()
369
revision_id=self.rev_id,
370
properties=self.revprops)
371
self.rev.parent_ids = self.parents
249
373
serializer_v5.write_revision(self.rev, rev_tmp)
251
self.branch.revision_store.add(rev_tmp, self.rev_id)
375
if self.config.signature_needed():
376
plaintext = Testament(self.rev, self.new_inv).as_short_text()
377
self.branch.repository.store_revision_signature(
378
gpg.GPGStrategy(self.config), plaintext, self.rev_id)
379
self.branch.repository.revision_store.add(rev_tmp, self.rev_id)
252
380
mutter('new revision_id is {%s}', self.rev_id)
255
382
def _remove_deleted(self):
256
"""Remove deleted files from the working and stored inventories."""
257
for path, id, kind in self.delta.removed:
258
if self.work_inv.has_id(id):
259
del self.work_inv[id]
260
if self.new_inv.has_id(id):
265
def _store_files(self):
266
"""Store new texts of modified/added files."""
267
# We must make sure that directories are added before anything
268
# inside them is added. the files within the delta report are
269
# sorted by path so we know the directory will come before its
271
for path, file_id, kind in self.delta.added:
273
ie = self.work_inv[file_id].copy()
383
"""Remove deleted files from the working inventories.
385
This is done prior to taking the working inventory as the
386
basis for the new committed inventory.
388
This returns true if any files
389
*that existed in the basis inventory* were deleted.
390
Files that were added and deleted
391
in the working copy don't matter.
393
specific = self.specific_files
395
for path, ie in self.work_inv.iter_entries():
396
if specific and not is_inside_any(specific, path):
398
if not self.work_tree.has_filename(path):
399
self.reporter.missing(path)
400
deleted_ids.append((path, ie.file_id))
402
deleted_ids.sort(reverse=True)
403
for path, file_id in deleted_ids:
404
del self.work_inv[file_id]
405
self.work_tree._write_inventory(self.work_inv)
407
def _store_snapshot(self):
408
"""Pass over inventory and record a snapshot.
410
Entries get a new revision when they are modified in
411
any way, which includes a merge with a new set of
412
parents that have the same entry.
414
# XXX: Need to think more here about when the user has
415
# made a specific decision on a particular value -- c.f.
417
for path, ie in self.new_inv.iter_entries():
418
previous_entries = ie.find_previous_heads(
420
self.weave_store.get_weave_prelude_or_empty(ie.file_id,
421
self.branch.get_transaction()))
422
if ie.revision is None:
423
change = ie.snapshot(self.rev_id, path, previous_entries,
424
self.work_tree, self.weave_store,
425
self.branch.get_transaction())
276
self._store_file_text(file_id)
278
for path, file_id, kind in self.delta.modified:
281
self._store_file_text(file_id)
283
for old_path, new_path, file_id, kind, text_modified in self.delta.renamed:
286
if not text_modified:
288
self._store_file_text(file_id)
291
def _store_file_text(self, file_id):
292
"""Store updated text for one modified or added file."""
293
note('store new text for {%s} in revision {%s}',
294
file_id, self.rev_id)
295
new_lines = self.work_tree.get_file(file_id).readlines()
296
if file_id in self.new_inv: # was in basis inventory
297
ie = self.new_inv[file_id]
298
assert ie.file_id == file_id
299
assert file_id in self.basis_inv
300
assert self.basis_inv[file_id].kind == 'file'
301
old_version = self.basis_inv[file_id].text_version
302
file_parents = [old_version]
303
else: # new in this revision
304
ie = self.work_inv[file_id].copy()
306
assert file_id not in self.basis_inv
308
assert ie.kind == 'file'
309
self._add_text_to_weave(file_id, new_lines, file_parents)
310
# make a new inventory entry for this file, using whatever
311
# it had in the working copy, plus details on the new text
312
ie.text_sha1 = _sha_strings(new_lines)
313
ie.text_size = sum(map(len, new_lines))
314
ie.text_version = self.rev_id
315
ie.entry_version = self.rev_id
318
def _add_text_to_weave(self, file_id, new_lines, parents):
319
weave_fn = self.branch.controlfilename(['weaves', file_id+'.weave'])
320
if os.path.exists(weave_fn):
321
w = read_weave(file(weave_fn, 'rb'))
324
# XXX: Should set the appropriate parents by looking for this file_id
325
# in all revision parents
326
parent_idxs = map(w.lookup, parents)
327
w.add(self.rev_id, parent_idxs, new_lines)
328
af = AtomicFile(weave_fn)
330
write_weave_v5(w, af)
336
def _gen_revision_id(branch, when):
428
self.reporter.snapshot_change(change, path)
430
def _populate_new_inv(self):
431
"""Build revision inventory.
433
This creates a new empty inventory. Depending on
434
which files are selected for commit, and what is present in the
435
current tree, the new inventory is populated. inventory entries
436
which are candidates for modification have their revision set to
437
None; inventory entries that are carried over untouched have their
438
revision set to their prior value.
440
mutter("Selecting files for commit with filter %s", self.specific_files)
441
self.new_inv = Inventory()
442
for path, new_ie in self.work_inv.iter_entries():
443
file_id = new_ie.file_id
444
mutter('check %s {%s}', path, new_ie.file_id)
445
if self.specific_files:
446
if not is_inside_any(self.specific_files, path):
447
mutter('%s not selected for commit', path)
448
self._carry_entry(file_id)
451
# this is selected, ensure its parents are too.
452
parent_id = new_ie.parent_id
453
while parent_id != ROOT_ID:
454
if not self.new_inv.has_id(parent_id):
455
ie = self._select_entry(self.work_inv[parent_id])
456
mutter('%s selected for commit because of %s',
457
self.new_inv.id2path(parent_id), path)
459
ie = self.new_inv[parent_id]
460
if ie.revision is not None:
462
mutter('%s selected for commit because of %s',
463
self.new_inv.id2path(parent_id), path)
464
parent_id = ie.parent_id
465
mutter('%s selected for commit', path)
466
self._select_entry(new_ie)
468
def _select_entry(self, new_ie):
469
"""Make new_ie be considered for committing."""
475
def _carry_entry(self, file_id):
476
"""Carry the file unchanged from the basis revision."""
477
if self.basis_inv.has_id(file_id):
478
self.new_inv.add(self.basis_inv[file_id].copy())
480
def _report_deletes(self):
481
for file_id in self.basis_inv:
482
if file_id not in self.new_inv:
483
self.reporter.deleted(self.basis_inv.id2path(file_id))
485
def _gen_revision_id(config, when):
337
486
"""Return new revision-id."""
338
s = '%s-%s-' % (user_email(branch), compact_date(when))
487
s = '%s-%s-' % (config.user_email(), compact_date(when))
339
488
s += hexlify(rand_bytes(8))
343
def _sha_strings(strings):
344
"""Return the sha-1 of concatenation of strings"""
346
map(s.update, strings)
added
removed
bzrlib/commit.py
