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
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.
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.
72
37
from binascii import hexlify
73
38
from cStringIO import StringIO
75
from bzrlib.atomicfile import AtomicFile
76
from bzrlib.osutils import (local_time_offset,
77
rand_bytes, compact_date,
40
from bzrlib.osutils import (local_time_offset, username,
41
rand_bytes, compact_date, user_email,
78
42
kind_marker, is_inside_any, quotefn,
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
43
sha_string, sha_file, isdir, isfile)
44
from bzrlib.branch import gen_file_id, INVENTORY_FILEID, ANCESTRY_FILEID
45
from bzrlib.errors import BzrError, PointlessCommit
46
from bzrlib.revision import Revision, RevisionReference
47
from bzrlib.trace import mutter, note
91
48
from bzrlib.xml5 import serializer_v5
92
from bzrlib.inventory import Inventory, ROOT_ID
49
from bzrlib.inventory import Inventory
50
from bzrlib.delta import compare_trees
93
51
from bzrlib.weave import Weave
94
52
from bzrlib.weavefile import read_weave, write_weave_v5
95
from bzrlib.workingtree import WorkingTree
53
from bzrlib.atomicfile import AtomicFile
98
56
def commit(*args, **kwargs):
103
61
New code should use the Commit class instead.
105
## XXX: Remove this in favor of Branch.commit?
106
63
Commit().commit(*args, **kwargs)
109
66
class NullCommitReporter(object):
110
67
"""I report on progress of a commit."""
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):
68
def added(self, path):
71
def removed(self, path):
74
def renamed(self, old_path, new_path):
128
78
class ReportCommitToLog(NullCommitReporter):
130
def snapshot_change(self, change, path):
131
note("%s %s", change, path)
133
def completed(self, revno, rev_id):
134
note('committed r%d {%s}', revno, rev_id)
136
def deleted(self, file_id):
137
note('deleted %s', file_id)
139
def escaped(self, escape_count, message):
140
note("replaced %d control characters in message", escape_count)
142
def missing(self, path):
143
note('missing %s', path)
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)
146
89
class Commit(object):
175
114
specific_files=None,
177
allow_pointless=True,
116
allow_pointless=True):
181
117
"""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.
183
136
timestamp -- if not None, seconds-since-epoch for a
184
137
postdated/predated commit.
186
specific_files -- If true, commit only those files.
140
If true, commit only those files.
188
rev_id -- If set, use this as the new revision id.
143
If set, use this as the new revision id.
189
144
Useful for test or import commands that need to tightly
190
145
control what revisions are assigned. If you duplicate
191
146
a revision id that exists elsewhere it is your own fault.
192
147
If null (default), a time/random revision id is generated.
194
allow_pointless -- If true (default), commit even if nothing
195
has changed and no merges are recorded.
197
strict -- If true, don't allow a commit if the working tree
198
contains unknown files.
200
revprops -- Properties for new revision
202
mutter('preparing to commit')
204
150
self.branch = branch
205
self.weave_store = branch.weave_store
151
self.branch.lock_write()
206
152
self.rev_id = rev_id
207
153
self.specific_files = specific_files
208
154
self.allow_pointless = allow_pointless
209
self.revprops = {'branch-nick': branch.nick}
211
self.revprops.update(revprops)
212
self.work_tree = WorkingTree(branch.base, branch)
215
# raise an exception as soon as we find a single unknown.
216
for unknown in self.work_tree.unknowns():
217
raise StrictCommitFailed()
219
156
if timestamp is None:
220
157
self.timestamp = time.time()
222
159
self.timestamp = long(timestamp)
224
if self.config is None:
225
self.config = bzrlib.config.BranchConfig(self.branch)
228
self.rev_id = _gen_revision_id(self.config, self.timestamp)
232
161
if committer is None:
233
self.committer = self.config.username()
162
self.committer = username(self.branch)
235
164
assert isinstance(committer, basestring), type(committer)
236
165
self.committer = committer
241
170
self.timezone = int(timezone)
243
if isinstance(message, str):
244
message = message.decode(bzrlib.user_encoding)
245
assert isinstance(message, unicode), type(message)
172
assert isinstance(message, basestring), type(message)
246
173
self.message = message
247
self._escape_commit_message()
249
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()
251
185
self.work_inv = self.work_tree.inventory
252
186
self.basis_tree = self.branch.basis_tree()
253
187
self.basis_inv = self.basis_tree.inventory
255
189
self._gather_parents()
256
if len(self.parents) > 1 and self.specific_files:
257
raise NotImplementedError('selected-file commit of merges is not supported yet')
258
self._check_parents_present()
191
if self.rev_id is None:
192
self.rev_id = _gen_revision_id(self.branch, time.time())
194
# TODO: update hashcache
195
self.delta = compare_trees(self.basis_tree, self.work_tree,
196
specific_files=self.specific_files)
198
if not (self.delta.has_changed()
199
or self.allow_pointless
200
or len(self.parents) != 1):
201
raise PointlessCommit()
203
self.new_inv = self.basis_inv.copy()
205
self.delta.show(sys.stdout)
260
207
self._remove_deleted()
261
self._populate_new_inv()
262
self._store_snapshot()
263
self._report_deletes()
265
if not (self.allow_pointless
266
or len(self.parents) > 1
267
or self.new_inv != self.basis_inv):
268
raise PointlessCommit()
270
if len(list(self.work_tree.iter_conflicts()))>0:
271
raise ConflictsInTree
210
self.branch._write_inventory(self.work_inv)
273
211
self._record_inventory()
212
self._record_ancestry()
274
214
self._make_revision()
215
note('committted r%d {%s}', (self.branch.revno() + 1),
275
217
self.branch.append_revision(self.rev_id)
276
self.work_tree.set_pending_merges([])
277
self.reporter.completed(self.branch.revno()+1, self.rev_id)
278
if self.config.post_commit() is not None:
279
hooks = self.config.post_commit().split(' ')
280
# this would be nicer with twisted.python.reflect.namedAny
282
result = eval(hook + '(branch, rev_id)',
283
{'branch':self.branch,
285
'rev_id':self.rev_id})
218
self.branch.set_pending_merges([])
287
220
self.branch.unlock()
289
223
def _record_inventory(self):
290
224
"""Store the inventory for the new revision."""
291
inv_text = serializer_v5.write_inventory_to_string(self.new_inv)
292
self.inv_sha1 = sha_string(inv_text)
293
s = self.branch.control_weaves
294
s.add_text('inventory', self.rev_id,
295
split_lines(inv_text), self.present_parents,
296
self.branch.get_transaction())
298
def _escape_commit_message(self):
299
"""Replace xml-incompatible control characters."""
300
# Python strings can include characters that can't be
301
# represented in well-formed XML; escape characters that
302
# aren't listed in the XML specification
303
# (http://www.w3.org/TR/REC-xml/#NT-Char).
304
self.message, escape_count = re.subn(
305
u'[^\x09\x0A\x0D\u0020-\uD7FF\uE000-\uFFFD]+',
306
lambda match: match.group(0).encode('unicode_escape'),
309
self.reporter.escaped(escape_count, self.message)
226
serializer_v5.write_inventory(self.new_inv, inv_tmp)
228
self.inv_sha1 = sha_string(inv_tmp.getvalue())
229
inv_lines = inv_tmp.readlines()
230
self.branch.weave_store.add_text(INVENTORY_FILEID, self.rev_id,
231
inv_lines, self.parents)
234
def _record_ancestry(self):
235
"""Append merged revision ancestry to the ancestry file."""
236
if len(self.parents) > 1:
237
raise NotImplementedError("sorry, can't commit merges yet")
238
w = self.branch.weave_store.get_weave_or_empty(ANCESTRY_FILEID)
240
lines = w.get(w.lookup(self.parents[0]))
243
lines.append(self.rev_id + '\n')
244
parent_idxs = map(w.lookup, self.parents)
245
w.add(self.rev_id, parent_idxs, lines)
246
self.branch.weave_store.put_weave(ANCESTRY_FILEID, w)
311
249
def _gather_parents(self):
312
"""Record the parents of a merge for merge detection."""
313
pending_merges = self.work_tree.pending_merges()
250
pending_merges = self.branch.pending_merges()
252
raise NotImplementedError("sorry, can't commit merges to the weave format yet")
314
253
self.parents = []
315
self.parent_invs = []
316
self.present_parents = []
317
precursor_id = self.branch.last_revision()
254
precursor_id = self.branch.last_patch()
319
256
self.parents.append(precursor_id)
320
257
self.parents += pending_merges
321
for revision in self.parents:
322
if self.branch.has_revision(revision):
323
self.parent_invs.append(self.branch.get_inventory(revision))
324
self.present_parents.append(revision)
326
def _check_parents_present(self):
327
for parent_id in self.parents:
328
mutter('commit parent revision {%s}', parent_id)
329
if not self.branch.has_revision(parent_id):
330
if parent_id == self.branch.last_revision():
331
warning("parent is missing %r", parent_id)
332
raise HistoryMissing(self.branch, 'revision', parent_id)
334
mutter("commit will ghost revision %r", parent_id)
336
260
def _make_revision(self):
337
261
"""Record a new revision object for this commit."""
338
262
self.rev = Revision(timestamp=self.timestamp,
340
264
committer=self.committer,
341
265
message=self.message,
342
266
inventory_sha1=self.inv_sha1,
343
revision_id=self.rev_id,
344
properties=self.revprops)
345
self.rev.parent_ids = self.parents
267
revision_id=self.rev_id)
268
self.rev.parents = map(RevisionReference, self.parents)
269
rev_tmp = tempfile.TemporaryFile()
347
270
serializer_v5.write_revision(self.rev, rev_tmp)
349
if self.config.signature_needed():
350
plaintext = Testament(self.rev, self.new_inv).as_short_text()
351
self.branch.store_revision_signature(gpg.GPGStrategy(self.config),
352
plaintext, self.rev_id)
353
272
self.branch.revision_store.add(rev_tmp, self.rev_id)
354
273
mutter('new revision_id is {%s}', self.rev_id)
356
276
def _remove_deleted(self):
357
"""Remove deleted files from the working inventories.
359
This is done prior to taking the working inventory as the
360
basis for the new committed inventory.
362
This returns true if any files
363
*that existed in the basis inventory* were deleted.
364
Files that were added and deleted
365
in the working copy don't matter.
367
specific = self.specific_files
369
for path, ie in self.work_inv.iter_entries():
370
if specific and not is_inside_any(specific, path):
372
if not self.work_tree.has_filename(path):
373
self.reporter.missing(path)
374
deleted_ids.append((path, ie.file_id))
376
deleted_ids.sort(reverse=True)
377
for path, file_id in deleted_ids:
378
del self.work_inv[file_id]
379
self.work_tree._write_inventory(self.work_inv)
381
def _store_snapshot(self):
382
"""Pass over inventory and record a snapshot.
384
Entries get a new revision when they are modified in
385
any way, which includes a merge with a new set of
386
parents that have the same entry.
388
# XXX: Need to think more here about when the user has
389
# made a specific decision on a particular value -- c.f.
391
for path, ie in self.new_inv.iter_entries():
392
previous_entries = ie.find_previous_heads(
394
self.weave_store.get_weave_or_empty(ie.file_id,
395
self.branch.get_transaction()))
396
if ie.revision is None:
397
change = ie.snapshot(self.rev_id, path, previous_entries,
398
self.work_tree, self.weave_store,
399
self.branch.get_transaction())
277
"""Remove deleted files from the working and stored inventories."""
278
for path, id, kind in self.delta.removed:
279
if self.work_inv.has_id(id):
280
del self.work_inv[id]
281
if self.new_inv.has_id(id):
286
def _store_files(self):
287
"""Store new texts of modified/added files."""
288
# We must make sure that directories are added before anything
289
# inside them is added. the files within the delta report are
290
# sorted by path so we know the directory will come before its
292
for path, file_id, kind in self.delta.added:
294
ie = self.work_inv[file_id].copy()
402
self.reporter.snapshot_change(change, path)
404
def _populate_new_inv(self):
405
"""Build revision inventory.
407
This creates a new empty inventory. Depending on
408
which files are selected for commit, and what is present in the
409
current tree, the new inventory is populated. inventory entries
410
which are candidates for modification have their revision set to
411
None; inventory entries that are carried over untouched have their
412
revision set to their prior value.
414
mutter("Selecting files for commit with filter %s", self.specific_files)
415
self.new_inv = Inventory()
416
for path, new_ie in self.work_inv.iter_entries():
417
file_id = new_ie.file_id
418
mutter('check %s {%s}', path, new_ie.file_id)
419
if self.specific_files:
420
if not is_inside_any(self.specific_files, path):
421
mutter('%s not selected for commit', path)
422
self._carry_entry(file_id)
425
# this is selected, ensure its parents are too.
426
parent_id = new_ie.parent_id
427
while parent_id != ROOT_ID:
428
if not self.new_inv.has_id(parent_id):
429
ie = self._select_entry(self.work_inv[parent_id])
430
mutter('%s selected for commit because of %s',
431
self.new_inv.id2path(parent_id), path)
433
ie = self.new_inv[parent_id]
434
if ie.revision is not None:
436
mutter('%s selected for commit because of %s',
437
self.new_inv.id2path(parent_id), path)
438
parent_id = ie.parent_id
439
mutter('%s selected for commit', path)
440
self._select_entry(new_ie)
442
def _select_entry(self, new_ie):
443
"""Make new_ie be considered for committing."""
449
def _carry_entry(self, file_id):
450
"""Carry the file unchanged from the basis revision."""
451
if self.basis_inv.has_id(file_id):
452
self.new_inv.add(self.basis_inv[file_id].copy())
454
def _report_deletes(self):
455
for file_id in self.basis_inv:
456
if file_id not in self.new_inv:
457
self.reporter.deleted(self.basis_inv.id2path(file_id))
459
def _gen_revision_id(config, when):
297
self._store_file_text(file_id)
299
for path, file_id, kind in self.delta.modified:
302
self._store_file_text(file_id)
304
for old_path, new_path, file_id, kind, text_modified in self.delta.renamed:
307
if not text_modified:
309
self._store_file_text(file_id)
312
def _store_file_text(self, file_id):
313
"""Store updated text for one modified or added file."""
314
note('store new text for {%s} in revision {%s}',
315
file_id, self.rev_id)
316
new_lines = self.work_tree.get_file(file_id).readlines()
317
if file_id in self.new_inv: # was in basis inventory
318
ie = self.new_inv[file_id]
319
assert ie.file_id == file_id
320
assert file_id in self.basis_inv
321
assert self.basis_inv[file_id].kind == 'file'
322
old_version = self.basis_inv[file_id].text_version
323
file_parents = [old_version]
324
else: # new in this revision
325
ie = self.work_inv[file_id].copy()
327
assert file_id not in self.basis_inv
329
assert ie.kind == 'file'
330
self._add_text_to_weave(file_id, new_lines, file_parents)
331
# make a new inventory entry for this file, using whatever
332
# it had in the working copy, plus details on the new text
333
ie.text_sha1 = _sha_strings(new_lines)
334
ie.text_size = sum(map(len, new_lines))
335
ie.text_version = self.rev_id
336
ie.entry_version = self.rev_id
339
def _add_text_to_weave(self, file_id, new_lines, parents):
340
if file_id.startswith('__'):
341
raise ValueError('illegal file-id %r for text file' % file_id)
342
self.branch.weave_store.add_text(file_id, self.rev_id, new_lines, parents)
345
def _gen_revision_id(branch, when):
460
346
"""Return new revision-id."""
461
s = '%s-%s-' % (config.user_email(), compact_date(when))
347
s = '%s-%s-' % (user_email(branch), compact_date(when))
462
348
s += hexlify(rand_bytes(8))
352
def _sha_strings(strings):
353
"""Return the sha-1 of concatenation of strings"""
355
map(s.update, strings)