1
# Copyright (C) 2007 Canonical Ltd
2
# Copyright (C) 2008-2010 Jelmer Vernooij <jelmer@samba.org>
3
# Copyright (C) 2008 John Carr
5
# This program is free software; you can redistribute it and/or modify
6
# it under the terms of the GNU General Public License as published by
7
# the Free Software Foundation; either version 2 of the License, or
8
# (at your option) any later version.
10
# This program is distributed in the hope that it will be useful,
11
# but WITHOUT ANY WARRANTY; without even the implied warranty of
12
# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
13
# GNU General Public License for more details.
15
# You should have received a copy of the GNU General Public License
16
# along with this program; if not, write to the Free Software
17
# Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
19
"""Converters, etc for going between Bazaar and Git ids."""
21
from __future__ import absolute_import
32
from ...bzr.inventory import (
35
from ...foreign import (
40
from ...revision import (
46
UnknownMercurialCommitExtra,
52
from .roundtrip import (
56
deserialize_fileid_map,
60
DEFAULT_FILE_MODE = stat.S_IFREG | 0644
61
HG_RENAME_SOURCE = "HG:rename-source"
64
# This HG extra is used to indicate the commit that this commit was based on.
65
HG_EXTRA_AMEND_SOURCE = "amend_source"
67
FILE_ID_PREFIX = b'git:'
70
def escape_file_id(file_id):
71
return file_id.replace('_', '__').replace(' ', '_s').replace('\x0c', '_c')
74
def unescape_file_id(file_id):
77
while i < len(file_id):
79
ret.append(file_id[i])
81
if file_id[i+1] == '_':
83
elif file_id[i+1] == 's':
85
elif file_id[i+1] == 'c':
88
raise ValueError("unknown escape character %s" %
95
def fix_person_identifier(text):
96
if "<" in text and ">" in text:
97
if not " <" in text and text.count("<") == 1:
98
text = text.replace("<", " <")
100
return "%s <%s>" % (text, text)
103
def warn_escaped(commit, num_escaped):
104
trace.warning("Escaped %d XML-invalid characters in %s. Will be unable "
105
"to regenerate the SHA map.", num_escaped, commit)
108
def warn_unusual_mode(commit, path, mode):
109
trace.mutter("Unusual file mode %o for %s in %s. Storing as revision "
110
"property. ", mode, path, commit)
113
class BzrGitMapping(foreign.VcsMapping):
114
"""Class that maps between Git and Bazaar semantics."""
117
BZR_FILE_IDS_FILE = None
119
BZR_DUMMY_FILE = None
121
def is_special_file(self, filename):
122
return (filename in (self.BZR_FILE_IDS_FILE, self.BZR_DUMMY_FILE))
125
super(BzrGitMapping, self).__init__(foreign_vcs_git)
127
def __eq__(self, other):
128
return (type(self) == type(other) and
129
self.revid_prefix == other.revid_prefix)
132
def revision_id_foreign_to_bzr(cls, git_rev_id):
133
"""Convert a git revision id handle to a Bazaar revision id."""
134
from dulwich.protocol import ZERO_SHA
135
if git_rev_id == ZERO_SHA:
137
return "%s:%s" % (cls.revid_prefix, git_rev_id)
140
def revision_id_bzr_to_foreign(cls, bzr_rev_id):
141
"""Convert a Bazaar revision id to a git revision id handle."""
142
if not bzr_rev_id.startswith("%s:" % cls.revid_prefix):
143
raise errors.InvalidRevisionId(bzr_rev_id, cls)
144
return bzr_rev_id[len(cls.revid_prefix)+1:], cls()
146
def generate_file_id(self, path):
147
# Git paths are just bytestrings
148
# We must just hope they are valid UTF-8..
151
if type(path) is unicode:
152
path = path.encode("utf-8")
153
return FILE_ID_PREFIX + escape_file_id(path)
155
def parse_file_id(self, file_id):
156
if file_id == ROOT_ID:
158
if not file_id.startswith(FILE_ID_PREFIX):
160
return unescape_file_id(file_id[len(FILE_ID_PREFIX):])
162
def revid_as_refname(self, revid):
164
return "refs/bzr/%s" % urllib.quote(revid)
166
def import_unusual_file_modes(self, rev, unusual_file_modes):
167
if unusual_file_modes:
168
ret = [(path, unusual_file_modes[path])
169
for path in sorted(unusual_file_modes.keys())]
170
rev.properties['file-modes'] = bencode.bencode(ret)
172
def export_unusual_file_modes(self, rev):
174
file_modes = rev.properties['file-modes']
178
return dict([(self.generate_file_id(path), mode) for (path, mode) in bencode.bdecode(file_modes.encode("utf-8"))])
180
def _generate_git_svn_metadata(self, rev, encoding):
182
git_svn_id = rev.properties["git-svn-id"]
186
return "\ngit-svn-id: %s\n" % git_svn_id.encode(encoding)
188
def _generate_hg_message_tail(self, rev):
192
for name in rev.properties:
193
if name == 'hg:extra:branch':
194
branch = rev.properties['hg:extra:branch']
195
elif name.startswith('hg:extra'):
196
extra[name[len('hg:extra:'):]] = base64.b64decode(
197
rev.properties[name])
198
elif name == 'hg:renames':
199
renames = bencode.bdecode(base64.b64decode(
200
rev.properties['hg:renames']))
201
# TODO: Export other properties as 'bzr:' extras?
202
ret = format_hg_metadata(renames, branch, extra)
203
assert isinstance(ret, str)
206
def _extract_git_svn_metadata(self, rev, message):
207
lines = message.split("\n")
208
if not (lines[-1] == "" and len(lines) >= 2 and lines[-2].startswith("git-svn-id:")):
210
git_svn_id = lines[-2].split(": ", 1)[1]
211
rev.properties['git-svn-id'] = git_svn_id
212
(url, rev, uuid) = parse_git_svn_id(git_svn_id)
213
# FIXME: Convert this to converted-from property somehow..
214
ret = "\n".join(lines[:-2])
215
assert isinstance(ret, str)
218
def _extract_hg_metadata(self, rev, message):
219
(message, renames, branch, extra) = extract_hg_metadata(message)
220
if branch is not None:
221
rev.properties['hg:extra:branch'] = branch
222
for name, value in extra.iteritems():
223
rev.properties['hg:extra:' + name] = base64.b64encode(value)
225
rev.properties['hg:renames'] = base64.b64encode(bencode.bencode(
226
[(new, old) for (old, new) in renames.iteritems()]))
229
def _extract_bzr_metadata(self, rev, message):
230
(message, metadata) = extract_bzr_metadata(message)
231
return message, metadata
233
def _decode_commit_message(self, rev, message, encoding):
234
return message.decode(encoding), CommitSupplement()
236
def _encode_commit_message(self, rev, message, encoding):
237
return message.encode(encoding)
239
def export_fileid_map(self, fileid_map):
240
"""Export a file id map to a fileid map.
242
:param fileid_map: File id map, mapping paths to file ids
243
:return: A Git blob object (or None if there are no entries)
245
from dulwich.objects import Blob
247
b.set_raw_chunks(serialize_fileid_map(fileid_map))
250
def export_commit(self, rev, tree_sha, parent_lookup, lossy,
252
"""Turn a Bazaar revision in to a Git commit
254
:param tree_sha: Tree sha for the commit
255
:param parent_lookup: Function for looking up the GIT sha equiv of a
257
:param lossy: Whether to store roundtripping information.
258
:param verifiers: Verifiers info
259
:return dulwich.objects.Commit represent the revision:
261
from dulwich.objects import Commit
263
commit.tree = tree_sha
265
metadata = CommitSupplement()
266
metadata.verifiers = verifiers
270
for p in rev.parent_ids:
272
git_p = parent_lookup(p)
275
if metadata is not None:
276
metadata.explicit_parent_ids = rev.parent_ids
277
if git_p is not None:
278
assert len(git_p) == 40, "unexpected length for %r" % git_p
279
parents.append(git_p)
280
commit.parents = parents
282
encoding = rev.properties['git-explicit-encoding']
284
encoding = rev.properties.get('git-implicit-encoding', 'utf-8')
286
commit.encoding = rev.properties['git-explicit-encoding'].encode('ascii')
289
commit.committer = fix_person_identifier(rev.committer.encode(
291
commit.author = fix_person_identifier(
292
rev.get_apparent_authors()[0].encode(encoding))
293
commit.commit_time = long(rev.timestamp)
294
if 'author-timestamp' in rev.properties:
295
commit.author_time = long(rev.properties['author-timestamp'])
297
commit.author_time = commit.commit_time
298
commit._commit_timezone_neg_utc = "commit-timezone-neg-utc" in rev.properties
299
commit.commit_timezone = rev.timezone
300
commit._author_timezone_neg_utc = "author-timezone-neg-utc" in rev.properties
301
if 'author-timezone' in rev.properties:
302
commit.author_timezone = int(rev.properties['author-timezone'])
304
commit.author_timezone = commit.commit_timezone
305
if 'git-gpg-signature' in rev.properties:
306
commit.gpgsig = rev.properties['git-gpg-signature'].encode('ascii')
307
commit.message = self._encode_commit_message(rev, rev.message,
309
assert type(commit.message) == str
310
if metadata is not None:
312
mapping_registry.parse_revision_id(rev.revision_id)
313
except errors.InvalidRevisionId:
314
metadata.revision_id = rev.revision_id
315
mapping_properties = set(
316
['author', 'author-timezone', 'author-timezone-neg-utc',
317
'commit-timezone-neg-utc', 'git-implicit-encoding',
318
'git-gpg-signature', 'git-explicit-encoding',
319
'author-timestamp', 'file-modes'])
320
for k, v in rev.properties.iteritems():
321
if not k in mapping_properties:
322
metadata.properties[k] = v
323
if not lossy and metadata:
324
if self.roundtripping:
325
commit.message = inject_bzr_metadata(commit.message, metadata,
328
raise NoPushSupport()
329
assert type(commit.message) == str
330
if 'git-extra' in rev.properties:
331
commit.extra.extend([l.split(' ', 1) for l in rev.properties['git-extra'].splitlines()])
334
def import_fileid_map(self, blob):
335
"""Convert a git file id map blob.
337
:param blob: Git blob object with fileid map
338
:return: Dictionary mapping paths to file ids
340
return deserialize_fileid_map(blob.data)
342
def import_commit(self, commit, lookup_parent_revid):
343
"""Convert a git commit to a bzr revision.
345
:return: a `breezy.revision.Revision` object, foreign revid and a
349
raise AssertionError("Commit object can't be None")
350
rev = ForeignRevision(commit.id, self,
351
self.revision_id_foreign_to_bzr(commit.id))
352
rev.git_metadata = None
353
def decode_using_encoding(rev, commit, encoding):
354
rev.committer = str(commit.committer).decode(encoding)
355
if commit.committer != commit.author:
356
rev.properties['author'] = str(commit.author).decode(encoding)
357
rev.message, rev.git_metadata = self._decode_commit_message(
358
rev, commit.message, encoding)
359
if commit.encoding is not None:
360
rev.properties['git-explicit-encoding'] = commit.encoding
361
decode_using_encoding(rev, commit, commit.encoding)
363
for encoding in ('utf-8', 'latin1'):
365
decode_using_encoding(rev, commit, encoding)
366
except UnicodeDecodeError:
369
if encoding != 'utf-8':
370
rev.properties['git-implicit-encoding'] = encoding
372
if commit.commit_time != commit.author_time:
373
rev.properties['author-timestamp'] = str(commit.author_time)
374
if commit.commit_timezone != commit.author_timezone:
375
rev.properties['author-timezone'] = "%d" % commit.author_timezone
376
if commit._author_timezone_neg_utc:
377
rev.properties['author-timezone-neg-utc'] = ""
378
if commit._commit_timezone_neg_utc:
379
rev.properties['commit-timezone-neg-utc'] = ""
381
rev.properties['git-gpg-signature'] = commit.gpgsig.decode('ascii')
382
rev.timestamp = commit.commit_time
383
rev.timezone = commit.commit_timezone
384
rev.parent_ids = None
385
if rev.git_metadata is not None:
386
md = rev.git_metadata
387
roundtrip_revid = md.revision_id
388
if md.explicit_parent_ids:
389
rev.parent_ids = md.explicit_parent_ids
390
rev.properties.update(md.properties)
391
verifiers = md.verifiers
393
roundtrip_revid = None
395
if rev.parent_ids is None:
396
rev.parent_ids = tuple([lookup_parent_revid(p) for p in commit.parents])
397
unknown_extra_fields = []
399
for k, v in commit.extra:
400
if k == HG_RENAME_SOURCE:
401
extra_lines.append(k + ' ' + v + '\n')
403
hgk, hgv = v.split(':', 1)
404
if hgk not in (HG_EXTRA_AMEND_SOURCE, ):
405
raise UnknownMercurialCommitExtra(commit, hgk)
406
extra_lines.append(k + ' ' + v + '\n')
408
unknown_extra_fields.append(k)
409
if unknown_extra_fields:
410
raise UnknownCommitExtra(commit, unknown_extra_fields)
412
rev.properties['git-extra'] = ''.join(extra_lines)
413
return rev, roundtrip_revid, verifiers
415
def get_fileid_map(self, lookup_object, tree_sha):
416
"""Obtain a fileid map for a particular tree.
418
:param lookup_object: Function for looking up an object
419
:param tree_sha: SHA of the root tree
420
:return: GitFileIdMap instance
423
file_id_map_sha = lookup_object(tree_sha)[self.BZR_FILE_IDS_FILE][1]
427
file_ids = self.import_fileid_map(lookup_object(file_id_map_sha))
428
return GitFileIdMap(file_ids, self)
431
class BzrGitMappingv1(BzrGitMapping):
432
revid_prefix = 'git-v1'
436
return self.revid_prefix
439
class BzrGitMappingExperimental(BzrGitMappingv1):
440
revid_prefix = 'git-experimental'
444
BZR_FILE_IDS_FILE = '.bzrfileids'
446
BZR_DUMMY_FILE = '.bzrdummy'
448
def _decode_commit_message(self, rev, message, encoding):
449
message = self._extract_hg_metadata(rev, message)
450
message = self._extract_git_svn_metadata(rev, message)
451
message, metadata = self._extract_bzr_metadata(rev, message)
452
return message.decode(encoding), metadata
454
def _encode_commit_message(self, rev, message, encoding):
455
ret = message.encode(encoding)
456
ret += self._generate_hg_message_tail(rev)
457
ret += self._generate_git_svn_metadata(rev, encoding)
460
def import_commit(self, commit, lookup_parent_revid):
461
rev, roundtrip_revid, verifiers = super(BzrGitMappingExperimental, self).import_commit(commit, lookup_parent_revid)
462
rev.properties['converted_revision'] = "git %s\n" % commit.id
463
return rev, roundtrip_revid, verifiers
466
class GitMappingRegistry(VcsMappingRegistry):
467
"""Registry with available git mappings."""
469
def revision_id_bzr_to_foreign(self, bzr_revid):
470
if bzr_revid == NULL_REVISION:
471
from dulwich.protocol import ZERO_SHA
472
return ZERO_SHA, None
473
if not bzr_revid.startswith("git-"):
474
raise errors.InvalidRevisionId(bzr_revid, None)
475
(mapping_version, git_sha) = bzr_revid.split(":", 1)
476
mapping = self.get(mapping_version)
477
return mapping.revision_id_bzr_to_foreign(bzr_revid)
479
parse_revision_id = revision_id_bzr_to_foreign
482
mapping_registry = GitMappingRegistry()
483
mapping_registry.register_lazy('git-v1', "breezy.plugins.git.mapping",
485
mapping_registry.register_lazy('git-experimental',
486
"breezy.plugins.git.mapping", "BzrGitMappingExperimental")
487
# Uncomment the next line to enable the experimental bzr-git mappings.
488
# This will make sure all bzr metadata is pushed into git, allowing for
489
# full roundtripping later.
490
# NOTE: THIS IS EXPERIMENTAL. IT MAY EAT YOUR DATA OR CORRUPT
491
# YOUR BZR OR GIT REPOSITORIES. USE WITH CARE.
492
#mapping_registry.set_default('git-experimental')
493
mapping_registry.set_default('git-v1')
496
class ForeignGit(ForeignVcs):
497
"""The Git Stupid Content Tracker"""
500
def branch_format(self):
501
from .branch import LocalGitBranchFormat
502
return LocalGitBranchFormat()
505
def repository_format(self):
506
from .repository import GitRepositoryFormat
507
return GitRepositoryFormat()
510
super(ForeignGit, self).__init__(mapping_registry)
511
self.abbreviation = "git"
514
def serialize_foreign_revid(self, foreign_revid):
518
def show_foreign_revid(cls, foreign_revid):
519
return { "git commit": foreign_revid }
522
foreign_vcs_git = ForeignGit()
523
default_mapping = mapping_registry.get_default()()
526
def symlink_to_blob(symlink_target):
527
from dulwich.objects import Blob
529
if type(symlink_target) == unicode:
530
symlink_target = symlink_target.encode('utf-8')
531
blob.data = symlink_target
535
def mode_is_executable(mode):
536
"""Check if mode should be considered executable."""
537
return bool(mode & 0111)
541
"""Determine the Bazaar inventory kind based on Unix file mode."""
544
entry_kind = (mode & 0700000) / 0100000
547
elif entry_kind == 1:
548
file_kind = (mode & 070000) / 010000
554
return 'tree-reference'
556
raise AssertionError(
557
"Unknown file kind %d, perms=%o." % (file_kind, mode,))
559
raise AssertionError(
560
"Unknown kind, perms=%r." % (mode,))
563
def object_mode(kind, executable):
564
if kind == 'directory':
566
elif kind == 'symlink':
572
mode = stat.S_IFREG | 0644
576
elif kind == 'tree-reference':
577
from dulwich.objects import S_IFGITLINK
583
def entry_mode(entry):
584
"""Determine the git file mode for an inventory entry."""
585
return object_mode(entry.kind, entry.executable)
588
def directory_to_tree(children, lookup_ie_sha1, unusual_modes, empty_file_name,
590
"""Create a Git Tree object from a Bazaar directory.
592
:param children: Children inventory entries
593
:param lookup_ie_sha1: Lookup the Git SHA1 for a inventory entry
594
:param unusual_modes: Dictionary with unusual file modes by file ids
595
:param empty_file_name: Name to use for dummy files in empty directories,
596
None to ignore empty directories.
598
from dulwich.objects import Blob, Tree
600
for name, value in children.iteritems():
603
mode = unusual_modes[ie.file_id]
605
mode = entry_mode(ie)
606
hexsha = lookup_ie_sha1(ie)
607
if hexsha is not None:
608
tree.add(name.encode("utf-8"), mode, hexsha)
609
if not allow_empty and len(tree) == 0:
610
# Only the root can be an empty tree
611
if empty_file_name is not None:
612
tree.add(empty_file_name, stat.S_IFREG | 0644, Blob().id)
618
def extract_unusual_modes(rev):
620
foreign_revid, mapping = mapping_registry.parse_revision_id(
622
except errors.InvalidRevisionId:
625
return mapping.export_unusual_file_modes(rev)
628
def parse_git_svn_id(text):
629
(head, uuid) = text.rsplit(" ", 1)
630
(full_url, rev) = head.rsplit("@", 1)
631
return (full_url, int(rev), uuid)
634
class GitFileIdMap(object):
636
def __init__(self, file_ids, mapping):
637
self.file_ids = file_ids
639
self.mapping = mapping
641
def all_file_ids(self):
642
return self.file_ids.values()
644
def set_file_id(self, path, file_id):
645
assert type(path) is str
646
assert type(file_id) is str
647
self.file_ids[path] = file_id
649
def lookup_file_id(self, path):
650
assert type(path) is str
652
file_id = self.file_ids[path]
654
file_id = self.mapping.generate_file_id(path)
655
assert type(file_id) is str
658
def lookup_path(self, file_id):
659
if self.paths is None:
661
for k, v in self.file_ids.iteritems():
664
path = self.paths[file_id]
666
return self.mapping.parse_file_id(file_id)
668
assert type(path) is str
672
return self.__class__(dict(self.file_ids), self.mapping)
675
def needs_roundtripping(repo, revid):
677
mapping_registry.parse_revision_id(revid)
678
except errors.InvalidRevisionId:
added
removed
mapping.py
