1
# Copyright (C) 2008-2018 Jelmer Vernooij <jelmer@jelmer.uk>
2
# Copyright (C) 2007 Canonical Ltd
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., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 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 not "<" in text and not ">" in text:
100
if text.rindex(">") < text.rindex("<"):
101
raise ValueError(text)
102
username, email = text.split("<", 2)[-2:]
103
email = email.split(">", 1)[0]
104
if username.endswith(" "):
105
username = username[:-1]
106
return "%s <%s>" % (username, email)
109
def warn_escaped(commit, num_escaped):
110
trace.warning("Escaped %d XML-invalid characters in %s. Will be unable "
111
"to regenerate the SHA map.", num_escaped, commit)
114
def warn_unusual_mode(commit, path, mode):
115
trace.mutter("Unusual file mode %o for %s in %s. Storing as revision "
116
"property. ", mode, path, commit)
119
class BzrGitMapping(foreign.VcsMapping):
120
"""Class that maps between Git and Bazaar semantics."""
123
BZR_FILE_IDS_FILE = None
125
BZR_DUMMY_FILE = None
127
def is_special_file(self, filename):
128
return (filename in (self.BZR_FILE_IDS_FILE, self.BZR_DUMMY_FILE))
131
super(BzrGitMapping, self).__init__(foreign_vcs_git)
133
def __eq__(self, other):
134
return (type(self) == type(other) and
135
self.revid_prefix == other.revid_prefix)
138
def revision_id_foreign_to_bzr(cls, git_rev_id):
139
"""Convert a git revision id handle to a Bazaar revision id."""
140
from dulwich.protocol import ZERO_SHA
141
if git_rev_id == ZERO_SHA:
143
return "%s:%s" % (cls.revid_prefix, git_rev_id)
146
def revision_id_bzr_to_foreign(cls, bzr_rev_id):
147
"""Convert a Bazaar revision id to a git revision id handle."""
148
if not bzr_rev_id.startswith("%s:" % cls.revid_prefix):
149
raise errors.InvalidRevisionId(bzr_rev_id, cls)
150
return bzr_rev_id[len(cls.revid_prefix)+1:], cls()
152
def generate_file_id(self, path):
153
# Git paths are just bytestrings
154
# We must just hope they are valid UTF-8..
157
if type(path) is unicode:
158
path = path.encode("utf-8")
159
return FILE_ID_PREFIX + escape_file_id(path)
161
def parse_file_id(self, file_id):
162
if file_id == ROOT_ID:
164
if not file_id.startswith(FILE_ID_PREFIX):
166
return unescape_file_id(file_id[len(FILE_ID_PREFIX):])
168
def revid_as_refname(self, revid):
170
return "refs/bzr/%s" % urllib.quote(revid)
172
def import_unusual_file_modes(self, rev, unusual_file_modes):
173
if unusual_file_modes:
174
ret = [(path, unusual_file_modes[path])
175
for path in sorted(unusual_file_modes.keys())]
176
rev.properties['file-modes'] = bencode.bencode(ret)
178
def export_unusual_file_modes(self, rev):
180
file_modes = rev.properties['file-modes']
184
return dict([(self.generate_file_id(path), mode) for (path, mode) in bencode.bdecode(file_modes.encode("utf-8"))])
186
def _generate_git_svn_metadata(self, rev, encoding):
188
git_svn_id = rev.properties["git-svn-id"]
192
return "\ngit-svn-id: %s\n" % git_svn_id.encode(encoding)
194
def _generate_hg_message_tail(self, rev):
198
for name in rev.properties:
199
if name == 'hg:extra:branch':
200
branch = rev.properties['hg:extra:branch']
201
elif name.startswith('hg:extra'):
202
extra[name[len('hg:extra:'):]] = base64.b64decode(
203
rev.properties[name])
204
elif name == 'hg:renames':
205
renames = bencode.bdecode(base64.b64decode(
206
rev.properties['hg:renames']))
207
# TODO: Export other properties as 'bzr:' extras?
208
ret = format_hg_metadata(renames, branch, extra)
209
if type(ret) is not str:
213
def _extract_git_svn_metadata(self, rev, message):
214
lines = message.split("\n")
215
if not (lines[-1] == "" and len(lines) >= 2 and lines[-2].startswith("git-svn-id:")):
217
git_svn_id = lines[-2].split(": ", 1)[1]
218
rev.properties['git-svn-id'] = git_svn_id
219
(url, rev, uuid) = parse_git_svn_id(git_svn_id)
220
# FIXME: Convert this to converted-from property somehow..
221
return "\n".join(lines[:-2])
223
def _extract_hg_metadata(self, rev, message):
224
(message, renames, branch, extra) = extract_hg_metadata(message)
225
if branch is not None:
226
rev.properties['hg:extra:branch'] = branch
227
for name, value in extra.iteritems():
228
rev.properties['hg:extra:' + name] = base64.b64encode(value)
230
rev.properties['hg:renames'] = base64.b64encode(bencode.bencode(
231
[(new, old) for (old, new) in renames.iteritems()]))
234
def _extract_bzr_metadata(self, rev, message):
235
(message, metadata) = extract_bzr_metadata(message)
236
return message, metadata
238
def _decode_commit_message(self, rev, message, encoding):
239
return message.decode(encoding), CommitSupplement()
241
def _encode_commit_message(self, rev, message, encoding):
242
return message.encode(encoding)
244
def export_fileid_map(self, fileid_map):
245
"""Export a file id map to a fileid map.
247
:param fileid_map: File id map, mapping paths to file ids
248
:return: A Git blob object (or None if there are no entries)
250
from dulwich.objects import Blob
252
b.set_raw_chunks(serialize_fileid_map(fileid_map))
255
def export_commit(self, rev, tree_sha, parent_lookup, lossy,
257
"""Turn a Bazaar revision in to a Git commit
259
:param tree_sha: Tree sha for the commit
260
:param parent_lookup: Function for looking up the GIT sha equiv of a
262
:param lossy: Whether to store roundtripping information.
263
:param verifiers: Verifiers info
264
:return dulwich.objects.Commit represent the revision:
266
from dulwich.objects import Commit
268
commit.tree = tree_sha
270
metadata = CommitSupplement()
271
metadata.verifiers = verifiers
275
for p in rev.parent_ids:
277
git_p = parent_lookup(p)
280
if metadata is not None:
281
metadata.explicit_parent_ids = rev.parent_ids
282
if git_p is not None:
284
raise AssertionError("unexpected length for %r" % git_p)
285
parents.append(git_p)
286
commit.parents = parents
288
encoding = rev.properties['git-explicit-encoding']
290
encoding = rev.properties.get('git-implicit-encoding', 'utf-8')
292
commit.encoding = rev.properties['git-explicit-encoding'].encode('ascii')
295
commit.committer = fix_person_identifier(rev.committer.encode(
297
commit.author = fix_person_identifier(
298
rev.get_apparent_authors()[0].encode(encoding))
299
commit.commit_time = long(rev.timestamp)
300
if 'author-timestamp' in rev.properties:
301
commit.author_time = long(rev.properties['author-timestamp'])
303
commit.author_time = commit.commit_time
304
commit._commit_timezone_neg_utc = "commit-timezone-neg-utc" in rev.properties
305
commit.commit_timezone = rev.timezone
306
commit._author_timezone_neg_utc = "author-timezone-neg-utc" in rev.properties
307
if 'author-timezone' in rev.properties:
308
commit.author_timezone = int(rev.properties['author-timezone'])
310
commit.author_timezone = commit.commit_timezone
311
if 'git-gpg-signature' in rev.properties:
312
commit.gpgsig = rev.properties['git-gpg-signature'].encode('ascii')
313
commit.message = self._encode_commit_message(rev, rev.message,
315
if type(commit.message) is not str:
316
raise TypeError(commit.message)
317
if metadata is not None:
319
mapping_registry.parse_revision_id(rev.revision_id)
320
except errors.InvalidRevisionId:
321
metadata.revision_id = rev.revision_id
322
mapping_properties = set(
323
['author', 'author-timezone', 'author-timezone-neg-utc',
324
'commit-timezone-neg-utc', 'git-implicit-encoding',
325
'git-gpg-signature', 'git-explicit-encoding',
326
'author-timestamp', 'file-modes'])
327
for k, v in rev.properties.iteritems():
328
if not k in mapping_properties:
329
metadata.properties[k] = v
330
if not lossy and metadata:
331
if self.roundtripping:
332
commit.message = inject_bzr_metadata(commit.message, metadata,
335
raise NoPushSupport()
336
if type(commit.message) is not str:
337
raise TypeError(commit.message)
338
if 'git-extra' in rev.properties:
339
commit.extra.extend([l.split(' ', 1) for l in rev.properties['git-extra'].splitlines()])
342
def import_fileid_map(self, blob):
343
"""Convert a git file id map blob.
345
:param blob: Git blob object with fileid map
346
:return: Dictionary mapping paths to file ids
348
return deserialize_fileid_map(blob.data)
350
def import_commit(self, commit, lookup_parent_revid):
351
"""Convert a git commit to a bzr revision.
353
:return: a `breezy.revision.Revision` object, foreign revid and a
357
raise AssertionError("Commit object can't be None")
358
rev = ForeignRevision(commit.id, self,
359
self.revision_id_foreign_to_bzr(commit.id))
360
rev.git_metadata = None
361
def decode_using_encoding(rev, commit, encoding):
362
rev.committer = str(commit.committer).decode(encoding)
363
if commit.committer != commit.author:
364
rev.properties['author'] = str(commit.author).decode(encoding)
365
rev.message, rev.git_metadata = self._decode_commit_message(
366
rev, commit.message, encoding)
367
if commit.encoding is not None:
368
rev.properties['git-explicit-encoding'] = commit.encoding
369
decode_using_encoding(rev, commit, commit.encoding)
371
for encoding in ('utf-8', 'latin1'):
373
decode_using_encoding(rev, commit, encoding)
374
except UnicodeDecodeError:
377
if encoding != 'utf-8':
378
rev.properties['git-implicit-encoding'] = encoding
380
if commit.commit_time != commit.author_time:
381
rev.properties['author-timestamp'] = str(commit.author_time)
382
if commit.commit_timezone != commit.author_timezone:
383
rev.properties['author-timezone'] = "%d" % commit.author_timezone
384
if commit._author_timezone_neg_utc:
385
rev.properties['author-timezone-neg-utc'] = ""
386
if commit._commit_timezone_neg_utc:
387
rev.properties['commit-timezone-neg-utc'] = ""
389
rev.properties['git-gpg-signature'] = commit.gpgsig.decode('ascii')
390
rev.timestamp = commit.commit_time
391
rev.timezone = commit.commit_timezone
392
rev.parent_ids = None
393
if rev.git_metadata is not None:
394
md = rev.git_metadata
395
roundtrip_revid = md.revision_id
396
if md.explicit_parent_ids:
397
rev.parent_ids = md.explicit_parent_ids
398
rev.properties.update(md.properties)
399
verifiers = md.verifiers
401
roundtrip_revid = None
403
if rev.parent_ids is None:
404
rev.parent_ids = tuple([lookup_parent_revid(p) for p in commit.parents])
405
unknown_extra_fields = []
407
for k, v in commit.extra:
408
if k == HG_RENAME_SOURCE:
409
extra_lines.append(k + ' ' + v + '\n')
411
hgk, hgv = v.split(':', 1)
412
if hgk not in (HG_EXTRA_AMEND_SOURCE, ):
413
raise UnknownMercurialCommitExtra(commit, hgk)
414
extra_lines.append(k + ' ' + v + '\n')
416
unknown_extra_fields.append(k)
417
if unknown_extra_fields:
418
raise UnknownCommitExtra(commit, unknown_extra_fields)
420
rev.properties['git-extra'] = ''.join(extra_lines)
421
return rev, roundtrip_revid, verifiers
423
def get_fileid_map(self, lookup_object, tree_sha):
424
"""Obtain a fileid map for a particular tree.
426
:param lookup_object: Function for looking up an object
427
:param tree_sha: SHA of the root tree
428
:return: GitFileIdMap instance
431
file_id_map_sha = lookup_object(tree_sha)[self.BZR_FILE_IDS_FILE][1]
435
file_ids = self.import_fileid_map(lookup_object(file_id_map_sha))
436
return GitFileIdMap(file_ids, self)
439
class BzrGitMappingv1(BzrGitMapping):
440
revid_prefix = 'git-v1'
444
return self.revid_prefix
447
class BzrGitMappingExperimental(BzrGitMappingv1):
448
revid_prefix = 'git-experimental'
452
BZR_FILE_IDS_FILE = '.bzrfileids'
454
BZR_DUMMY_FILE = '.bzrdummy'
456
def _decode_commit_message(self, rev, message, encoding):
457
message = self._extract_hg_metadata(rev, message)
458
message = self._extract_git_svn_metadata(rev, message)
459
message, metadata = self._extract_bzr_metadata(rev, message)
460
return message.decode(encoding), metadata
462
def _encode_commit_message(self, rev, message, encoding):
463
ret = message.encode(encoding)
464
ret += self._generate_hg_message_tail(rev)
465
ret += self._generate_git_svn_metadata(rev, encoding)
468
def import_commit(self, commit, lookup_parent_revid):
469
rev, roundtrip_revid, verifiers = super(BzrGitMappingExperimental, self).import_commit(commit, lookup_parent_revid)
470
rev.properties['converted_revision'] = "git %s\n" % commit.id
471
return rev, roundtrip_revid, verifiers
474
class GitMappingRegistry(VcsMappingRegistry):
475
"""Registry with available git mappings."""
477
def revision_id_bzr_to_foreign(self, bzr_revid):
478
if bzr_revid == NULL_REVISION:
479
from dulwich.protocol import ZERO_SHA
480
return ZERO_SHA, None
481
if not bzr_revid.startswith("git-"):
482
raise errors.InvalidRevisionId(bzr_revid, None)
483
(mapping_version, git_sha) = bzr_revid.split(":", 1)
484
mapping = self.get(mapping_version)
485
return mapping.revision_id_bzr_to_foreign(bzr_revid)
487
parse_revision_id = revision_id_bzr_to_foreign
490
mapping_registry = GitMappingRegistry()
491
mapping_registry.register_lazy('git-v1', "breezy.plugins.git.mapping",
493
mapping_registry.register_lazy('git-experimental',
494
"breezy.plugins.git.mapping", "BzrGitMappingExperimental")
495
# Uncomment the next line to enable the experimental bzr-git mappings.
496
# This will make sure all bzr metadata is pushed into git, allowing for
497
# full roundtripping later.
498
# NOTE: THIS IS EXPERIMENTAL. IT MAY EAT YOUR DATA OR CORRUPT
499
# YOUR BZR OR GIT REPOSITORIES. USE WITH CARE.
500
#mapping_registry.set_default('git-experimental')
501
mapping_registry.set_default('git-v1')
504
class ForeignGit(ForeignVcs):
505
"""The Git Stupid Content Tracker"""
508
def branch_format(self):
509
from .branch import LocalGitBranchFormat
510
return LocalGitBranchFormat()
513
def repository_format(self):
514
from .repository import GitRepositoryFormat
515
return GitRepositoryFormat()
518
super(ForeignGit, self).__init__(mapping_registry)
519
self.abbreviation = "git"
522
def serialize_foreign_revid(self, foreign_revid):
526
def show_foreign_revid(cls, foreign_revid):
527
return { "git commit": foreign_revid }
530
foreign_vcs_git = ForeignGit()
531
default_mapping = mapping_registry.get_default()()
534
def symlink_to_blob(symlink_target):
535
from dulwich.objects import Blob
537
if type(symlink_target) == unicode:
538
symlink_target = symlink_target.encode('utf-8')
539
blob.data = symlink_target
543
def mode_is_executable(mode):
544
"""Check if mode should be considered executable."""
545
return bool(mode & 0111)
549
"""Determine the Bazaar inventory kind based on Unix file mode."""
552
entry_kind = (mode & 0700000) / 0100000
555
elif entry_kind == 1:
556
file_kind = (mode & 070000) / 010000
562
return 'tree-reference'
564
raise AssertionError(
565
"Unknown file kind %d, perms=%o." % (file_kind, mode,))
567
raise AssertionError(
568
"Unknown kind, perms=%r." % (mode,))
571
def object_mode(kind, executable):
572
if kind == 'directory':
574
elif kind == 'symlink':
580
mode = stat.S_IFREG | 0644
584
elif kind == 'tree-reference':
585
from dulwich.objects import S_IFGITLINK
591
def entry_mode(entry):
592
"""Determine the git file mode for an inventory entry."""
593
return object_mode(entry.kind, entry.executable)
596
def directory_to_tree(children, lookup_ie_sha1, unusual_modes, empty_file_name,
598
"""Create a Git Tree object from a Bazaar directory.
600
:param children: Children inventory entries
601
:param lookup_ie_sha1: Lookup the Git SHA1 for a inventory entry
602
:param unusual_modes: Dictionary with unusual file modes by file ids
603
:param empty_file_name: Name to use for dummy files in empty directories,
604
None to ignore empty directories.
606
from dulwich.objects import Blob, Tree
608
for name, value in children.iteritems():
611
mode = unusual_modes[ie.file_id]
613
mode = entry_mode(ie)
614
hexsha = lookup_ie_sha1(ie)
615
if hexsha is not None:
616
tree.add(name.encode("utf-8"), mode, hexsha)
617
if not allow_empty and len(tree) == 0:
618
# Only the root can be an empty tree
619
if empty_file_name is not None:
620
tree.add(empty_file_name, stat.S_IFREG | 0644, Blob().id)
626
def extract_unusual_modes(rev):
628
foreign_revid, mapping = mapping_registry.parse_revision_id(
630
except errors.InvalidRevisionId:
633
return mapping.export_unusual_file_modes(rev)
636
def parse_git_svn_id(text):
637
(head, uuid) = text.rsplit(" ", 1)
638
(full_url, rev) = head.rsplit("@", 1)
639
return (full_url, int(rev), uuid)
642
class GitFileIdMap(object):
644
def __init__(self, file_ids, mapping):
645
self.file_ids = file_ids
647
self.mapping = mapping
649
def all_file_ids(self):
650
return self.file_ids.values()
652
def set_file_id(self, path, file_id):
653
if type(path) is not str:
654
raise TypeError(path)
655
if type(file_id) is not str:
656
raise TypeError(file_id)
657
self.file_ids[path] = file_id
659
def lookup_file_id(self, path):
660
if type(path) is not str:
661
raise TypeError(path)
663
file_id = self.file_ids[path]
665
file_id = self.mapping.generate_file_id(path)
666
if type(file_id) is not str:
667
raise TypeError(file_id)
670
def lookup_path(self, file_id):
671
if self.paths is None:
673
for k, v in self.file_ids.iteritems():
676
path = self.paths[file_id]
678
return self.mapping.parse_file_id(file_id)
680
if type(path) is not str:
681
raise TypeError(path)
685
return self.__class__(dict(self.file_ids), self.mapping)
688
def needs_roundtripping(repo, revid):
690
mapping_registry.parse_revision_id(revid)
691
except errors.InvalidRevisionId:
added
removed
mapping.py
