1
# Copyright (C) 2007 Canonical Ltd
2
# Copyright (C) 2008-2009 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."""
30
from bzrlib import bencode
32
from bzrlib.util import bencode
33
from bzrlib.inventory import (
36
from bzrlib.foreign import (
41
from bzrlib.revision import (
44
from bzrlib.plugins.git.hg import (
48
from bzrlib.plugins.git.roundtrip import (
51
BzrGitRevisionMetadata,
52
deserialize_fileid_map,
56
DEFAULT_FILE_MODE = stat.S_IFREG | 0644
59
def escape_file_id(file_id):
60
return file_id.replace('_', '__').replace(' ', '_s')
63
def unescape_file_id(file_id):
66
while i < len(file_id):
68
ret.append(file_id[i])
70
if file_id[i+1] == '_':
72
elif file_id[i+1] == 's':
75
raise AssertionError("unknown escape character %s" %
82
def fix_person_identifier(text):
83
if "<" in text and ">" in text:
85
return "%s <%s>" % (text, text)
88
def warn_escaped(commit, num_escaped):
89
trace.warning("Escaped %d XML-invalid characters in %s. Will be unable "
90
"to regenerate the SHA map.", num_escaped, commit)
93
def warn_unusual_mode(commit, path, mode):
94
trace.mutter("Unusual file mode %o for %s in %s. Storing as revision "
95
"property. ", mode, path, commit)
98
def squash_revision(target_repo, rev):
99
"""Remove characters that can't be stored from a revision, if necessary.
101
:param target_repo: Repository in which the revision will be stored
102
:param rev: Revision object, will be modified in-place
104
if not getattr(target_repo._serializer, "squashes_xml_invalid_characters", True):
106
from bzrlib.xml_serializer import escape_invalid_chars
107
rev.message, num_escaped = escape_invalid_chars(rev.message)
109
warn_escaped(rev.foreign_revid, num_escaped)
110
if 'author' in rev.properties:
111
rev.properties['author'], num_escaped = escape_invalid_chars(
112
rev.properties['author'])
114
warn_escaped(rev.foreign_revid, num_escaped)
115
rev.committer, num_escaped = escape_invalid_chars(rev.committer)
117
warn_escaped(rev.foreign_revid, num_escaped)
120
class BzrGitMapping(foreign.VcsMapping):
121
"""Class that maps between Git and Bazaar semantics."""
124
BZR_FILE_IDS_FILE = '.bzrfileids'
127
super(BzrGitMapping, self).__init__(foreign_git)
129
def __eq__(self, other):
130
return (type(self) == type(other) and
131
self.revid_prefix == other.revid_prefix)
134
def revision_id_foreign_to_bzr(cls, git_rev_id):
135
"""Convert a git revision id handle to a Bazaar revision id."""
136
from dulwich.protocol import ZERO_SHA
137
if git_rev_id == ZERO_SHA:
139
return "%s:%s" % (cls.revid_prefix, git_rev_id)
142
def revision_id_bzr_to_foreign(cls, bzr_rev_id):
143
"""Convert a Bazaar revision id to a git revision id handle."""
144
if not bzr_rev_id.startswith("%s:" % cls.revid_prefix):
145
raise errors.InvalidRevisionId(bzr_rev_id, cls)
146
return bzr_rev_id[len(cls.revid_prefix)+1:], cls()
148
def generate_file_id(self, path):
149
# Git paths are just bytestrings
150
# We must just hope they are valid UTF-8..
153
return escape_file_id(path)
155
def parse_file_id(self, file_id):
156
if file_id == ROOT_ID:
158
return unescape_file_id(file_id)
160
def import_unusual_file_modes(self, rev, unusual_file_modes):
161
if unusual_file_modes:
162
ret = [(path, unusual_file_modes[path])
163
for path in sorted(unusual_file_modes.keys())]
164
rev.properties['file-modes'] = bencode.bencode(ret)
166
def export_unusual_file_modes(self, rev):
168
file_modes = rev.properties['file-modes']
172
return dict([(self.generate_file_id(path), mode) for (path, mode) in bencode.bdecode(file_modes.encode("utf-8"))])
174
def _generate_git_svn_metadata(self, rev, encoding):
176
git_svn_id = rev.properties["git-svn-id"]
180
return "\ngit-svn-id: %s\n" % git_svn_id.encode(encoding)
182
def _generate_hg_message_tail(self, rev):
186
for name in rev.properties:
187
if name == 'hg:extra:branch':
188
branch = rev.properties['hg:extra:branch']
189
elif name.startswith('hg:extra'):
190
extra[name[len('hg:extra:'):]] = base64.b64decode(
191
rev.properties[name])
192
elif name == 'hg:renames':
193
renames = bencode.bdecode(base64.b64decode(
194
rev.properties['hg:renames']))
195
# TODO: Export other properties as 'bzr:' extras?
196
ret = format_hg_metadata(renames, branch, extra)
197
assert isinstance(ret, str)
200
def _extract_git_svn_metadata(self, rev, message):
201
lines = message.split("\n")
202
if not (lines[-1] == "" and lines[-2].startswith("git-svn-id:")):
204
git_svn_id = lines[-2].split(": ", 1)[1]
205
rev.properties['git-svn-id'] = git_svn_id
206
(url, rev, uuid) = parse_git_svn_id(git_svn_id)
207
# FIXME: Convert this to converted-from property somehow..
208
ret = "\n".join(lines[:-2])
209
assert isinstance(ret, str)
212
def _extract_hg_metadata(self, rev, message):
213
(message, renames, branch, extra) = extract_hg_metadata(message)
214
if branch is not None:
215
rev.properties['hg:extra:branch'] = branch
216
for name, value in extra.iteritems():
217
rev.properties['hg:extra:' + name] = base64.b64encode(value)
219
rev.properties['hg:renames'] = base64.b64encode(bencode.bencode(
220
[(new, old) for (old, new) in renames.iteritems()]))
223
def _extract_bzr_metadata(self, rev, message):
224
(message, metadata) = extract_bzr_metadata(message)
225
return message, metadata
227
def _decode_commit_message(self, rev, message, encoding):
228
message, metadata = self._extract_bzr_metadata(rev, message)
229
return message.decode(encoding), metadata
231
def _encode_commit_message(self, rev, message, encoding):
232
return message.encode(encoding)
234
def export_fileid_map(self, fileid_map):
235
"""Export a file id map to a fileid map.
237
:param fileid_map: File id map, mapping paths to file ids
238
:return: A Git blob object
240
from dulwich.objects import Blob
242
b.set_raw_chunks(serialize_fileid_map(fileid_map))
245
def export_commit(self, rev, tree_sha, parent_lookup, roundtrip):
246
"""Turn a Bazaar revision in to a Git commit
248
:param tree_sha: Tree sha for the commit
249
:param parent_lookup: Function for looking up the GIT sha equiv of a
251
:return dulwich.objects.Commit represent the revision:
253
from dulwich.objects import Commit
255
commit.tree = tree_sha
257
metadata = BzrGitRevisionMetadata()
260
for p in rev.parent_ids:
262
git_p = parent_lookup(p)
265
if metadata is not None:
266
metadata.explicit_parent_ids = rev.parent_ids
267
if git_p is not None:
268
assert len(git_p) == 40, "unexpected length for %r" % git_p
269
commit.parents.append(git_p)
271
encoding = rev.properties['git-explicit-encoding']
273
encoding = rev.properties.get('git-implicit-encoding', 'utf-8')
274
commit.encoding = rev.properties.get('git-explicit-encoding')
275
commit.committer = fix_person_identifier(rev.committer.encode(
277
commit.author = fix_person_identifier(
278
rev.get_apparent_authors()[0].encode(encoding))
279
commit.commit_time = long(rev.timestamp)
280
if 'author-timestamp' in rev.properties:
281
commit.author_time = long(rev.properties['author-timestamp'])
283
commit.author_time = commit.commit_time
284
commit._commit_timezone_neg_utc = "commit-timezone-neg-utc" in rev.properties
285
commit.commit_timezone = rev.timezone
286
commit._author_timezone_neg_utc = "author-timezone-neg-utc" in rev.properties
287
if 'author-timezone' in rev.properties:
288
commit.author_timezone = int(rev.properties['author-timezone'])
290
commit.author_timezone = commit.commit_timezone
291
commit.message = self._encode_commit_message(rev, rev.message,
293
if metadata is not None:
295
mapping_registry.parse_revision_id(rev.revision_id)
296
except errors.InvalidRevisionId:
297
metadata.revision_id = rev.revision_id
298
mapping_properties = set(
299
['author', 'author-timezone', 'author-timezone-neg-utc',
300
'commit-timezone-neg-utc', 'git-implicit-encoding',
301
'git-explicit-encoding', 'author-timestamp', 'file-modes'])
302
for k, v in rev.properties.iteritems():
303
if not k in mapping_properties:
304
metadata.properties[k] = v
305
commit.message = inject_bzr_metadata(commit.message, metadata)
308
def import_fileid_map(self, blob):
309
"""Convert a git file id map blob.
311
:param blob: Git blob object with fileid map
312
:return: Dictionary mapping paths to file ids
314
return deserialize_fileid_map(blob.text)
316
def import_commit(self, commit):
317
"""Convert a git commit to a bzr revision.
319
:return: a `bzrlib.revision.Revision` object and a
320
dictionary of path -> file ids
323
raise AssertionError("Commit object can't be None")
324
rev = ForeignRevision(commit.id, self,
325
self.revision_id_foreign_to_bzr(commit.id))
326
rev.parent_ids = tuple([self.revision_id_foreign_to_bzr(p) for p in commit.parents])
327
rev.git_metadata = None
328
def decode_using_encoding(rev, commit, encoding):
329
rev.committer = str(commit.committer).decode(encoding)
330
if commit.committer != commit.author:
331
rev.properties['author'] = str(commit.author).decode(encoding)
332
rev.message, rev.git_metadata = self._decode_commit_message(
333
rev, commit.message, encoding)
334
if commit.encoding is not None:
335
rev.properties['git-explicit-encoding'] = commit.encoding
336
decode_using_encoding(rev, commit, commit.encoding)
338
for encoding in ('utf-8', 'latin1'):
340
decode_using_encoding(rev, commit, encoding)
341
except UnicodeDecodeError:
344
if encoding != 'utf-8':
345
rev.properties['git-implicit-encoding'] = encoding
347
if commit.commit_time != commit.author_time:
348
rev.properties['author-timestamp'] = str(commit.author_time)
349
if commit.commit_timezone != commit.author_timezone:
350
rev.properties['author-timezone'] = "%d" % commit.author_timezone
351
if commit._author_timezone_neg_utc:
352
rev.properties['author-timezone-neg-utc'] = ""
353
if commit._commit_timezone_neg_utc:
354
rev.properties['commit-timezone-neg-utc'] = ""
355
rev.timestamp = commit.commit_time
356
rev.timezone = commit.commit_timezone
357
if rev.git_metadata is not None:
358
md = rev.git_metadata
360
rev.revision_id = md.revision_id
361
if md.explicit_parent_ids:
362
rev.parent_ids = md.explicit_parent_ids
363
rev.properties.update(md.properties)
367
class BzrGitMappingv1(BzrGitMapping):
368
revid_prefix = 'git-v1'
372
return self.revid_prefix
375
class BzrGitMappingExperimental(BzrGitMappingv1):
376
revid_prefix = 'git-experimental'
379
def _decode_commit_message(self, rev, message, encoding):
380
message = self._extract_hg_metadata(rev, message)
381
message = self._extract_git_svn_metadata(rev, message)
382
message, metadata = self._extract_bzr_metadata(rev, message)
383
return message.decode(encoding), metadata
385
def _encode_commit_message(self, rev, message, encoding):
386
ret = message.encode(encoding)
387
ret += self._generate_hg_message_tail(rev)
388
ret += self._generate_git_svn_metadata(rev, encoding)
391
def import_commit(self, commit):
392
rev, file_ids = super(BzrGitMappingExperimental, self).import_commit(commit)
393
rev.properties['converted_revision'] = "git %s\n" % commit.id
397
class GitMappingRegistry(VcsMappingRegistry):
398
"""Registry with available git mappings."""
400
def revision_id_bzr_to_foreign(self, bzr_revid):
401
if bzr_revid == NULL_REVISION:
402
from dulwich.protocol import ZERO_SHA
403
return ZERO_SHA, None
404
if not bzr_revid.startswith("git-"):
405
raise errors.InvalidRevisionId(bzr_revid, None)
406
(mapping_version, git_sha) = bzr_revid.split(":", 1)
407
mapping = self.get(mapping_version)
408
return mapping.revision_id_bzr_to_foreign(bzr_revid)
410
parse_revision_id = revision_id_bzr_to_foreign
413
mapping_registry = GitMappingRegistry()
414
mapping_registry.register_lazy('git-v1', "bzrlib.plugins.git.mapping",
416
mapping_registry.register_lazy('git-experimental',
417
"bzrlib.plugins.git.mapping", "BzrGitMappingExperimental")
418
mapping_registry.set_default('git-v1')
421
class ForeignGit(ForeignVcs):
422
"""The Git Stupid Content Tracker"""
425
def branch_format(self):
426
from bzrlib.plugins.git.branch import GitBranchFormat
427
return GitBranchFormat()
430
def repository_format(self):
431
from bzrlib.plugins.git.repository import GitRepositoryFormat
432
return GitRepositoryFormat()
435
super(ForeignGit, self).__init__(mapping_registry)
436
self.abbreviation = "git"
439
def serialize_foreign_revid(self, foreign_revid):
443
def show_foreign_revid(cls, foreign_revid):
444
return { "git commit": foreign_revid }
447
foreign_git = ForeignGit()
448
default_mapping = mapping_registry.get_default()()
451
def symlink_to_blob(entry):
452
from dulwich.objects import Blob
454
symlink_target = entry.symlink_target
455
if type(symlink_target) == unicode:
456
symlink_target = symlink_target.encode('utf-8')
457
blob.data = symlink_target
461
def mode_is_executable(mode):
462
"""Check if mode should be considered executable."""
463
return bool(mode & 0111)
467
"""Determine the Bazaar inventory kind based on Unix file mode."""
468
entry_kind = (mode & 0700000) / 0100000
471
elif entry_kind == 1:
472
file_kind = (mode & 070000) / 010000
478
return 'tree-reference'
480
raise AssertionError(
481
"Unknown file kind %d, perms=%o." % (file_kind, mode,))
483
raise AssertionError(
484
"Unknown kind, perms=%r." % (mode,))
487
def object_mode(kind, executable):
488
if kind == 'directory':
490
elif kind == 'symlink':
496
mode = stat.S_IFREG | 0644
500
elif kind == 'tree-reference':
501
from dulwich.objects import S_IFGITLINK
507
def entry_mode(entry):
508
"""Determine the git file mode for an inventory entry."""
509
return object_mode(entry.kind, entry.executable)
512
def directory_to_tree(entry, lookup_ie_sha1, unusual_modes):
513
from dulwich.objects import Tree
515
for name, value in entry.children.iteritems():
516
ie = entry.children[name]
518
mode = unusual_modes[ie.file_id]
520
mode = entry_mode(ie)
521
hexsha = lookup_ie_sha1(ie)
522
if hexsha is not None:
523
tree.add(mode, name.encode("utf-8"), hexsha)
524
if entry.parent_id is not None and len(tree) == 0:
525
# Only the root can be an empty tree
530
def extract_unusual_modes(rev):
532
foreign_revid, mapping = mapping_registry.parse_revision_id(
534
except errors.InvalidRevisionId:
537
return mapping.export_unusual_file_modes(rev)
540
def parse_git_svn_id(text):
541
(head, uuid) = text.rsplit(" ", 1)
542
(full_url, rev) = head.rsplit("@", 1)
543
return (full_url, int(rev), uuid)
added
removed
mapping.py
