15
15
# You should have received a copy of the GNU General Public License
16
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
17
# Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
19
19
"""Converters, etc for going between Bazaar and Git ids."""
30
from bzrlib.inventory import (
33
from bzrlib.foreign import (
31
from ..foreign import (
35
33
VcsMappingRegistry,
38
from bzrlib.revision import (
36
from ..revision import (
41
from bzrlib.plugins.git.hg import (
42
44
format_hg_metadata,
43
45
extract_hg_metadata,
45
from bzrlib.plugins.git.roundtrip import (
47
from .roundtrip import (
46
48
extract_bzr_metadata,
47
49
inject_bzr_metadata,
48
BzrGitRevisionMetadata,
49
deserialize_fileid_map,
53
DEFAULT_FILE_MODE = stat.S_IFREG | 0644
54
DEFAULT_FILE_MODE = stat.S_IFREG | 0o644
55
HG_RENAME_SOURCE = b"HG:rename-source"
56
HG_EXTRA = b"HG:extra"
58
# This HG extra is used to indicate the commit that this commit was based on.
59
HG_EXTRA_AMEND_SOURCE = b"amend_source"
61
FILE_ID_PREFIX = b'git:'
64
ROOT_ID = b"TREE_ROOT"
67
class UnknownCommitExtra(errors.BzrError):
68
_fmt = "Unknown extra fields in %(object)r: %(fields)r."
70
def __init__(self, object, fields):
71
errors.BzrError.__init__(self)
73
self.fields = ",".join(fields)
76
class UnknownMercurialCommitExtra(errors.BzrError):
77
_fmt = "Unknown mercurial extra fields in %(object)r: %(fields)r."
79
def __init__(self, object, fields):
80
errors.BzrError.__init__(self)
82
self.fields = b",".join(fields)
56
85
def escape_file_id(file_id):
57
return file_id.replace('_', '__').replace(' ', '_s')
86
file_id = file_id.replace(b'_', b'__')
87
file_id = file_id.replace(b' ', b'_s')
88
file_id = file_id.replace(b'\x0c', b'_c')
60
92
def unescape_file_id(file_id):
63
95
while i < len(file_id):
96
if file_id[i:i + 1] != b'_':
65
97
ret.append(file_id[i])
67
if file_id[i+1] == '_':
69
elif file_id[i+1] == 's':
99
if file_id[i + 1:i + 2] == b'_':
101
elif file_id[i + 1:i + 2] == b's':
103
elif file_id[i + 1:i + 2] == b'c':
104
ret.append(b"\x0c"[0])
72
raise AssertionError("unknown escape character %s" %
106
raise ValueError("unknown escape character %s" %
107
file_id[i + 1:i + 2])
79
113
def fix_person_identifier(text):
80
if "<" in text and ">" in text:
82
return "%s <%s>" % (text, text)
114
if b"<" not in text and b">" not in text:
117
elif b">" not in text:
120
if text.rindex(b">") < text.rindex(b"<"):
121
raise ValueError(text)
122
username, email = text.split(b"<", 2)[-2:]
123
email = email.split(b">", 1)[0]
124
if username.endswith(b" "):
125
username = username[:-1]
126
return b"%s <%s>" % (username, email)
129
def decode_git_path(path):
130
"""Take a git path and decode it."""
132
return path.decode('utf-8')
133
except UnicodeDecodeError:
135
return path.decode('utf-8', 'surrogateescape')
139
def encode_git_path(path):
140
"""Take a regular path and encode it for git."""
142
return path.encode('utf-8')
143
except UnicodeEncodeError:
145
return path.encode('utf-8', 'surrogateescape')
85
149
def warn_escaped(commit, num_escaped):
113
178
from dulwich.protocol import ZERO_SHA
114
179
if git_rev_id == ZERO_SHA:
115
180
return NULL_REVISION
116
return "%s:%s" % (cls.revid_prefix, git_rev_id)
181
return b"%s:%s" % (cls.revid_prefix, git_rev_id)
119
184
def revision_id_bzr_to_foreign(cls, bzr_rev_id):
120
185
"""Convert a Bazaar revision id to a git revision id handle."""
121
if not bzr_rev_id.startswith("%s:" % cls.revid_prefix):
186
if not bzr_rev_id.startswith(b"%s:" % cls.revid_prefix):
122
187
raise errors.InvalidRevisionId(bzr_rev_id, cls)
123
return bzr_rev_id[len(cls.revid_prefix)+1:], cls()
188
return bzr_rev_id[len(cls.revid_prefix) + 1:], cls()
125
190
def generate_file_id(self, path):
126
191
# Git paths are just bytestrings
127
192
# We must just hope they are valid UTF-8..
193
if isinstance(path, str):
194
path = path.encode("utf-8")
130
if type(path) is unicode:
131
path = path.encode("utf-8")
132
return escape_file_id(path)
134
def is_control_file(self, path):
135
return path in (self.BZR_FILE_IDS_FILE, self.BZR_DUMMY_FILE)
197
return FILE_ID_PREFIX + escape_file_id(path)
137
199
def parse_file_id(self, file_id):
138
200
if file_id == ROOT_ID:
140
return unescape_file_id(file_id)
142
def revid_as_refname(self, revid):
144
return "refs/bzr/%s" % urllib.quote(revid)
202
if not file_id.startswith(FILE_ID_PREFIX):
204
return decode_git_path(unescape_file_id(file_id[len(FILE_ID_PREFIX):]))
146
206
def import_unusual_file_modes(self, rev, unusual_file_modes):
147
207
if unusual_file_modes:
148
208
ret = [(path, unusual_file_modes[path])
149
209
for path in sorted(unusual_file_modes.keys())]
150
rev.properties['file-modes'] = bencode.bencode(ret)
210
rev.properties[u'file-modes'] = bencode.bencode(ret)
152
212
def export_unusual_file_modes(self, rev):
154
file_modes = rev.properties['file-modes']
214
file_modes = rev.properties[u'file-modes']
158
return dict([(self.generate_file_id(path), mode) for (path, mode) in bencode.bdecode(file_modes.encode("utf-8"))])
218
return dict(bencode.bdecode(file_modes.encode("utf-8")))
160
220
def _generate_git_svn_metadata(self, rev, encoding):
162
git_svn_id = rev.properties["git-svn-id"]
222
git_svn_id = rev.properties[u"git-svn-id"]
171
231
branch = 'default'
172
232
for name in rev.properties:
173
if name == 'hg:extra:branch':
174
branch = rev.properties['hg:extra:branch']
175
elif name.startswith('hg:extra'):
176
extra[name[len('hg:extra:'):]] = base64.b64decode(
233
if name == u'hg:extra:branch':
234
branch = rev.properties[u'hg:extra:branch']
235
elif name.startswith(u'hg:extra'):
236
extra[name[len(u'hg:extra:'):]] = base64.b64decode(
177
237
rev.properties[name])
178
elif name == 'hg:renames':
238
elif name == u'hg:renames':
179
239
renames = bencode.bdecode(base64.b64decode(
180
rev.properties['hg:renames']))
240
rev.properties[u'hg:renames']))
181
241
# TODO: Export other properties as 'bzr:' extras?
182
242
ret = format_hg_metadata(renames, branch, extra)
183
assert isinstance(ret, str)
243
if not isinstance(ret, bytes):
186
247
def _extract_git_svn_metadata(self, rev, message):
187
248
lines = message.split("\n")
188
if not (lines[-1] == "" and len(lines) >= 2 and lines[-2].startswith("git-svn-id:")):
249
if not (lines[-1] == "" and len(lines) >= 2 and
250
lines[-2].startswith("git-svn-id:")):
190
252
git_svn_id = lines[-2].split(": ", 1)[1]
191
rev.properties['git-svn-id'] = git_svn_id
253
rev.properties[u'git-svn-id'] = git_svn_id
192
254
(url, rev, uuid) = parse_git_svn_id(git_svn_id)
193
255
# FIXME: Convert this to converted-from property somehow..
194
ret = "\n".join(lines[:-2])
195
assert isinstance(ret, str)
256
return "\n".join(lines[:-2])
198
258
def _extract_hg_metadata(self, rev, message):
199
259
(message, renames, branch, extra) = extract_hg_metadata(message)
200
260
if branch is not None:
201
rev.properties['hg:extra:branch'] = branch
202
for name, value in extra.iteritems():
203
rev.properties['hg:extra:' + name] = base64.b64encode(value)
261
rev.properties[u'hg:extra:branch'] = branch
262
for name, value in extra.items():
263
rev.properties[u'hg:extra:' + name] = base64.b64encode(value)
205
rev.properties['hg:renames'] = base64.b64encode(bencode.bencode(
206
[(new, old) for (old, new) in renames.iteritems()]))
265
rev.properties[u'hg:renames'] = base64.b64encode(bencode.bencode(
266
[(new, old) for (old, new) in renames.items()]))
209
269
def _extract_bzr_metadata(self, rev, message):
211
271
return message, metadata
213
273
def _decode_commit_message(self, rev, message, encoding):
214
return message.decode(encoding), BzrGitRevisionMetadata()
274
return message.decode(encoding), CommitSupplement()
216
276
def _encode_commit_message(self, rev, message, encoding):
217
277
return message.encode(encoding)
219
def export_fileid_map(self, fileid_map):
220
"""Export a file id map to a fileid map.
222
:param fileid_map: File id map, mapping paths to file ids
223
:return: A Git blob object
225
from dulwich.objects import Blob
227
b.set_raw_chunks(serialize_fileid_map(fileid_map))
230
def export_commit(self, rev, tree_sha, parent_lookup, roundtrip,
279
def export_commit(self, rev, tree_sha, parent_lookup, lossy,
232
281
"""Turn a Bazaar revision in to a Git commit
234
283
:param tree_sha: Tree sha for the commit
235
284
:param parent_lookup: Function for looking up the GIT sha equiv of a
237
:param roundtrip: Whether to store roundtripping information.
286
:param lossy: Whether to store roundtripping information.
238
287
:param verifiers: Verifiers info
239
288
:return dulwich.objects.Commit represent the revision:
241
from dulwich.objects import Commit
290
from dulwich.objects import Commit, Tag
242
291
commit = Commit()
243
292
commit.tree = tree_sha
245
metadata = BzrGitRevisionMetadata()
294
metadata = CommitSupplement()
246
295
metadata.verifiers = verifiers
255
304
if metadata is not None:
256
305
metadata.explicit_parent_ids = rev.parent_ids
257
306
if git_p is not None:
258
assert len(git_p) == 40, "unexpected length for %r" % git_p
308
raise AssertionError("unexpected length for %r" % git_p)
259
309
parents.append(git_p)
260
310
commit.parents = parents
262
encoding = rev.properties['git-explicit-encoding']
264
encoding = rev.properties.get('git-implicit-encoding', 'utf-8')
265
commit.encoding = rev.properties.get('git-explicit-encoding')
312
encoding = rev.properties[u'git-explicit-encoding']
314
encoding = rev.properties.get(u'git-implicit-encoding', 'utf-8')
316
commit.encoding = rev.properties[u'git-explicit-encoding'].encode(
266
320
commit.committer = fix_person_identifier(rev.committer.encode(
268
322
commit.author = fix_person_identifier(
269
323
rev.get_apparent_authors()[0].encode(encoding))
324
# TODO(jelmer): Don't use this hack.
325
long = getattr(__builtins__, 'long', int)
270
326
commit.commit_time = long(rev.timestamp)
271
if 'author-timestamp' in rev.properties:
272
commit.author_time = long(rev.properties['author-timestamp'])
327
if u'author-timestamp' in rev.properties:
328
commit.author_time = long(rev.properties[u'author-timestamp'])
274
330
commit.author_time = commit.commit_time
275
commit._commit_timezone_neg_utc = "commit-timezone-neg-utc" in rev.properties
331
commit._commit_timezone_neg_utc = (
332
u"commit-timezone-neg-utc" in rev.properties)
276
333
commit.commit_timezone = rev.timezone
277
commit._author_timezone_neg_utc = "author-timezone-neg-utc" in rev.properties
278
if 'author-timezone' in rev.properties:
279
commit.author_timezone = int(rev.properties['author-timezone'])
334
commit._author_timezone_neg_utc = (
335
u"author-timezone-neg-utc" in rev.properties)
336
if u'author-timezone' in rev.properties:
337
commit.author_timezone = int(rev.properties[u'author-timezone'])
281
339
commit.author_timezone = commit.commit_timezone
282
commit.message = self._encode_commit_message(rev, rev.message,
284
assert type(commit.message) == str
340
if u'git-gpg-signature' in rev.properties:
341
commit.gpgsig = rev.properties[u'git-gpg-signature'].encode(
342
'utf-8', 'surrogateescape')
343
commit.message = self._encode_commit_message(rev, rev.message,
345
if not isinstance(commit.message, bytes):
346
raise TypeError(commit.message)
285
347
if metadata is not None:
287
349
mapping_registry.parse_revision_id(rev.revision_id)
288
350
except errors.InvalidRevisionId:
289
351
metadata.revision_id = rev.revision_id
290
352
mapping_properties = set(
291
['author', 'author-timezone', 'author-timezone-neg-utc',
292
'commit-timezone-neg-utc', 'git-implicit-encoding',
293
'git-explicit-encoding', 'author-timestamp', 'file-modes'])
294
for k, v in rev.properties.iteritems():
295
if not k in mapping_properties:
353
[u'author', u'author-timezone', u'author-timezone-neg-utc',
354
u'commit-timezone-neg-utc', u'git-implicit-encoding',
355
u'git-gpg-signature', u'git-explicit-encoding',
356
u'author-timestamp', u'file-modes'])
357
for k, v in rev.properties.items():
358
if k not in mapping_properties:
296
359
metadata.properties[k] = v
297
if self.roundtripping:
298
commit.message = inject_bzr_metadata(commit.message, metadata,
300
assert type(commit.message) == str
360
if not lossy and metadata:
361
if self.roundtripping:
362
commit.message = inject_bzr_metadata(commit.message, metadata,
366
None, None, self, revision_id=rev.revision_id)
367
if not isinstance(commit.message, bytes):
368
raise TypeError(commit.message)
370
propname = u'git-mergetag-0'
371
while propname in rev.properties:
372
commit.mergetag.append(Tag.from_string(rev.properties[propname]))
374
propname = u'git-mergetag-%d' % i
375
if u'git-extra' in rev.properties:
378
for l in rev.properties[u'git-extra'].splitlines()])
303
def import_fileid_map(self, blob):
304
"""Convert a git file id map blob.
306
:param blob: Git blob object with fileid map
307
:return: Dictionary mapping paths to file ids
309
return deserialize_fileid_map(blob.data)
311
def import_commit(self, commit, lookup_parent_revid):
381
def get_revision_id(self, commit):
383
encoding = commit.encoding.decode('ascii')
387
message, metadata = self._decode_commit_message(
388
None, commit.message, encoding)
389
except UnicodeDecodeError:
392
if metadata.revision_id:
393
return metadata.revision_id
394
return self.revision_id_foreign_to_bzr(commit.id)
396
def import_commit(self, commit, lookup_parent_revid, strict=True):
312
397
"""Convert a git commit to a bzr revision.
314
:return: a `bzrlib.revision.Revision` object, foreign revid and a
399
:return: a `breezy.revision.Revision` object, foreign revid and a
317
402
if commit is None:
318
403
raise AssertionError("Commit object can't be None")
319
404
rev = ForeignRevision(commit.id, self,
320
self.revision_id_foreign_to_bzr(commit.id))
321
rev.parent_ids = tuple([lookup_parent_revid(p) for p in commit.parents])
405
self.revision_id_foreign_to_bzr(commit.id))
322
406
rev.git_metadata = None
323
408
def decode_using_encoding(rev, commit, encoding):
324
rev.committer = str(commit.committer).decode(encoding)
409
rev.committer = commit.committer.decode(encoding)
325
410
if commit.committer != commit.author:
326
rev.properties['author'] = str(commit.author).decode(encoding)
411
rev.properties[u'author'] = commit.author.decode(encoding)
327
412
rev.message, rev.git_metadata = self._decode_commit_message(
328
413
rev, commit.message, encoding)
329
414
if commit.encoding is not None:
330
rev.properties['git-explicit-encoding'] = commit.encoding
331
decode_using_encoding(rev, commit, commit.encoding)
415
rev.properties[u'git-explicit-encoding'] = commit.encoding.decode(
417
decode_using_encoding(rev, commit, commit.encoding.decode('ascii'))
333
419
for encoding in ('utf-8', 'latin1'):
339
425
if encoding != 'utf-8':
340
rev.properties['git-implicit-encoding'] = encoding
426
rev.properties[u'git-implicit-encoding'] = encoding
342
428
if commit.commit_time != commit.author_time:
343
rev.properties['author-timestamp'] = str(commit.author_time)
429
rev.properties[u'author-timestamp'] = str(commit.author_time)
344
430
if commit.commit_timezone != commit.author_timezone:
345
rev.properties['author-timezone'] = "%d" % commit.author_timezone
431
rev.properties[u'author-timezone'] = "%d" % commit.author_timezone
346
432
if commit._author_timezone_neg_utc:
347
rev.properties['author-timezone-neg-utc'] = ""
433
rev.properties[u'author-timezone-neg-utc'] = ""
348
434
if commit._commit_timezone_neg_utc:
349
rev.properties['commit-timezone-neg-utc'] = ""
435
rev.properties[u'commit-timezone-neg-utc'] = ""
437
rev.properties[u'git-gpg-signature'] = commit.gpgsig.decode(
438
'utf-8', 'surrogateescape')
440
for i, tag in enumerate(commit.mergetag):
441
rev.properties[u'git-mergetag-%d' % i] = tag.as_raw_string()
350
442
rev.timestamp = commit.commit_time
351
443
rev.timezone = commit.commit_timezone
444
rev.parent_ids = None
352
445
if rev.git_metadata is not None:
353
446
md = rev.git_metadata
354
447
roundtrip_revid = md.revision_id
360
453
roundtrip_revid = None
455
if rev.parent_ids is None:
457
for p in commit.parents:
459
parents.append(lookup_parent_revid(p))
461
parents.append(self.revision_id_foreign_to_bzr(p))
462
rev.parent_ids = list(parents)
463
unknown_extra_fields = []
465
for k, v in commit.extra:
466
if k == HG_RENAME_SOURCE:
467
extra_lines.append(k + b' ' + v + b'\n')
469
hgk, hgv = v.split(b':', 1)
470
if hgk not in (HG_EXTRA_AMEND_SOURCE, ) and strict:
471
raise UnknownMercurialCommitExtra(commit, [hgk])
472
extra_lines.append(k + b' ' + v + b'\n')
474
unknown_extra_fields.append(k)
475
if unknown_extra_fields and strict:
476
raise UnknownCommitExtra(
478
[f.decode('ascii', 'replace') for f in unknown_extra_fields])
480
rev.properties[u'git-extra'] = b''.join(extra_lines)
362
481
return rev, roundtrip_revid, verifiers
364
def get_fileid_map(self, lookup_object, tree_sha):
365
"""Obtain a fileid map for a particular tree.
367
:param lookup_object: Function for looking up an object
368
:param tree_sha: SHA of the root tree
369
:return: GitFileIdMap instance
372
file_id_map_sha = lookup_object(tree_sha)[self.BZR_FILE_IDS_FILE][1]
376
file_ids = self.import_fileid_map(lookup_object(file_id_map_sha))
377
return GitFileIdMap(file_ids, self)
380
484
class BzrGitMappingv1(BzrGitMapping):
381
revid_prefix = 'git-v1'
485
revid_prefix = b'git-v1'
382
486
experimental = False
384
488
def __str__(self):
461
573
def show_foreign_revid(cls, foreign_revid):
462
return { "git commit": foreign_revid }
465
foreign_git = ForeignGit()
574
return {"git commit": foreign_revid.decode('utf-8')}
577
foreign_vcs_git = ForeignGit()
466
578
default_mapping = mapping_registry.get_default()()
469
def symlink_to_blob(entry):
581
def symlink_to_blob(symlink_target):
470
582
from dulwich.objects import Blob
472
symlink_target = entry.symlink_target
473
if type(symlink_target) == unicode:
474
symlink_target = symlink_target.encode('utf-8')
584
if isinstance(symlink_target, str):
585
symlink_target = encode_git_path(symlink_target)
475
586
blob.data = symlink_target
479
590
def mode_is_executable(mode):
480
591
"""Check if mode should be considered executable."""
481
return bool(mode & 0111)
592
return bool(mode & 0o111)
484
595
def mode_kind(mode):
485
596
"""Determine the Bazaar inventory kind based on Unix file mode."""
486
entry_kind = (mode & 0700000) / 0100000
599
entry_kind = (mode & 0o700000) / 0o100000
487
600
if entry_kind == 0:
488
601
return 'directory'
489
602
elif entry_kind == 1:
490
file_kind = (mode & 070000) / 010000
603
file_kind = (mode & 0o70000) / 0o10000
491
604
if file_kind == 0:
493
606
elif file_kind == 2:
525
638
def entry_mode(entry):
526
639
"""Determine the git file mode for an inventory entry."""
527
return object_mode(entry.kind, entry.executable)
530
def directory_to_tree(entry, lookup_ie_sha1, unusual_modes, empty_file_name):
531
"""Create a Git Tree object from a Bazaar directory.
533
:param entry: Inventory entry
534
:param lookup_ie_sha1: Lookup the Git SHA1 for a inventory entry
535
:param unusual_modes: Dictionary with unusual file modes by file ids
536
:param empty_file_name: Name to use for dummy files in empty directories,
537
None to ignore empty directories.
539
from dulwich.objects import Blob, Tree
541
for name, value in entry.children.iteritems():
542
ie = entry.children[name]
544
mode = unusual_modes[ie.file_id]
546
mode = entry_mode(ie)
547
hexsha = lookup_ie_sha1(ie)
548
if hexsha is not None:
549
tree.add(mode, name.encode("utf-8"), hexsha)
550
if entry.parent_id is not None and len(tree) == 0:
551
# Only the root can be an empty tree
552
if empty_file_name is not None:
553
tree.add(stat.S_IFREG | 0644, empty_file_name,
640
return object_mode(entry.kind, getattr(entry, 'executable', False))
560
643
def extract_unusual_modes(rev):
added
removed
breezy/git/mapping.py
