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."""
21
from __future__ import absolute_import
30
from bzrlib.inventory import (
33
from bzrlib.foreign import (
33
from ..foreign import (
35
35
VcsMappingRegistry,
38
from bzrlib.revision import (
38
from ..revision import (
41
from bzrlib.plugins.git.hg import (
42
from ..sixish import (
42
51
format_hg_metadata,
43
52
extract_hg_metadata,
45
from bzrlib.plugins.git.roundtrip import (
54
from .roundtrip import (
46
55
extract_bzr_metadata,
47
56
inject_bzr_metadata,
48
BzrGitRevisionMetadata,
49
deserialize_fileid_map,
53
DEFAULT_FILE_MODE = stat.S_IFREG | 0644
61
DEFAULT_FILE_MODE = stat.S_IFREG | 0o644
62
HG_RENAME_SOURCE = b"HG:rename-source"
63
HG_EXTRA = b"HG:extra"
65
# This HG extra is used to indicate the commit that this commit was based on.
66
HG_EXTRA_AMEND_SOURCE = b"amend_source"
68
FILE_ID_PREFIX = b'git:'
71
ROOT_ID = b"TREE_ROOT"
74
class UnknownCommitExtra(errors.BzrError):
75
_fmt = "Unknown extra fields in %(object)r: %(fields)r."
77
def __init__(self, object, fields):
78
errors.BzrError.__init__(self)
80
self.fields = ",".join(fields)
83
class UnknownMercurialCommitExtra(errors.BzrError):
84
_fmt = "Unknown mercurial extra fields in %(object)r: %(fields)r."
86
def __init__(self, object, fields):
87
errors.BzrError.__init__(self)
89
self.fields = b",".join(fields)
56
92
def escape_file_id(file_id):
57
return file_id.replace('_', '__').replace(' ', '_s')
93
file_id = file_id.replace(b'_', b'__')
94
file_id = file_id.replace(b' ', b'_s')
95
file_id = file_id.replace(b'\x0c', b'_c')
60
99
def unescape_file_id(file_id):
63
102
while i < len(file_id):
103
if file_id[i:i + 1] != b'_':
65
104
ret.append(file_id[i])
67
if file_id[i+1] == '_':
69
elif file_id[i+1] == 's':
106
if file_id[i + 1:i + 2] == b'_':
108
elif file_id[i + 1:i + 2] == b's':
110
elif file_id[i + 1:i + 2] == b'c':
111
ret.append(b"\x0c"[0])
72
raise AssertionError("unknown escape character %s" %
113
raise ValueError("unknown escape character %s" %
114
file_id[i + 1:i + 2])
79
120
def fix_person_identifier(text):
80
if "<" in text and ">" in text:
82
return "%s <%s>" % (text, text)
121
if b"<" not in text and b">" not in text:
124
elif b">" not in text:
127
if text.rindex(b">") < text.rindex(b"<"):
128
raise ValueError(text)
129
username, email = text.split(b"<", 2)[-2:]
130
email = email.split(b">", 1)[0]
131
if username.endswith(b" "):
132
username = username[:-1]
133
return b"%s <%s>" % (username, email)
136
def decode_git_path(path):
137
"""Take a git path and decode it."""
139
return path.decode('utf-8')
140
except UnicodeDecodeError:
142
return path.decode('utf-8', 'surrogateescape')
146
def encode_git_path(path):
147
"""Take a regular path and encode it for git."""
149
return path.encode('utf-8')
150
except UnicodeEncodeError:
152
return path.encode('utf-8', 'surrogateescape')
85
156
def warn_escaped(commit, num_escaped):
113
185
from dulwich.protocol import ZERO_SHA
114
186
if git_rev_id == ZERO_SHA:
115
187
return NULL_REVISION
116
return "%s:%s" % (cls.revid_prefix, git_rev_id)
188
return b"%s:%s" % (cls.revid_prefix, git_rev_id)
119
191
def revision_id_bzr_to_foreign(cls, bzr_rev_id):
120
192
"""Convert a Bazaar revision id to a git revision id handle."""
121
if not bzr_rev_id.startswith("%s:" % cls.revid_prefix):
193
if not bzr_rev_id.startswith(b"%s:" % cls.revid_prefix):
122
194
raise errors.InvalidRevisionId(bzr_rev_id, cls)
123
return bzr_rev_id[len(cls.revid_prefix)+1:], cls()
195
return bzr_rev_id[len(cls.revid_prefix) + 1:], cls()
125
197
def generate_file_id(self, path):
126
198
# Git paths are just bytestrings
127
199
# We must just hope they are valid UTF-8..
200
if isinstance(path, text_type):
201
path = encode_git_path(path)
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)
204
return FILE_ID_PREFIX + escape_file_id(path)
137
206
def parse_file_id(self, file_id):
138
207
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)
209
if not file_id.startswith(FILE_ID_PREFIX):
211
return decode_git_path(unescape_file_id(file_id[len(FILE_ID_PREFIX):]))
146
213
def import_unusual_file_modes(self, rev, unusual_file_modes):
147
214
if unusual_file_modes:
148
215
ret = [(path, unusual_file_modes[path])
149
216
for path in sorted(unusual_file_modes.keys())]
150
rev.properties['file-modes'] = bencode.bencode(ret)
217
rev.properties[u'file-modes'] = bencode.bencode(ret)
152
219
def export_unusual_file_modes(self, rev):
154
file_modes = rev.properties['file-modes']
221
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"))])
225
return dict(bencode.bdecode(file_modes.encode("utf-8")))
160
227
def _generate_git_svn_metadata(self, rev, encoding):
162
git_svn_id = rev.properties["git-svn-id"]
229
git_svn_id = rev.properties[u"git-svn-id"]
171
238
branch = 'default'
172
239
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(
240
if name == u'hg:extra:branch':
241
branch = rev.properties[u'hg:extra:branch']
242
elif name.startswith(u'hg:extra'):
243
extra[name[len(u'hg:extra:'):]] = base64.b64decode(
177
244
rev.properties[name])
178
elif name == 'hg:renames':
245
elif name == u'hg:renames':
179
246
renames = bencode.bdecode(base64.b64decode(
180
rev.properties['hg:renames']))
247
rev.properties[u'hg:renames']))
181
248
# TODO: Export other properties as 'bzr:' extras?
182
249
ret = format_hg_metadata(renames, branch, extra)
183
assert isinstance(ret, str)
250
if not isinstance(ret, bytes):
186
254
def _extract_git_svn_metadata(self, rev, message):
187
255
lines = message.split("\n")
188
if not (lines[-1] == "" and len(lines) >= 2 and lines[-2].startswith("git-svn-id:")):
256
if not (lines[-1] == "" and len(lines) >= 2 and
257
lines[-2].startswith("git-svn-id:")):
190
259
git_svn_id = lines[-2].split(": ", 1)[1]
191
rev.properties['git-svn-id'] = git_svn_id
260
rev.properties[u'git-svn-id'] = git_svn_id
192
261
(url, rev, uuid) = parse_git_svn_id(git_svn_id)
193
262
# FIXME: Convert this to converted-from property somehow..
194
ret = "\n".join(lines[:-2])
195
assert isinstance(ret, str)
263
return "\n".join(lines[:-2])
198
265
def _extract_hg_metadata(self, rev, message):
199
266
(message, renames, branch, extra) = extract_hg_metadata(message)
200
267
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)
268
rev.properties[u'hg:extra:branch'] = branch
269
for name, value in viewitems(extra):
270
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()]))
272
rev.properties[u'hg:renames'] = base64.b64encode(bencode.bencode(
273
[(new, old) for (old, new) in viewitems(renames)]))
209
276
def _extract_bzr_metadata(self, rev, message):
211
278
return message, metadata
213
280
def _decode_commit_message(self, rev, message, encoding):
214
return message.decode(encoding), BzrGitRevisionMetadata()
281
return message.decode(encoding), CommitSupplement()
216
283
def _encode_commit_message(self, rev, message, encoding):
217
284
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,
286
def export_commit(self, rev, tree_sha, parent_lookup, lossy,
232
288
"""Turn a Bazaar revision in to a Git commit
234
290
:param tree_sha: Tree sha for the commit
235
291
:param parent_lookup: Function for looking up the GIT sha equiv of a
237
:param roundtrip: Whether to store roundtripping information.
293
:param lossy: Whether to store roundtripping information.
238
294
:param verifiers: Verifiers info
239
295
:return dulwich.objects.Commit represent the revision:
241
from dulwich.objects import Commit
297
from dulwich.objects import Commit, Tag
242
298
commit = Commit()
243
299
commit.tree = tree_sha
245
metadata = BzrGitRevisionMetadata()
301
metadata = CommitSupplement()
246
302
metadata.verifiers = verifiers
255
311
if metadata is not None:
256
312
metadata.explicit_parent_ids = rev.parent_ids
257
313
if git_p is not None:
258
assert len(git_p) == 40, "unexpected length for %r" % git_p
315
raise AssertionError("unexpected length for %r" % git_p)
259
316
parents.append(git_p)
260
317
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')
319
encoding = rev.properties[u'git-explicit-encoding']
321
encoding = rev.properties.get(u'git-implicit-encoding', 'utf-8')
323
commit.encoding = rev.properties[u'git-explicit-encoding'].encode(
266
327
commit.committer = fix_person_identifier(rev.committer.encode(
268
329
commit.author = fix_person_identifier(
269
330
rev.get_apparent_authors()[0].encode(encoding))
331
# TODO(jelmer): Don't use this hack.
332
long = getattr(__builtins__, 'long', int)
270
333
commit.commit_time = long(rev.timestamp)
271
if 'author-timestamp' in rev.properties:
272
commit.author_time = long(rev.properties['author-timestamp'])
334
if u'author-timestamp' in rev.properties:
335
commit.author_time = long(rev.properties[u'author-timestamp'])
274
337
commit.author_time = commit.commit_time
275
commit._commit_timezone_neg_utc = "commit-timezone-neg-utc" in rev.properties
338
commit._commit_timezone_neg_utc = (
339
u"commit-timezone-neg-utc" in rev.properties)
276
340
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'])
341
commit._author_timezone_neg_utc = (
342
u"author-timezone-neg-utc" in rev.properties)
343
if u'author-timezone' in rev.properties:
344
commit.author_timezone = int(rev.properties[u'author-timezone'])
281
346
commit.author_timezone = commit.commit_timezone
282
commit.message = self._encode_commit_message(rev, rev.message,
284
assert type(commit.message) == str
347
if u'git-gpg-signature' in rev.properties:
348
commit.gpgsig = rev.properties[u'git-gpg-signature'].encode(
349
'utf-8', 'surrogateescape')
350
commit.message = self._encode_commit_message(rev, rev.message,
352
if not isinstance(commit.message, bytes):
353
raise TypeError(commit.message)
285
354
if metadata is not None:
287
356
mapping_registry.parse_revision_id(rev.revision_id)
288
357
except errors.InvalidRevisionId:
289
358
metadata.revision_id = rev.revision_id
290
359
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:
360
[u'author', u'author-timezone', u'author-timezone-neg-utc',
361
u'commit-timezone-neg-utc', u'git-implicit-encoding',
362
u'git-gpg-signature', u'git-explicit-encoding',
363
u'author-timestamp', u'file-modes'])
364
for k, v in viewitems(rev.properties):
365
if k not in mapping_properties:
296
366
metadata.properties[k] = v
297
if self.roundtripping:
298
commit.message = inject_bzr_metadata(commit.message, metadata,
300
assert type(commit.message) == str
367
if not lossy and metadata:
368
if self.roundtripping:
369
commit.message = inject_bzr_metadata(commit.message, metadata,
373
None, None, self, revision_id=rev.revision_id)
374
if not isinstance(commit.message, bytes):
375
raise TypeError(commit.message)
377
propname = u'git-mergetag-0'
378
while propname in rev.properties:
379
commit.mergetag.append(Tag.from_string(rev.properties[propname]))
381
propname = u'git-mergetag-%d' % i
382
if u'git-extra' in rev.properties:
385
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):
388
def get_revision_id(self, commit):
390
encoding = commit.encoding.decode('ascii')
394
message, metadata = self._decode_commit_message(
395
None, commit.message, encoding)
396
except UnicodeDecodeError:
399
if metadata.revision_id:
400
return metadata.revision_id
401
return self.revision_id_foreign_to_bzr(commit.id)
403
def import_commit(self, commit, lookup_parent_revid, strict=True):
312
404
"""Convert a git commit to a bzr revision.
314
:return: a `bzrlib.revision.Revision` object, foreign revid and a
406
:return: a `breezy.revision.Revision` object, foreign revid and a
317
409
if commit is None:
318
410
raise AssertionError("Commit object can't be None")
319
411
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])
412
self.revision_id_foreign_to_bzr(commit.id))
322
413
rev.git_metadata = None
323
415
def decode_using_encoding(rev, commit, encoding):
324
rev.committer = str(commit.committer).decode(encoding)
416
rev.committer = commit.committer.decode(encoding)
325
417
if commit.committer != commit.author:
326
rev.properties['author'] = str(commit.author).decode(encoding)
418
rev.properties[u'author'] = commit.author.decode(encoding)
327
419
rev.message, rev.git_metadata = self._decode_commit_message(
328
420
rev, commit.message, encoding)
329
421
if commit.encoding is not None:
330
rev.properties['git-explicit-encoding'] = commit.encoding
331
decode_using_encoding(rev, commit, commit.encoding)
422
rev.properties[u'git-explicit-encoding'] = commit.encoding.decode(
424
decode_using_encoding(rev, commit, commit.encoding.decode('ascii'))
333
426
for encoding in ('utf-8', 'latin1'):
339
432
if encoding != 'utf-8':
340
rev.properties['git-implicit-encoding'] = encoding
433
rev.properties[u'git-implicit-encoding'] = encoding
342
435
if commit.commit_time != commit.author_time:
343
rev.properties['author-timestamp'] = str(commit.author_time)
436
rev.properties[u'author-timestamp'] = str(commit.author_time)
344
437
if commit.commit_timezone != commit.author_timezone:
345
rev.properties['author-timezone'] = "%d" % commit.author_timezone
438
rev.properties[u'author-timezone'] = "%d" % commit.author_timezone
346
439
if commit._author_timezone_neg_utc:
347
rev.properties['author-timezone-neg-utc'] = ""
440
rev.properties[u'author-timezone-neg-utc'] = ""
348
441
if commit._commit_timezone_neg_utc:
349
rev.properties['commit-timezone-neg-utc'] = ""
442
rev.properties[u'commit-timezone-neg-utc'] = ""
444
rev.properties[u'git-gpg-signature'] = commit.gpgsig.decode(
445
'utf-8', 'surrogateescape')
447
for i, tag in enumerate(commit.mergetag):
448
rev.properties[u'git-mergetag-%d' % i] = tag.as_raw_string()
350
449
rev.timestamp = commit.commit_time
351
450
rev.timezone = commit.commit_timezone
451
rev.parent_ids = None
352
452
if rev.git_metadata is not None:
353
453
md = rev.git_metadata
354
454
roundtrip_revid = md.revision_id
360
460
roundtrip_revid = None
462
if rev.parent_ids is None:
464
for p in commit.parents:
466
parents.append(lookup_parent_revid(p))
468
parents.append(self.revision_id_foreign_to_bzr(p))
469
rev.parent_ids = list(parents)
470
unknown_extra_fields = []
472
for k, v in commit.extra:
473
if k == HG_RENAME_SOURCE:
474
extra_lines.append(k + b' ' + v + b'\n')
476
hgk, hgv = v.split(b':', 1)
477
if hgk not in (HG_EXTRA_AMEND_SOURCE, ) and strict:
478
raise UnknownMercurialCommitExtra(commit, [hgk])
479
extra_lines.append(k + b' ' + v + b'\n')
481
unknown_extra_fields.append(k)
482
if unknown_extra_fields and strict:
483
raise UnknownCommitExtra(
485
[f.decode('ascii', 'replace') for f in unknown_extra_fields])
487
rev.properties[u'git-extra'] = b''.join(extra_lines)
362
488
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
491
class BzrGitMappingv1(BzrGitMapping):
381
revid_prefix = 'git-v1'
492
revid_prefix = b'git-v1'
382
493
experimental = False
384
495
def __str__(self):
461
580
def show_foreign_revid(cls, foreign_revid):
462
return { "git commit": foreign_revid }
465
foreign_git = ForeignGit()
581
return {"git commit": foreign_revid.decode('utf-8')}
584
foreign_vcs_git = ForeignGit()
466
585
default_mapping = mapping_registry.get_default()()
469
def symlink_to_blob(entry):
588
def symlink_to_blob(symlink_target):
470
589
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')
591
if isinstance(symlink_target, text_type):
592
symlink_target = encode_git_path(symlink_target)
475
593
blob.data = symlink_target
479
597
def mode_is_executable(mode):
480
598
"""Check if mode should be considered executable."""
481
return bool(mode & 0111)
599
return bool(mode & 0o111)
484
602
def mode_kind(mode):
485
603
"""Determine the Bazaar inventory kind based on Unix file mode."""
486
entry_kind = (mode & 0700000) / 0100000
606
entry_kind = (mode & 0o700000) / 0o100000
487
607
if entry_kind == 0:
488
608
return 'directory'
489
609
elif entry_kind == 1:
490
file_kind = (mode & 070000) / 010000
610
file_kind = (mode & 0o70000) / 0o10000
491
611
if file_kind == 0:
493
613
elif file_kind == 2:
525
645
def entry_mode(entry):
526
646
"""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,
647
return object_mode(entry.kind, getattr(entry, 'executable', False))
560
650
def extract_unusual_modes(rev):
added
removed
breezy/git/mapping.py
