17
19
"""Converters, etc for going between Bazaar and Git ids."""
19
from bzrlib import errors, foreign, urlutils
20
from bzrlib.inventory import ROOT_ID
32
from bzrlib import bencode
34
from bzrlib.util import bencode
35
from bzrlib.inventory import (
21
38
from bzrlib.foreign import (
43
from bzrlib.revision import (
46
from bzrlib.plugins.git.hg import (
51
DEFAULT_FILE_MODE = stat.S_IFREG | 0644
27
54
def escape_file_id(file_id):
28
55
return file_id.replace('_', '__').replace(' ', '_s')
31
58
def unescape_file_id(file_id):
32
return file_id.replace("_s", " ").replace("__", "_")
61
while i < len(file_id):
63
ret.append(file_id[i])
65
if file_id[i+1] == '_':
67
elif file_id[i+1] == 's':
70
raise AssertionError("unknown escape character %s" % file_id[i+1])
76
def fix_person_identifier(text):
77
if "<" in text and ">" in text:
79
return "%s <%s>" % (text, text)
82
def warn_escaped(commit, num_escaped):
83
trace.warning("Escaped %d XML-invalid characters in %s. Will be unable "
84
"to regenerate the SHA map.", num_escaped, commit)
87
def warn_unusual_mode(commit, path, mode):
88
trace.mutter("Unusual file mode %o for %s in %s. Storing as revision property. ",
92
def squash_revision(target_repo, rev):
93
"""Remove characters that can't be stored from a revision, if necessary.
95
:param target_repo: Repository in which the revision will be stored
96
:param rev: Revision object, will be modified in-place
98
if not getattr(target_repo._serializer, "squashes_xml_invalid_characters", True):
100
from bzrlib.xml_serializer import escape_invalid_chars
101
rev.message, num_escaped = escape_invalid_chars(rev.message)
103
warn_escaped(rev.foreign_revid, num_escaped)
104
if 'author' in rev.properties:
105
rev.properties['author'], num_escaped = escape_invalid_chars(
106
rev.properties['author'])
108
warn_escaped(rev.foreign_revid, num_escaped)
109
rev.committer, num_escaped = escape_invalid_chars(rev.committer)
111
warn_escaped(rev.foreign_revid, num_escaped)
35
114
class BzrGitMapping(foreign.VcsMapping):
55
137
return bzr_rev_id[len(cls.revid_prefix)+1:], cls()
57
139
def generate_file_id(self, path):
140
# Git paths are just bytestrings
141
# We must just hope they are valid UTF-8..
60
return escape_file_id(path.encode('utf-8'))
144
return escape_file_id(path)
146
def parse_file_id(self, file_id):
147
if file_id == ROOT_ID:
149
return unescape_file_id(file_id)
151
def import_unusual_file_modes(self, rev, unusual_file_modes):
152
if unusual_file_modes:
153
ret = [(name, unusual_file_modes[name])
154
for name in sorted(unusual_file_modes.keys())]
155
rev.properties['file-modes'] = bencode.bencode(ret)
157
def export_unusual_file_modes(self, rev):
159
return dict([(self.generate_file_id(path), mode) for (path, mode) in bencode.bdecode(rev.properties['file-modes'].encode("utf-8"))])
163
def _generate_git_svn_metadata(self, rev, encoding):
165
return "\ngit-svn-id: %s\n" % rev.properties["git-svn-id"].encode(encoding)
169
def _generate_hg_message_tail(self, rev):
173
for name in rev.properties:
174
if name == 'hg:extra:branch':
175
branch = rev.properties['hg:extra:branch']
176
elif name.startswith('hg:extra'):
177
extra[name[len('hg:extra:'):]] = base64.b64decode(rev.properties[name])
178
elif name == 'hg:renames':
179
renames = bencode.bdecode(base64.b64decode(rev.properties['hg:renames']))
180
# TODO: Export other properties as 'bzr:' extras?
181
ret = format_hg_metadata(renames, branch, extra)
182
assert isinstance(ret, str)
185
def _extract_git_svn_metadata(self, rev, message):
186
lines = message.split("\n")
187
if not (lines[-1] == "" and lines[-2].startswith("git-svn-id:")):
189
git_svn_id = lines[-2].split(": ", 1)[1]
190
rev.properties['git-svn-id'] = git_svn_id
191
(url, rev, uuid) = parse_git_svn_id(git_svn_id)
192
# FIXME: Convert this to converted-from property somehow..
193
ret = "\n".join(lines[:-2])
194
assert isinstance(ret, str)
197
def _extract_hg_metadata(self, rev, message):
198
(message, renames, branch, extra) = extract_hg_metadata(message)
199
if branch is not None:
200
rev.properties['hg:extra:branch'] = branch
201
for name, value in extra.iteritems():
202
rev.properties['hg:extra:' + name] = base64.b64encode(value)
204
rev.properties['hg:renames'] = base64.b64encode(bencode.bencode([(new, old) for (old, new) in renames.iteritems()]))
207
def _decode_commit_message(self, rev, message, encoding):
208
return message.decode(encoding)
210
def _encode_commit_message(self, rev, message, encoding):
211
return message.encode(encoding)
213
def export_commit(self, rev, tree_sha, parent_lookup):
214
"""Turn a Bazaar revision in to a Git commit
216
:param tree_sha: Tree sha for the commit
217
:param parent_lookup: Function for looking up the GIT sha equiv of a bzr revision
218
:return dulwich.objects.Commit represent the revision:
220
from dulwich.objects import Commit
222
commit.tree = tree_sha
223
for p in rev.parent_ids:
225
git_p = parent_lookup(p)
228
if git_p is not None:
229
assert len(git_p) == 40, "unexpected length for %r" % git_p
230
commit.parents.append(git_p)
232
encoding = rev.properties['git-explicit-encoding']
234
encoding = rev.properties.get('git-implicit-encoding', 'utf-8')
235
commit.encoding = rev.properties.get('git-explicit-encoding')
236
commit.committer = fix_person_identifier(rev.committer.encode(
238
commit.author = fix_person_identifier(
239
rev.get_apparent_authors()[0].encode(encoding))
240
commit.commit_time = long(rev.timestamp)
241
if 'author-timestamp' in rev.properties:
242
commit.author_time = long(rev.properties['author-timestamp'])
244
commit.author_time = commit.commit_time
245
commit.commit_timezone = rev.timezone
246
if 'author-timezone' in rev.properties:
247
commit.author_timezone = int(rev.properties['author-timezone'])
249
commit.author_timezone = commit.commit_timezone
250
commit.message = self._encode_commit_message(rev, rev.message,
62
254
def import_commit(self, commit):
63
255
"""Convert a git commit to a bzr revision.
68
260
raise AssertionError("Commit object can't be None")
69
261
rev = ForeignRevision(commit.id, self, self.revision_id_foreign_to_bzr(commit.id))
70
262
rev.parent_ids = tuple([self.revision_id_foreign_to_bzr(p) for p in commit.parents])
71
rev.message = commit.message.decode("utf-8", "replace")
72
rev.committer = str(commit.committer).decode("utf-8", "replace")
73
if commit.committer != commit.author:
74
rev.properties['author'] = str(commit.author).decode("utf-8", "replace")
263
def decode_using_encoding(rev, commit, encoding):
264
rev.committer = str(commit.committer).decode(encoding)
265
if commit.committer != commit.author:
266
rev.properties['author'] = str(commit.author).decode(encoding)
267
rev.message = self._decode_commit_message(rev, commit.message,
269
if commit.encoding is not None:
270
rev.properties['git-explicit-encoding'] = commit.encoding
271
decode_using_encoding(rev, commit, commit.encoding)
273
for encoding in ('utf-8', 'latin1'):
275
decode_using_encoding(rev, commit, encoding)
276
except UnicodeDecodeError:
279
if encoding != 'utf-8':
280
rev.properties['git-implicit-encoding'] = encoding
282
if commit.commit_time != commit.author_time:
283
rev.properties['author-timestamp'] = str(commit.author_time)
284
if commit.commit_timezone != commit.author_timezone:
285
rev.properties['author-timezone'] = "%d" % (commit.author_timezone, )
75
286
rev.timestamp = commit.commit_time
287
rev.timezone = commit.commit_timezone
81
292
revid_prefix = 'git-v1'
82
293
experimental = False
296
return self.revid_prefix
85
299
class BzrGitMappingExperimental(BzrGitMappingv1):
86
300
revid_prefix = 'git-experimental'
87
301
experimental = True
303
def _decode_commit_message(self, rev, message, encoding):
304
message = self._extract_hg_metadata(rev, message)
305
message = self._extract_git_svn_metadata(rev, message)
306
return message.decode(encoding)
308
def _encode_commit_message(self, rev, message, encoding):
309
ret = message.encode(encoding)
310
ret += self._generate_hg_message_tail(rev)
311
ret += self._generate_git_svn_metadata(rev, encoding)
314
def import_commit(self, commit):
315
rev = super(BzrGitMappingExperimental, self).import_commit(commit)
316
rev.properties['converted_revision'] = "git %s\n" % commit.id
90
320
class GitMappingRegistry(VcsMappingRegistry):
321
"""Registry with available git mappings."""
92
323
def revision_id_bzr_to_foreign(self, bzr_revid):
324
if bzr_revid == NULL_REVISION:
325
return "0" * 20, None
93
326
if not bzr_revid.startswith("git-"):
94
327
raise errors.InvalidRevisionId(bzr_revid, None)
95
328
(mapping_version, git_sha) = bzr_revid.split(":", 1)
120
369
foreign_git = ForeignGit()
121
default_mapping = BzrGitMappingv1()
124
def inventory_to_tree_and_blobs(repo, mapping, revision_id):
125
from dulwich.objects import Tree, Blob
126
from bzrlib.inventory import InventoryDirectory, InventoryFile
370
default_mapping = mapping_registry.get_default()()
373
def text_to_blob(texts, entry):
374
from dulwich.objects import Blob
375
text = texts.get_record_stream([(entry.file_id, entry.revision)], 'unordered', True).next().get_bytes_as('fulltext')
381
def symlink_to_blob(entry):
382
from dulwich.objects import Blob
384
blob._text = entry.symlink_target
388
def mode_is_executable(mode):
389
"""Check if mode should be considered executable."""
390
return bool(mode & 0111)
394
"""Determine the Bazaar inventory kind based on Unix file mode."""
395
entry_kind = (mode & 0700000) / 0100000
398
elif entry_kind == 1:
399
file_kind = (mode & 070000) / 010000
405
return 'tree-reference'
407
raise AssertionError(
408
"Unknown file kind %d, perms=%o." % (file_kind, mode,))
410
raise AssertionError(
411
"Unknown kind, perms=%r." % (mode,))
414
def object_mode(kind, executable):
415
if kind == 'directory':
417
elif kind == 'symlink':
423
mode = stat.S_IFREG | 0644
427
elif kind == 'tree-reference':
428
from dulwich.objects import S_IFGITLINK
434
def entry_mode(entry):
435
"""Determine the git file mode for an inventory entry."""
436
return object_mode(entry.kind, entry.executable)
439
def directory_to_tree(entry, lookup_ie_sha1, unusual_modes):
440
from dulwich.objects import Tree
442
for name in sorted(entry.children.keys()):
443
ie = entry.children[name]
445
mode = unusual_modes[ie.file_id]
447
mode = entry_mode(ie)
448
if ie.kind == 'directory':
449
subtree = directory_to_tree(ie, lookup_ie_sha1, unusual_modes)
455
hexsha = lookup_ie_sha1(ie)
456
if hexsha is not None:
457
tree.add(mode, name.encode("utf-8"), hexsha)
458
if entry.parent_id is not None and len(tree) == 0:
459
# Only the root can be an empty tree
465
def extract_unusual_modes(rev):
467
foreign_revid, mapping = mapping_registry.parse_revision_id(rev.revision_id)
468
except errors.InvalidRevisionId:
471
return mapping.export_unusual_file_modes(rev)
474
def inventory_to_tree_and_blobs(inventory, texts, mapping, unusual_modes, cur=None):
475
"""Convert a Bazaar tree to a Git tree.
477
:return: Yields tuples with object sha1, object and path
479
from dulwich.objects import Tree
132
inv = repo.get_inventory(revision_id)
134
# stack contains the set of trees that we haven't
486
# stack contains the set of trees that we haven't
135
487
# finished constructing
137
for path, entry in inv.iter_entries():
138
while stack and not path.startswith(cur):
488
for path, entry in inventory.iter_entries():
489
while stack and not path.startswith(osutils.pathjoin(cur, "")):
490
# We've hit a file that's not a child of the previous path
140
sha = tree.sha().hexdigest()
142
t = (stat.S_IFDIR, urlutils.basename(cur).encode('UTF-8'), sha)
493
yield sha, tree, cur.encode("utf-8")
494
mode = unusual_modes.get(cur.encode("utf-8"), stat.S_IFDIR)
495
t = (mode, urlutils.basename(cur).encode('UTF-8'), sha)
143
496
cur, tree = stack.pop()
146
if type(entry) == InventoryDirectory:
499
if entry.kind == "directory":
147
500
stack.append((cur, tree))
151
if type(entry) == InventoryFile:
152
#FIXME: We can make potentially make this Lazy to avoid shaing lots of stuff
153
# and having all these objects in memory at once
155
_, blob._text = repo.iter_files_bytes([(entry.file_id, entry.revision, path)]).next()
156
sha = blob.sha().hexdigest()
157
yield sha, blob, path
504
if entry.kind == "file":
505
blob = text_to_blob(texts, entry)
506
elif entry.kind == "symlink":
507
blob = symlink_to_blob(entry)
509
raise AssertionError("Unknown kind %s" % entry.kind)
511
yield sha, blob, path.encode("utf-8")
159
512
name = urlutils.basename(path).encode("utf-8")
160
mode = stat.S_IFREG | 0644
513
mode = unusual_modes.get(path.encode("utf-8"), entry_mode(entry))
163
514
tree.add(mode, name, sha)
165
516
while len(stack) > 1:
167
sha = tree.sha().hexdigest()
169
t = (stat.S_IFDIR, urlutils.basename(cur).encode('UTF-8'), sha)
519
yield sha, tree, cur.encode("utf-8")
520
mode = unusual_modes.get(cur.encode('utf-8'), stat.S_IFDIR)
521
t = (mode, urlutils.basename(cur).encode('UTF-8'), sha)
170
522
cur, tree = stack.pop()
174
yield tree.sha().hexdigest(), tree, cur
177
def revision_to_commit(rev, tree_sha, parent_lookup):
178
"""Turn a Bazaar revision in to a Git commit
180
:param tree_sha: Tree sha for the commit
181
:param parent_lookup: Function for looking up the GIT sha equiv of a bzr revision
182
:return dulwich.objects.Commit represent the revision:
184
from dulwich.objects import Commit
186
commit._tree = tree_sha
187
for p in rev.parent_ids:
188
git_p = parent_lookup(p)
189
if git_p is not None:
190
commit._parents.append(git_p)
191
commit._message = rev.message.encode("utf-8")
192
commit._committer = rev.committer.encode("utf-8")
193
commit._author = rev.get_apparent_author().encode("utf-8")
194
commit._commit_time = long(rev.timestamp)
526
yield tree.id, tree, cur.encode("utf-8")
529
def parse_git_svn_id(text):
530
(head, uuid) = text.rsplit(" ", 1)
531
(full_url, rev) = head.rsplit("@", 1)
532
return (full_url, int(rev), uuid)