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."""
32
from bzrlib import bencode
34
from bzrlib.util import bencode
35
from bzrlib.inventory import (
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
54
def escape_file_id(file_id):
55
return file_id.replace('_', '__').replace(' ', '_s')
58
def unescape_file_id(file_id):
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)
114
class BzrGitMapping(foreign.VcsMapping):
115
"""Class that maps between Git and Bazaar semantics."""
119
super(BzrGitMapping, self).__init__(foreign_git)
121
def __eq__(self, other):
122
return (type(self) == type(other) and
123
self.revid_prefix == other.revid_prefix)
126
def revision_id_foreign_to_bzr(cls, git_rev_id):
127
"""Convert a git revision id handle to a Bazaar revision id."""
128
return "%s:%s" % (cls.revid_prefix, git_rev_id)
131
def revision_id_bzr_to_foreign(cls, bzr_rev_id):
132
"""Convert a Bazaar revision id to a git revision id handle."""
133
if not bzr_rev_id.startswith("%s:" % cls.revid_prefix):
134
raise errors.InvalidRevisionId(bzr_rev_id, cls)
135
return bzr_rev_id[len(cls.revid_prefix)+1:], cls()
137
def generate_file_id(self, path):
138
# Git paths are just bytestrings
139
# We must just hope they are valid UTF-8..
142
return escape_file_id(path)
144
def parse_file_id(self, file_id):
145
if file_id == ROOT_ID:
147
return unescape_file_id(file_id)
149
def import_unusual_file_modes(self, rev, unusual_file_modes):
150
if unusual_file_modes:
151
ret = [(name, unusual_file_modes[name])
152
for name in sorted(unusual_file_modes.keys())]
153
rev.properties['file-modes'] = bencode.bencode(ret)
155
def export_unusual_file_modes(self, rev):
157
return dict([(self.generate_file_id(path), mode) for (path, mode) in bencode.bdecode(rev.properties['file-modes'].encode("utf-8"))])
161
def _generate_git_svn_metadata(self, rev, encoding):
163
return "\ngit-svn-id: %s\n" % rev.properties["git-svn-id"].encode(encoding)
167
def _generate_hg_message_tail(self, rev):
171
for name in rev.properties:
172
if name == 'hg:extra:branch':
173
branch = rev.properties['hg:extra:branch']
174
elif name.startswith('hg:extra'):
175
extra[name[len('hg:extra:'):]] = base64.b64decode(rev.properties[name])
176
elif name == 'hg:renames':
177
renames = bencode.bdecode(base64.b64decode(rev.properties['hg:renames']))
178
# TODO: Export other properties as 'bzr:' extras?
179
ret = format_hg_metadata(renames, branch, extra)
180
assert isinstance(ret, str)
183
def _extract_git_svn_metadata(self, rev, message):
184
lines = message.split("\n")
185
if not (lines[-1] == "" and lines[-2].startswith("git-svn-id:")):
187
git_svn_id = lines[-2].split(": ", 1)[1]
188
rev.properties['git-svn-id'] = git_svn_id
189
(url, rev, uuid) = parse_git_svn_id(git_svn_id)
190
# FIXME: Convert this to converted-from property somehow..
191
ret = "\n".join(lines[:-2])
192
assert isinstance(ret, str)
195
def _extract_hg_metadata(self, rev, message):
196
(message, renames, branch, extra) = extract_hg_metadata(message)
197
if branch is not None:
198
rev.properties['hg:extra:branch'] = branch
199
for name, value in extra.iteritems():
200
rev.properties['hg:extra:' + name] = base64.b64encode(value)
202
rev.properties['hg:renames'] = base64.b64encode(bencode.bencode([(new, old) for (old, new) in renames.iteritems()]))
205
def _decode_commit_message(self, rev, message, encoding):
206
return message.decode(encoding)
208
def _encode_commit_message(self, rev, message, encoding):
209
return message.encode(encoding)
211
def export_commit(self, rev, tree_sha, parent_lookup):
212
"""Turn a Bazaar revision in to a Git commit
214
:param tree_sha: Tree sha for the commit
215
:param parent_lookup: Function for looking up the GIT sha equiv of a bzr revision
216
:return dulwich.objects.Commit represent the revision:
218
from dulwich.objects import Commit
220
commit.tree = tree_sha
221
for p in rev.parent_ids:
223
git_p = parent_lookup(p)
226
if git_p is not None:
227
assert len(git_p) == 40, "unexpected length for %r" % git_p
228
commit.parents.append(git_p)
230
encoding = rev.properties['git-explicit-encoding']
232
encoding = rev.properties.get('git-implicit-encoding', 'utf-8')
233
commit.encoding = rev.properties.get('git-explicit-encoding')
234
commit.committer = fix_person_identifier(rev.committer.encode(
236
commit.author = fix_person_identifier(
237
rev.get_apparent_authors()[0].encode(encoding))
238
commit.commit_time = long(rev.timestamp)
239
if 'author-timestamp' in rev.properties:
240
commit.author_time = long(rev.properties['author-timestamp'])
242
commit.author_time = commit.commit_time
243
commit.commit_timezone = rev.timezone
244
if 'author-timezone' in rev.properties:
245
commit.author_timezone = int(rev.properties['author-timezone'])
247
commit.author_timezone = commit.commit_timezone
248
commit.message = self._encode_commit_message(rev, rev.message,
252
def import_commit(self, commit):
253
"""Convert a git commit to a bzr revision.
255
:return: a `bzrlib.revision.Revision` object.
258
raise AssertionError("Commit object can't be None")
259
rev = ForeignRevision(commit.id, self, self.revision_id_foreign_to_bzr(commit.id))
260
rev.parent_ids = tuple([self.revision_id_foreign_to_bzr(p) for p in commit.parents])
261
def decode_using_encoding(rev, commit, encoding):
262
rev.committer = str(commit.committer).decode(encoding)
263
if commit.committer != commit.author:
264
rev.properties['author'] = str(commit.author).decode(encoding)
265
rev.message = self._decode_commit_message(rev, commit.message,
267
if commit.encoding is not None:
268
rev.properties['git-explicit-encoding'] = commit.encoding
269
decode_using_encoding(rev, commit, commit.encoding)
271
for encoding in ('utf-8', 'latin1'):
273
decode_using_encoding(rev, commit, encoding)
274
except UnicodeDecodeError:
277
if encoding != 'utf-8':
278
rev.properties['git-implicit-encoding'] = encoding
280
if commit.commit_time != commit.author_time:
281
rev.properties['author-timestamp'] = str(commit.author_time)
282
if commit.commit_timezone != commit.author_timezone:
283
rev.properties['author-timezone'] = "%d" % (commit.author_timezone, )
284
rev.timestamp = commit.commit_time
285
rev.timezone = commit.commit_timezone
289
class BzrGitMappingv1(BzrGitMapping):
290
revid_prefix = 'git-v1'
294
return self.revid_prefix
297
class BzrGitMappingExperimental(BzrGitMappingv1):
298
revid_prefix = 'git-experimental'
301
def _decode_commit_message(self, rev, message, encoding):
302
message = self._extract_hg_metadata(rev, message)
303
message = self._extract_git_svn_metadata(rev, message)
304
return message.decode(encoding)
306
def _encode_commit_message(self, rev, message, encoding):
307
ret = message.encode(encoding)
308
ret += self._generate_hg_message_tail(rev)
309
ret += self._generate_git_svn_metadata(rev, encoding)
312
def import_commit(self, commit):
313
rev = super(BzrGitMappingExperimental, self).import_commit(commit)
314
rev.properties['converted_revision'] = "git %s\n" % commit.id
318
class GitMappingRegistry(VcsMappingRegistry):
319
"""Registry with available git mappings."""
321
def revision_id_bzr_to_foreign(self, bzr_revid):
322
if bzr_revid == NULL_REVISION:
323
return "0" * 20, None
324
if not bzr_revid.startswith("git-"):
325
raise errors.InvalidRevisionId(bzr_revid, None)
326
(mapping_version, git_sha) = bzr_revid.split(":", 1)
327
mapping = self.get(mapping_version)
328
return mapping.revision_id_bzr_to_foreign(bzr_revid)
330
parse_revision_id = revision_id_bzr_to_foreign
333
mapping_registry = GitMappingRegistry()
334
mapping_registry.register_lazy('git-v1', "bzrlib.plugins.git.mapping",
336
mapping_registry.register_lazy('git-experimental', "bzrlib.plugins.git.mapping",
337
"BzrGitMappingExperimental")
338
mapping_registry.set_default('git-v1')
341
class ForeignGit(ForeignVcs):
342
"""The Git Stupid Content Tracker"""
345
def branch_format(self):
346
from bzrlib.plugins.git.branch import GitBranchFormat
347
return GitBranchFormat()
350
def repository_format(self):
351
from bzrlib.plugins.git.repository import GitRepositoryFormat
352
return GitRepositoryFormat()
355
super(ForeignGit, self).__init__(mapping_registry)
356
self.abbreviation = "git"
359
def serialize_foreign_revid(self, foreign_revid):
363
def show_foreign_revid(cls, foreign_revid):
364
return { "git commit": foreign_revid }
367
foreign_git = ForeignGit()
368
default_mapping = mapping_registry.get_default()()
371
def text_to_blob(texts, entry):
372
from dulwich.objects import Blob
373
text = texts.get_record_stream([(entry.file_id, entry.revision)], 'unordered', True).next().get_bytes_as('fulltext')
379
def symlink_to_blob(entry):
380
from dulwich.objects import Blob
382
blob._text = entry.symlink_target
386
def mode_is_executable(mode):
387
"""Check if mode should be considered executable."""
388
return bool(mode & 0111)
392
"""Determine the Bazaar inventory kind based on Unix file mode."""
393
entry_kind = (mode & 0700000) / 0100000
396
elif entry_kind == 1:
397
file_kind = (mode & 070000) / 010000
403
return 'tree-reference'
405
raise AssertionError(
406
"Unknown file kind %d, perms=%o." % (file_kind, mode,))
408
raise AssertionError(
409
"Unknown kind, perms=%r." % (mode,))
412
def object_mode(kind, executable):
413
if kind == 'directory':
415
elif kind == 'symlink':
421
mode = stat.S_IFREG | 0644
425
elif kind == 'tree-reference':
426
from dulwich.objects import S_IFGITLINK
432
def entry_mode(entry):
433
"""Determine the git file mode for an inventory entry."""
434
return object_mode(entry.kind, entry.executable)
437
def directory_to_tree(entry, lookup_ie_sha1, unusual_modes):
438
from dulwich.objects import Tree
440
for name in sorted(entry.children.keys()):
441
ie = entry.children[name]
443
mode = unusual_modes[ie.file_id]
445
mode = entry_mode(ie)
446
hexsha = lookup_ie_sha1(ie)
447
if hexsha is not None:
448
tree.add(mode, name.encode("utf-8"), hexsha)
449
if entry.parent_id is not None and len(tree) == 0:
450
# Only the root can be an empty tree
456
def extract_unusual_modes(rev):
458
foreign_revid, mapping = mapping_registry.parse_revision_id(rev.revision_id)
459
except errors.InvalidRevisionId:
462
return mapping.export_unusual_file_modes(rev)
465
def inventory_to_tree_and_blobs(inventory, texts, mapping, unusual_modes, cur=None):
466
"""Convert a Bazaar tree to a Git tree.
468
:return: Yields tuples with object sha1, object and path
470
from dulwich.objects import Tree
477
# stack contains the set of trees that we haven't
478
# finished constructing
479
for path, entry in inventory.iter_entries():
480
while stack and not path.startswith(osutils.pathjoin(cur, "")):
481
# We've hit a file that's not a child of the previous path
484
yield sha, tree, cur.encode("utf-8")
485
mode = unusual_modes.get(cur.encode("utf-8"), stat.S_IFDIR)
486
t = (mode, urlutils.basename(cur).encode('UTF-8'), sha)
487
cur, tree = stack.pop()
490
if entry.kind == "directory":
491
stack.append((cur, tree))
495
if entry.kind == "file":
496
blob = text_to_blob(texts, entry)
497
elif entry.kind == "symlink":
498
blob = symlink_to_blob(entry)
500
raise AssertionError("Unknown kind %s" % entry.kind)
502
yield sha, blob, path.encode("utf-8")
503
name = urlutils.basename(path).encode("utf-8")
504
mode = unusual_modes.get(path.encode("utf-8"), entry_mode(entry))
505
tree.add(mode, name, sha)
507
while len(stack) > 1:
510
yield sha, tree, cur.encode("utf-8")
511
mode = unusual_modes.get(cur.encode('utf-8'), stat.S_IFDIR)
512
t = (mode, urlutils.basename(cur).encode('UTF-8'), sha)
513
cur, tree = stack.pop()
517
yield tree.id, tree, cur.encode("utf-8")
520
def parse_git_svn_id(text):
521
(head, uuid) = text.rsplit(" ", 1)
522
(full_url, rev) = head.rsplit("@", 1)
523
return (full_url, int(rev), uuid)
added
removed
mapping.py
