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
if git_rev_id == "0" * 40:
130
return "%s:%s" % (cls.revid_prefix, git_rev_id)
133
def revision_id_bzr_to_foreign(cls, bzr_rev_id):
134
"""Convert a Bazaar revision id to a git revision id handle."""
135
if not bzr_rev_id.startswith("%s:" % cls.revid_prefix):
136
raise errors.InvalidRevisionId(bzr_rev_id, cls)
137
return bzr_rev_id[len(cls.revid_prefix)+1:], cls()
139
def generate_file_id(self, path):
140
# Git paths are just bytestrings
141
# We must just hope they are valid 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,
254
def import_commit(self, commit):
255
"""Convert a git commit to a bzr revision.
257
:return: a `bzrlib.revision.Revision` object.
260
raise AssertionError("Commit object can't be None")
261
rev = ForeignRevision(commit.id, self, self.revision_id_foreign_to_bzr(commit.id))
262
rev.parent_ids = tuple([self.revision_id_foreign_to_bzr(p) for p in commit.parents])
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, )
286
rev.timestamp = commit.commit_time
287
rev.timezone = commit.commit_timezone
291
class BzrGitMappingv1(BzrGitMapping):
292
revid_prefix = 'git-v1'
296
return self.revid_prefix
299
class BzrGitMappingExperimental(BzrGitMappingv1):
300
revid_prefix = 'git-experimental'
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
320
class GitMappingRegistry(VcsMappingRegistry):
321
"""Registry with available git mappings."""
323
def revision_id_bzr_to_foreign(self, bzr_revid):
324
if bzr_revid == NULL_REVISION:
325
return "0" * 20, None
326
if not bzr_revid.startswith("git-"):
327
raise errors.InvalidRevisionId(bzr_revid, None)
328
(mapping_version, git_sha) = bzr_revid.split(":", 1)
329
mapping = self.get(mapping_version)
330
return mapping.revision_id_bzr_to_foreign(bzr_revid)
332
parse_revision_id = revision_id_bzr_to_foreign
335
mapping_registry = GitMappingRegistry()
336
mapping_registry.register_lazy('git-v1', "bzrlib.plugins.git.mapping",
338
mapping_registry.register_lazy('git-experimental', "bzrlib.plugins.git.mapping",
339
"BzrGitMappingExperimental")
340
mapping_registry.set_default('git-v1')
343
class ForeignGit(ForeignVcs):
344
"""The Git Stupid Content Tracker"""
347
def branch_format(self):
348
from bzrlib.plugins.git.branch import GitBranchFormat
349
return GitBranchFormat()
352
def repository_format(self):
353
from bzrlib.plugins.git.repository import GitRepositoryFormat
354
return GitRepositoryFormat()
357
super(ForeignGit, self).__init__(mapping_registry)
358
self.abbreviation = "git"
361
def serialize_foreign_revid(self, foreign_revid):
365
def show_foreign_revid(cls, foreign_revid):
366
return { "git commit": foreign_revid }
369
foreign_git = ForeignGit()
370
default_mapping = mapping_registry.get_default()()
373
def symlink_to_blob(entry):
374
from dulwich.objects import Blob
376
symlink_target = entry.symlink_target
377
if type(symlink_target) == unicode:
378
symlink_target = symlink_target.encode('utf-8')
379
blob.data = symlink_target
383
def mode_is_executable(mode):
384
"""Check if mode should be considered executable."""
385
return bool(mode & 0111)
389
"""Determine the Bazaar inventory kind based on Unix file mode."""
390
entry_kind = (mode & 0700000) / 0100000
393
elif entry_kind == 1:
394
file_kind = (mode & 070000) / 010000
400
return 'tree-reference'
402
raise AssertionError(
403
"Unknown file kind %d, perms=%o." % (file_kind, mode,))
405
raise AssertionError(
406
"Unknown kind, perms=%r." % (mode,))
409
def object_mode(kind, executable):
410
if kind == 'directory':
412
elif kind == 'symlink':
418
mode = stat.S_IFREG | 0644
422
elif kind == 'tree-reference':
423
from dulwich.objects import S_IFGITLINK
429
def entry_mode(entry):
430
"""Determine the git file mode for an inventory entry."""
431
return object_mode(entry.kind, entry.executable)
434
def directory_to_tree(entry, lookup_ie_sha1, unusual_modes):
435
from dulwich.objects import Tree
437
for name, value in entry.children.iteritems():
438
ie = entry.children[name]
440
mode = unusual_modes[ie.file_id]
442
mode = entry_mode(ie)
443
hexsha = lookup_ie_sha1(ie)
444
if hexsha is not None:
445
tree.add(mode, name.encode("utf-8"), hexsha)
446
if entry.parent_id is not None and len(tree) == 0:
447
# Only the root can be an empty tree
452
def extract_unusual_modes(rev):
454
foreign_revid, mapping = mapping_registry.parse_revision_id(rev.revision_id)
455
except errors.InvalidRevisionId:
458
return mapping.export_unusual_file_modes(rev)
461
def inventory_to_tree_and_blobs(inventory, texts, mapping, unusual_modes, cur=None):
462
"""Convert a Bazaar tree to a Git tree.
464
:return: Yields tuples with object sha1, object and path
466
from dulwich.objects import Tree
473
# stack contains the set of trees that we haven't
474
# finished constructing
475
for path, entry in inventory.iter_entries():
476
while stack and not path.startswith(osutils.pathjoin(cur, "")):
477
# We've hit a file that's not a child of the previous path
479
yield sha, tree, cur.encode("utf-8")
480
mode = unusual_modes.get(cur.encode("utf-8"), stat.S_IFDIR)
481
t = (mode, urlutils.basename(cur).encode('UTF-8'), sha)
482
cur, tree = stack.pop()
485
if entry.kind == "directory":
486
stack.append((cur, tree))
490
if entry.kind == "file":
491
from dulwich.objects import Blob
492
text = texts.get_record_stream([(entry.file_id, entry.revision)], 'unordered', True).next().get_bytes_as('fulltext')
495
elif entry.kind == "symlink":
496
blob = symlink_to_blob(entry)
498
raise AssertionError("Unknown kind %s" % entry.kind)
500
yield sha, blob, path.encode("utf-8")
501
name = urlutils.basename(path).encode("utf-8")
502
mode = unusual_modes.get(path.encode("utf-8"), entry_mode(entry))
503
tree.add(mode, name, sha)
505
while len(stack) > 1:
507
yield sha, tree, cur.encode("utf-8")
508
mode = unusual_modes.get(cur.encode('utf-8'), stat.S_IFDIR)
509
t = (mode, urlutils.basename(cur).encode('UTF-8'), sha)
510
cur, tree = stack.pop()
513
yield tree.id, tree, cur.encode("utf-8")
516
def parse_git_svn_id(text):
517
(head, uuid) = text.rsplit(" ", 1)
518
(full_url, rev) = head.rsplit("@", 1)
519
return (full_url, int(rev), uuid)
added
removed
mapping.py
