/brz/remove-bazaar

To get this branch, use:
bzr branch http://gegoxaren.bato24.eu/bzr/brz/remove-bazaar

« back to all changes in this revision

Viewing changes to mapping.py

Update docs.

Show diffs side-by-side

added added

removed removed

Lines of Context:
mapping.py
18
18
 
19
19
"""Converters, etc for going between Bazaar and Git ids."""
20
20
 
21
 
import base64
22
21
import stat
23
22
 
24
23
from bzrlib import (
25
24
    errors,
26
25
    foreign,
27
26
    osutils,
28
 
    trace,
29
27
    urlutils,
30
28
    )
31
 
try:
32
 
    from bzrlib import bencode
33
 
except ImportError:
34
 
    from bzrlib.util import bencode
35
29
from bzrlib.inventory import (
36
30
    ROOT_ID,
37
31
    )
38
32
from bzrlib.foreign import (
39
 
    ForeignVcs,
40
 
    VcsMappingRegistry,
 
33
    ForeignVcs, 
 
34
    VcsMappingRegistry, 
41
35
    ForeignRevision,
42
36
    )
43
 
from bzrlib.revision import (
44
 
    NULL_REVISION,
45
 
    )
46
 
from bzrlib.plugins.git.hg import (
47
 
    format_hg_metadata,
48
 
    extract_hg_metadata,
 
37
from bzrlib.xml_serializer import (
 
38
    escape_invalid_chars,
49
39
    )
50
40
 
51
41
DEFAULT_FILE_MODE = stat.S_IFREG | 0644
67
57
            elif file_id[i+1] == 's':
68
58
                ret.append(" ")
69
59
            else:
70
 
                raise AssertionError("unknown escape character %s" %
71
 
                    file_id[i+1])
 
60
                raise AssertionError("unknown escape character %s" % file_id[i+1])
72
61
            i += 1
73
62
        i += 1
74
63
    return "".join(ret)
80
69
    return "%s <%s>" % (text, text)
81
70
 
82
71
 
83
 
def warn_escaped(commit, num_escaped):
84
 
    trace.warning("Escaped %d XML-invalid characters in %s. Will be unable "
85
 
                  "to regenerate the SHA map.", num_escaped, commit)
86
 
 
87
 
 
88
 
def warn_unusual_mode(commit, path, mode):
89
 
    trace.mutter("Unusual file mode %o for %s in %s. Storing as revision "
90
 
                 "property. ", mode, path, commit)
91
 
 
92
 
 
93
 
def squash_revision(target_repo, rev):
94
 
    """Remove characters that can't be stored from a revision, if necessary.
95
 
 
96
 
    :param target_repo: Repository in which the revision will be stored
97
 
    :param rev: Revision object, will be modified in-place
98
 
    """
99
 
    if not getattr(target_repo._serializer, "squashes_xml_invalid_characters", True):
100
 
        return
101
 
    from bzrlib.xml_serializer import escape_invalid_chars
102
 
    rev.message, num_escaped = escape_invalid_chars(rev.message)
103
 
    if num_escaped:
104
 
        warn_escaped(rev.foreign_revid, num_escaped)
105
 
    if 'author' in rev.properties:
106
 
        rev.properties['author'], num_escaped = escape_invalid_chars(
107
 
            rev.properties['author'])
108
 
        if num_escaped:
109
 
            warn_escaped(rev.foreign_revid, num_escaped)
110
 
    rev.committer, num_escaped = escape_invalid_chars(rev.committer)
111
 
    if num_escaped:
112
 
        warn_escaped(rev.foreign_revid, num_escaped)
113
 
 
114
 
 
115
72
class BzrGitMapping(foreign.VcsMapping):
116
73
    """Class that maps between Git and Bazaar semantics."""
117
74
    experimental = False
120
77
        super(BzrGitMapping, self).__init__(foreign_git)
121
78
 
122
79
    def __eq__(self, other):
123
 
        return (type(self) == type(other) and 
124
 
                self.revid_prefix == other.revid_prefix)
 
80
        return type(self) == type(other) and self.revid_prefix == other.revid_prefix
125
81
 
126
82
    @classmethod
127
83
    def revision_id_foreign_to_bzr(cls, git_rev_id):
128
84
        """Convert a git revision id handle to a Bazaar revision id."""
129
 
        if git_rev_id == "0" * 40:
130
 
            return NULL_REVISION
131
85
        return "%s:%s" % (cls.revid_prefix, git_rev_id)
132
86
 
133
87
    @classmethod
149
103
            return ""
150
104
        return unescape_file_id(file_id)
151
105
 
152
 
    def import_unusual_file_modes(self, rev, unusual_file_modes):
153
 
        if unusual_file_modes:
154
 
            ret = [(name, unusual_file_modes[name])
155
 
                   for name in sorted(unusual_file_modes.keys())]
156
 
            rev.properties['file-modes'] = bencode.bencode(ret)
157
 
 
158
 
    def export_unusual_file_modes(self, rev):
159
 
        try:
160
 
            return dict([(self.generate_file_id(path), mode) for (path, mode) in bencode.bdecode(rev.properties['file-modes'].encode("utf-8"))])
161
 
        except KeyError:
162
 
            return {}
163
 
 
164
 
    def _generate_git_svn_metadata(self, rev, encoding):
165
 
        try:
166
 
            return "\ngit-svn-id: %s\n" % rev.properties["git-svn-id"].encode(
167
 
                encoding)
168
 
        except KeyError:
169
 
            return ""
170
 
 
171
 
    def _generate_hg_message_tail(self, rev):
172
 
        extra = {}
173
 
        renames = []
174
 
        branch = 'default'
175
 
        for name in rev.properties:
176
 
            if name == 'hg:extra:branch':
177
 
                branch = rev.properties['hg:extra:branch']
178
 
            elif name.startswith('hg:extra'):
179
 
                extra[name[len('hg:extra:'):]] = base64.b64decode(
180
 
                    rev.properties[name])
181
 
            elif name == 'hg:renames':
182
 
                renames = bencode.bdecode(base64.b64decode(
183
 
                    rev.properties['hg:renames']))
184
 
            # TODO: Export other properties as 'bzr:' extras?
185
 
        ret = format_hg_metadata(renames, branch, extra)
186
 
        assert isinstance(ret, str)
187
 
        return ret
188
 
 
189
 
    def _extract_git_svn_metadata(self, rev, message):
190
 
        lines = message.split("\n")
191
 
        if not (lines[-1] == "" and lines[-2].startswith("git-svn-id:")):
192
 
            return message
193
 
        git_svn_id = lines[-2].split(": ", 1)[1]
194
 
        rev.properties['git-svn-id'] = git_svn_id
195
 
        (url, rev, uuid) = parse_git_svn_id(git_svn_id)
196
 
        # FIXME: Convert this to converted-from property somehow..
197
 
        ret = "\n".join(lines[:-2])
198
 
        assert isinstance(ret, str)
199
 
        return ret
200
 
 
201
 
    def _extract_hg_metadata(self, rev, message):
202
 
        (message, renames, branch, extra) = extract_hg_metadata(message)
203
 
        if branch is not None:
204
 
            rev.properties['hg:extra:branch'] = branch
205
 
        for name, value in extra.iteritems():
206
 
            rev.properties['hg:extra:' + name] = base64.b64encode(value)
207
 
        if renames:
208
 
            rev.properties['hg:renames'] = base64.b64encode(bencode.bencode(
209
 
                [(new, old) for (old, new) in renames.iteritems()]))
210
 
        return message
211
 
 
212
 
    def _decode_commit_message(self, rev, message, encoding):
213
 
        return message.decode(encoding)
214
 
 
215
 
    def _encode_commit_message(self, rev, message, encoding):
216
 
        return message.encode(encoding)
217
 
 
218
 
    def export_commit(self, rev, tree_sha, parent_lookup):
219
 
        """Turn a Bazaar revision in to a Git commit
220
 
 
221
 
        :param tree_sha: Tree sha for the commit
222
 
        :param parent_lookup: Function for looking up the GIT sha equiv of a
223
 
            bzr revision
224
 
        :return dulwich.objects.Commit represent the revision:
225
 
        """
226
 
        from dulwich.objects import Commit
227
 
        commit = Commit()
228
 
        commit.tree = tree_sha
229
 
        for p in rev.parent_ids:
230
 
            try:
231
 
                git_p = parent_lookup(p)
232
 
            except KeyError:
233
 
                git_p = None
234
 
            if git_p is not None:
235
 
                assert len(git_p) == 40, "unexpected length for %r" % git_p
236
 
                commit.parents.append(git_p)
237
 
        try:
238
 
            encoding = rev.properties['git-explicit-encoding']
239
 
        except KeyError:
240
 
            encoding = rev.properties.get('git-implicit-encoding', 'utf-8')
241
 
        commit.encoding = rev.properties.get('git-explicit-encoding')
242
 
        commit.committer = fix_person_identifier(rev.committer.encode(
243
 
            encoding))
244
 
        commit.author = fix_person_identifier(
245
 
            rev.get_apparent_authors()[0].encode(encoding))
246
 
        commit.commit_time = long(rev.timestamp)
247
 
        if 'author-timestamp' in rev.properties:
248
 
            commit.author_time = long(rev.properties['author-timestamp'])
249
 
        else:
250
 
            commit.author_time = commit.commit_time
251
 
        commit.commit_timezone = rev.timezone
252
 
        if 'author-timezone' in rev.properties:
253
 
            commit.author_timezone = int(rev.properties['author-timezone'])
254
 
        else:
255
 
            commit.author_timezone = commit.commit_timezone
256
 
        commit.message = self._encode_commit_message(rev, rev.message, 
257
 
            encoding)
258
 
        return commit
259
 
 
260
106
    def import_commit(self, commit):
261
107
        """Convert a git commit to a bzr revision.
262
108
 
264
110
        """
265
111
        if commit is None:
266
112
            raise AssertionError("Commit object can't be None")
267
 
        rev = ForeignRevision(commit.id, self,
268
 
                self.revision_id_foreign_to_bzr(commit.id))
 
113
        rev = ForeignRevision(commit.id, self, self.revision_id_foreign_to_bzr(commit.id))
269
114
        rev.parent_ids = tuple([self.revision_id_foreign_to_bzr(p) for p in commit.parents])
270
 
        def decode_using_encoding(rev, commit, encoding):
271
 
            rev.committer = str(commit.committer).decode(encoding)
272
 
            if commit.committer != commit.author:
273
 
                rev.properties['author'] = str(commit.author).decode(encoding)
274
 
            rev.message = self._decode_commit_message(rev, commit.message, 
275
 
                encoding)
276
 
        if commit.encoding is not None:
277
 
            rev.properties['git-explicit-encoding'] = commit.encoding
278
 
            decode_using_encoding(rev, commit, commit.encoding)
279
 
        else:
280
 
            for encoding in ('utf-8', 'latin1'):
281
 
                try:
282
 
                    decode_using_encoding(rev, commit, encoding)
283
 
                except UnicodeDecodeError:
284
 
                    pass
285
 
                else:
286
 
                    if encoding != 'utf-8':
287
 
                        rev.properties['git-implicit-encoding'] = encoding
288
 
                    break
 
115
        rev.message = escape_invalid_chars(commit.message.decode("utf-8", "replace"))[0]
 
116
        rev.committer = escape_invalid_chars(str(commit.committer).decode("utf-8", "replace"))[0]
 
117
        if commit.committer != commit.author:
 
118
            rev.properties['author'] = escape_invalid_chars(str(commit.author).decode("utf-8", "replace"))[0]
 
119
 
289
120
        if commit.commit_time != commit.author_time:
290
121
            rev.properties['author-timestamp'] = str(commit.author_time)
291
122
        if commit.commit_timezone != commit.author_timezone:
292
 
            rev.properties['author-timezone'] = "%d" % commit.author_timezone
 
123
            rev.properties['author-timezone'] = "%d" % (commit.author_timezone, )
293
124
        rev.timestamp = commit.commit_time
294
125
        rev.timezone = commit.commit_timezone
295
126
        return rev
307
138
    revid_prefix = 'git-experimental'
308
139
    experimental = True
309
140
 
310
 
    def _decode_commit_message(self, rev, message, encoding):
311
 
        message = self._extract_hg_metadata(rev, message)
312
 
        message = self._extract_git_svn_metadata(rev, message)
313
 
        return message.decode(encoding)
314
 
 
315
 
    def _encode_commit_message(self, rev, message, encoding):
316
 
        ret = message.encode(encoding)
317
 
        ret += self._generate_hg_message_tail(rev)
318
 
        ret += self._generate_git_svn_metadata(rev, encoding)
319
 
        return ret
320
 
 
321
 
    def import_commit(self, commit):
322
 
        rev = super(BzrGitMappingExperimental, self).import_commit(commit)
323
 
        rev.properties['converted_revision'] = "git %s\n" % commit.id
324
 
        return rev
325
 
 
326
141
 
327
142
class GitMappingRegistry(VcsMappingRegistry):
328
 
    """Registry with available git mappings."""
329
143
 
330
144
    def revision_id_bzr_to_foreign(self, bzr_revid):
331
 
        if bzr_revid == NULL_REVISION:
332
 
            return "0" * 20, None
333
145
        if not bzr_revid.startswith("git-"):
334
146
            raise errors.InvalidRevisionId(bzr_revid, None)
335
147
        (mapping_version, git_sha) = bzr_revid.split(":", 1)
341
153
 
342
154
mapping_registry = GitMappingRegistry()
343
155
mapping_registry.register_lazy('git-v1', "bzrlib.plugins.git.mapping",
344
 
    "BzrGitMappingv1")
345
 
mapping_registry.register_lazy('git-experimental',
346
 
    "bzrlib.plugins.git.mapping", "BzrGitMappingExperimental")
347
 
mapping_registry.set_default('git-v1')
 
156
                                   "BzrGitMappingv1")
 
157
mapping_registry.register_lazy('git-experimental', "bzrlib.plugins.git.mapping",
 
158
                                   "BzrGitMappingExperimental")
348
159
 
349
160
 
350
161
class ForeignGit(ForeignVcs):
351
162
    """The Git Stupid Content Tracker"""
352
163
 
353
 
    @property
354
 
    def branch_format(self):
355
 
        from bzrlib.plugins.git.branch import GitBranchFormat
356
 
        return GitBranchFormat()
357
 
 
358
 
    @property
359
 
    def repository_format(self):
360
 
        from bzrlib.plugins.git.repository import GitRepositoryFormat
361
 
        return GitRepositoryFormat()
362
 
 
363
164
    def __init__(self):
364
165
        super(ForeignGit, self).__init__(mapping_registry)
365
 
        self.abbreviation = "git"
366
 
 
367
 
    @classmethod
368
 
    def serialize_foreign_revid(self, foreign_revid):
369
 
        return foreign_revid
370
166
 
371
167
    @classmethod
372
168
    def show_foreign_revid(cls, foreign_revid):
374
170
 
375
171
 
376
172
foreign_git = ForeignGit()
377
 
default_mapping = mapping_registry.get_default()()
 
173
default_mapping = BzrGitMappingv1()
 
174
 
 
175
 
 
176
def text_to_blob(texts, entry):
 
177
    from dulwich.objects import Blob
 
178
    text = texts.get_record_stream([(entry.file_id, entry.revision)], 'unordered', True).next().get_bytes_as('fulltext')
 
179
    blob = Blob()
 
180
    blob._text = text
 
181
    return blob
378
182
 
379
183
 
380
184
def symlink_to_blob(entry):
381
185
    from dulwich.objects import Blob
382
186
    blob = Blob()
383
 
    symlink_target = entry.symlink_target
384
 
    if type(symlink_target) == unicode:
385
 
        symlink_target = symlink_target.encode('utf-8')
386
 
    blob.data = symlink_target
 
187
    blob._text = entry.symlink_target
387
188
    return blob
388
189
 
389
190
 
390
 
def mode_is_executable(mode):
391
 
    """Check if mode should be considered executable."""
392
 
    return bool(mode & 0111)
393
 
 
394
 
 
395
 
def mode_kind(mode):
396
 
    """Determine the Bazaar inventory kind based on Unix file mode."""
397
 
    entry_kind = (mode & 0700000) / 0100000
398
 
    if entry_kind == 0:
399
 
        return 'directory'
400
 
    elif entry_kind == 1:
401
 
        file_kind = (mode & 070000) / 010000
402
 
        if file_kind == 0:
403
 
            return 'file'
404
 
        elif file_kind == 2:
405
 
            return 'symlink'
406
 
        elif file_kind == 6:
407
 
            return 'tree-reference'
408
 
        else:
409
 
            raise AssertionError(
410
 
                "Unknown file kind %d, perms=%o." % (file_kind, mode,))
411
 
    else:
412
 
        raise AssertionError(
413
 
            "Unknown kind, perms=%r." % (mode,))
414
 
 
415
 
 
416
 
def object_mode(kind, executable):
417
 
    if kind == 'directory':
 
191
def entry_mode(entry):
 
192
    if entry.kind == 'directory':
418
193
        return stat.S_IFDIR
419
 
    elif kind == 'symlink':
420
 
        mode = stat.S_IFLNK
421
 
        if executable:
422
 
            mode |= 0111
423
 
        return mode
424
 
    elif kind == 'file':
 
194
    elif entry.kind == 'symlink':
 
195
        return stat.S_IFLNK
 
196
    elif entry.kind == 'file':
425
197
        mode = stat.S_IFREG | 0644
426
 
        if executable:
 
198
        if entry.executable:
427
199
            mode |= 0111
428
200
        return mode
429
 
    elif kind == 'tree-reference':
430
 
        from dulwich.objects import S_IFGITLINK
431
 
        return S_IFGITLINK
432
201
    else:
433
202
        raise AssertionError
434
203
 
435
204
 
436
 
def entry_mode(entry):
437
 
    """Determine the git file mode for an inventory entry."""
438
 
    return object_mode(entry.kind, entry.executable)
439
 
 
440
 
 
441
 
def directory_to_tree(entry, lookup_ie_sha1, unusual_modes):
 
205
def directory_to_tree(entry, lookup_ie_sha1):
442
206
    from dulwich.objects import Tree
443
207
    tree = Tree()
444
 
    for name, value in entry.children.iteritems():
 
208
    for name in sorted(entry.children.keys()):
445
209
        ie = entry.children[name]
446
 
        try:
447
 
            mode = unusual_modes[ie.file_id]
448
 
        except KeyError:
449
 
            mode = entry_mode(ie)
450
 
        hexsha = lookup_ie_sha1(ie)
451
 
        if hexsha is not None:
452
 
            tree.add(mode, name.encode("utf-8"), hexsha)
453
 
    if entry.parent_id is not None and len(tree) == 0:
454
 
        # Only the root can be an empty tree
455
 
        return None
 
210
        tree.add(entry_mode(ie), name.encode("utf-8"), lookup_ie_sha1(ie))
 
211
    tree.serialize()
456
212
    return tree
457
213
 
458
214
 
459
 
def extract_unusual_modes(rev):
460
 
    try:
461
 
        foreign_revid, mapping = mapping_registry.parse_revision_id(
462
 
            rev.revision_id)
463
 
    except errors.InvalidRevisionId:
464
 
        return {}
465
 
    else:
466
 
        return mapping.export_unusual_file_modes(rev)
467
 
 
468
 
 
469
 
def inventory_to_tree_and_blobs(inventory, texts, mapping, unusual_modes,
470
 
                                cur=None):
 
215
def inventory_to_tree_and_blobs(inventory, texts, mapping, cur=None):
471
216
    """Convert a Bazaar tree to a Git tree.
472
217
 
473
218
    :return: Yields tuples with object sha1, object and path
479
224
        cur = ""
480
225
    tree = Tree()
481
226
 
482
 
    # stack contains the set of trees that we haven't
 
227
    # stack contains the set of trees that we haven't 
483
228
    # finished constructing
484
229
    for path, entry in inventory.iter_entries():
485
230
        while stack and not path.startswith(osutils.pathjoin(cur, "")):
486
231
            # We've hit a file that's not a child of the previous path
 
232
            tree.serialize()
487
233
            sha = tree.id
488
234
            yield sha, tree, cur.encode("utf-8")
489
 
            mode = unusual_modes.get(cur.encode("utf-8"), stat.S_IFDIR)
490
 
            t = (mode, urlutils.basename(cur).encode('UTF-8'), sha)
 
235
            t = (stat.S_IFDIR, urlutils.basename(cur).encode('UTF-8'), sha)
491
236
            cur, tree = stack.pop()
492
237
            tree.add(*t)
493
238
 
497
242
            tree = Tree()
498
243
        else:
499
244
            if entry.kind == "file":
500
 
                from dulwich.objects import Blob
501
 
                stream = texts.get_record_stream(
502
 
                    [(entry.file_id, entry.revision)], 'unordered', True)
503
 
                blob = Blob()
504
 
                blob.chunked = stream.next().get_bytes_as('chunks')
 
245
                blob = text_to_blob(texts, entry)
505
246
            elif entry.kind == "symlink":
506
247
                blob = symlink_to_blob(entry)
507
248
            else:
509
250
            sha = blob.id
510
251
            yield sha, blob, path.encode("utf-8")
511
252
            name = urlutils.basename(path).encode("utf-8")
512
 
            mode = unusual_modes.get(path.encode("utf-8"), entry_mode(entry))
513
 
            tree.add(mode, name, sha)
 
253
            tree.add(entry_mode(entry), name, sha)
514
254
 
515
255
    while len(stack) > 1:
 
256
        tree.serialize()
516
257
        sha = tree.id
517
258
        yield sha, tree, cur.encode("utf-8")
518
 
        mode = unusual_modes.get(cur.encode('utf-8'), stat.S_IFDIR)
519
 
        t = (mode, urlutils.basename(cur).encode('UTF-8'), sha)
 
259
        t = (stat.S_IFDIR, urlutils.basename(cur).encode('UTF-8'), sha)
520
260
        cur, tree = stack.pop()
521
261
        tree.add(*t)
522
262
 
 
263
    tree.serialize()
523
264
    yield tree.id, tree, cur.encode("utf-8")
524
265
 
525
266
 
526
 
def parse_git_svn_id(text):
527
 
    (head, uuid) = text.rsplit(" ", 1)
528
 
    (full_url, rev) = head.rsplit("@", 1)
529
 
    return (full_url, int(rev), uuid)
 
267
def revision_to_commit(rev, tree_sha, parent_lookup):
 
268
    """Turn a Bazaar revision in to a Git commit
 
269
 
 
270
    :param tree_sha: Tree sha for the commit
 
271
    :param parent_lookup: Function for looking up the GIT sha equiv of a bzr revision
 
272
    :return dulwich.objects.Commit represent the revision:
 
273
    """
 
274
    from dulwich.objects import Commit
 
275
    commit = Commit()
 
276
    commit.tree = tree_sha
 
277
    for p in rev.parent_ids:
 
278
        git_p = parent_lookup(p)
 
279
        if git_p is not None:
 
280
            assert len(git_p) == 40, "unexpected length for %r" % git_p
 
281
            commit.parents.append(git_p)
 
282
    commit.message = rev.message.encode("utf-8")
 
283
    commit.committer = fix_person_identifier(rev.committer.encode("utf-8"))
 
284
    commit.author = fix_person_identifier(rev.get_apparent_authors()[0].encode("utf-8"))
 
285
    commit.commit_time = long(rev.timestamp)
 
286
    if 'author-timestamp' in rev.properties:
 
287
        commit.author_time = long(rev.properties['author-timestamp'])
 
288
    else:
 
289
        commit.author_time = commit.commit_time
 
290
    commit.commit_timezone = rev.timezone
 
291
    if 'author-timezone' in rev.properties:
 
292
        commit.author_timezone = int(rev.properties['author-timezone'])
 
293
    else:
 
294
        commit.author_timezone = commit.commit_timezone 
 
295
    return commit