/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

  • Committer: Jelmer Vernooij
  • Date: 2009-05-16 21:13:17 UTC
  • mto: (0.200.527 trunk)
  • mto: This revision was merged to the branch mainline in revision 6960.
  • Revision ID: jelmer@samba.org-20090516211317-a0s14tsrf2qusapf
Fix tests.

Show diffs side-by-side

added added

removed removed

Lines of Context:
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 (
28
27
    trace,
29
28
    urlutils,
30
29
    )
31
 
try:
32
 
    from bzrlib import bencode
33
 
except ImportError:
34
 
    from bzrlib.util import bencode
35
30
from bzrlib.inventory import (
36
31
    ROOT_ID,
37
32
    )
38
33
from bzrlib.foreign import (
39
 
    ForeignVcs,
40
 
    VcsMappingRegistry,
 
34
    ForeignVcs, 
 
35
    VcsMappingRegistry, 
41
36
    ForeignRevision,
42
37
    )
43
 
from bzrlib.revision import (
44
 
    NULL_REVISION,
45
 
    )
46
 
from bzrlib.plugins.git.hg import (
47
 
    format_hg_metadata,
48
 
    extract_hg_metadata,
 
38
from bzrlib.xml_serializer import (
 
39
    escape_invalid_chars,
49
40
    )
50
41
 
51
42
DEFAULT_FILE_MODE = stat.S_IFREG | 0644
85
76
 
86
77
 
87
78
def warn_unusual_mode(commit, path, mode):
88
 
    trace.mutter("Unusual file mode %o for %s in %s. Storing as revision property. ",
89
 
                 mode, path, commit)
90
 
 
91
 
 
92
 
def squash_revision(target_repo, rev):
93
 
    """Remove characters that can't be stored from a revision, if necessary.
94
 
 
95
 
    :param target_repo: Repository in which the revision will be stored
96
 
    :param rev: Revision object, will be modified in-place
97
 
    """
98
 
    if not getattr(target_repo._serializer, "squashes_xml_invalid_characters", True):
99
 
        return
100
 
    from bzrlib.xml_serializer import escape_invalid_chars
101
 
    rev.message, num_escaped = escape_invalid_chars(rev.message)
102
 
    if num_escaped:
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'])
107
 
        if num_escaped:
108
 
            warn_escaped(rev.foreign_revid, num_escaped)
109
 
    rev.committer, num_escaped = escape_invalid_chars(rev.committer)
110
 
    if num_escaped:
111
 
        warn_escaped(rev.foreign_revid, num_escaped)
 
79
    trace.warning("Unusual file mode %o for %s in %s. Will be unable to "
 
80
                  "regenerate the SHA map.", mode, path, commit)
112
81
 
113
82
 
114
83
class BzrGitMapping(foreign.VcsMapping):
145
114
            return ""
146
115
        return unescape_file_id(file_id)
147
116
 
148
 
    def import_unusual_file_modes(self, rev, unusual_file_modes):
149
 
        if unusual_file_modes:
150
 
            ret = [(name, unusual_file_modes[name]) for name in sorted(unusual_file_modes.keys())]
151
 
            rev.properties['file-modes'] = bencode.bencode(ret)
152
 
 
153
 
    def export_unusual_file_modes(self, rev):
154
 
        try:
155
 
            return dict([(self.generate_file_id(path), mode) for (path, mode) in bencode.bdecode(rev.properties['file-modes'].encode("utf-8"))])
156
 
        except KeyError:
157
 
            return {}
158
 
 
159
 
    def _generate_git_svn_metadata(self, rev):
160
 
        try:
161
 
            return "\ngit-svn-id: %s\n" % rev.properties["git-svn-id"].encode("utf-8")
162
 
        except KeyError:
163
 
            return ""
164
 
 
165
 
    def _generate_hg_message_tail(self, rev):
166
 
        extra = {}
167
 
        renames = []
168
 
        branch = 'default'
169
 
        for name in rev.properties:
170
 
            if name == 'hg:extra:branch':
171
 
                branch = rev.properties['hg:extra:branch']
172
 
            elif name.startswith('hg:extra'):
173
 
                extra[name[len('hg:extra:'):]] = base64.b64decode(rev.properties[name])
174
 
            elif name == 'hg:renames':
175
 
                renames = bencode.bdecode(base64.b64decode(rev.properties['hg:renames']))
176
 
            # TODO: Export other properties as 'bzr:' extras?
177
 
        ret = format_hg_metadata(renames, branch, extra)
178
 
        assert isinstance(ret, str)
179
 
        return ret
180
 
 
181
 
    def _extract_git_svn_metadata(self, rev, message):
182
 
        lines = message.split("\n")
183
 
        if not (lines[-1] == "" and lines[-2].startswith("git-svn-id:")):
184
 
            return message
185
 
        git_svn_id = lines[-2].split(": ", 1)[1]
186
 
        rev.properties['git-svn-id'] = git_svn_id
187
 
        (url, rev, uuid) = parse_git_svn_id(git_svn_id)
188
 
        # FIXME: Convert this to converted-from property somehow..
189
 
        ret = "\n".join(lines[:-2])
190
 
        assert isinstance(ret, str)
191
 
        return ret
192
 
 
193
 
    def _extract_hg_metadata(self, rev, message):
194
 
        (message, renames, branch, extra) = extract_hg_metadata(message)
195
 
        if branch is not None:
196
 
            rev.properties['hg:extra:branch'] = branch
197
 
        for name, value in extra.iteritems():
198
 
            rev.properties['hg:extra:' + name] = base64.b64encode(value)
199
 
        if renames:
200
 
            rev.properties['hg:renames'] = base64.b64encode(bencode.bencode([(new, old) for (old, new) in renames.iteritems()]))
201
 
        return message
202
 
 
203
 
    def _decode_commit_message(self, rev, message):
204
 
        return message.decode("utf-8", "replace")
205
 
 
206
 
    def _encode_commit_message(self, rev, message):
207
 
        return message.encode("utf-8")
208
 
 
209
 
    def export_commit(self, rev, tree_sha, parent_lookup):
210
 
        """Turn a Bazaar revision in to a Git commit
211
 
 
212
 
        :param tree_sha: Tree sha for the commit
213
 
        :param parent_lookup: Function for looking up the GIT sha equiv of a bzr revision
214
 
        :return dulwich.objects.Commit represent the revision:
215
 
        """
216
 
        from dulwich.objects import Commit
217
 
        commit = Commit()
218
 
        commit.tree = tree_sha
219
 
        for p in rev.parent_ids:
220
 
            try:
221
 
                git_p = parent_lookup(p)
222
 
            except KeyError:
223
 
                git_p = None
224
 
            if git_p is not None:
225
 
                assert len(git_p) == 40, "unexpected length for %r" % git_p
226
 
                commit.parents.append(git_p)
227
 
        commit.committer = fix_person_identifier(rev.committer.encode("utf-8"))
228
 
        commit.author = fix_person_identifier(rev.get_apparent_authors()[0].encode("utf-8"))
229
 
        commit.commit_time = long(rev.timestamp)
230
 
        if 'author-timestamp' in rev.properties:
231
 
            commit.author_time = long(rev.properties['author-timestamp'])
232
 
        else:
233
 
            commit.author_time = commit.commit_time
234
 
        commit.commit_timezone = rev.timezone
235
 
        if 'author-timezone' in rev.properties:
236
 
            commit.author_timezone = int(rev.properties['author-timezone'])
237
 
        else:
238
 
            commit.author_timezone = commit.commit_timezone
239
 
        commit.message = self._encode_commit_message(rev, rev.message)
240
 
        return commit
241
 
 
242
117
    def import_commit(self, commit):
243
118
        """Convert a git commit to a bzr revision.
244
119
 
248
123
            raise AssertionError("Commit object can't be None")
249
124
        rev = ForeignRevision(commit.id, self, self.revision_id_foreign_to_bzr(commit.id))
250
125
        rev.parent_ids = tuple([self.revision_id_foreign_to_bzr(p) for p in commit.parents])
251
 
        rev.committer = str(commit.committer).decode("utf-8", "replace")
 
126
        rev.message, num_escaped = escape_invalid_chars(commit.message.decode("utf-8", "replace"))
 
127
        if num_escaped:
 
128
            warn_escaped(commit.id, num_escaped)
 
129
        rev.committer, num_escaped = escape_invalid_chars(str(commit.committer).decode("utf-8", "replace"))
 
130
        if num_escaped:
 
131
            warn_escaped(commit.id, num_escaped)
252
132
        if commit.committer != commit.author:
253
 
            rev.properties['author'] = str(commit.author).decode("utf-8", "replace")
 
133
            rev.properties['author'], num_escaped = escape_invalid_chars(str(commit.author).decode("utf-8", "replace"))
 
134
            if num_escaped:
 
135
                warn_escaped(commit.id, num_escaped)
254
136
 
255
137
        if commit.commit_time != commit.author_time:
256
138
            rev.properties['author-timestamp'] = str(commit.author_time)
258
140
            rev.properties['author-timezone'] = "%d" % (commit.author_timezone, )
259
141
        rev.timestamp = commit.commit_time
260
142
        rev.timezone = commit.commit_timezone
261
 
        rev.message = self._decode_commit_message(rev, commit.message)
262
143
        return rev
263
144
 
264
145
 
274
155
    revid_prefix = 'git-experimental'
275
156
    experimental = True
276
157
 
277
 
    def _decode_commit_message(self, rev, message):
278
 
        message = self._extract_hg_metadata(rev, message)
279
 
        message = self._extract_git_svn_metadata(rev, message)
280
 
        return message.decode("utf-8", "replace")
281
 
 
282
 
    def _encode_commit_message(self, rev, message):
283
 
        ret = message.encode("utf-8")
284
 
        ret += self._generate_hg_message_tail(rev)
285
 
        ret += self._generate_git_svn_metadata(rev)
286
 
        return ret
287
 
 
288
 
    def import_commit(self, commit):
289
 
        rev = super(BzrGitMappingExperimental, self).import_commit(commit)
290
 
        rev.properties['converted_revision'] = "git %s\n" % commit.id
291
 
        return rev
292
 
 
293
158
 
294
159
class GitMappingRegistry(VcsMappingRegistry):
295
 
    """Registry with available git mappings."""
296
160
 
297
161
    def revision_id_bzr_to_foreign(self, bzr_revid):
298
 
        if bzr_revid == NULL_REVISION:
299
 
            return "0" * 20, None
300
162
        if not bzr_revid.startswith("git-"):
301
163
            raise errors.InvalidRevisionId(bzr_revid, None)
302
164
        (mapping_version, git_sha) = bzr_revid.split(":", 1)
311
173
                                   "BzrGitMappingv1")
312
174
mapping_registry.register_lazy('git-experimental', "bzrlib.plugins.git.mapping",
313
175
                                   "BzrGitMappingExperimental")
314
 
mapping_registry.set_default('git-v1')
315
176
 
316
177
 
317
178
class ForeignGit(ForeignVcs):
318
179
    """The Git Stupid Content Tracker"""
319
180
 
320
 
    @property
321
 
    def branch_format(self):
322
 
        from bzrlib.plugins.git.branch import GitBranchFormat
323
 
        return GitBranchFormat()
324
 
 
325
 
    @property
326
 
    def repository_format(self):
327
 
        from bzrlib.plugins.git.repository import GitRepositoryFormat
328
 
        return GitRepositoryFormat()
329
 
 
330
181
    def __init__(self):
331
182
        super(ForeignGit, self).__init__(mapping_registry)
332
 
        self.abbreviation = "git"
333
 
 
334
 
    @classmethod
335
 
    def serialize_foreign_revid(self, foreign_revid):
336
 
        return foreign_revid
337
183
 
338
184
    @classmethod
339
185
    def show_foreign_revid(cls, foreign_revid):
341
187
 
342
188
 
343
189
foreign_git = ForeignGit()
344
 
default_mapping = mapping_registry.get_default()()
 
190
default_mapping = BzrGitMappingv1()
345
191
 
346
192
 
347
193
def text_to_blob(texts, entry):
358
204
    blob._text = entry.symlink_target
359
205
    return blob
360
206
 
361
 
 
362
207
def mode_is_executable(mode):
363
 
    """Check if mode should be considered executable."""
364
208
    return bool(mode & 0111)
365
209
 
366
 
 
367
210
def mode_kind(mode):
368
 
    """Determine the Bazaar inventory kind based on Unix file mode."""
369
211
    entry_kind = (mode & 0700000) / 0100000
370
212
    if entry_kind == 0:
371
213
        return 'directory'
385
227
            "Unknown kind, perms=%r." % (mode,))
386
228
 
387
229
 
388
 
def object_mode(kind, executable):
389
 
    if kind == 'directory':
 
230
def entry_mode(entry):
 
231
    if entry.kind == 'directory':
390
232
        return stat.S_IFDIR
391
 
    elif kind == 'symlink':
392
 
        mode = stat.S_IFLNK
393
 
        if executable:
394
 
            mode |= 0111
395
 
        return mode
396
 
    elif kind == 'file':
 
233
    elif entry.kind == 'symlink':
 
234
        return stat.S_IFLNK
 
235
    elif entry.kind == 'file':
397
236
        mode = stat.S_IFREG | 0644
398
 
        if executable:
 
237
        if entry.executable:
399
238
            mode |= 0111
400
239
        return mode
401
 
    elif kind == 'tree-reference':
402
 
        from dulwich.objects import S_IFGITLINK
403
 
        return S_IFGITLINK
404
240
    else:
405
241
        raise AssertionError
406
242
 
407
243
 
408
 
def entry_mode(entry):
409
 
    """Determine the git file mode for an inventory entry."""
410
 
    return object_mode(entry.kind, entry.executable)
411
 
 
412
 
 
413
 
def directory_to_tree(entry, lookup_ie_sha1, unusual_modes):
 
244
def directory_to_tree(entry, lookup_ie_sha1):
414
245
    from dulwich.objects import Tree
415
246
    tree = Tree()
416
247
    for name in sorted(entry.children.keys()):
417
248
        ie = entry.children[name]
418
 
        try:
419
 
            mode = unusual_modes[ie.file_id]
420
 
        except KeyError:
421
 
            mode = entry_mode(ie)
422
 
        hexsha = lookup_ie_sha1(ie)
423
 
        if hexsha is not None:
424
 
            tree.add(mode, name.encode("utf-8"), hexsha)
425
 
    if entry.parent_id is not None and len(tree) == 0:
426
 
        # Only the root can be an empty tree
427
 
        return None
 
249
        tree.add(entry_mode(ie), name.encode("utf-8"), lookup_ie_sha1(ie))
428
250
    tree.serialize()
429
251
    return tree
430
252
 
431
253
 
432
 
def extract_unusual_modes(rev):
433
 
    try:
434
 
        foreign_revid, mapping = mapping_registry.parse_revision_id(rev.revision_id)
435
 
    except errors.InvalidRevisionId:
436
 
        return {}
437
 
    else:
438
 
        return mapping.export_unusual_file_modes(rev)
439
 
 
440
 
 
441
 
def inventory_to_tree_and_blobs(inventory, texts, mapping, unusual_modes, cur=None):
 
254
def inventory_to_tree_and_blobs(inventory, texts, mapping, cur=None):
442
255
    """Convert a Bazaar tree to a Git tree.
443
256
 
444
257
    :return: Yields tuples with object sha1, object and path
450
263
        cur = ""
451
264
    tree = Tree()
452
265
 
453
 
    # stack contains the set of trees that we haven't
 
266
    # stack contains the set of trees that we haven't 
454
267
    # finished constructing
455
268
    for path, entry in inventory.iter_entries():
456
269
        while stack and not path.startswith(osutils.pathjoin(cur, "")):
458
271
            tree.serialize()
459
272
            sha = tree.id
460
273
            yield sha, tree, cur.encode("utf-8")
461
 
            mode = unusual_modes.get(cur.encode("utf-8"), stat.S_IFDIR)
462
 
            t = (mode, urlutils.basename(cur).encode('UTF-8'), sha)
 
274
            t = (stat.S_IFDIR, urlutils.basename(cur).encode('UTF-8'), sha)
463
275
            cur, tree = stack.pop()
464
276
            tree.add(*t)
465
277
 
477
289
            sha = blob.id
478
290
            yield sha, blob, path.encode("utf-8")
479
291
            name = urlutils.basename(path).encode("utf-8")
480
 
            mode = unusual_modes.get(path.encode("utf-8"), entry_mode(entry))
481
 
            tree.add(mode, name, sha)
 
292
            tree.add(entry_mode(entry), name, sha)
482
293
 
483
294
    while len(stack) > 1:
484
295
        tree.serialize()
485
296
        sha = tree.id
486
297
        yield sha, tree, cur.encode("utf-8")
487
 
        mode = unusual_modes.get(cur.encode('utf-8'), stat.S_IFDIR)
488
 
        t = (mode, urlutils.basename(cur).encode('UTF-8'), sha)
 
298
        t = (stat.S_IFDIR, urlutils.basename(cur).encode('UTF-8'), sha)
489
299
        cur, tree = stack.pop()
490
300
        tree.add(*t)
491
301
 
493
303
    yield tree.id, tree, cur.encode("utf-8")
494
304
 
495
305
 
496
 
def parse_git_svn_id(text):
497
 
    (head, uuid) = text.rsplit(" ", 1)
498
 
    (full_url, rev) = head.rsplit("@", 1)
499
 
    return (full_url, int(rev), uuid)
 
306
def revision_to_commit(rev, tree_sha, parent_lookup):
 
307
    """Turn a Bazaar revision in to a Git commit
 
308
 
 
309
    :param tree_sha: Tree sha for the commit
 
310
    :param parent_lookup: Function for looking up the GIT sha equiv of a bzr revision
 
311
    :return dulwich.objects.Commit represent the revision:
 
312
    """
 
313
    from dulwich.objects import Commit
 
314
    commit = Commit()
 
315
    commit.tree = tree_sha
 
316
    for p in rev.parent_ids:
 
317
        git_p = parent_lookup(p)
 
318
        if git_p is not None:
 
319
            assert len(git_p) == 40, "unexpected length for %r" % git_p
 
320
            commit.parents.append(git_p)
 
321
    commit.message = rev.message.encode("utf-8")
 
322
    commit.committer = fix_person_identifier(rev.committer.encode("utf-8"))
 
323
    commit.author = fix_person_identifier(rev.get_apparent_authors()[0].encode("utf-8"))
 
324
    commit.commit_time = long(rev.timestamp)
 
325
    if 'author-timestamp' in rev.properties:
 
326
        commit.author_time = long(rev.properties['author-timestamp'])
 
327
    else:
 
328
        commit.author_time = commit.commit_time
 
329
    commit.commit_timezone = rev.timezone
 
330
    if 'author-timezone' in rev.properties:
 
331
        commit.author_timezone = int(rev.properties['author-timezone'])
 
332
    else:
 
333
        commit.author_timezone = commit.commit_timezone 
 
334
    return commit