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 repository.py
- browse files at revision 0.200.330
- compare with another revision
- download diff
- download tarball
- view history from revision 0.200.330
- Committer: Jelmer Vernooij
- Date: 2009-04-10 02:39:41 UTC
- mto: (0.312.1 master) (6883.23.1 bundle-git)
- mto: This revision was merged to the branch mainline in revision 6960.
- Revision ID: jelmer@samba.org-20090410023941-1nzhsbes38afdbc7
Add description of git-v1 mapping.
- files added:
- HACKING
- foreign
- notes
- files removed:
- cache.py
- files renamed:
- git_branch.py => branch.py
- git_dir.py => dir.py
- ids.py => mapping.py
- git_repository.py => repository.py
- tests/test_git_branch.py => tests/test_branch.py
- tests/test_git_dir.py => tests/test_dir.py
- tests/test_ids.py => tests/test_mapping.py
- tests/test_git_repository.py => tests/test_repository.py
- files modified:
- .bzrignore
added
removed
repository.py
1
1
# Copyright (C) 2007 Canonical Ltd
2
# Copyright (C) 2008-2009 Jelmer Vernooij <jelmer@samba.org>
2
3
#
3
4
# This program is free software; you can redistribute it and/or modify
4
5
# it under the terms of the GNU General Public License as published by
16
17
17
18
"""An adapter between a Git Repository and a Bazaar Branch"""
18
19
19
import git
20
import os
21
import time
22
23
20
import bzrlib
24
21
from bzrlib import (
25
deprecated_graph,
26
22
errors,
23
graph,
27
24
inventory,
28
25
osutils,
29
26
repository,
30
27
revision,
31
28
revisiontree,
29
ui,
32
30
urlutils,
33
versionedfile,
34
)
35
from bzrlib.transport import get_transport
36
37
from bzrlib.plugins.git import (
38
cache,
39
ids,
40
)
41
42
43
cachedbs = {}
44
45
46
class GitRepository(repository.Repository):
31
)
32
from bzrlib.foreign import (
33
ForeignRepository,
34
)
35
from bzrlib.trace import (
36
mutter,
37
)
38
from bzrlib.transport import (
39
get_transport,
40
)
41
42
from bzrlib.plugins.git.foreign import (
43
versionedfiles,
44
)
45
from bzrlib.plugins.git.mapping import (
46
default_mapping,
47
inventory_to_tree_and_blobs,
48
mapping_registry,
49
revision_to_commit,
50
)
51
from bzrlib.plugins.git.versionedfiles import (
52
GitTexts,
53
)
54
55
import dulwich as git
56
import os
57
import time
58
59
60
class GitTags(object):
61
62
def __init__(self, tags):
63
self._tags = tags
64
65
def __iter__(self):
66
return iter(self._tags)
67
68
69
class GitRepository(ForeignRepository):
47
70
"""An adapter to git repositories for bzr."""
48
71
49
72
_serializer = None
50
73
51
74
def __init__(self, gitdir, lockfiles):
75
ForeignRepository.__init__(self, GitRepositoryFormat(), gitdir,
76
lockfiles)
77
from bzrlib.plugins.git import fetch, push
78
for optimiser in [fetch.InterRemoteGitNonGitRepository,
79
fetch.InterLocalGitNonGitRepository,
80
fetch.InterGitRepository,
81
push.InterToGitRepository]:
82
repository.InterRepository.register_optimiser(optimiser)
83
84
def is_shared(self):
85
return True
86
87
def supports_rich_root(self):
88
return True
89
90
def _warn_if_deprecated(self):
91
# This class isn't deprecated
92
pass
93
94
def get_mapping(self):
95
return default_mapping
96
97
def make_working_trees(self):
98
return True
99
100
101
class LocalGitRepository(GitRepository):
102
"""Git repository on the file system."""
103
104
def __init__(self, gitdir, lockfiles):
105
# FIXME: This also caches negatives. Need to be more careful
106
# about this once we start writing to git
107
self._parents_provider = graph.CachingParentsProvider(self)
108
GitRepository.__init__(self, gitdir, lockfiles)
52
109
self.base = gitdir.root_transport.base
53
self.bzrdir = gitdir
54
self.control_files = lockfiles
55
self._git = git.repo.Repo(gitdir.root_transport.local_abspath("."))
56
self._blob_cache = {}
57
self._blob_info_cache = {}
58
cache_dir = cache.create_cache_dir()
59
cachedir_transport = get_transport(cache_dir)
60
cache_file = os.path.join(cache_dir, 'cache-%s' % ids.NAMESPACE)
61
if not cachedbs.has_key(cache_file):
62
cachedbs[cache_file] = cache.sqlite3.connect(cache_file)
63
self.cachedb = cachedbs[cache_file]
64
self._init_cachedb()
110
self._git = gitdir._git
65
111
self.texts = None
66
self.signatures = versionedfile.VirtualSignatureTexts(self)
67
self.revisions = None
68
self._format = GitFormat()
69
self._fallback_repositories = []
70
71
def _init_cachedb(self):
72
self.cachedb.executescript("""
73
create table if not exists inventory (
74
revid blob);
75
create unique index if not exists inventory_revid
76
on inventory (revid);
77
create table if not exists entry_revision (
78
inventory blob,
79
path blob,
80
gitid blob,
81
executable integer,
82
revision blob);
83
create unique index if not exists entry_revision_revid_path
84
on entry_revision (inventory, path);
85
""")
86
self.cachedb.commit()
87
88
def is_shared(self):
89
return True
90
91
def supports_rich_root(self):
92
return False
93
94
def get_ancestry(self, revision_id):
95
revision_id = revision.ensure_null(revision_id)
96
ret = []
97
if revision_id != revision.NULL_REVISION:
98
skip = 0
99
max_count = 1000
100
cms = None
101
while cms != []:
102
cms = self._git.commits(ids.convert_revision_id_bzr_to_git(revision_id), max_count=max_count, skip=skip)
103
skip += max_count
104
ret += [ids.convert_revision_id_git_to_bzr(cm.id) for cm in cms]
105
return [None] + ret
112
self.signatures = versionedfiles.VirtualSignatureTexts(self)
113
self.revisions = versionedfiles.VirtualRevisionTexts(self)
114
self.inventories = versionedfiles.VirtualInventoryTexts(self)
115
self.texts = GitTexts(self)
116
self.tags = GitTags(self._git.get_tags())
117
118
def all_revision_ids(self):
119
ret = set([revision.NULL_REVISION])
120
heads = self._git.heads()
121
if heads == {}:
122
return ret
123
bzr_heads = [self.get_mapping().revision_id_foreign_to_bzr(h) for h in heads.itervalues()]
124
ret = set(bzr_heads)
125
graph = self.get_graph()
126
for rev, parents in graph.iter_ancestry(bzr_heads):
127
ret.add(rev)
128
return ret
129
130
#def get_revision_delta(self, revision_id):
131
# parent_revid = self.get_revision(revision_id).parent_ids[0]
132
# diff = self._git.diff(ids.convert_revision_id_bzr_to_git(parent_revid),
133
# ids.convert_revision_id_bzr_to_git(revision_id))
134
135
def _make_parents_provider(self):
136
"""See Repository._make_parents_provider()."""
137
return self._parents_provider
138
139
def get_parent_map(self, revids):
140
parent_map = {}
141
mutter("get_parent_map(%r)", revids)
142
for revision_id in revids:
143
assert isinstance(revision_id, str)
144
if revision_id == revision.NULL_REVISION:
145
parent_map[revision_id] = ()
146
continue
147
hexsha, mapping = self.lookup_git_revid(revision_id)
148
commit = self._git.commit(hexsha)
149
if commit is None:
150
continue
151
else:
152
parent_map[revision_id] = [mapping.revision_id_foreign_to_bzr(p) for p in commit.parents]
153
return parent_map
154
155
def get_ancestry(self, revision_id, topo_sorted=True):
156
"""See Repository.get_ancestry().
157
"""
158
if revision_id is None:
159
return [None, revision.NULL_REVISION] + self._all_revision_ids()
160
assert isinstance(revision_id, str)
161
ancestry = []
162
graph = self.get_graph()
163
for rev, parents in graph.iter_ancestry([revision_id]):
164
ancestry.append(rev)
165
ancestry.reverse()
166
return [None] + ancestry
167
168
def import_revision_gist(self, source, revid, parent_lookup):
169
"""Import the gist of a revision into this Git repository.
170
171
"""
172
objects = []
173
rev = source.get_revision(revid)
174
for sha, object, path in inventory_to_tree_and_blobs(source, None, revid):
175
if path == "":
176
tree_sha = sha
177
objects.append((object, path))
178
commit = revision_to_commit(rev, tree_sha, parent_lookup)
179
objects.append((commit, None))
180
self._git.object_store.add_objects(objects)
181
return commit.sha().hexdigest()
182
183
def dfetch(self, source, stop_revision):
184
"""Import the gist of the ancestry of a particular revision."""
185
if stop_revision is None:
186
raise NotImplementedError
187
revidmap = {}
188
gitidmap = {}
189
def parent_lookup(revid):
190
try:
191
return gitidmap[revid]
192
except KeyError:
193
return self.lookup_git_revid(revid)[0]
194
todo = []
195
source.lock_write()
196
try:
197
graph = source.get_graph()
198
ancestry = [x for x in source.get_ancestry(stop_revision) if x is not None]
199
for revid in graph.iter_topo_order(ancestry):
200
if not self.has_revision(revid):
201
todo.append(revid)
202
pb = ui.ui_factory.nested_progress_bar()
203
try:
204
for i, revid in enumerate(todo):
205
pb.update("pushing revisions", i, len(todo))
206
git_commit = self.import_revision_gist(source, revid,
207
parent_lookup)
208
gitidmap[revid] = git_commit
209
git_revid = self.get_mapping().revision_id_foreign_to_bzr(
210
git_commit)
211
revidmap[revid] = git_revid
212
finally:
213
pb.finished()
214
source.fetch(self, revision_id=revidmap[stop_revision])
215
finally:
216
source.unlock()
217
return revidmap
106
218
107
219
def get_signature_text(self, revision_id):
108
220
raise errors.NoSuchRevision(self, revision_id)
109
221
222
def lookup_revision_id(self, revid):
223
"""Lookup a revision id.
224
225
:param revid: Bazaar revision id.
226
:return: Tuple with git revisionid and mapping.
227
"""
228
# Yes, this doesn't really work, but good enough as a stub
229
return osutils.sha(rev_id).hexdigest(), self.get_mapping()
230
110
231
def has_signature_for_revision_id(self, revision_id):
111
232
return False
112
233
113
def get_parent_map(self, revision_ids):
114
ret = {}
115
for revid in revision_ids:
116
commit = self._git.commit(ids.convert_revision_id_bzr_to_git(revid))
117
ret[revid] = tuple([ids.convert_revision_id_git_to_bzr(p.id) for p in commit.parents])
118
return ret
234
def lookup_git_revid(self, bzr_revid):
235
try:
236
return mapping_registry.revision_id_bzr_to_foreign(bzr_revid)
237
except errors.InvalidRevisionId:
238
raise errors.NoSuchRevision(self, bzr_revid)
119
239
120
240
def get_revision(self, revision_id):
121
git_commit_id = ids.convert_revision_id_bzr_to_git(revision_id)
122
commit = self._git.commit(git_commit_id)
241
git_commit_id, mapping = self.lookup_git_revid(revision_id)
242
try:
243
commit = self._git.commit(git_commit_id)
244
except KeyError:
245
raise errors.NoSuchRevision(self, revision_id)
123
246
# print "fetched revision:", git_commit_id
124
revision = self._parse_rev(commit)
247
revision = mapping.import_commit(commit)
248
assert revision is not None
125
249
return revision
126
250
127
251
def has_revision(self, revision_id):
128
252
try:
129
253
self.get_revision(revision_id)
130
except NoSuchRevision:
254
except errors.NoSuchRevision:
131
255
return False
132
256
else:
133
257
return True
134
258
135
def get_revisions(self, revisions):
136
return [self.get_revision(r) for r in revisions]
137
138
@classmethod
139
def _parse_rev(klass, commit):
140
"""Convert a git commit to a bzr revision.
141
142
:return: a `bzrlib.revision.Revision` object.
143
"""
144
rev = revision.Revision(ids.convert_revision_id_git_to_bzr(commit.id))
145
rev.parent_ids = tuple([ids.convert_revision_id_git_to_bzr(p.id) for p in commit.parents])
146
rev.inventory_sha1 = ""
147
rev.message = commit.message.decode("utf-8", "replace")
148
rev.committer = str(commit.committer)
149
rev.properties['author'] = str(commit.author)
150
rev.timestamp = time.mktime(commit.committed_date)
151
rev.timezone = 0
152
return rev
259
def get_revisions(self, revids):
260
return [self.get_revision(r) for r in revids]
153
261
154
262
def revision_trees(self, revids):
155
263
for revid in revids:
157
265
158
266
def revision_tree(self, revision_id):
159
267
revision_id = revision.ensure_null(revision_id)
160
161
268
if revision_id == revision.NULL_REVISION:
162
269
inv = inventory.Inventory(root_id=None)
163
270
inv.revision_id = revision_id
164
271
return revisiontree.RevisionTree(self, inv, revision_id)
165
166
272
return GitRevisionTree(self, revision_id)
167
273
168
def _fetch_blob(self, git_id):
169
lines = self._git.cat_file('blob', git_id)
170
# print "fetched blob:", git_id
171
if self._building_inventory is not None:
172
self._building_inventory.git_file_data[git_id] = lines
173
return lines
174
175
def _get_blob(self, git_id):
176
try:
177
return self._blob_cache[git_id]
178
except KeyError:
179
return self._fetch_blob(git_id)
180
181
def _get_blob_caching(self, git_id):
182
try:
183
return self._blob_cache[git_id]
184
except KeyError:
185
lines = self._fetch_blob(git_id)
186
self._blob_cache[git_id] = lines
187
return lines
188
189
def _get_blob_info(self, git_id):
190
try:
191
return self._blob_info_cache[git_id]
192
except KeyError:
193
lines = self._get_blob(git_id)
194
size = sum(len(line) for line in lines)
195
sha1 = osutils.sha_strings(lines)
196
self._blob_info_cache[git_id] = (size, sha1)
197
return size, sha1
198
199
274
def get_inventory(self, revision_id):
200
275
assert revision_id != None
201
276
return self.revision_tree(revision_id).inventory
202
277
203
def _set_entry_text_info(self, inv, entry, git_id):
204
if entry.kind == 'directory':
205
return
206
size, sha1 = self._get_blob_info(git_id)
207
entry.text_size = size
208
entry.text_sha1 = sha1
209
if entry.kind == 'symlink':
210
lines = self._get_blob_caching(git_id)
211
entry.symlink_target = ''.join(lines)
212
213
def _get_file_revision(self, revision_id, path):
214
lines = self._git.rev_list(
215
[ids.convert_revision_id_bzr_to_git(revision_id)],
216
max_count=1, topo_order=True, paths=[path])
217
[line] = lines
218
result = ids.convert_revision_id_git_to_bzr(line[:-1])
219
# print "fetched file revision", line[:-1], path
220
return result
221
222
def _get_entry_revision_from_db(self, revid, path, git_id, executable):
223
result = self.cachedb.execute(
224
"select revision from entry_revision where"
225
" inventory=? and path=? and gitid=? and executable=?",
226
(revid, path, git_id, executable)).fetchone()
227
if result is None:
228
return None
229
[revision] = result
230
return revision
231
232
def _set_entry_revision_in_db(self, revid, path, git_id, executable, revision):
233
self.cachedb.execute(
234
"insert into entry_revision"
235
" (inventory, path, gitid, executable, revision)"
236
" values (?, ?, ?, ?, ?)",
237
(revid, path, git_id, executable, revision))
238
239
def _all_inventories_in_db(self, revids):
240
for revid in revids:
241
result = self.cachedb.execute(
242
"select count(*) from inventory where revid = ?",
243
(revid,)).fetchone()
244
if result is None:
245
return False
246
return True
247
248
def _set_entry_revision(self, entry, revid, path, git_id):
249
# If a revision is in the cache, we assume it contains entries for the
250
# whole inventory. So if all parent revisions are in the cache, but no
251
# parent entry is present, then the entry revision is the current
252
# revision. That amortizes the number of _get_file_revision calls for
253
# large pulls to a "small number".
254
entry_rev = self._get_entry_revision_from_db(
255
revid, path, git_id, entry.executable)
256
if entry_rev is not None:
257
entry.revision = entry_rev
258
return
259
260
revision = self.get_revision(revid)
261
for parent_id in revision.parent_ids:
262
entry_rev = self._get_entry_revision_from_db(
263
parent_id, path, git_id, entry.executable)
264
if entry_rev is not None:
265
break
266
else:
267
if self._all_inventories_in_db(revision.parent_ids):
268
entry_rev = revid
269
else:
270
entry_rev = self._get_file_revision(revid, path)
271
self._set_entry_revision_in_db(
272
revid, path, git_id, entry.executable, entry_rev)
273
#self.cachedb.commit()
274
entry.revision = entry_rev
275
276
277
def escape_file_id(file_id):
278
return file_id.replace('_', '__').replace(' ', '_s')
278
def set_make_working_trees(self, trees):
279
pass
280
281
def fetch_objects(self, determine_wants, graph_walker, resolve_ext_ref,
282
progress=None):
283
return self._git.fetch_objects(determine_wants, graph_walker, progress)
279
284
280
285
281
286
class GitRevisionTree(revisiontree.RevisionTree):
282
287
283
288
def __init__(self, repository, revision_id):
284
289
self._repository = repository
285
self.revision_id = revision_id
286
git_id = ids.convert_revision_id_bzr_to_git(revision_id)
287
self.tree = repository._git.commit(git_id).tree
290
self._revision_id = revision_id
291
assert isinstance(revision_id, str)
292
git_id, self.mapping = repository.lookup_git_revid(revision_id)
293
try:
294
commit = repository._git.commit(git_id)
295
except KeyError, r:
296
raise errors.NoSuchRevision(repository, revision_id)
297
self.tree = commit.tree
288
298
self._inventory = inventory.Inventory(revision_id=revision_id)
289
299
self._inventory.root.revision = revision_id
290
300
self._build_inventory(self.tree, self._inventory.root, "")
291
301
292
def get_file_lines(self, file_id):
302
def get_revision_id(self):
303
return self._revision_id
304
305
def get_file_text(self, file_id):
293
306
entry = self._inventory[file_id]
294
if entry.kind == 'directory': return []
295
git_id = self._inventory.git_ids[file_id]
296
if git_id in self._inventory.git_file_data:
297
return self._inventory.git_file_data[git_id]
298
return self._repository._get_blob(git_id)
307
if entry.kind == 'directory': return ""
308
return self._repository._git.get_blob(entry.text_id).data
299
309
300
def _build_inventory(self, tree, ie, path):
310
def _build_inventory(self, tree_id, ie, path):
301
311
assert isinstance(path, str)
302
for b in tree.contents:
303
basename = b.name.decode("utf-8")
312
tree = self._repository._git.tree(tree_id)
313
for mode, name, hexsha in tree.entries():
314
basename = name.decode("utf-8")
304
315
if path == "":
305
child_path = b.name
316
child_path = name
306
317
else:
307
child_path = urlutils.join(path, b.name)
308
file_id = escape_file_id(child_path.encode('utf-8'))
309
if b.mode[0] == '0':
318
child_path = urlutils.join(path, name)
319
file_id = self.mapping.generate_file_id(child_path)
320
entry_kind = (mode & 0700000) / 0100000
321
if entry_kind == 0:
310
322
child_ie = inventory.InventoryDirectory(file_id, basename, ie.file_id)
311
elif b.mode[0] == '1':
312
if b.mode[1] == '0':
323
elif entry_kind == 1:
324
file_kind = (mode & 070000) / 010000
325
b = self._repository._git.get_blob(hexsha)
326
if file_kind == 0:
313
327
child_ie = inventory.InventoryFile(file_id, basename, ie.file_id)
314
328
child_ie.text_sha1 = osutils.sha_string(b.data)
315
elif b.mode[1] == '2':
329
elif file_kind == 2:
316
330
child_ie = inventory.InventoryLink(file_id, basename, ie.file_id)
331
child_ie.symlink_target = b.data
317
332
child_ie.text_sha1 = osutils.sha_string("")
318
333
else:
319
334
raise AssertionError(
320
"Unknown file kind, perms=%r." % (b.mode,))
321
child_ie.text_size = b.size
335
"Unknown file kind, perms=%o." % (mode,))
336
child_ie.text_id = b.id
337
child_ie.text_size = len(b.data)
322
338
else:
323
339
raise AssertionError(
324
"Unknown blob kind, perms=%r." % (b.mode,))
325
child_ie.executable = bool(int(b.mode[3:], 8) & 0111)
326
child_ie.revision = self.revision_id
327
assert not basename in ie.children
328
ie.children[basename] = child_ie
329
if b.mode[0] == '0':
330
self._build_inventory(b, child_ie, child_path)
331
332
333
class GitFormat(object):
340
"Unknown blob kind, perms=%r." % (mode,))
341
fs_mode = mode & 0777
342
child_ie.executable = bool(fs_mode & 0111)
343
# TODO: This should be set to the revision id in which
344
# child_ie was last changed instead.
345
child_ie.revision = self._revision_id
346
self._inventory.add(child_ie)
347
if entry_kind == 0:
348
self._build_inventory(hexsha, child_ie, child_path)
349
350
351
class GitRepositoryFormat(repository.RepositoryFormat):
334
352
335
353
supports_tree_reference = False
354
rich_root_data = True
336
355
337
356
def get_format_description(self):
338
357
return "Git Repository"
358
359
def initialize(self, url, shared=False, _internal=False):
360
raise bzr_errors.UninitializableFormat(self)
361
362
def check_conversion_target(self, target_repo_format):
363
return target_repo_format.rich_root_data
Loggerhead is a web-based interface for Breezy
Version: 2.0.2.dev0