35
from bzrlib.foreign import (
38
from bzrlib.trace import mutter
35
39
from bzrlib.transport import get_transport
37
from bzrlib.plugins.git import (
41
from bzrlib.plugins.git.foreign import (
46
class GitRepository(repository.Repository):
45
from bzrlib.plugins.git.mapping import default_mapping
47
from bzrlib.plugins.git import git
50
class GitTags(object):
52
def __init__(self, tags):
56
return iter(self._tags)
59
class GitRepository(ForeignRepository):
47
60
"""An adapter to git repositories for bzr."""
51
64
def __init__(self, gitdir, lockfiles):
65
# FIXME: This also caches negatives. Need to be more careful
66
# about this once we start writing to git
67
self._parents_provider = graph.CachingParentsProvider(self)
68
ForeignRepository.__init__(self, GitFormat(), gitdir, lockfiles)
52
69
self.base = gitdir.root_transport.base
53
70
self.bzrdir = gitdir
54
self.control_files = lockfiles
55
self._git = git.repo.Repo(gitdir.root_transport.local_abspath("."))
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]
71
self._git = gitdir._git
66
self.signatures = versionedfile.VirtualSignatureTexts(self)
68
self._format = GitFormat()
69
self._fallback_repositories = []
73
self.signatures = versionedfiles.VirtualSignatureTexts(self)
74
self.revisions = versionedfiles.VirtualRevisionTexts(self)
75
self.tags = GitTags(self._git.get_tags())
76
from bzrlib.plugins.git import fetch
77
repository.InterRepository.register_optimiser(fetch.InterFromGitRepository)
71
def _init_cachedb(self):
72
self.cachedb.executescript("""
73
create table if not exists inventory (
75
create unique index if not exists inventory_revid
77
create table if not exists entry_revision (
83
create unique index if not exists entry_revision_revid_path
84
on entry_revision (inventory, path);
79
def all_revision_ids(self):
80
ret = set([revision.NULL_REVISION])
81
if self._git.heads() == []:
83
bzr_heads = [self.get_mapping().revision_id_foreign_to_bzr(h) for h in self._git.heads()]
85
graph = self.get_graph()
86
for rev, parents in graph.iter_ancestry(bzr_heads):
88
90
def is_shared(self):
91
93
def supports_rich_root(self):
94
def get_ancestry(self, revision_id):
95
revision_id = revision.ensure_null(revision_id)
97
if revision_id != revision.NULL_REVISION:
102
cms = self._git.commits(ids.convert_revision_id_bzr_to_git(revision_id), max_count=max_count, skip=skip)
104
ret += [ids.convert_revision_id_git_to_bzr(cm.id) for cm in cms]
96
#def get_revision_delta(self, revision_id):
97
# parent_revid = self.get_revision(revision_id).parent_ids[0]
98
# diff = self._git.diff(ids.convert_revision_id_bzr_to_git(parent_revid),
99
# ids.convert_revision_id_bzr_to_git(revision_id))
101
def _make_parents_provider(self):
102
"""See Repository._make_parents_provider()."""
103
return self._parents_provider
105
def get_parent_map(self, revids):
107
mutter("get_parent_map(%r)", revids)
108
for revision_id in revids:
109
assert isinstance(revision_id, str)
110
if revision_id == revision.NULL_REVISION:
111
parent_map[revision_id] = ()
113
hexsha = self.lookup_git_revid(revision_id, self.get_mapping())
114
commit = self._git.commit(hexsha)
118
parent_map[revision_id] = [self.get_mapping().revision_id_foreign_to_bzr(p) for p in commit.parents]
121
def get_ancestry(self, revision_id, topo_sorted=True):
122
"""See Repository.get_ancestry().
124
if revision_id is None:
125
return self._all_revision_ids()
126
assert isinstance(revision_id, str)
128
graph = self.get_graph()
129
for rev, parents in graph.iter_ancestry([revision_id]):
130
if rev == revision.NULL_REVISION:
136
def _warn_if_deprecated(self):
137
# This class isn't deprecated
107
140
def get_signature_text(self, revision_id):
108
141
raise errors.NoSuchRevision(self, revision_id)
143
def lookup_revision_id(self, revid):
144
"""Lookup a revision id.
146
:param revid: Bazaar revision id.
147
:return: Tuple with git revisionid and mapping.
149
# Yes, this doesn't really work, but good enough as a stub
150
return osutils.sha(rev_id).hexdigest(), self.get_mapping()
110
152
def has_signature_for_revision_id(self, revision_id):
113
def get_parent_map(self, revision_ids):
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])
155
def get_mapping(self):
156
return default_mapping
158
def lookup_git_revid(self, bzr_revid, mapping):
160
return mapping.revision_id_bzr_to_foreign(bzr_revid)
161
except errors.InvalidRevisionId:
162
raise errors.NoSuchRevision(self, bzr_revid)
120
164
def get_revision(self, revision_id):
121
git_commit_id = ids.convert_revision_id_bzr_to_git(revision_id)
165
git_commit_id = self.lookup_git_revid(revision_id, self.get_mapping())
122
166
commit = self._git.commit(git_commit_id)
123
167
# print "fetched revision:", git_commit_id
124
revision = self._parse_rev(commit)
169
raise errors.NoSuchRevision(self, revision_id)
170
revision = self._parse_rev(commit, self.get_mapping())
171
assert revision is not None
127
174
def has_revision(self, revision_id):
129
176
self.get_revision(revision_id)
130
except NoSuchRevision:
177
except errors.NoSuchRevision:
135
def get_revisions(self, revisions):
136
return [self.get_revision(r) for r in revisions]
182
def get_revisions(self, revids):
183
return [self.get_revision(r) for r in revids]
139
def _parse_rev(klass, commit):
186
def _parse_rev(klass, commit, mapping):
140
187
"""Convert a git commit to a bzr revision.
142
189
:return: a `bzrlib.revision.Revision` object.
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 = ""
192
raise AssertionError("Commit object can't be None")
193
rev = ForeignRevision(commit.id, mapping, mapping.revision_id_foreign_to_bzr(commit.id))
194
rev.parent_ids = tuple([mapping.revision_id_foreign_to_bzr(p) for p in commit.parents])
147
195
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)
196
rev.committer = str(commit.committer).decode("utf-8", "replace")
197
if commit.committer != commit.author:
198
rev.properties['author'] = str(commit.author).decode("utf-8", "replace")
199
rev.timestamp = commit.commit_time
166
215
return GitRevisionTree(self, revision_id)
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
175
def _get_blob(self, git_id):
177
return self._blob_cache[git_id]
179
return self._fetch_blob(git_id)
181
def _get_blob_caching(self, git_id):
183
return self._blob_cache[git_id]
185
lines = self._fetch_blob(git_id)
186
self._blob_cache[git_id] = lines
189
def _get_blob_info(self, git_id):
191
return self._blob_info_cache[git_id]
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)
199
217
def get_inventory(self, revision_id):
200
218
assert revision_id != None
201
219
return self.revision_tree(revision_id).inventory
203
def _set_entry_text_info(self, inv, entry, git_id):
204
if entry.kind == 'directory':
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)
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])
218
result = ids.convert_revision_id_git_to_bzr(line[:-1])
219
# print "fetched file revision", line[:-1], path
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()
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))
239
def _all_inventories_in_db(self, revids):
241
result = self.cachedb.execute(
242
"select count(*) from inventory where revid = ?",
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
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:
267
if self._all_inventories_in_db(revision.parent_ids):
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
221
def set_make_working_trees(self, trees):
277
225
def escape_file_id(file_id):
278
226
return file_id.replace('_', '__').replace(' ', '_s')
229
def unescape_file_id(file_id):
230
return file_id.replace("_s", " ").replace("__", "_")
281
233
class GitRevisionTree(revisiontree.RevisionTree):
283
235
def __init__(self, repository, revision_id):
284
236
self._repository = repository
285
237
self.revision_id = revision_id
286
git_id = ids.convert_revision_id_bzr_to_git(revision_id)
238
git_id = repository.lookup_git_revid(revision_id, repository.get_mapping())
287
239
self.tree = repository._git.commit(git_id).tree
288
240
self._inventory = inventory.Inventory(revision_id=revision_id)
289
241
self._inventory.root.revision = revision_id
290
242
self._build_inventory(self.tree, self._inventory.root, "")
292
def get_file_lines(self, file_id):
244
def get_revision_id(self):
245
return self.revision_id
247
def get_file_text(self, file_id):
293
248
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)
249
if entry.kind == 'directory': return ""
250
return self._repository._git.get_blob(entry.text_id).data
300
def _build_inventory(self, tree, ie, path):
252
def _build_inventory(self, tree_id, ie, path):
301
253
assert isinstance(path, str)
302
for b in tree.contents:
303
basename = b.name.decode("utf-8")
254
tree = self._repository._git.get_tree(tree_id)
255
for mode, name, hexsha in tree.entries():
256
basename = name.decode("utf-8")
307
child_path = urlutils.join(path, b.name)
260
child_path = urlutils.join(path, name)
308
261
file_id = escape_file_id(child_path.encode('utf-8'))
262
entry_kind = (mode & 0700000) / 0100000
310
264
child_ie = inventory.InventoryDirectory(file_id, basename, ie.file_id)
311
elif b.mode[0] == '1':
265
elif entry_kind == 1:
266
file_kind = (mode & 070000) / 010000
267
b = self._repository._git.get_blob(hexsha)
313
269
child_ie = inventory.InventoryFile(file_id, basename, ie.file_id)
314
270
child_ie.text_sha1 = osutils.sha_string(b.data)
315
elif b.mode[1] == '2':
316
272
child_ie = inventory.InventoryLink(file_id, basename, ie.file_id)
317
273
child_ie.text_sha1 = osutils.sha_string("")
319
275
raise AssertionError(
320
"Unknown file kind, perms=%r." % (b.mode,))
321
child_ie.text_size = b.size
276
"Unknown file kind, perms=%o." % (mode,))
277
child_ie.text_id = b.id
278
child_ie.text_size = len(b.data)
323
280
raise AssertionError(
324
"Unknown blob kind, perms=%r." % (b.mode,))
325
child_ie.executable = bool(int(b.mode[3:], 8) & 0111)
281
"Unknown blob kind, perms=%r." % (mode,))
282
fs_mode = mode & 0777
283
child_ie.executable = bool(fs_mode & 0111)
326
284
child_ie.revision = self.revision_id
327
assert not basename in ie.children
328
ie.children[basename] = child_ie
330
self._build_inventory(b, child_ie, child_path)
285
self._inventory.add(child_ie)
287
self._build_inventory(hexsha, child_ie, child_path)
333
290
class GitFormat(object):
335
292
supports_tree_reference = False
293
rich_root_data = True
337
295
def get_format_description(self):
338
296
return "Git Repository"
298
def initialize(self, url, shared=False, _internal=False):
299
raise bzr_errors.UninitializableFormat(self)
301
def check_conversion_target(self, target_repo_format):
302
return target_repo_format.rich_root_data
added
removed
repository.py
