23
24
from bzrlib import (
35
from bzrlib.foreign import (
38
from bzrlib.trace import mutter
39
34
from bzrlib.transport import get_transport
41
from bzrlib.plugins.git.foreign import (
36
from bzrlib.plugins.git import (
44
from bzrlib.plugins.git.mapping import default_mapping
46
from bzrlib.plugins.git import git
49
class GitTags(object):
51
def __init__(self, tags):
55
return iter(self._tags)
58
class GitRepository(ForeignRepository):
45
class GitRepository(repository.Repository):
59
46
"""An adapter to git repositories for bzr."""
63
50
def __init__(self, gitdir, lockfiles):
64
ForeignRepository.__init__(self, GitFormat(), gitdir, lockfiles)
65
from bzrlib.plugins.git import fetch
66
repository.InterRepository.register_optimiser(fetch.InterGitRepository)
52
self.control_files = lockfiles
53
self._git = git.repo.Repo(gitdir.root_transport.local_abspath("."))
54
self._revision_cache = {}
56
self._blob_info_cache = {}
57
cache_dir = cache.create_cache_dir()
58
cachedir_transport = get_transport(cache_dir)
59
cache_file = os.path.join(cache_dir, 'cache-%s' % ids.NAMESPACE)
60
if not cachedbs.has_key(cache_file):
61
cachedbs[cache_file] = cache.sqlite3.connect(cache_file)
62
self.cachedb = cachedbs[cache_file]
65
self.signatures = None
67
self._format = GitFormat()
68
self._fallback_repositories = []
70
def _init_cachedb(self):
71
self.cachedb.executescript("""
72
create table if not exists inventory (
74
create unique index if not exists inventory_revid
76
create table if not exists entry_revision (
82
create unique index if not exists entry_revision_revid_path
83
on entry_revision (inventory, path);
68
87
def is_shared(self):
71
90
def supports_rich_root(self):
74
def _warn_if_deprecated(self):
75
# This class isn't deprecated
78
def get_mapping(self):
79
return default_mapping
81
def make_working_trees(self):
85
class LocalGitRepository(GitRepository):
87
def __init__(self, gitdir, lockfiles):
88
# FIXME: This also caches negatives. Need to be more careful
89
# about this once we start writing to git
90
self._parents_provider = graph.CachingParentsProvider(self)
91
GitRepository.__init__(self, gitdir, lockfiles)
92
self.base = gitdir.root_transport.base
93
self._git = gitdir._git
95
self.signatures = versionedfiles.VirtualSignatureTexts(self)
96
self.revisions = versionedfiles.VirtualRevisionTexts(self)
97
self.tags = GitTags(self._git.get_tags())
99
def all_revision_ids(self):
100
ret = set([revision.NULL_REVISION])
101
if self._git.heads() == []:
103
bzr_heads = [self.get_mapping().revision_id_foreign_to_bzr(h) for h in self._git.heads()]
105
graph = self.get_graph()
106
for rev, parents in graph.iter_ancestry(bzr_heads):
110
#def get_revision_delta(self, revision_id):
111
# parent_revid = self.get_revision(revision_id).parent_ids[0]
112
# diff = self._git.diff(ids.convert_revision_id_bzr_to_git(parent_revid),
113
# ids.convert_revision_id_bzr_to_git(revision_id))
115
def _make_parents_provider(self):
116
"""See Repository._make_parents_provider()."""
117
return self._parents_provider
119
def get_parent_map(self, revids):
121
mutter("get_parent_map(%r)", revids)
122
for revision_id in revids:
123
assert isinstance(revision_id, str)
124
if revision_id == revision.NULL_REVISION:
125
parent_map[revision_id] = ()
127
hexsha = self.lookup_git_revid(revision_id, self.get_mapping())
128
commit = self._git.commit(hexsha)
132
parent_map[revision_id] = [self.get_mapping().revision_id_foreign_to_bzr(p) for p in commit.parents]
135
def get_ancestry(self, revision_id, topo_sorted=True):
136
"""See Repository.get_ancestry().
138
if revision_id is None:
139
return self._all_revision_ids()
140
assert isinstance(revision_id, str)
142
graph = self.get_graph()
143
for rev, parents in graph.iter_ancestry([revision_id]):
144
if rev == revision.NULL_REVISION:
93
def get_ancestry(self, revision_id):
94
param = [ids.convert_revision_id_bzr_to_git(revision_id)]
95
git_ancestry = self._git.get_ancestry(param)
96
# print "fetched ancestry:", param
98
ids.convert_revision_id_git_to_bzr(git_id)
99
for git_id in git_ancestry]
150
101
def get_signature_text(self, revision_id):
151
102
raise errors.NoSuchRevision(self, revision_id)
153
def lookup_revision_id(self, revid):
154
"""Lookup a revision id.
156
:param revid: Bazaar revision id.
157
:return: Tuple with git revisionid and mapping.
159
# Yes, this doesn't really work, but good enough as a stub
160
return osutils.sha(rev_id).hexdigest(), self.get_mapping()
162
def has_signature_for_revision_id(self, revision_id):
165
def lookup_git_revid(self, bzr_revid, mapping):
167
return mapping.revision_id_bzr_to_foreign(bzr_revid)
168
except errors.InvalidRevisionId:
169
raise errors.NoSuchRevision(self, bzr_revid)
104
def get_parent_map(self, revision_ids):
106
for revid in revision_ids:
107
commit = self._git.commit(ids.convert_revision_id_bzr_to_git(revid))
108
ret[revid] = tuple([ids.convert_revision_id_git_to_bzr(p.id) for p in commit.parents])
171
111
def get_revision(self, revision_id):
172
git_commit_id = self.lookup_git_revid(revision_id, self.get_mapping())
174
commit = self._git.commit(git_commit_id)
176
raise errors.NoSuchRevision(self, revision_id)
112
if revision_id in self._revision_cache:
113
return self._revision_cache[revision_id]
114
git_commit_id = ids.convert_revision_id_bzr_to_git(revision_id)
115
commit = self._git.commit(git_commit_id)
177
116
# print "fetched revision:", git_commit_id
178
revision = self.get_mapping().import_commit(commit)
179
assert revision is not None
117
revision = self._parse_rev(commit)
118
self._revision_cache[revision_id] = revision
182
121
def has_revision(self, revision_id):
184
123
self.get_revision(revision_id)
185
except errors.NoSuchRevision:
124
except NoSuchRevision:
190
def get_revisions(self, revids):
191
return [self.get_revision(r) for r in revids]
129
def get_revisions(self, revisions):
130
return [self.get_revision(r) for r in revisions]
133
def _parse_rev(klass, commit):
134
"""Convert a git commit to a bzr revision.
136
:return: a `bzrlib.revision.Revision` object.
138
rev = revision.Revision(ids.convert_revision_id_git_to_bzr(commit.id))
139
rev.parent_ids = tuple([ids.convert_revision_id_git_to_bzr(p.id) for p in commit.parents])
140
rev.inventory_sha1 = ""
141
rev.message = commit.message
142
rev.committer = str(commit.committer)
143
rev.properties['author'] = str(commit.author)
144
rev.timestamp = time.mktime(commit.committed_date)
193
148
def revision_trees(self, revids):
194
149
for revid in revids:
202
157
inv.revision_id = revision_id
203
158
return revisiontree.RevisionTree(self, inv, revision_id)
205
return GitRevisionTree(self, self.get_mapping(), revision_id)
160
return GitRevisionTree(self, revision_id)
162
def _fetch_blob(self, git_id):
163
lines = self._git.cat_file('blob', git_id)
164
# print "fetched blob:", git_id
165
if self._building_inventory is not None:
166
self._building_inventory.git_file_data[git_id] = lines
169
def _get_blob(self, git_id):
171
return self._blob_cache[git_id]
173
return self._fetch_blob(git_id)
175
def _get_blob_caching(self, git_id):
177
return self._blob_cache[git_id]
179
lines = self._fetch_blob(git_id)
180
self._blob_cache[git_id] = lines
183
def _get_blob_info(self, git_id):
185
return self._blob_info_cache[git_id]
187
lines = self._get_blob(git_id)
188
size = sum(len(line) for line in lines)
189
sha1 = osutils.sha_strings(lines)
190
self._blob_info_cache[git_id] = (size, sha1)
207
193
def get_inventory(self, revision_id):
208
194
assert revision_id != None
209
195
return self.revision_tree(revision_id).inventory
211
def set_make_working_trees(self, trees):
214
def fetch_objects(self, determine_wants, graph_walker, progress=None):
215
return self._git.fetch_objects(determine_wants, graph_walker, progress)
197
def _set_entry_text_info(self, inv, entry, git_id):
198
if entry.kind == 'directory':
200
size, sha1 = self._get_blob_info(git_id)
201
entry.text_size = size
202
entry.text_sha1 = sha1
203
if entry.kind == 'symlink':
204
lines = self._get_blob_caching(git_id)
205
entry.symlink_target = ''.join(lines)
207
def _get_file_revision(self, revision_id, path):
208
lines = self._git.rev_list(
209
[ids.convert_revision_id_bzr_to_git(revision_id)],
210
max_count=1, topo_order=True, paths=[path])
212
result = ids.convert_revision_id_git_to_bzr(line[:-1])
213
# print "fetched file revision", line[:-1], path
216
def _get_entry_revision_from_db(self, revid, path, git_id, executable):
217
result = self.cachedb.execute(
218
"select revision from entry_revision where"
219
" inventory=? and path=? and gitid=? and executable=?",
220
(revid, path, git_id, executable)).fetchone()
226
def _set_entry_revision_in_db(self, revid, path, git_id, executable, revision):
227
self.cachedb.execute(
228
"insert into entry_revision"
229
" (inventory, path, gitid, executable, revision)"
230
" values (?, ?, ?, ?, ?)",
231
(revid, path, git_id, executable, revision))
233
def _all_inventories_in_db(self, revids):
235
result = self.cachedb.execute(
236
"select count(*) from inventory where revid = ?",
242
def _set_entry_revision(self, entry, revid, path, git_id):
243
# If a revision is in the cache, we assume it contains entries for the
244
# whole inventory. So if all parent revisions are in the cache, but no
245
# parent entry is present, then the entry revision is the current
246
# revision. That amortizes the number of _get_file_revision calls for
247
# large pulls to a "small number".
248
entry_rev = self._get_entry_revision_from_db(
249
revid, path, git_id, entry.executable)
250
if entry_rev is not None:
251
entry.revision = entry_rev
254
revision = self.get_revision(revid)
255
for parent_id in revision.parent_ids:
256
entry_rev = self._get_entry_revision_from_db(
257
parent_id, path, git_id, entry.executable)
258
if entry_rev is not None:
261
if self._all_inventories_in_db(revision.parent_ids):
264
entry_rev = self._get_file_revision(revid, path)
265
self._set_entry_revision_in_db(
266
revid, path, git_id, entry.executable, entry_rev)
267
#self.cachedb.commit()
268
entry.revision = entry_rev
271
def escape_file_id(file_id):
272
return file_id.replace('_', '__').replace(' ', '_s')
218
275
class GitRevisionTree(revisiontree.RevisionTree):
220
def __init__(self, repository, mapping, revision_id):
277
def __init__(self, repository, revision_id):
221
278
self._repository = repository
222
279
self.revision_id = revision_id
223
assert isinstance(revision_id, str)
224
self.mapping = mapping
225
git_id = repository.lookup_git_revid(revision_id, self.mapping)
227
commit = repository._git.commit(git_id)
229
raise errors.NoSuchRevision(repository, revision_id)
230
self.tree = commit.tree
280
git_id = ids.convert_revision_id_bzr_to_git(revision_id)
281
self.tree = repository._git.commit(git_id).tree
231
282
self._inventory = inventory.Inventory(revision_id=revision_id)
232
283
self._inventory.root.revision = revision_id
233
284
self._build_inventory(self.tree, self._inventory.root, "")
235
def get_revision_id(self):
236
return self.revision_id
238
def get_file_text(self, file_id):
286
def get_file_lines(self, file_id):
239
287
entry = self._inventory[file_id]
240
if entry.kind == 'directory': return ""
241
return self._repository._git.get_blob(entry.text_id).data
288
if entry.kind == 'directory': return []
289
git_id = self._inventory.git_ids[file_id]
290
if git_id in self._inventory.git_file_data:
291
return self._inventory.git_file_data[git_id]
292
return self._repository._get_blob(git_id)
243
def _build_inventory(self, tree_id, ie, path):
294
def _build_inventory(self, tree, ie, path):
244
295
assert isinstance(path, str)
245
tree = self._repository._git.tree(tree_id)
246
for mode, name, hexsha in tree.entries():
247
basename = name.decode("utf-8")
296
for b in tree.contents:
297
basename = b.name.decode("utf-8")
251
child_path = urlutils.join(path, name)
252
file_id = self.mapping.generate_file_id(child_path)
253
entry_kind = (mode & 0700000) / 0100000
301
child_path = urlutils.join(path, b.name)
302
file_id = escape_file_id(child_path.encode('utf-8'))
255
304
child_ie = inventory.InventoryDirectory(file_id, basename, ie.file_id)
256
elif entry_kind == 1:
257
file_kind = (mode & 070000) / 010000
258
b = self._repository._git.get_blob(hexsha)
305
elif b.mode[0] == '1':
260
307
child_ie = inventory.InventoryFile(file_id, basename, ie.file_id)
261
308
child_ie.text_sha1 = osutils.sha_string(b.data)
309
elif b.mode[1] == '2':
263
310
child_ie = inventory.InventoryLink(file_id, basename, ie.file_id)
264
311
child_ie.text_sha1 = osutils.sha_string("")
266
313
raise AssertionError(
267
"Unknown file kind, perms=%o." % (mode,))
268
child_ie.text_id = b.id
269
child_ie.text_size = len(b.data)
314
"Unknown file kind, perms=%r." % (b.mode,))
315
child_ie.text_size = b.size
271
317
raise AssertionError(
272
"Unknown blob kind, perms=%r." % (mode,))
273
fs_mode = mode & 0777
274
child_ie.executable = bool(fs_mode & 0111)
318
"Unknown blob kind, perms=%r." % (b.mode,))
319
child_ie.executable = bool(int(b.mode[3:], 8) & 0111)
275
320
child_ie.revision = self.revision_id
276
self._inventory.add(child_ie)
278
self._build_inventory(hexsha, child_ie, child_path)
321
assert not basename in ie.children
322
ie.children[basename] = child_ie
324
self._build_inventory(b, child_ie, child_path)
281
327
class GitFormat(object):
283
329
supports_tree_reference = False
284
rich_root_data = True
286
def get_format_description(self):
287
return "Git Repository"
289
def initialize(self, url, shared=False, _internal=False):
290
raise bzr_errors.UninitializableFormat(self)
292
def check_conversion_target(self, target_repo_format):
293
return target_repo_format.rich_root_data
added
removed
git_repository.py
