17
17
"""An adapter between a Git Repository and a Bazaar Branch"""
23
19
from bzrlib import (
35
from bzrlib.foreign import (
38
from bzrlib.trace import mutter
39
from bzrlib.transport import get_transport
41
from bzrlib.plugins.git.foreign 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):
30
from bzrlib.plugins.git import (
36
class GitRepository(repository.Repository):
59
37
"""An adapter to git repositories for bzr."""
39
# To make bzrlib happy
63
42
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)
44
self.control_files = lockfiles
45
gitdirectory = gitdir.transport.local_abspath('.')
46
self._git = model.GitModel(gitdirectory)
47
self._revision_cache = {}
49
self._entry_revision_cache = {}
51
def _ancestor_revisions(self, revision_ids):
52
if revision_ids is not None:
53
git_revisions = [gitrevid_from_bzr(r) for r in revision_ids]
56
for lines in self._git.ancestor_lines(git_revisions):
57
yield self._parse_rev(lines)
58
# print "fetched ancestors:", git_revisions
68
60
def is_shared(self):
71
63
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:
66
def get_revision_graph(self, revision_id=None):
68
if revision_id is not None:
69
param = [ids.convert_revision_id_bzr_to_git(revision_id)]
72
git_graph = self._git.get_revision_graph(param)
73
# print "fetched revision graph:", param
74
for node, parents in git_graph.iteritems():
75
bzr_node = ids.convert_revision_id_git_to_bzr(node)
76
bzr_parents = [ids.convert_revision_id_git_to_bzr(n)
78
result[bzr_node] = bzr_parents
81
def get_revision_graph_with_ghosts(self, revision_ids=None):
82
graph = deprecated_graph.Graph()
83
if revision_ids is not None:
84
revision_ids = [ids.convert_revision_id_bzr_to_git(r)
85
for r in revision_ids]
86
git_graph = self._git.get_revision_graph(revision_ids)
87
# print "fetched revision graph (ghosts):", revision_ids
88
for node, parents in git_graph.iteritems():
89
bzr_node = ids.convert_revision_id_git_to_bzr(node)
90
bzr_parents = [ids.convert_revision_id_git_to_bzr(n)
93
graph.add_node(bzr_node, bzr_parents)
96
def get_ancestry(self, revision_id):
97
param = [ids.convert_revision_id_bzr_to_git(revision_id)]
98
git_ancestry = self._git.get_ancestry(param)
99
# print "fetched ancestry:", param
101
ids.convert_revision_id_git_to_bzr(git_id)
102
for git_id in git_ancestry]
150
104
def get_signature_text(self, revision_id):
151
105
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)
107
def get_inventory_xml(self, revision_id):
108
"""See Repository.get_inventory_xml()."""
109
return bzrlib.xml5.serializer_v5.write_inventory_to_string(
110
self.get_inventory(revision_id))
112
def get_inventory_sha1(self, revision_id):
113
"""Get the sha1 for the XML representation of an inventory.
115
:param revision_id: Revision id of the inventory for which to return
120
return osutils.sha_string(self.get_inventory_xml(revision_id))
122
def get_revision_xml(self, revision_id):
123
"""Return the XML representation of a revision.
125
:param revision_id: Revision for which to return the XML.
128
return bzrlib.xml5.serializer_v5.write_revision_to_string(
129
self.get_revision(revision_id))
171
131
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)
132
if revision_id in self._revision_cache:
133
return self._revision_cache[revision_id]
134
git_commit_id = ids.convert_revision_id_bzr_to_git(revision_id)
135
raw = self._git.rev_list([git_commit_id], max_count=1, header=True)
177
136
# print "fetched revision:", git_commit_id
178
revision = self.get_mapping().import_commit(commit)
179
assert revision is not None
137
revision = self._parse_rev(raw)
138
self._revision_cache[revision_id] = revision
182
141
def has_revision(self, revision_id):
184
143
self.get_revision(revision_id)
185
except errors.NoSuchRevision:
144
except NoSuchRevision:
190
def get_revisions(self, revids):
191
return [self.get_revision(r) for r in revids]
149
def get_revisions(self, revisions):
150
return [self.get_revision(r) for r in revisions]
153
def _parse_rev(klass, raw):
154
"""Parse a single git revision.
156
* The first line is the git commit id.
157
* Following lines conform to the 'name value' structure, until the
159
* All lines after the first blank line and until the NULL line have 4
160
leading spaces and constitute the commit message.
162
:param raw: sequence of newline-terminated strings, its last item is a
163
single NULL character.
164
:return: a `bzrlib.revision.Revision` object.
169
committer_was_set = False
170
revision_id = ids.convert_revision_id_git_to_bzr(raw[0][:-1])
171
rev = revision.Revision(revision_id)
172
rev.inventory_sha1 = ""
173
assert raw[-1] == '\x00', (
174
"Last item of raw was not a single NULL character.")
175
for line in raw[1:-1]:
177
assert line[:4] == ' ', (
178
"Unexpected line format in commit message: %r" % line)
179
message_lines.append(line[4:])
184
name, value = line[:-1].split(' ', 1)
186
rev.parent_ids.append(
187
ids.convert_revision_id_git_to_bzr(value))
190
author, timestamp, timezone = value.rsplit(' ', 2)
191
rev.properties['author'] = author
192
rev.properties['git-author-timestamp'] = timestamp
193
rev.properties['git-author-timezone'] = timezone
194
if not committer_was_set:
195
rev.committer = author
196
rev.timestamp = float(timestamp)
197
rev.timezone = klass._parse_tz(timezone)
199
if name == 'committer':
200
committer_was_set = True
201
committer, timestamp, timezone = value.rsplit(' ', 2)
202
rev.committer = committer
203
rev.timestamp = float(timestamp)
204
rev.timezone = klass._parse_tz(timezone)
207
rev.properties['git-tree-id'] = value
210
rev.message = ''.join(message_lines)
214
def _parse_tz(klass, tz):
215
"""Parse a timezone specification in the [+|-]HHMM format.
217
:return: the timezone offset in seconds.
220
sign = {'+': +1, '-': -1}[tz[0]]
222
minutes = int(tz[3:])
223
return sign * 60 * (60 * hours + minutes)
193
225
def revision_trees(self, revids):
194
226
for revid in revids:
195
227
yield self.revision_tree(revid)
197
229
def revision_tree(self, revision_id):
198
revision_id = revision.ensure_null(revision_id)
230
return GitRevisionTree(self, revision_id)
232
def _get_blob(self, git_id):
234
return self._blob_cache[git_id]
236
blob = self._git.cat_file('blob', git_id)
237
# print "fetched blob:", git_id
238
self._blob_cache[git_id] = blob
241
def get_inventory(self, revision_id):
242
if revision_id is None:
243
revision_id = revision.NULL_REVISION
200
244
if revision_id == revision.NULL_REVISION:
201
inv = inventory.Inventory(root_id=None)
202
inv.revision_id = revision_id
203
return revisiontree.RevisionTree(self, inv, revision_id)
205
return GitRevisionTree(self, self.get_mapping(), revision_id)
207
def get_inventory(self, revision_id):
208
assert revision_id != None
209
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)
245
return inventory.Inventory(
246
revision_id=revision_id, root_id=None)
248
# First pass at building the inventory. We need this one to get the
249
# git ids, so we do not have to cache the entire tree text. Ideally,
250
# this should be all we need to do.
251
git_commit = ids.convert_revision_id_bzr_to_git(revision_id)
252
git_inventory = self._git.get_inventory(git_commit)
253
# print "fetched inventory:", git_commit
254
inv = self._parse_inventory(revision_id, git_inventory)
256
# Second pass at building the inventory. There we retrieve additional
257
# data that bzrlib requires: text sizes, sha1s, symlink targets and
258
# revisions that introduced inventory entries
259
inv.git_file_data = {}
260
for file_id in sorted(inv.git_ids.iterkeys()):
261
git_id = inv.git_ids[file_id]
263
self._set_entry_text_info(inv, entry, git_id)
264
for file_id in sorted(inv.git_ids.iterkeys()):
265
git_id = inv.git_ids[file_id]
267
path = inv.id2path(file_id)
268
self._set_entry_revision(entry, revision_id, path, git_id)
272
def _parse_inventory(klass, revid, git_inv):
273
# For now, git inventory do not have root ids. It is not clear that we
274
# can reliably support root ids. -- David Allouche 2007-12-28
275
inv = inventory.Inventory(revision_id=revid)
277
for perms, git_kind, git_id, path in git_inv:
280
if git_kind == 'blob':
283
executable = bool(int(perms[-3:], 8) & 0111)
284
elif perms[1] == '2':
287
raise AssertionError(
288
"Unknown blob kind, perms=%r." % (perms,))
289
elif git_kind == 'tree':
292
raise AssertionError(
293
"Unknown git entry kind: %r" % (git_kind,))
294
# XXX: Maybe the file id should be prefixed by file kind, so when
295
# the kind of path changes, the id changes too.
296
# -- David Allouche 2007-12-28.
297
file_id = escape_file_id(path.encode('utf-8'))
298
entry = inv.add_path(path, kind, file_id=file_id)
299
entry.executable = executable
300
inv.git_ids[file_id] = git_id
301
inv.root.revision = revid
304
def _set_entry_text_info(self, inv, entry, git_id):
305
if entry.kind == 'directory':
307
lines = self._get_blob(git_id)
308
entry.text_size = sum(len(line) for line in lines)
309
entry.text_sha1 = osutils.sha_strings(lines)
310
if entry.kind == 'symlink':
311
entry.symlink_target = ''.join(lines)
312
inv.git_file_data[entry.file_id] = lines
314
def _get_file_revision(self, revision_id, path):
315
lines = self._git.rev_list(
316
[ids.convert_revision_id_bzr_to_git(revision_id)],
317
max_count=1, topo_order=True, paths=[path])
319
result = ids.convert_revision_id_git_to_bzr(line[:-1])
320
# print "fetched file revision", line[:-1], path
323
def _set_entry_revision(self, entry, revid, path, git_id):
324
# If a revision is in the cache, we assume it contains entries for the
325
# whole inventory. So if all parent revisions are in the cache, but no
326
# parent entry is present, then the entry revision is the current
327
# revision. That amortizes the number of _get_file_revision calls for
328
# large pulls to a "small number".
329
cached = self._entry_revision_cache.get(revid, {}).get(
330
(path, git_id, entry.executable))
331
if cached is not None:
332
entry.revision = cached
334
revision = self.get_revision(revid)
335
all_parents_in_cache = True
336
for parent_id in revision.parent_ids:
337
if parent_id not in self._entry_revision_cache:
338
all_parents_in_cache = False
340
entry_rev = self._entry_revision_cache[parent_id].get(
341
(path, git_id, entry.executable))
342
if entry_rev is not None:
345
if all_parents_in_cache:
348
entry_rev = self._get_file_revision(revid, path)
349
self._entry_revision_cache.setdefault(
350
revid, {})[(path, git_id, entry.executable)] = entry_rev
351
entry.revision = entry_rev
354
def escape_file_id(file_id):
355
return file_id.replace('_', '__').replace(' ', '_s')
218
357
class GitRevisionTree(revisiontree.RevisionTree):
220
def __init__(self, repository, mapping, revision_id):
359
def __init__(self, repository, revision_id):
360
if revision_id is None:
361
revision_id = revision.NULL_REVISION
362
self._inventory = repository.get_inventory(revision_id)
221
363
self._repository = repository
222
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
231
self._inventory = inventory.Inventory(revision_id=revision_id)
232
self._inventory.root.revision = revision_id
233
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):
364
self._revision_id = revision_id
366
def get_file_lines(self, file_id):
239
367
entry = self._inventory[file_id]
240
if entry.kind == 'directory': return ""
241
return self._repository._git.get_blob(entry.text_id).data
243
def _build_inventory(self, tree_id, ie, path):
244
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")
251
child_path = urlutils.join(path, name)
252
file_id = self.mapping.generate_file_id(child_path)
253
entry_kind = (mode & 0700000) / 0100000
255
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)
260
child_ie = inventory.InventoryFile(file_id, basename, ie.file_id)
261
child_ie.text_sha1 = osutils.sha_string(b.data)
263
child_ie = inventory.InventoryLink(file_id, basename, ie.file_id)
264
child_ie.text_sha1 = osutils.sha_string("")
266
raise AssertionError(
267
"Unknown file kind, perms=%o." % (mode,))
268
child_ie.text_id = b.id
269
child_ie.text_size = len(b.data)
271
raise AssertionError(
272
"Unknown blob kind, perms=%r." % (mode,))
273
fs_mode = mode & 0777
274
child_ie.executable = bool(fs_mode & 0111)
275
child_ie.revision = self.revision_id
276
self._inventory.add(child_ie)
278
self._build_inventory(hexsha, child_ie, child_path)
281
class GitFormat(object):
283
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
368
if entry.kind == 'directory': return []
369
return self._inventory.git_file_data[file_id]
371
obj_id = self._inventory.git_ids[file_id]
372
assert obj_id is not None, (
373
"git_id must not be None: %r" % (self._inventory[file_id],))
374
return self._repository._git.cat_file('blob', obj_id)
added
removed
git_repository.py
