14
14
# along with this program; if not, write to the Free Software
15
15
# Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
17
from bzrlib import osutils, ui, urlutils
18
from bzrlib.errors import InvalidRevisionId
19
from bzrlib.inventory import Inventory
20
from bzrlib.repository import InterRepository
21
from bzrlib.trace import info
22
from bzrlib.tsort import topo_sort
24
from bzrlib.plugins.git import git
17
from cStringIO import (
21
from dulwich.objects import (
26
from dulwich.object_store import (
38
from bzrlib.errors import (
43
from bzrlib.inventory import (
49
from bzrlib.lru_cache import (
52
from bzrlib.repository import (
55
from bzrlib.revision import (
58
from bzrlib.tsort import (
61
from bzrlib.versionedfile import (
62
FulltextContentFactory,
65
from bzrlib.plugins.git.mapping import (
67
inventory_to_tree_and_blobs,
73
from bzrlib.plugins.git.object_store import (
76
from bzrlib.plugins.git.remote import (
25
79
from bzrlib.plugins.git.repository import (
30
from bzrlib.plugins.git.remote import RemoteGitRepository
32
from dulwich.objects import Commit
34
from cStringIO import StringIO
37
class BzrFetchGraphWalker(object):
39
def __init__(self, repository, mapping):
40
self.repository = repository
41
self.mapping = mapping
43
self.heads = set(repository.all_revision_ids())
47
revid = self.mapping.revision_id_foreign_to_bzr(sha)
50
def remove(self, revid):
53
self.heads.remove(revid)
54
if revid in self.parents:
55
for p in self.parents[revid]:
60
ret = self.heads.pop()
61
ps = self.repository.get_parent_map([ret])[ret]
62
self.parents[ret] = ps
63
self.heads.update([p for p in ps if not p in self.done])
66
return self.mapping.revision_id_bzr_to_foreign(ret)
67
except InvalidRevisionId:
72
def import_git_blob(repo, mapping, path, blob, inv, parent_invs, executable):
86
def import_git_blob(texts, mapping, path, hexsha, base_inv, parent_id,
87
revision_id, parent_invs, shagitmap, lookup_object, executable, symlink):
73
88
"""Import a git blob object into a bzr repository.
75
:param repo: bzr repository
90
:param texts: VersionedFiles to add to
76
91
:param path: Path in the tree
77
92
:param blob: A git blob
93
:return: Inventory delta for this file
79
95
file_id = mapping.generate_file_id(path)
80
text_revision = inv.revision_id
81
repo.texts.add_lines((file_id, text_revision),
82
[(file_id, p[file_id].revision) for p in parent_invs if file_id in p],
83
osutils.split_lines(blob.data))
84
ie = inv.add_path(path, "file", file_id)
85
ie.revision = text_revision
86
ie.text_size = len(blob.data)
87
ie.text_sha1 = osutils.sha_string(blob.data)
100
# We just have to hope this is indeed utf-8:
101
ie = cls(file_id, urlutils.basename(path).decode("utf-8"), parent_id)
88
102
ie.executable = executable
91
def import_git_tree(repo, mapping, path, tree, inv, parent_invs, lookup_object):
104
# See if this has changed at all
106
base_ie = base_inv[file_id]
111
base_sha = base_ie.text_id
114
base_sha = shagitmap.lookup_blob(file_id, base_ie.revision)
118
if (base_sha == hexsha and base_ie.executable == ie.executable
119
and base_ie.kind == ie.kind):
120
# If nothing has changed since the base revision, we're done
122
if base_sha == hexsha and base_ie.kind == ie.kind:
123
ie.text_size = base_ie.text_size
124
ie.text_sha1 = base_ie.text_sha1
125
ie.symlink_target = base_ie.symlink_target
126
if ie.executable == base_ie.executable:
127
ie.revision = base_ie.revision
129
blob = lookup_object(hexsha)
131
blob = lookup_object(hexsha)
132
if ie.kind == "symlink":
134
ie.symlink_target = blob.data
138
ie.text_size = len(blob.data)
139
ie.text_sha1 = osutils.sha_string(blob.data)
140
# Check what revision we should store
142
for pinv in parent_invs:
143
if pinv.revision_id == base_inv.revision_id:
152
if pie.text_sha1 == ie.text_sha1 and pie.executable == ie.executable and pie.symlink_target == ie.symlink_target:
153
# found a revision in one of the parents to use
154
ie.revision = pie.revision
156
parent_keys.append((file_id, pie.revision))
157
if ie.revision is None:
158
# Need to store a new revision
159
ie.revision = revision_id
160
assert file_id is not None
161
assert ie.revision is not None
162
texts.insert_record_stream([FulltextContentFactory((file_id, ie.revision), tuple(parent_keys), ie.text_sha1, blob.data)])
163
shamap = [(hexsha, "blob", (ie.file_id, ie.revision))]
166
if file_id in base_inv:
167
old_path = base_inv.id2path(file_id)
170
return ([(old_path, path, file_id, ie)], shamap)
173
def import_git_submodule(texts, mapping, path, hexsha, base_inv, parent_id,
174
revision_id, parent_invs, shagitmap, lookup_object):
175
raise NotImplementedError(import_git_submodule)
178
def import_git_tree(texts, mapping, path, hexsha, base_inv, parent_id,
179
revision_id, parent_invs, shagitmap, lookup_object):
92
180
"""Import a git tree object into a bzr repository.
94
:param repo: A Bzr repository object
182
:param texts: VersionedFiles object to add to
95
183
:param path: Path in the tree
96
184
:param tree: A git tree object
97
:param inv: Inventory object
185
:param base_inv: Base inventory against which to return inventory delta
186
:return: Inventory delta for this subtree
99
189
file_id = mapping.generate_file_id(path)
100
text_revision = inv.revision_id
101
repo.texts.add_lines((file_id, text_revision),
102
[(file_id, p[file_id].revision) for p in parent_invs if file_id in p],
104
ie = inv.add_path(path, "directory", file_id)
105
ie.revision = text_revision
106
for mode, name, hexsha in tree.entries():
107
entry_kind = (mode & 0700000) / 0100000
190
# We just have to hope this is indeed utf-8:
191
ie = InventoryDirectory(file_id, urlutils.basename(path.decode("utf-8")),
195
base_ie = base_inv[file_id]
197
# Newly appeared here
199
ie.revision = revision_id
200
texts.add_lines((file_id, ie.revision), (), [])
201
invdelta.append((None, path, file_id, ie))
203
base_sha = base_ie.text_id
204
# See if this has changed at all
207
base_sha = shagitmap.lookup_tree(file_id, base_inv.revision_id)
211
if base_sha == hexsha:
212
# If nothing has changed since the base revision, we're done
214
# Remember for next time
215
existing_children = set()
218
tree = lookup_object(hexsha)
219
for mode, name, child_hexsha in tree.entries():
108
220
basename = name.decode("utf-8")
112
child_path = urlutils.join(path, name)
114
tree = lookup_object(hexsha)
115
import_git_tree(repo, mapping, child_path, tree, inv, parent_invs, lookup_object)
116
elif entry_kind == 1:
117
blob = lookup_object(hexsha)
118
fs_mode = mode & 0777
119
import_git_blob(repo, mapping, child_path, blob, inv, parent_invs, bool(fs_mode & 0111))
121
raise AssertionError("Unknown blob kind, perms=%r." % (mode,))
124
def import_git_objects(repo, mapping, object_iter, pb=None):
221
existing_children.add(basename)
222
child_path = osutils.pathjoin(path, name)
223
if stat.S_ISDIR(mode):
224
subinvdelta, grandchildmodes, subshamap = import_git_tree(texts,
225
mapping, child_path, child_hexsha, base_inv, file_id,
226
revision_id, parent_invs, shagitmap, lookup_object)
227
invdelta.extend(subinvdelta)
228
child_modes.update(grandchildmodes)
229
shamap.extend(subshamap)
230
elif S_ISGITLINK(mode): # submodule
231
subinvdelta, grandchildmodes, subshamap = import_git_submodule(
232
texts, mapping, child_path, child_hexsha, base_inv,
233
file_id, revision_id, parent_invs, shagitmap, lookup_object)
234
invdelta.extend(subinvdelta)
235
child_modes.update(grandchildmodes)
236
shamap.extend(subshamap)
238
subinvdelta, subshamap = import_git_blob(texts, mapping,
239
child_path, child_hexsha, base_inv, file_id, revision_id,
240
parent_invs, shagitmap, lookup_object,
241
mode_is_executable(mode), stat.S_ISLNK(mode))
242
invdelta.extend(subinvdelta)
243
shamap.extend(subshamap)
244
if mode not in (stat.S_IFDIR, DEFAULT_FILE_MODE,
245
stat.S_IFLNK, DEFAULT_FILE_MODE|0111):
246
child_modes[child_path] = mode
247
# Remove any children that have disappeared
248
if base_ie is not None and base_ie.kind == 'directory':
249
deletable = [v for k,v in base_ie.children.iteritems() if k not in existing_children]
252
invdelta.append((base_inv.id2path(ie.file_id), None, ie.file_id, None))
253
if ie.kind == "directory":
254
deletable.extend(ie.children.values())
255
shamap.append((hexsha, "tree", (file_id, revision_id)))
256
return invdelta, child_modes, shamap
259
def import_git_objects(repo, mapping, object_iter, target_git_object_retriever,
125
261
"""Import a set of git objects into a bzr repository.
127
:param repo: Bazaar repository
263
:param repo: Target Bazaar repository
128
264
:param mapping: Mapping to use
129
265
:param object_iter: Iterator over Git objects.
267
def lookup_object(sha):
269
return object_iter[sha]
271
return target_git_object_retriever[sha]
131
272
# TODO: a more (memory-)efficient implementation of this
133
for i, o in enumerate(object_iter):
135
pb.update("fetching objects", i)
278
parent_invs_cache = LRUCache(50)
140
279
# Find and convert commit objects
141
for o in objects.itervalues():
282
pb.update("finding revisions to fetch", len(graph), None)
284
assert isinstance(head, str)
286
o = lookup_object(head)
142
289
if isinstance(o, Commit):
143
290
rev = mapping.import_commit(o)
144
root_trees[rev.revision_id] = objects[o.tree]
291
if repo.has_revision(rev.revision_id):
293
squash_revision(repo, rev)
294
root_trees[rev.revision_id] = o.tree
145
295
revisions[rev.revision_id] = rev
146
296
graph.append((rev.revision_id, rev.parent_ids))
297
target_git_object_retriever._idmap.add_entry(o.id, "commit",
298
(rev.revision_id, o.tree))
299
heads.extend([p for p in o.parents if p not in checked])
300
elif isinstance(o, Tag):
301
heads.append(o.object[1])
303
trace.warning("Unable to import head object %r" % o)
147
305
# Order the revisions
148
306
# Create the inventory objects
149
307
for i, revid in enumerate(topo_sort(graph)):
150
308
if pb is not None:
151
309
pb.update("fetching revisions", i, len(graph))
152
root_tree = root_trees[revid]
153
310
rev = revisions[revid]
154
311
# We have to do this here, since we have to walk the tree and
155
# we need to make sure to import the blobs / trees with the riht
312
# we need to make sure to import the blobs / trees with the right
156
313
# path; this may involve adding them more than once.
158
inv.revision_id = rev.revision_id
159
def lookup_object(sha):
162
return reconstruct_git_object(repo, mapping, sha)
163
parent_invs = [repo.get_inventory(r) for r in rev.parent_ids]
164
import_git_tree(repo, mapping, "", root_tree, inv, parent_invs, lookup_object)
165
repo.add_revision(rev.revision_id, rev, inv)
168
def reconstruct_git_commit(repo, rev):
169
raise NotImplementedError(self.reconstruct_git_commit)
172
def reconstruct_git_object(repo, mapping, sha):
174
revid = mapping.revision_id_foreign_to_bzr(sha)
176
rev = repo.get_revision(revid)
177
except NoSuchRevision:
180
return reconstruct_git_commit(rev)
184
raise KeyError("No such object %s" % sha)
315
for parent_id in rev.parent_ids:
317
parent_invs.append(parent_invs_cache[parent_id])
319
parent_inv = repo.get_inventory(parent_id)
320
parent_invs.append(parent_inv)
321
parent_invs_cache[parent_id] = parent_inv
322
if parent_invs == []:
323
base_inv = Inventory(root_id=None)
325
base_inv = parent_invs[0]
326
inv_delta, unusual_modes, shamap = import_git_tree(repo.texts,
327
mapping, "", root_trees[revid], base_inv, None, revid,
328
parent_invs, target_git_object_retriever._idmap, lookup_object)
329
target_git_object_retriever._idmap.add_entries(shamap)
330
if unusual_modes != {}:
331
for path, mode in unusual_modes.iteritems():
332
warn_unusual_mode(rev.foreign_revid, path, mode)
333
mapping.import_unusual_file_modes(rev, unusual_modes)
335
basis_id = rev.parent_ids[0]
337
basis_id = NULL_REVISION
338
rev.inventory_sha1, inv = repo.add_inventory_by_delta(basis_id,
339
inv_delta, rev.revision_id, rev.parent_ids)
340
parent_invs_cache[rev.revision_id] = inv
341
repo.add_revision(rev.revision_id, rev)
342
if "verify" in debug.debug_flags:
343
new_unusual_modes = mapping.export_unusual_file_modes(rev)
344
if new_unusual_modes != unusual_modes:
345
raise AssertionError("unusual modes don't match: %r != %r" % (unusual_modes, new_unusual_modes))
346
objs = inventory_to_tree_and_blobs(inv, repo.texts, mapping, unusual_modes)
347
for sha1, newobj, path in objs:
348
assert path is not None
349
oldobj = tree_lookup_path(lookup_object, root_trees[revid], path)
351
raise AssertionError("%r != %r in %s" % (oldobj, newobj, path))
353
target_git_object_retriever._idmap.commit()
187
356
class InterGitRepository(InterRepository):
189
_matching_repo_format = GitFormat()
358
_matching_repo_format = GitRepositoryFormat()
192
361
def _get_repo_format_to_test():
196
365
"""See InterRepository.copy_content."""
197
366
self.fetch(revision_id, pb, find_ghosts=False)
199
def fetch(self, revision_id=None, pb=None, find_ghosts=False,
368
def fetch(self, revision_id=None, pb=None, find_ghosts=False, mapping=None,
370
self.fetch_refs(revision_id=revision_id, pb=pb, find_ghosts=find_ghosts,
371
mapping=mapping, fetch_spec=fetch_spec)
374
class InterGitNonGitRepository(InterGitRepository):
375
"""Base InterRepository that copies revisions from a Git into a non-Git
378
def fetch_refs(self, revision_id=None, pb=None, find_ghosts=False,
379
mapping=None, fetch_spec=None):
201
380
if mapping is None:
202
381
mapping = self.source.get_mapping()
204
pb.note("git: %s", text)
205
def determine_wants(heads):
206
if revision_id is None:
382
if revision_id is not None:
383
interesting_heads = [revision_id]
384
elif fetch_spec is not None:
385
interesting_heads = fetch_spec.heads
387
interesting_heads = None
389
def determine_wants(refs):
391
if interesting_heads is None:
392
ret = [sha for (ref, sha) in refs.iteritems() if not ref.endswith("^{}")]
209
ret = [mapping.revision_id_bzr_to_foreign(revision_id)]
394
ret = [mapping.revision_id_bzr_to_foreign(revid)[0] for revid in interesting_heads if revid not in (None, NULL_REVISION)]
210
395
return [rev for rev in ret if not self.target.has_revision(mapping.revision_id_foreign_to_bzr(rev))]
211
graph_walker = BzrFetchGraphWalker(self.target, mapping)
396
self.fetch_objects(determine_wants, mapping, pb)
400
class InterRemoteGitNonGitRepository(InterGitNonGitRepository):
401
"""InterRepository that copies revisions from a remote Git into a non-Git
404
def fetch_objects(self, determine_wants, mapping, pb=None):
406
pb.update("git: %s" % text.rstrip("\r\n"), 0, 0)
407
store = BazaarObjectStore(self.target, mapping)
408
self.target.lock_write()
410
heads = self.target.get_graph().heads(self.target.all_revision_ids())
411
graph_walker = store.get_graph_walker(
412
[store._lookup_revision_sha1(head) for head in heads])
415
def record_determine_wants(heads):
416
wants = determine_wants(heads)
417
recorded_wants.extend(wants)
422
create_pb = pb = ui.ui_factory.nested_progress_bar()
424
self.target.start_write_group()
426
objects_iter = self.source.fetch_objects(
427
record_determine_wants, graph_walker,
428
store.get_raw, progress)
429
import_git_objects(self.target, mapping, objects_iter,
430
store, recorded_wants, pb)
432
self.target.commit_write_group()
440
def is_compatible(source, target):
441
"""Be compatible with GitRepository."""
442
# FIXME: Also check target uses VersionedFile
443
return (isinstance(source, RemoteGitRepository) and
444
target.supports_rich_root() and
445
not isinstance(target, GitRepository))
448
class InterLocalGitNonGitRepository(InterGitNonGitRepository):
449
"""InterRepository that copies revisions from a local Git into a non-Git
452
def fetch_objects(self, determine_wants, mapping, pb=None):
453
wants = determine_wants(self.source._git.get_refs())
214
456
create_pb = pb = ui.ui_factory.nested_progress_bar()
457
target_git_object_retriever = BazaarObjectStore(self.target, mapping)
216
459
self.target.lock_write()
218
461
self.target.start_write_group()
220
import_git_objects(self.target, mapping,
221
iter(self.source.fetch_objects(determine_wants, graph_walker,
463
import_git_objects(self.target, mapping,
464
self.source._git.object_store,
465
target_git_object_retriever, wants, pb)
224
467
self.target.commit_write_group()