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
invdelta = [(old_path, path, file_id, ie)]
171
invdelta.extend(remove_disappeared_children(base_inv, base_ie, []))
172
return (invdelta, shamap)
175
def import_git_submodule(texts, mapping, path, hexsha, base_inv, parent_id,
176
revision_id, parent_invs, shagitmap, lookup_object):
177
raise NotImplementedError(import_git_submodule)
180
def remove_disappeared_children(base_inv, base_ie, existing_children):
181
if base_ie is None or base_ie.kind != 'directory':
184
deletable = [v for k,v in base_ie.children.iteritems() if k not in existing_children]
187
ret.append((base_inv.id2path(ie.file_id), None, ie.file_id, None))
188
if ie.kind == "directory":
189
deletable.extend(ie.children.values())
193
def import_git_tree(texts, mapping, path, hexsha, base_inv, parent_id,
194
revision_id, parent_invs, shagitmap, lookup_object):
92
195
"""Import a git tree object into a bzr repository.
94
:param repo: A Bzr repository object
197
:param texts: VersionedFiles object to add to
95
198
:param path: Path in the tree
96
199
:param tree: A git tree object
97
:param inv: Inventory object
200
:param base_inv: Base inventory against which to return inventory delta
201
:return: Inventory delta for this subtree
99
204
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
205
# We just have to hope this is indeed utf-8:
206
ie = InventoryDirectory(file_id, urlutils.basename(path.decode("utf-8")),
210
base_ie = base_inv[file_id]
212
# Newly appeared here
214
ie.revision = revision_id
215
texts.add_lines((file_id, ie.revision), (), [])
216
invdelta.append((None, path, file_id, ie))
218
base_sha = base_ie.text_id
219
# See if this has changed at all
222
base_sha = shagitmap.lookup_tree(file_id, base_inv.revision_id)
226
if base_sha == hexsha:
227
# If nothing has changed since the base revision, we're done
229
if base_ie.kind != "directory":
230
ie.revision = revision_id
231
texts.add_lines((file_id, ie.revision), (), [])
232
invdelta.append((base_inv.id2path(file_id), path, file_id, ie))
233
# Remember for next time
234
existing_children = set()
237
tree = lookup_object(hexsha)
238
for mode, name, child_hexsha in tree.entries():
108
239
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):
240
existing_children.add(basename)
241
child_path = osutils.pathjoin(path, name)
242
if stat.S_ISDIR(mode):
243
subinvdelta, grandchildmodes, subshamap = import_git_tree(texts,
244
mapping, child_path, child_hexsha, base_inv, file_id,
245
revision_id, parent_invs, shagitmap, lookup_object)
246
invdelta.extend(subinvdelta)
247
child_modes.update(grandchildmodes)
248
shamap.extend(subshamap)
249
elif S_ISGITLINK(mode): # submodule
250
subinvdelta, grandchildmodes, subshamap = import_git_submodule(
251
texts, mapping, child_path, child_hexsha, base_inv,
252
file_id, revision_id, parent_invs, shagitmap, lookup_object)
253
invdelta.extend(subinvdelta)
254
child_modes.update(grandchildmodes)
255
shamap.extend(subshamap)
257
subinvdelta, subshamap = import_git_blob(texts, mapping,
258
child_path, child_hexsha, base_inv, file_id, revision_id,
259
parent_invs, shagitmap, lookup_object,
260
mode_is_executable(mode), stat.S_ISLNK(mode))
261
invdelta.extend(subinvdelta)
262
shamap.extend(subshamap)
263
if mode not in (stat.S_IFDIR, DEFAULT_FILE_MODE,
264
stat.S_IFLNK, DEFAULT_FILE_MODE|0111):
265
child_modes[child_path] = mode
266
# Remove any children that have disappeared
267
invdelta.extend(remove_disappeared_children(base_inv, base_ie, existing_children))
268
shamap.append((hexsha, "tree", (file_id, revision_id)))
269
return invdelta, child_modes, shamap
272
def import_git_objects(repo, mapping, object_iter, target_git_object_retriever,
125
274
"""Import a set of git objects into a bzr repository.
127
:param repo: Bazaar repository
276
:param repo: Target Bazaar repository
128
277
:param mapping: Mapping to use
129
278
:param object_iter: Iterator over Git objects.
280
def lookup_object(sha):
282
return object_iter[sha]
284
return target_git_object_retriever[sha]
131
285
# TODO: a more (memory-)efficient implementation of this
133
for i, o in enumerate(object_iter):
135
pb.update("fetching objects", i)
291
parent_invs_cache = LRUCache(50)
140
292
# Find and convert commit objects
141
for o in objects.itervalues():
295
pb.update("finding revisions to fetch", len(graph), None)
297
assert isinstance(head, str)
299
o = lookup_object(head)
142
302
if isinstance(o, Commit):
143
303
rev = mapping.import_commit(o)
144
root_trees[rev.revision_id] = objects[o.tree]
304
if repo.has_revision(rev.revision_id):
306
squash_revision(repo, rev)
307
root_trees[rev.revision_id] = o.tree
145
308
revisions[rev.revision_id] = rev
146
309
graph.append((rev.revision_id, rev.parent_ids))
310
target_git_object_retriever._idmap.add_entry(o.id, "commit",
311
(rev.revision_id, o.tree))
312
heads.extend([p for p in o.parents if p not in checked])
313
elif isinstance(o, Tag):
314
heads.append(o.object[1])
316
trace.warning("Unable to import head object %r" % o)
147
318
# Order the revisions
148
319
# Create the inventory objects
149
320
for i, revid in enumerate(topo_sort(graph)):
150
321
if pb is not None:
151
322
pb.update("fetching revisions", i, len(graph))
152
root_tree = root_trees[revid]
153
323
rev = revisions[revid]
154
324
# 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
325
# we need to make sure to import the blobs / trees with the right
156
326
# 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)
328
for parent_id in rev.parent_ids:
330
parent_invs.append(parent_invs_cache[parent_id])
332
parent_inv = repo.get_inventory(parent_id)
333
parent_invs.append(parent_inv)
334
parent_invs_cache[parent_id] = parent_inv
335
if parent_invs == []:
336
base_inv = Inventory(root_id=None)
338
base_inv = parent_invs[0]
339
inv_delta, unusual_modes, shamap = import_git_tree(repo.texts,
340
mapping, "", root_trees[revid], base_inv, None, revid,
341
parent_invs, target_git_object_retriever._idmap, lookup_object)
342
target_git_object_retriever._idmap.add_entries(shamap)
343
if unusual_modes != {}:
344
for path, mode in unusual_modes.iteritems():
345
warn_unusual_mode(rev.foreign_revid, path, mode)
346
mapping.import_unusual_file_modes(rev, unusual_modes)
348
basis_id = rev.parent_ids[0]
350
basis_id = NULL_REVISION
351
rev.inventory_sha1, inv = repo.add_inventory_by_delta(basis_id,
352
inv_delta, rev.revision_id, rev.parent_ids)
353
parent_invs_cache[rev.revision_id] = inv
354
repo.add_revision(rev.revision_id, rev)
355
if "verify" in debug.debug_flags:
356
new_unusual_modes = mapping.export_unusual_file_modes(rev)
357
if new_unusual_modes != unusual_modes:
358
raise AssertionError("unusual modes don't match: %r != %r" % (unusual_modes, new_unusual_modes))
359
objs = inventory_to_tree_and_blobs(inv, repo.texts, mapping, unusual_modes)
360
for sha1, newobj, path in objs:
361
assert path is not None
362
oldobj = tree_lookup_path(lookup_object, root_trees[revid], path)
364
raise AssertionError("%r != %r in %s" % (oldobj, newobj, path))
366
target_git_object_retriever._idmap.commit()
187
369
class InterGitRepository(InterRepository):
189
_matching_repo_format = GitFormat()
371
_matching_repo_format = GitRepositoryFormat()
192
374
def _get_repo_format_to_test():
196
378
"""See InterRepository.copy_content."""
197
379
self.fetch(revision_id, pb, find_ghosts=False)
199
def fetch(self, revision_id=None, pb=None, find_ghosts=False,
381
def fetch(self, revision_id=None, pb=None, find_ghosts=False, mapping=None,
383
self.fetch_refs(revision_id=revision_id, pb=pb, find_ghosts=find_ghosts,
384
mapping=mapping, fetch_spec=fetch_spec)
387
class InterGitNonGitRepository(InterGitRepository):
388
"""Base InterRepository that copies revisions from a Git into a non-Git
391
def fetch_refs(self, revision_id=None, pb=None, find_ghosts=False,
392
mapping=None, fetch_spec=None):
201
393
if mapping is None:
202
394
mapping = self.source.get_mapping()
204
pb.note("git: %s", text)
205
def determine_wants(heads):
206
if revision_id is None:
395
if revision_id is not None:
396
interesting_heads = [revision_id]
397
elif fetch_spec is not None:
398
interesting_heads = fetch_spec.heads
400
interesting_heads = None
402
def determine_wants(refs):
404
if interesting_heads is None:
405
ret = [sha for (ref, sha) in refs.iteritems() if not ref.endswith("^{}")]
209
ret = [mapping.revision_id_bzr_to_foreign(revision_id)]
407
ret = [mapping.revision_id_bzr_to_foreign(revid)[0] for revid in interesting_heads if revid not in (None, NULL_REVISION)]
210
408
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)
409
self.fetch_objects(determine_wants, mapping, pb)
413
class InterRemoteGitNonGitRepository(InterGitNonGitRepository):
414
"""InterRepository that copies revisions from a remote Git into a non-Git
417
def fetch_objects(self, determine_wants, mapping, pb=None):
419
pb.update("git: %s" % text.rstrip("\r\n"), 0, 0)
420
store = BazaarObjectStore(self.target, mapping)
421
self.target.lock_write()
423
heads = self.target.get_graph().heads(self.target.all_revision_ids())
424
graph_walker = store.get_graph_walker(
425
[store._lookup_revision_sha1(head) for head in heads])
428
def record_determine_wants(heads):
429
wants = determine_wants(heads)
430
recorded_wants.extend(wants)
435
create_pb = pb = ui.ui_factory.nested_progress_bar()
437
self.target.start_write_group()
439
objects_iter = self.source.fetch_objects(
440
record_determine_wants, graph_walker,
441
store.get_raw, progress)
442
import_git_objects(self.target, mapping, objects_iter,
443
store, recorded_wants, pb)
445
self.target.commit_write_group()
453
def is_compatible(source, target):
454
"""Be compatible with GitRepository."""
455
# FIXME: Also check target uses VersionedFile
456
return (isinstance(source, RemoteGitRepository) and
457
target.supports_rich_root() and
458
not isinstance(target, GitRepository))
461
class InterLocalGitNonGitRepository(InterGitNonGitRepository):
462
"""InterRepository that copies revisions from a local Git into a non-Git
465
def fetch_objects(self, determine_wants, mapping, pb=None):
466
wants = determine_wants(self.source._git.get_refs())
214
469
create_pb = pb = ui.ui_factory.nested_progress_bar()
470
target_git_object_retriever = BazaarObjectStore(self.target, mapping)
216
472
self.target.lock_write()
218
474
self.target.start_write_group()
220
import_git_objects(self.target, mapping,
221
iter(self.source.fetch_objects(determine_wants, graph_walker,
476
import_git_objects(self.target, mapping,
477
self.source._git.object_store,
478
target_git_object_retriever, wants, pb)
224
480
self.target.commit_write_group()