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
18
from bzrlib.errors import InvalidRevisionId
19
from bzrlib.inventory import Inventory
20
from bzrlib.repository import InterRepository
21
from bzrlib.trace import info
23
from bzrlib.plugins.git import git
24
from bzrlib.plugins.git.repository import LocalGitRepository, GitRepository, GitFormat
25
from bzrlib.plugins.git.remote import RemoteGitRepository
27
from dulwich.objects import Commit
29
from cStringIO import StringIO
17
from cStringIO import (
21
from dulwich.client import (
22
SimpleFetchGraphWalker,
24
from dulwich.objects import (
28
from dulwich.object_store import (
40
from bzrlib.errors import (
45
from bzrlib.inventory import (
51
from bzrlib.lru_cache import (
54
from bzrlib.repository import (
57
from bzrlib.revision import (
60
from bzrlib.tsort import (
64
from bzrlib.plugins.git.converter import (
67
from bzrlib.plugins.git.mapping import (
69
inventory_to_tree_and_blobs,
72
from bzrlib.plugins.git.repository import (
77
from bzrlib.plugins.git.remote import (
32
82
class BzrFetchGraphWalker(object):
83
"""GraphWalker implementation that uses a Bazaar repository."""
34
85
def __init__(self, repository, mapping):
35
86
self.repository = repository
58
112
self.heads.update([p for p in ps if not p in self.done])
60
114
self.done.add(ret)
61
return self.mapping.revision_id_bzr_to_foreign(ret)
115
return self.mapping.revision_id_bzr_to_foreign(ret)[0]
62
116
except InvalidRevisionId:
67
def import_git_blob(repo, mapping, path, blob):
121
def import_git_blob(texts, mapping, path, hexsha, base_inv, parent_id,
122
revision_id, parent_invs, shagitmap, lookup_object, executable, symlink):
68
123
"""Import a git blob object into a bzr repository.
70
:param repo: bzr repository
125
:param texts: VersionedFiles to add to
71
126
:param path: Path in the tree
72
127
:param blob: A git blob
128
:return: Inventory delta for this file
74
130
file_id = mapping.generate_file_id(path)
75
repo.texts.add_lines((file_id, blob.id),
77
osutils.split_lines(blob.data))
78
inv.add_path(path, "file", file_id)
81
def import_git_tree(repo, mapping, path, tree, inv, lookup_object):
135
# We just have to hope this is indeed utf-8:
136
ie = cls(file_id, urlutils.basename(path).decode("utf-8"), parent_id)
137
ie.executable = executable
139
# See if this has changed at all
141
base_ie = base_inv[file_id]
146
if base_ie.text_id == hexsha:
149
base_sha = shagitmap.lookup_blob(file_id, base_ie.revision)
153
if (base_sha == hexsha and base_ie.executable == ie.executable
154
and base_ie.kind == ie.kind):
155
# If nothing has changed since the base revision, we're done
157
if base_sha == hexsha:
158
ie.text_size = base_ie.text_size
159
ie.text_sha1 = base_ie.text_sha1
160
ie.symlink_target = base_ie.symlink_target
161
ie.revision = base_ie.revision
163
blob = lookup_object(hexsha)
164
if ie.kind == "symlink":
165
ie.symlink_target = blob.data
169
ie.text_size = len(blob.data)
170
ie.text_sha1 = osutils.sha_string(blob.data)
171
# Check what revision we should store
173
for pinv in parent_invs:
174
if pinv.revision_id == base_inv.revision_id:
183
if pie.text_sha1 == ie.text_sha1:
184
# found a revision in one of the parents to use
185
ie.revision = pie.revision
187
parent_keys.append((file_id, pie.revision))
188
if ie.revision is None:
189
# Need to store a new revision
190
ie.revision = revision_id
191
assert file_id is not None
192
assert ie.revision is not None
193
texts.add_lines((file_id, ie.revision), tuple(parent_keys),
194
osutils.split_lines(blob.data))
195
if file_id in base_inv:
196
old_path = base_inv.id2path(file_id)
199
return ([(old_path, path, file_id, ie)],
200
[(hexsha, "blob", (ie.file_id, ie.revision))])
203
def import_git_tree(texts, mapping, path, hexsha, base_inv, parent_id,
204
revision_id, parent_invs, shagitmap, lookup_object):
82
205
"""Import a git tree object into a bzr repository.
84
:param repo: A Bzr repository object
207
:param texts: VersionedFiles object to add to
85
208
:param path: Path in the tree
86
209
:param tree: A git tree object
87
:param inv: Inventory object
210
:param base_inv: Base inventory against which to return inventory delta
211
:return: Inventory delta for this subtree
89
214
file_id = mapping.generate_file_id(path)
90
repo.texts.add_lines((file_id, tree.id),
93
inv.add_path(path, "directory", file_id)
94
for mode, name, hexsha in tree.entries():
95
entry_kind = (mode & 0700000) / 0100000
215
# We just have to hope this is indeed utf-8:
216
ie = InventoryDirectory(file_id, urlutils.basename(path.decode("utf-8")),
220
base_ie = base_inv[file_id]
222
# Newly appeared here
224
ie.revision = revision_id
225
texts.add_lines((file_id, ie.revision), (), [])
226
invdelta.append((None, path, file_id, ie))
228
if base_ie.text_id == hexsha:
230
# See if this has changed at all
232
base_sha = shagitmap.lookup_tree(file_id, base_inv.revision_id)
236
if base_sha == hexsha:
237
# If nothing has changed since the base revision, we're done
239
# Remember for next time
240
existing_children = set()
243
tree = lookup_object(hexsha)
244
for mode, name, child_hexsha in tree.entries():
96
245
basename = name.decode("utf-8")
100
child_path = urlutils.join(path, name)
102
import_git_tree(repo, mapping, child_path, lookup_object, inv)
103
elif entry_kind == 1:
104
import_git_blob(repo, mapping, child_path, lookup_object, inv)
106
raise AssertionError("Unknown blob kind, perms=%r." % (mode,))
109
def import_git_objects(repo, mapping, object_iter):
246
existing_children.add(basename)
247
child_path = osutils.pathjoin(path, name)
248
if stat.S_ISDIR(mode):
249
subinvdelta, grandchildmodes, subshamap = import_git_tree(texts,
250
mapping, child_path, child_hexsha, base_inv, file_id,
251
revision_id, parent_invs, shagitmap, lookup_object)
252
invdelta.extend(subinvdelta)
253
child_modes.update(grandchildmodes)
254
shamap.extend(subshamap)
256
fs_mode = stat.S_IMODE(mode)
257
symlink = stat.S_ISLNK(mode)
258
subinvdelta, subshamap = import_git_blob(texts, mapping,
259
child_path, child_hexsha, base_inv, file_id, revision_id,
260
parent_invs, shagitmap, lookup_object,
261
bool(fs_mode & 0111), symlink)
262
invdelta.extend(subinvdelta)
263
shamap.extend(subshamap)
264
if mode not in (stat.S_IFDIR, DEFAULT_FILE_MODE,
265
stat.S_IFLNK, DEFAULT_FILE_MODE|0111):
266
child_modes[child_path] = mode
267
# Remove any children that have disappeared
268
if base_ie is not None:
269
deletable = [v for k,v in base_ie.children.iteritems() if k not in existing_children]
272
invdelta.append((base_inv.id2path(ie.file_id), None, ie.file_id, None))
273
if ie.kind == "directory":
274
deletable.extend(ie.children.values())
275
shamap.append((hexsha, "tree", (file_id, revision_id)))
276
return invdelta, child_modes, shamap
279
def import_git_objects(repo, mapping, object_iter, target_git_object_retriever,
110
281
"""Import a set of git objects into a bzr repository.
112
283
:param repo: Bazaar repository
114
285
:param object_iter: Iterator over Git objects.
116
287
# TODO: a more (memory-)efficient implementation of this
118
for o in object_iter:
293
parent_invs_cache = LRUCache(50)
121
294
# Find and convert commit objects
122
for o in objects.iterkeys():
297
pb.update("finding revisions to fetch", len(graph), None)
299
assert isinstance(head, str)
301
o = object_iter[head]
123
304
if isinstance(o, Commit):
124
305
rev = mapping.import_commit(o)
125
root_trees[rev] = objects[o.tree_sha]
306
if repo.has_revision(rev.revision_id):
308
root_trees[rev.revision_id] = o.tree
309
revisions[rev.revision_id] = rev
310
graph.append((rev.revision_id, rev.parent_ids))
311
target_git_object_retriever._idmap.add_entry(o.id, "commit",
312
(rev.revision_id, o.tree))
313
heads.extend([p for p in o.parents if p not in checked])
314
elif isinstance(o, Tag):
315
heads.append(o.object[1])
317
trace.warning("Unable to import head object %r" % o)
319
# Order the revisions
126
320
# Create the inventory objects
127
for rev, root_tree in root_trees.iteritems():
321
for i, revid in enumerate(topo_sort(graph)):
323
pb.update("fetching revisions", i, len(graph))
324
rev = revisions[revid]
128
325
# We have to do this here, since we have to walk the tree and
129
# we need to make sure to import the blobs / trees with the riht
326
# we need to make sure to import the blobs / trees with the right
130
327
# path; this may involve adding them more than once.
132
328
def lookup_object(sha):
135
return reconstruct_git_object(repo, mapping, sha)
136
import_git_tree(repo, mapping, "", tree, inv, lookup_object)
137
repo.add_revision(rev.revision_id, rev, inv)
140
def reconstruct_git_commit(repo, rev):
141
raise NotImplementedError(self.reconstruct_git_commit)
144
def reconstruct_git_object(repo, mapping, sha):
146
revid = mapping.revision_id_foreign_to_bzr(sha)
148
rev = repo.get_revision(revid)
149
except NoSuchRevision:
152
return reconstruct_git_commit(rev)
156
raise KeyError("No such object %s" % sha)
159
class InterGitRepository(InterRepository):
161
_matching_repo_format = GitFormat()
330
return object_iter[sha]
332
return target_git_object_retriever[sha]
334
for parent_id in rev.parent_ids:
336
parent_invs.append(parent_invs_cache[parent_id])
338
parent_inv = repo.get_inventory(parent_id)
339
parent_invs.append(parent_inv)
340
parent_invs_cache[parent_id] = parent_inv
341
if parent_invs == []:
342
base_inv = Inventory(root_id=None)
344
base_inv = parent_invs[0]
345
inv_delta, unusual_modes, shamap = import_git_tree(repo.texts,
346
mapping, "", root_trees[revid], base_inv, None, revid,
347
parent_invs, target_git_object_retriever._idmap, lookup_object)
348
target_git_object_retriever._idmap.add_entries(shamap)
349
if unusual_modes != {}:
350
ret = "unusual modes: \n"
351
for item in unusual_modes.iteritems():
352
ret += "\t%s: %o\n" % item
353
raise AssertionError(ret)
355
basis_id = rev.parent_ids[0]
357
basis_id = NULL_REVISION
358
rev.inventory_sha1, inv = repo.add_inventory_by_delta(basis_id,
359
inv_delta, rev.revision_id, rev.parent_ids)
360
parent_invs_cache[rev.revision_id] = inv
361
repo.add_revision(rev.revision_id, rev)
362
if "verify" in debug.debug_flags:
363
objs = inventory_to_tree_and_blobs(inv, repo.texts, mapping)
364
for sha1, newobj, path in objs:
365
assert path is not None
366
oldobj = tree_lookup_path(lookup_object, root_trees[revid], path)
367
assert oldobj == newobj, "%r != %r in %s" % (oldobj, newobj, path)
369
target_git_object_retriever._idmap.commit()
372
class InterGitNonGitRepository(InterRepository):
373
"""Base InterRepository that copies revisions from a Git into a non-Git
376
_matching_repo_format = GitRepositoryFormat()
164
379
def _get_repo_format_to_test():
168
383
"""See InterRepository.copy_content."""
169
384
self.fetch(revision_id, pb, find_ghosts=False)
171
def fetch(self, revision_id=None, pb=None, find_ghosts=False,
386
def fetch(self, revision_id=None, pb=None, find_ghosts=False, mapping=None,
388
self.fetch_refs(revision_id=revision_id, pb=pb, find_ghosts=find_ghosts,
389
mapping=mapping, fetch_spec=fetch_spec)
391
def fetch_refs(self, revision_id=None, pb=None, find_ghosts=False,
392
mapping=None, fetch_spec=None):
173
393
if mapping is None:
174
394
mapping = self.source.get_mapping()
177
pb.note("git: %s" % text)
179
info("git: %s" % text)
180
def determine_wants(heads):
181
if revision_id is None:
184
ret = [mapping.revision_id_bzr_to_foreign(revision_id)]
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("^{}")]
407
ret = [mapping.revision_id_bzr_to_foreign(revid)[0] for revid in interesting_heads if revid != NULL_REVISION]
185
408
return [rev for rev in ret if not self.target.has_revision(mapping.revision_id_foreign_to_bzr(rev))]
409
self.fetch_objects(determine_wants, mapping, pb)
414
class InterRemoteGitNonGitRepository(InterGitNonGitRepository):
415
"""InterRepository that copies revisions from a remote Git into a non-Git
418
def fetch_objects(self, determine_wants, mapping, pb=None):
420
pb.update("git: %s" % text.rstrip("\r\n"), 0, 0)
186
421
graph_walker = BzrFetchGraphWalker(self.target, mapping)
187
self.target.lock_write()
189
import_git_objects(self.target, mapping,
190
self.source.fetch_objects(determine_wants, graph_walker,
424
create_pb = pb = ui.ui_factory.nested_progress_bar()
425
target_git_object_retriever = BazaarObjectStore(self.target, mapping)
428
def record_determine_wants(heads):
429
wants = determine_wants(heads)
430
recorded_wants.extend(wants)
434
self.target.lock_write()
436
self.target.start_write_group()
438
objects_iter = self.source.fetch_objects(
439
record_determine_wants,
441
target_git_object_retriever.get_raw,
443
import_git_objects(self.target, mapping, objects_iter,
444
target_git_object_retriever, recorded_wants, pb)
446
self.target.commit_write_group()
454
def is_compatible(source, target):
455
"""Be compatible with GitRepository."""
456
# FIXME: Also check target uses VersionedFile
457
return (isinstance(source, RemoteGitRepository) and
458
target.supports_rich_root() and
459
not isinstance(target, GitRepository))
462
class InterLocalGitNonGitRepository(InterGitNonGitRepository):
463
"""InterRepository that copies revisions from a remote Git into a non-Git
466
def fetch_objects(self, determine_wants, mapping, pb=None):
467
wants = determine_wants(self.source._git.get_refs())
470
create_pb = pb = ui.ui_factory.nested_progress_bar()
471
target_git_object_retriever = BazaarObjectStore(self.target, mapping)
473
self.target.lock_write()
475
self.target.start_write_group()
477
import_git_objects(self.target, mapping,
478
self.source._git.object_store,
479
target_git_object_retriever, wants, pb)
481
self.target.commit_write_group()
196
489
def is_compatible(source, target):
197
490
"""Be compatible with GitRepository."""
198
491
# FIXME: Also check target uses VersionedFile
199
492
return (isinstance(source, LocalGitRepository) and
200
target.supports_rich_root())
493
target.supports_rich_root() and
494
not isinstance(target, GitRepository))
497
class InterGitRepository(InterRepository):
498
"""InterRepository that copies between Git repositories."""
500
_matching_repo_format = GitRepositoryFormat()
503
def _get_repo_format_to_test():
506
def copy_content(self, revision_id=None, pb=None):
507
"""See InterRepository.copy_content."""
508
self.fetch(revision_id, pb, find_ghosts=False)
510
def fetch(self, revision_id=None, pb=None, find_ghosts=False,
511
mapping=None, fetch_spec=None):
513
mapping = self.source.get_mapping()
515
trace.info("git: %s", text)
517
if revision_id is not None:
518
args = [mapping.revision_id_bzr_to_foreign(revision_id)[0]]
519
elif fetch_spec is not None:
520
args = [mapping.revision_id_bzr_to_foreign(revid)[0] for revid in fetch_spec.heads]
521
if fetch_spec is None and revision_id is None:
522
determine_wants = r.object_store.determine_wants_all
524
determine_wants = lambda x: [y for y in args if not y in r.object_store]
526
graphwalker = SimpleFetchGraphWalker(r.heads().values(), r.get_parents)
527
f, commit = r.object_store.add_thin_pack()
529
self.source.fetch_pack(determine_wants, graphwalker, f.write, progress)
536
def is_compatible(source, target):
537
"""Be compatible with GitRepository."""
538
return (isinstance(source, GitRepository) and
539
isinstance(target, GitRepository))