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, NoSuchRevision
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
17
from cStringIO import (
21
from dulwich.objects import (
26
from dulwich.object_store import (
39
from bzrlib.errors import (
44
from bzrlib.inventory import (
50
from bzrlib.lru_cache import (
53
from bzrlib.repository import (
56
from bzrlib.revision import (
59
from bzrlib.tsort import (
62
from bzrlib.versionedfile import (
63
FulltextContentFactory,
66
from bzrlib.plugins.git.mapping import (
68
inventory_to_tree_and_blobs,
74
from bzrlib.plugins.git.object_store import (
77
from bzrlib.plugins.git.remote import (
24
80
from bzrlib.plugins.git.repository import (
29
from bzrlib.plugins.git.converter import GitObjectConverter
30
from bzrlib.plugins.git.remote import RemoteGitRepository
33
from dulwich.client import SimpleFetchGraphWalker
34
from dulwich.objects import Commit
36
from cStringIO import StringIO
39
class BzrFetchGraphWalker(object):
40
"""GraphWalker implementation that uses a Bazaar repository."""
42
def __init__(self, repository, mapping):
43
self.repository = repository
44
self.mapping = mapping
46
self.heads = set(repository.all_revision_ids())
50
return iter(self.next, None)
53
revid = self.mapping.revision_id_foreign_to_bzr(sha)
56
def remove(self, revid):
58
if revid in self.heads:
59
self.heads.remove(revid)
60
if revid in self.parents:
61
for p in self.parents[revid]:
66
ret = self.heads.pop()
67
ps = self.repository.get_parent_map([ret])[ret]
68
self.parents[ret] = ps
69
self.heads.update([p for p in ps if not p in self.done])
72
return self.mapping.revision_id_bzr_to_foreign(ret)[0]
73
except InvalidRevisionId:
78
def import_git_blob(repo, mapping, path, blob, inv, parent_invs, gitmap, executable):
87
def import_git_blob(texts, mapping, path, hexsha, base_inv, parent_id,
88
revision_id, parent_invs, shagitmap, lookup_object, executable, symlink):
79
89
"""Import a git blob object into a bzr repository.
81
:param repo: bzr repository
91
:param texts: VersionedFiles to add to
82
92
:param path: Path in the tree
83
93
:param blob: A git blob
94
:return: Inventory delta for this file
85
96
file_id = mapping.generate_file_id(path)
86
text_revision = inv.revision_id
87
repo.texts.add_lines((file_id, text_revision),
88
[(file_id, p[file_id].revision) for p in parent_invs if file_id in p],
89
osutils.split_lines(blob.data))
90
ie = inv.add_path(path, "file", file_id)
91
ie.revision = text_revision
92
ie.text_size = len(blob.data)
93
ie.text_sha1 = osutils.sha_string(blob.data)
101
# We just have to hope this is indeed utf-8:
102
ie = cls(file_id, urlutils.basename(path).decode("utf-8"), parent_id)
94
103
ie.executable = executable
95
gitmap._idmap.add_entry(blob.sha().hexdigest(), "blob", (ie.file_id, ie.revision))
98
def import_git_tree(repo, mapping, path, tree, inv, parent_invs,
99
gitmap, lookup_object):
104
# See if this has changed at all
106
base_ie = base_inv[file_id]
112
base_sha = shagitmap.lookup_blob(file_id, base_ie.revision)
116
if (base_sha == hexsha and base_ie.executable == ie.executable
117
and base_ie.kind == ie.kind):
118
# If nothing has changed since the base revision, we're done
120
if base_sha == hexsha and base_ie.kind == ie.kind:
121
ie.text_size = base_ie.text_size
122
ie.text_sha1 = base_ie.text_sha1
123
ie.symlink_target = base_ie.symlink_target
124
if ie.executable == base_ie.executable:
125
ie.revision = base_ie.revision
127
blob = lookup_object(hexsha)
129
blob = lookup_object(hexsha)
130
if ie.kind == "symlink":
132
ie.symlink_target = blob.data
136
ie.text_size = len(blob.data)
137
ie.text_sha1 = osutils.sha_string(blob.data)
138
# Check what revision we should store
140
for pinv in parent_invs:
141
if pinv.revision_id == base_inv.revision_id:
150
if pie.text_sha1 == ie.text_sha1 and pie.executable == ie.executable and pie.symlink_target == ie.symlink_target:
151
# found a revision in one of the parents to use
152
ie.revision = pie.revision
154
parent_keys.append((file_id, pie.revision))
155
if ie.revision is None:
156
# Need to store a new revision
157
ie.revision = revision_id
158
assert file_id is not None
159
assert ie.revision is not None
160
texts.insert_record_stream([FulltextContentFactory((file_id, ie.revision), tuple(parent_keys), ie.text_sha1, blob.data)])
161
shamap = [(hexsha, "blob", (ie.file_id, ie.revision))]
164
if file_id in base_inv:
165
old_path = base_inv.id2path(file_id)
168
invdelta = [(old_path, path, file_id, ie)]
169
invdelta.extend(remove_disappeared_children(base_inv, base_ie, []))
170
return (invdelta, 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 remove_disappeared_children(base_inv, base_ie, existing_children):
179
if base_ie is None or base_ie.kind != 'directory':
182
deletable = [v for k,v in base_ie.children.iteritems() if k not in existing_children]
185
ret.append((base_inv.id2path(ie.file_id), None, ie.file_id, None))
186
if ie.kind == "directory":
187
deletable.extend(ie.children.values())
191
def import_git_tree(texts, mapping, path, hexsha, base_inv, parent_id,
192
revision_id, parent_invs, shagitmap, lookup_object):
100
193
"""Import a git tree object into a bzr repository.
102
:param repo: A Bzr repository object
195
:param texts: VersionedFiles object to add to
103
196
:param path: Path in the tree
104
197
:param tree: A git tree object
105
:param inv: Inventory object
198
:param base_inv: Base inventory against which to return inventory delta
199
:return: Inventory delta for this subtree
107
202
file_id = mapping.generate_file_id(path)
108
text_revision = inv.revision_id
109
repo.texts.add_lines((file_id, text_revision),
110
[(file_id, p[file_id].revision) for p in parent_invs if file_id in p],
112
ie = inv.add_path(path, "directory", file_id)
113
ie.revision = text_revision
114
gitmap._idmap.add_entry(tree.sha().hexdigest(), "tree", (file_id, text_revision))
115
for mode, name, hexsha in tree.entries():
116
entry_kind = (mode & 0700000) / 0100000
203
# We just have to hope this is indeed utf-8:
204
ie = InventoryDirectory(file_id, urlutils.basename(path.decode("utf-8")),
207
base_ie = base_inv[file_id]
209
# Newly appeared here
211
ie.revision = revision_id
212
texts.add_lines((file_id, ie.revision), (), [])
213
invdelta.append((None, path, file_id, ie))
215
# See if this has changed at all
217
base_sha = shagitmap.lookup_tree(file_id, base_inv.revision_id)
221
if base_sha == hexsha:
222
# If nothing has changed since the base revision, we're done
224
if base_ie.kind != "directory":
225
ie.revision = revision_id
226
texts.add_lines((ie.file_id, ie.revision), (), [])
227
invdelta.append((base_inv.id2path(ie.file_id), path, ie.file_id, ie))
228
# Remember for next time
229
existing_children = set()
232
tree = lookup_object(hexsha)
233
for mode, name, child_hexsha in tree.entries():
117
234
basename = name.decode("utf-8")
121
child_path = urlutils.join(path, name)
123
tree = lookup_object(hexsha)
124
import_git_tree(repo, mapping, child_path, tree, inv, parent_invs, gitmap, lookup_object)
125
elif entry_kind == 1:
126
blob = lookup_object(hexsha)
127
fs_mode = mode & 0777
128
import_git_blob(repo, mapping, child_path, blob, inv, parent_invs, gitmap, bool(fs_mode & 0111))
130
raise AssertionError("Unknown blob kind, perms=%r." % (mode,))
235
existing_children.add(basename)
236
child_path = osutils.pathjoin(path, name)
237
if stat.S_ISDIR(mode):
238
subinvdelta, grandchildmodes, subshamap = import_git_tree(
239
texts, mapping, child_path, child_hexsha, base_inv,
240
file_id, revision_id, parent_invs, shagitmap, lookup_object)
241
invdelta.extend(subinvdelta)
242
child_modes.update(grandchildmodes)
243
shamap.extend(subshamap)
244
elif S_ISGITLINK(mode): # submodule
245
subinvdelta, grandchildmodes, subshamap = import_git_submodule(
246
texts, mapping, child_path, child_hexsha, base_inv,
247
file_id, revision_id, parent_invs, shagitmap, lookup_object)
248
invdelta.extend(subinvdelta)
249
child_modes.update(grandchildmodes)
250
shamap.extend(subshamap)
252
subinvdelta, subshamap = import_git_blob(texts, mapping,
253
child_path, child_hexsha, base_inv, file_id, revision_id,
254
parent_invs, shagitmap, lookup_object,
255
mode_is_executable(mode), stat.S_ISLNK(mode))
256
invdelta.extend(subinvdelta)
257
shamap.extend(subshamap)
258
if mode not in (stat.S_IFDIR, DEFAULT_FILE_MODE,
259
stat.S_IFLNK, DEFAULT_FILE_MODE|0111):
260
child_modes[child_path] = mode
261
# Remove any children that have disappeared
262
invdelta.extend(remove_disappeared_children(base_inv, base_ie, existing_children))
263
shamap.append((hexsha, "tree", (file_id, revision_id)))
264
return invdelta, child_modes, shamap
133
267
def import_git_objects(repo, mapping, object_iter, target_git_object_retriever,
135
269
"""Import a set of git objects into a bzr repository.
137
:param repo: Bazaar repository
271
:param repo: Target Bazaar repository
138
272
:param mapping: Mapping to use
139
273
:param object_iter: Iterator over Git objects.
275
def lookup_object(sha):
277
return object_iter[sha]
279
return target_git_object_retriever[sha]
141
280
# TODO: a more (memory-)efficient implementation of this
286
parent_invs_cache = LRUCache(50)
145
287
# Find and convert commit objects
146
for o in object_iter.iterobjects():
290
pb.update("finding revisions to fetch", len(graph), None)
292
assert isinstance(head, str)
294
o = lookup_object(head)
147
297
if isinstance(o, Commit):
148
298
rev = mapping.import_commit(o)
149
root_trees[rev.revision_id] = object_iter[o.tree]
299
if repo.has_revision(rev.revision_id):
301
squash_revision(repo, rev)
302
root_trees[rev.revision_id] = o.tree
150
303
revisions[rev.revision_id] = rev
151
304
graph.append((rev.revision_id, rev.parent_ids))
152
target_git_object_retriever._idmap.add_entry(o.sha().hexdigest(), "commit", (rev.revision_id, o._tree))
305
target_git_object_retriever._idmap.add_entry(o.id, "commit",
306
(rev.revision_id, o.tree))
307
heads.extend([p for p in o.parents if p not in checked])
308
elif isinstance(o, Tag):
309
heads.append(o.object[1])
311
trace.warning("Unable to import head object %r" % o)
153
313
# Order the revisions
154
314
# Create the inventory objects
155
315
for i, revid in enumerate(topo_sort(graph)):
156
316
if pb is not None:
157
317
pb.update("fetching revisions", i, len(graph))
158
root_tree = root_trees[revid]
159
318
rev = revisions[revid]
160
319
# We have to do this here, since we have to walk the tree and
161
# we need to make sure to import the blobs / trees with the riht
320
# we need to make sure to import the blobs / trees with the right
162
321
# path; this may involve adding them more than once.
164
inv.revision_id = rev.revision_id
165
def lookup_object(sha):
166
if sha in object_iter:
167
return object_iter[sha]
168
return target_git_object_retriever[sha]
169
parent_invs = [repo.get_inventory(r) for r in rev.parent_ids]
170
import_git_tree(repo, mapping, "", root_tree, inv, parent_invs,
171
target_git_object_retriever, lookup_object)
172
repo.add_revision(rev.revision_id, rev, inv)
175
class InterGitNonGitRepository(InterRepository):
177
_matching_repo_format = GitFormat()
323
for parent_id in rev.parent_ids:
325
parent_invs.append(parent_invs_cache[parent_id])
327
parent_inv = repo.get_inventory(parent_id)
328
parent_invs.append(parent_inv)
329
parent_invs_cache[parent_id] = parent_inv
330
if parent_invs == []:
331
base_inv = Inventory(root_id=None)
333
base_inv = parent_invs[0]
334
inv_delta, unusual_modes, shamap = import_git_tree(repo.texts,
335
mapping, "", root_trees[revid], base_inv, None, revid,
336
parent_invs, target_git_object_retriever._idmap, lookup_object)
337
target_git_object_retriever._idmap.add_entries(shamap)
338
if unusual_modes != {}:
339
for path, mode in unusual_modes.iteritems():
340
warn_unusual_mode(rev.foreign_revid, path, mode)
341
mapping.import_unusual_file_modes(rev, unusual_modes)
343
basis_id = rev.parent_ids[0]
345
basis_id = NULL_REVISION
346
rev.inventory_sha1, inv = repo.add_inventory_by_delta(basis_id,
347
inv_delta, rev.revision_id, rev.parent_ids)
348
parent_invs_cache[rev.revision_id] = inv
349
repo.add_revision(rev.revision_id, rev)
350
if "verify" in debug.debug_flags:
351
new_unusual_modes = mapping.export_unusual_file_modes(rev)
352
if new_unusual_modes != unusual_modes:
353
raise AssertionError("unusual modes don't match: %r != %r" % (unusual_modes, new_unusual_modes))
354
objs = inventory_to_tree_and_blobs(inv, repo.texts, mapping, unusual_modes)
355
for sha1, newobj, path in objs:
356
assert path is not None
357
oldobj = tree_lookup_path(lookup_object, root_trees[revid], path)
359
raise AssertionError("%r != %r in %s" % (oldobj, newobj, path))
361
target_git_object_retriever._idmap.commit()
364
class InterGitRepository(InterRepository):
366
_matching_repo_format = GitRepositoryFormat()
180
369
def _get_repo_format_to_test():
184
373
"""See InterRepository.copy_content."""
185
374
self.fetch(revision_id, pb, find_ghosts=False)
376
def fetch(self, revision_id=None, pb=None, find_ghosts=False, mapping=None,
378
self.fetch_refs(revision_id=revision_id, pb=pb, find_ghosts=find_ghosts,
379
mapping=mapping, fetch_spec=fetch_spec)
382
class InterGitNonGitRepository(InterGitRepository):
383
"""Base InterRepository that copies revisions from a Git into a non-Git
386
def fetch_refs(self, revision_id=None, pb=None, find_ghosts=False,
387
mapping=None, fetch_spec=None):
389
mapping = self.source.get_mapping()
390
if revision_id is not None:
391
interesting_heads = [revision_id]
392
elif fetch_spec is not None:
393
interesting_heads = fetch_spec.heads
395
interesting_heads = None
397
def determine_wants(refs):
399
if interesting_heads is None:
400
ret = [sha for (ref, sha) in refs.iteritems() if not ref.endswith("^{}")]
402
ret = [mapping.revision_id_bzr_to_foreign(revid)[0] for revid in interesting_heads if revid not in (None, NULL_REVISION)]
403
return [rev for rev in ret if not self.target.has_revision(mapping.revision_id_foreign_to_bzr(rev))]
404
self.fetch_objects(determine_wants, mapping, pb)
408
_GIT_PROGRESS_RE = re.compile(r"(.*?): +(\d+)% \((\d+)/(\d+)\)")
409
def report_git_progress(pb, text):
410
text = text.rstrip("\r\n")
411
g = _GIT_PROGRESS_RE.match(text)
413
(text, pct, current, total) = g.groups()
414
pb.update(text, int(current), int(total))
416
pb.update(text, 0, 0)
419
class InterRemoteGitNonGitRepository(InterGitNonGitRepository):
420
"""InterRepository that copies revisions from a remote Git into a non-Git
187
423
def fetch_objects(self, determine_wants, mapping, pb=None):
188
424
def progress(text):
189
pb.update("git: %s" % text.rstrip("\r\n"), 0, 0)
190
graph_walker = BzrFetchGraphWalker(self.target, mapping)
425
report_git_progress(pb, text)
426
store = BazaarObjectStore(self.target, mapping)
427
self.target.lock_write()
429
# FIXME: This should be more efficient
430
heads = self.target.get_graph().heads(self.target.all_revision_ids())
431
graph_walker = store.get_graph_walker(
432
[store._lookup_revision_sha1(head) for head in heads])
435
def record_determine_wants(heads):
436
wants = determine_wants(heads)
437
recorded_wants.extend(wants)
442
create_pb = pb = ui.ui_factory.nested_progress_bar()
444
self.target.start_write_group()
446
objects_iter = self.source.fetch_objects(
447
record_determine_wants, graph_walker,
448
store.get_raw, progress)
449
import_git_objects(self.target, mapping, objects_iter,
450
store, recorded_wants, pb)
452
self.target.commit_write_group()
460
def is_compatible(source, target):
461
"""Be compatible with GitRepository."""
462
# FIXME: Also check target uses VersionedFile
463
return (isinstance(source, RemoteGitRepository) and
464
target.supports_rich_root() and
465
not isinstance(target, GitRepository))
468
class InterLocalGitNonGitRepository(InterGitNonGitRepository):
469
"""InterRepository that copies revisions from a local Git into a non-Git
472
def fetch_objects(self, determine_wants, mapping, pb=None):
473
wants = determine_wants(self.source._git.get_refs())
193
476
create_pb = pb = ui.ui_factory.nested_progress_bar()
194
target_git_object_retriever = GitObjectConverter(self.target, mapping)
477
target_git_object_retriever = BazaarObjectStore(self.target, mapping)
197
479
self.target.lock_write()
199
481
self.target.start_write_group()
201
objects_iter = self.source.fetch_objects(determine_wants,
203
target_git_object_retriever.__getitem__,
205
import_git_objects(self.target, mapping, objects_iter,
206
target_git_object_retriever, pb)
483
import_git_objects(self.target, mapping,
484
self.source._git.object_store,
485
target_git_object_retriever, wants, pb)
208
487
self.target.commit_write_group()