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 (
25
from dulwich.object_store import (
37
from bzrlib.errors import (
42
from bzrlib.inventory import (
48
from bzrlib.lru_cache import (
51
from bzrlib.repository import (
54
from bzrlib.revision import (
57
from bzrlib.tsort import (
60
from bzrlib.versionedfile import (
61
FulltextContentFactory,
64
from bzrlib.plugins.git.mapping import (
66
inventory_to_tree_and_blobs,
69
from bzrlib.plugins.git.object_store import (
72
from bzrlib.plugins.git.remote import (
25
75
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):
82
def import_git_blob(texts, mapping, path, hexsha, base_inv, parent_id,
83
revision_id, parent_invs, shagitmap, lookup_object, executable, symlink):
73
84
"""Import a git blob object into a bzr repository.
75
:param repo: bzr repository
86
:param texts: VersionedFiles to add to
76
87
:param path: Path in the tree
77
88
:param blob: A git blob
89
:return: Inventory delta for this file
79
91
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)
96
# We just have to hope this is indeed utf-8:
97
ie = cls(file_id, urlutils.basename(path).decode("utf-8"), parent_id)
88
98
ie.executable = executable
91
def import_git_tree(repo, mapping, path, tree, inv, parent_invs, lookup_object):
100
# See if this has changed at all
102
base_ie = base_inv[file_id]
107
base_sha = base_ie.text_id
110
base_sha = shagitmap.lookup_blob(file_id, base_ie.revision)
114
if (base_sha == hexsha and base_ie.executable == ie.executable
115
and base_ie.kind == ie.kind):
116
# If nothing has changed since the base revision, we're done
118
if base_sha == hexsha:
119
ie.text_size = base_ie.text_size
120
ie.text_sha1 = base_ie.text_sha1
121
ie.symlink_target = base_ie.symlink_target
122
ie.revision = base_ie.revision
124
blob = lookup_object(hexsha)
125
if ie.kind == "symlink":
126
ie.symlink_target = blob.data
130
ie.text_size = len(blob.data)
131
ie.text_sha1 = osutils.sha_string(blob.data)
132
# Check what revision we should store
134
for pinv in parent_invs:
135
if pinv.revision_id == base_inv.revision_id:
144
if pie.text_sha1 == ie.text_sha1:
145
# found a revision in one of the parents to use
146
ie.revision = pie.revision
148
parent_keys.append((file_id, pie.revision))
149
if ie.revision is None:
150
# Need to store a new revision
151
ie.revision = revision_id
152
assert file_id is not None
153
assert ie.revision is not None
154
texts.insert_record_stream([FulltextContentFactory((file_id, ie.revision), tuple(parent_keys), ie.text_sha1, blob.data)])
155
shamap = [(hexsha, "blob", (ie.file_id, ie.revision))]
158
if file_id in base_inv:
159
old_path = base_inv.id2path(file_id)
162
return ([(old_path, path, file_id, ie)], shamap)
165
def import_git_tree(texts, mapping, path, hexsha, base_inv, parent_id,
166
revision_id, parent_invs, shagitmap, lookup_object):
92
167
"""Import a git tree object into a bzr repository.
94
:param repo: A Bzr repository object
169
:param texts: VersionedFiles object to add to
95
170
:param path: Path in the tree
96
171
:param tree: A git tree object
97
:param inv: Inventory object
172
:param base_inv: Base inventory against which to return inventory delta
173
:return: Inventory delta for this subtree
99
176
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
177
# We just have to hope this is indeed utf-8:
178
ie = InventoryDirectory(file_id, urlutils.basename(path.decode("utf-8")),
182
base_ie = base_inv[file_id]
184
# Newly appeared here
186
ie.revision = revision_id
187
texts.add_lines((file_id, ie.revision), (), [])
188
invdelta.append((None, path, file_id, ie))
190
base_sha = base_ie.text_id
191
# See if this has changed at all
194
base_sha = shagitmap.lookup_tree(file_id, base_inv.revision_id)
198
if base_sha == hexsha:
199
# If nothing has changed since the base revision, we're done
201
# Remember for next time
202
existing_children = set()
205
tree = lookup_object(hexsha)
206
for mode, name, child_hexsha in tree.entries():
108
207
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):
208
existing_children.add(basename)
209
child_path = osutils.pathjoin(path, name)
210
if stat.S_ISDIR(mode):
211
subinvdelta, grandchildmodes, subshamap = import_git_tree(texts,
212
mapping, child_path, child_hexsha, base_inv, file_id,
213
revision_id, parent_invs, shagitmap, lookup_object)
214
invdelta.extend(subinvdelta)
215
child_modes.update(grandchildmodes)
216
shamap.extend(subshamap)
218
fs_mode = stat.S_IMODE(mode)
219
symlink = stat.S_ISLNK(mode)
220
subinvdelta, subshamap = import_git_blob(texts, mapping,
221
child_path, child_hexsha, base_inv, file_id, revision_id,
222
parent_invs, shagitmap, lookup_object,
223
bool(fs_mode & 0111), symlink)
224
invdelta.extend(subinvdelta)
225
shamap.extend(subshamap)
226
if mode not in (stat.S_IFDIR, DEFAULT_FILE_MODE,
227
stat.S_IFLNK, DEFAULT_FILE_MODE|0111):
228
child_modes[child_path] = mode
229
# Remove any children that have disappeared
230
if base_ie is not None:
231
deletable = [v for k,v in base_ie.children.iteritems() if k not in existing_children]
234
invdelta.append((base_inv.id2path(ie.file_id), None, ie.file_id, None))
235
if ie.kind == "directory":
236
deletable.extend(ie.children.values())
237
shamap.append((hexsha, "tree", (file_id, revision_id)))
238
return invdelta, child_modes, shamap
241
def import_git_objects(repo, mapping, object_iter, target_git_object_retriever,
125
243
"""Import a set of git objects into a bzr repository.
127
245
:param repo: Bazaar repository
129
247
:param object_iter: Iterator over Git objects.
131
249
# TODO: a more (memory-)efficient implementation of this
133
for i, o in enumerate(object_iter):
135
pb.update("fetching objects", i)
255
parent_invs_cache = LRUCache(50)
140
256
# Find and convert commit objects
141
for o in objects.itervalues():
259
pb.update("finding revisions to fetch", len(graph), None)
261
assert isinstance(head, str)
263
o = object_iter[head]
142
266
if isinstance(o, Commit):
143
267
rev = mapping.import_commit(o)
144
root_trees[rev.revision_id] = objects[o.tree]
268
if repo.has_revision(rev.revision_id):
270
root_trees[rev.revision_id] = o.tree
145
271
revisions[rev.revision_id] = rev
146
272
graph.append((rev.revision_id, rev.parent_ids))
273
target_git_object_retriever._idmap.add_entry(o.id, "commit",
274
(rev.revision_id, o.tree))
275
heads.extend([p for p in o.parents if p not in checked])
276
elif isinstance(o, Tag):
277
heads.append(o.object[1])
279
trace.warning("Unable to import head object %r" % o)
147
281
# Order the revisions
148
282
# Create the inventory objects
149
283
for i, revid in enumerate(topo_sort(graph)):
150
284
if pb is not None:
151
285
pb.update("fetching revisions", i, len(graph))
152
root_tree = root_trees[revid]
153
286
rev = revisions[revid]
154
287
# 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
288
# we need to make sure to import the blobs / trees with the right
156
289
# path; this may involve adding them more than once.
158
inv.revision_id = rev.revision_id
159
290
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)
292
return object_iter[sha]
294
return target_git_object_retriever[sha]
296
for parent_id in rev.parent_ids:
298
parent_invs.append(parent_invs_cache[parent_id])
300
parent_inv = repo.get_inventory(parent_id)
301
parent_invs.append(parent_inv)
302
parent_invs_cache[parent_id] = parent_inv
303
if parent_invs == []:
304
base_inv = Inventory(root_id=None)
306
base_inv = parent_invs[0]
307
inv_delta, unusual_modes, shamap = import_git_tree(repo.texts,
308
mapping, "", root_trees[revid], base_inv, None, revid,
309
parent_invs, target_git_object_retriever._idmap, lookup_object)
310
target_git_object_retriever._idmap.add_entries(shamap)
311
if unusual_modes != {}:
312
ret = "unusual modes: \n"
313
for item in unusual_modes.iteritems():
314
ret += "\t%s: %o\n" % item
315
raise AssertionError(ret)
317
basis_id = rev.parent_ids[0]
319
basis_id = NULL_REVISION
320
rev.inventory_sha1, inv = repo.add_inventory_by_delta(basis_id,
321
inv_delta, rev.revision_id, rev.parent_ids)
322
parent_invs_cache[rev.revision_id] = inv
323
repo.add_revision(rev.revision_id, rev)
324
if "verify" in debug.debug_flags:
325
objs = inventory_to_tree_and_blobs(inv, repo.texts, mapping)
326
for sha1, newobj, path in objs:
327
assert path is not None
328
oldobj = tree_lookup_path(lookup_object, root_trees[revid], path)
329
assert oldobj == newobj, "%r != %r in %s" % (oldobj, newobj, path)
331
target_git_object_retriever._idmap.commit()
187
334
class InterGitRepository(InterRepository):
189
_matching_repo_format = GitFormat()
336
_matching_repo_format = GitRepositoryFormat()
192
339
def _get_repo_format_to_test():
196
343
"""See InterRepository.copy_content."""
197
344
self.fetch(revision_id, pb, find_ghosts=False)
199
def fetch(self, revision_id=None, pb=None, find_ghosts=False,
346
def fetch(self, revision_id=None, pb=None, find_ghosts=False, mapping=None,
348
self.fetch_refs(revision_id=revision_id, pb=pb, find_ghosts=find_ghosts,
349
mapping=mapping, fetch_spec=fetch_spec)
352
class InterGitNonGitRepository(InterGitRepository):
353
"""Base InterRepository that copies revisions from a Git into a non-Git
356
def fetch_refs(self, revision_id=None, pb=None, find_ghosts=False,
357
mapping=None, fetch_spec=None):
201
358
if mapping is None:
202
359
mapping = self.source.get_mapping()
204
pb.note("git: %s", text)
205
def determine_wants(heads):
206
if revision_id is None:
360
if revision_id is not None:
361
interesting_heads = [revision_id]
362
elif fetch_spec is not None:
363
interesting_heads = fetch_spec.heads
365
interesting_heads = None
367
def determine_wants(refs):
369
if interesting_heads is None:
370
ret = [sha for (ref, sha) in refs.iteritems() if not ref.endswith("^{}")]
209
ret = [mapping.revision_id_bzr_to_foreign(revision_id)]
372
ret = [mapping.revision_id_bzr_to_foreign(revid)[0] for revid in interesting_heads if revid not in (None, NULL_REVISION)]
210
373
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)
214
create_pb = pb = ui.ui_factory.nested_progress_bar()
216
self.target.lock_write()
218
self.target.start_write_group()
220
import_git_objects(self.target, mapping,
221
iter(self.source.fetch_objects(determine_wants, graph_walker,
224
self.target.commit_write_group()
232
def is_compatible(source, target):
233
"""Be compatible with GitRepository."""
234
# FIXME: Also check target uses VersionedFile
374
self.fetch_objects(determine_wants, mapping, pb)
378
class InterRemoteGitNonGitRepository(InterGitNonGitRepository):
379
"""InterRepository that copies revisions from a remote Git into a non-Git
382
def fetch_objects(self, determine_wants, mapping, pb=None):
384
pb.update("git: %s" % text.rstrip("\r\n"), 0, 0)
385
store = BazaarObjectStore(self.target, mapping)
386
self.target.lock_read()
388
heads = self.target.get_graph().heads(self.target.all_revision_ids())
391
graph_walker = store.get_graph_walker(
392
[store._lookup_revision_sha1(head) for head in heads])
395
create_pb = pb = ui.ui_factory.nested_progress_bar()
398
def record_determine_wants(heads):
399
wants = determine_wants(heads)
400
recorded_wants.extend(wants)
404
self.target.lock_write()
406
self.target.start_write_group()
408
objects_iter = self.source.fetch_objects(
409
record_determine_wants,
413
import_git_objects(self.target, mapping, objects_iter,
414
store, recorded_wants, pb)
416
self.target.commit_write_group()
424
def is_compatible(source, target):
425
"""Be compatible with GitRepository."""
426
# FIXME: Also check target uses VersionedFile
427
return (isinstance(source, RemoteGitRepository) and
428
target.supports_rich_root() and
429
not isinstance(target, GitRepository))
432
class InterLocalGitNonGitRepository(InterGitNonGitRepository):
433
"""InterRepository that copies revisions from a remote Git into a non-Git
436
def fetch_objects(self, determine_wants, mapping, pb=None):
437
wants = determine_wants(self.source._git.get_refs())
440
create_pb = pb = ui.ui_factory.nested_progress_bar()
441
target_git_object_retriever = BazaarObjectStore(self.target, mapping)
443
self.target.lock_write()
445
self.target.start_write_group()
447
import_git_objects(self.target, mapping,
448
self.source._git.object_store,
449
target_git_object_retriever, wants, pb)
451
self.target.commit_write_group()
459
def is_compatible(source, target):
460
"""Be compatible with GitRepository."""
461
# FIXME: Also check target uses VersionedFile
462
return (isinstance(source, LocalGitRepository) and
463
target.supports_rich_root() and
464
not isinstance(target, GitRepository))
467
class InterGitGitRepository(InterGitRepository):
468
"""InterRepository that copies between Git repositories."""
470
def fetch_refs(self, revision_id=None, pb=None, find_ghosts=False,
471
mapping=None, fetch_spec=None, branches=None):
473
mapping = self.source.get_mapping()
475
trace.info("git: %s", text)
477
if revision_id is not None:
478
args = [mapping.revision_id_bzr_to_foreign(revision_id)[0]]
479
elif fetch_spec is not None:
480
args = [mapping.revision_id_bzr_to_foreign(revid)[0] for revid in fetch_spec.heads]
481
if branches is not None:
482
determine_wants = lambda x: [x[y] for y in branches if not x[y] in r.object_store]
483
elif fetch_spec is None and revision_id is None:
484
determine_wants = r.object_store.determine_wants_all
486
determine_wants = lambda x: [y for y in args if not y in r.object_store]
488
graphwalker = r.get_graph_walker()
489
f, commit = r.object_store.add_thin_pack()
491
refs = self.source.fetch_pack(determine_wants, graphwalker,
500
def is_compatible(source, target):
501
"""Be compatible with GitRepository."""
235
502
return (isinstance(source, GitRepository) and
236
target.supports_rich_root())
503
isinstance(target, GitRepository))