126
113
if ie.kind == "symlink":
127
114
ie.revision = None
128
115
ie.symlink_target = blob.data
130
ie.text_size = sum(imap(len, blob.chunked))
131
ie.text_sha1 = osutils.sha_strings(blob.chunked)
119
ie.text_size = len(blob.data)
120
ie.text_sha1 = osutils.sha_string(blob.data)
132
121
# Check what revision we should store
134
for pinv in parent_invs:
139
if (pie.text_sha1 == ie.text_sha1 and
140
pie.executable == ie.executable and
141
pie.symlink_target == ie.symlink_target):
123
for pinv in parent_invs[1:]:
124
if pinv.revision_id == base_inv.revision_id:
133
if pie.text_sha1 == ie.text_sha1 and pie.executable == ie.executable and pie.symlink_target == ie.symlink_target:
142
134
# found a revision in one of the parents to use
143
135
ie.revision = pie.revision
145
parent_key = (file_id, pie.revision)
146
if not parent_key in parent_keys:
147
parent_keys.append(parent_key)
137
parent_keys.append((file_id, pie.revision))
148
138
if ie.revision is None:
149
139
# Need to store a new revision
150
140
ie.revision = revision_id
141
assert file_id is not None
151
142
assert ie.revision is not None
152
143
if ie.kind == 'symlink':
155
chunks = blob.chunked
156
texts.insert_record_stream([
157
ChunkedContentFactory((file_id, ie.revision),
158
tuple(parent_keys), ie.text_sha1, chunks)])
147
chunks = blob.chunked
148
except AttributeError: # older version of dulwich
150
texts.insert_record_stream([ChunkedContentFactory((file_id, ie.revision), tuple(parent_keys), ie.text_sha1, chunks)])
151
shamap = { ie.file_id: hexsha }
160
153
if base_hexsha is not None:
161
old_path = path.decode("utf-8") # Renames are not supported yet
154
old_path = path # Renames are not supported yet
162
155
if stat.S_ISDIR(base_mode):
163
invdelta.extend(remove_disappeared_children(base_inv, old_path,
164
lookup_object(base_hexsha), [], lookup_object))
156
invdelta.extend(remove_disappeared_children(base_inv, old_path, lookup_object(base_hexsha), [], lookup_object))
167
new_path = path.decode("utf-8")
168
invdelta.append((old_path, new_path, file_id, ie))
169
if base_hexsha != hexsha:
170
store_updater.add_object(blob, ie, path)
159
invdelta.append((old_path, path, file_id, ie))
160
return (invdelta, shamap)
174
163
class SubmodulesRequireSubtrees(BzrError):
191
180
ie.reference_revision = mapping.revision_id_foreign_to_bzr(hexsha)
192
texts.insert_record_stream([
193
ChunkedContentFactory((file_id, ie.revision), (), None, [])])
181
texts.insert_record_stream([ChunkedContentFactory((file_id, ie.revision), (), None, [])])
194
182
invdelta = [(oldpath, path, file_id, ie)]
183
return invdelta, {}, {}
198
186
def remove_disappeared_children(base_inv, path, base_tree, existing_children,
200
"""Generate an inventory delta for removed children.
202
:param base_inv: Base inventory against which to generate the
204
:param path: Path to process (unicode)
205
:param base_tree: Git Tree base object
206
:param existing_children: Children that still exist
207
:param lookup_object: Lookup a git object by its SHA1
208
:return: Inventory delta, as list
210
assert type(path) is unicode
212
189
for name, mode, hexsha in base_tree.iteritems():
213
190
if name in existing_children:
215
192
c_path = posixpath.join(path, name.decode("utf-8"))
216
file_id = base_inv.path2id(c_path)
217
assert file_id is not None
218
ret.append((c_path, None, file_id, None))
193
ret.append((c_path, None, base_inv.path2id(c_path), None))
219
194
if stat.S_ISDIR(mode):
220
195
ret.extend(remove_disappeared_children(
221
196
base_inv, c_path, lookup_object(hexsha), [], lookup_object))
225
200
def import_git_tree(texts, mapping, path, name, (base_hexsha, hexsha),
226
base_inv, parent_id, revision_id, parent_invs,
227
lookup_object, (base_mode, mode), store_updater,
228
lookup_file_id, allow_submodules=False):
201
base_inv, base_inv_shamap, parent_id, revision_id, parent_invs,
202
lookup_object, (base_mode, mode), allow_submodules=False):
229
203
"""Import a git tree object into a bzr repository.
231
205
:param texts: VersionedFiles object to add to
232
:param path: Path in the tree (str)
233
:param name: Name of the tree (str)
206
:param path: Path in the tree
234
207
:param tree: A git tree object
235
208
:param base_inv: Base inventory against which to return inventory delta
236
209
:return: Inventory delta for this subtree
238
assert type(path) is str
239
assert type(name) is str
240
211
if base_hexsha == hexsha and base_mode == mode:
241
212
# If nothing has changed since the base revision, we're done
244
file_id = lookup_file_id(path)
215
file_id = mapping.generate_file_id(path)
245
216
# We just have to hope this is indeed utf-8:
246
217
ie = InventoryDirectory(file_id, name.decode("utf-8"), parent_id)
247
218
tree = lookup_object(hexsha)
248
219
if base_hexsha is None:
250
old_path = None # Newly appeared here
252
222
base_tree = lookup_object(base_hexsha)
253
old_path = path.decode("utf-8") # Renames aren't supported yet
254
new_path = path.decode("utf-8")
255
if base_tree is None or type(base_tree) is not Tree:
256
ie.revision = revision_id
257
invdelta.append((old_path, new_path, ie.file_id, ie))
258
texts.insert_record_stream([
259
ChunkedContentFactory((ie.file_id, ie.revision), (), None, [])])
223
if base_tree is None:
224
# Newly appeared here
225
ie.revision = revision_id
226
texts.insert_record_stream([ChunkedContentFactory((file_id, ie.revision), (), None, [])])
228
invdelta.append((None, path, file_id, ie))
229
elif type(base_tree) is not Tree:
230
ie.revision = revision_id
231
old_path = path # Renames aren't supported yet
232
texts.insert_record_stream([ChunkedContentFactory((ie.file_id, ie.revision), (), None, [])])
233
invdelta.append((old_path, path, ie.file_id, ie))
235
old_path = path # Renames aren't supported yet
260
236
# Remember for next time
261
237
existing_children = set()
263
240
for child_mode, name, child_hexsha in tree.entries():
264
241
existing_children.add(name)
265
242
child_path = posixpath.join(path, name)
273
250
child_base_hexsha = None
274
251
child_base_mode = 0
275
252
if stat.S_ISDIR(child_mode):
276
subinvdelta, grandchildmodes = import_git_tree(texts, mapping,
277
child_path, name, (child_base_hexsha, child_hexsha), base_inv,
278
file_id, revision_id, parent_invs, lookup_object,
279
(child_base_mode, child_mode), store_updater, lookup_file_id,
280
allow_submodules=allow_submodules)
253
subinvdelta, grandchildmodes, subshamap = import_git_tree(
254
texts, mapping, child_path, name,
255
(child_base_hexsha, child_hexsha),
256
base_inv, base_inv_shamap,
257
file_id, revision_id, parent_invs, lookup_object,
258
(child_base_mode, child_mode),
259
allow_submodules=allow_submodules)
281
260
elif S_ISGITLINK(child_mode): # submodule
282
261
if not allow_submodules:
283
262
raise SubmodulesRequireSubtrees()
284
subinvdelta, grandchildmodes = import_git_submodule(texts, mapping,
285
child_path, name, (child_base_hexsha, child_hexsha), base_inv,
286
file_id, revision_id, parent_invs, lookup_object,
287
(child_base_mode, child_mode), store_updater, lookup_file_id)
263
subinvdelta, grandchildmodes, subshamap = import_git_submodule(
264
texts, mapping, child_path, name,
265
(child_base_hexsha, child_hexsha),
266
base_inv, file_id, revision_id, parent_invs, lookup_object,
267
(child_base_mode, child_mode))
289
subinvdelta = import_git_blob(texts, mapping, child_path, name,
290
(child_base_hexsha, child_hexsha), base_inv, file_id,
291
revision_id, parent_invs, lookup_object,
292
(child_base_mode, child_mode), store_updater, lookup_file_id)
269
subinvdelta, subshamap = import_git_blob(texts, mapping,
270
child_path, name, (child_base_hexsha, child_hexsha),
271
base_inv, base_inv_shamap,
273
revision_id, parent_invs, lookup_object,
274
(child_base_mode, child_mode))
293
275
grandchildmodes = {}
294
276
child_modes.update(grandchildmodes)
295
277
invdelta.extend(subinvdelta)
278
shamap.update(subshamap)
296
279
if child_mode not in (stat.S_IFDIR, DEFAULT_FILE_MODE,
297
280
stat.S_IFLNK, DEFAULT_FILE_MODE|0111):
298
281
child_modes[child_path] = child_mode
299
282
# Remove any children that have disappeared
300
283
if base_tree is not None and type(base_tree) is Tree:
301
invdelta.extend(remove_disappeared_children(base_inv, old_path,
284
invdelta.extend(remove_disappeared_children(base_inv, old_path,
302
285
base_tree, existing_children, lookup_object))
303
store_updater.add_object(tree, ie, path)
304
return invdelta, child_modes
307
def verify_commit_reconstruction(target_git_object_retriever, lookup_object,
308
o, rev, ret_tree, parent_trees, mapping, unusual_modes):
309
new_unusual_modes = mapping.export_unusual_file_modes(rev)
310
if new_unusual_modes != unusual_modes:
311
raise AssertionError("unusual modes don't match: %r != %r" % (
312
unusual_modes, new_unusual_modes))
313
# Verify that we can reconstruct the commit properly
314
rec_o = target_git_object_retriever._reconstruct_commit(rev, o.tree, True)
316
raise AssertionError("Reconstructed commit differs: %r != %r" % (
320
for path, obj, ie in _tree_to_objects(ret_tree, parent_trees,
321
target_git_object_retriever._cache.idmap, unusual_modes, mapping.BZR_DUMMY_FILE):
322
old_obj_id = tree_lookup_path(lookup_object, o.tree, path)[1]
324
if obj.id != old_obj_id:
325
diff.append((path, lookup_object(old_obj_id), obj))
326
for (path, old_obj, new_obj) in diff:
327
while (old_obj.type_name == "tree" and
328
new_obj.type_name == "tree" and
329
sorted(old_obj) == sorted(new_obj)):
331
if old_obj[name][0] != new_obj[name][0]:
332
raise AssertionError("Modes for %s differ: %o != %o" %
333
(path, old_obj[name][0], new_obj[name][0]))
334
if old_obj[name][1] != new_obj[name][1]:
335
# Found a differing child, delve deeper
336
path = posixpath.join(path, name)
337
old_obj = lookup_object(old_obj[name][1])
338
new_obj = new_objs[path]
340
raise AssertionError("objects differ for %s: %r != %r" % (path,
286
shamap[file_id] = hexsha
287
return invdelta, child_modes, shamap
344
290
def import_git_commit(repo, mapping, head, lookup_object,
345
target_git_object_retriever, trees_cache):
291
target_git_object_retriever, parent_invs_cache):
346
292
o = lookup_object(head)
347
rev, roundtrip_revid, verifiers = mapping.import_commit(o,
348
lambda x: target_git_object_retriever.lookup_git_sha(x)[1][0])
293
rev = mapping.import_commit(o)
349
294
# We have to do this here, since we have to walk the tree and
350
295
# we need to make sure to import the blobs / trees with the right
351
296
# path; this may involve adding them more than once.
352
parent_trees = trees_cache.revision_trees(rev.parent_ids)
353
if parent_trees == []:
297
parent_invs = parent_invs_cache.get_inventories(rev.parent_ids)
298
if parent_invs == []:
354
299
base_inv = Inventory(root_id=None)
300
base_inv_shamap = None # Should never be accessed
358
base_inv = parent_trees[0].inventory
304
base_inv = parent_invs[0]
305
base_inv_shamap = target_git_object_retriever._idmap.get_inventory_sha_map(base_inv.revision_id)
359
306
base_tree = lookup_object(o.parents[0]).tree
360
307
base_mode = stat.S_IFDIR
361
store_updater = target_git_object_retriever._get_updater(rev)
362
fileid_map = mapping.get_fileid_map(lookup_object, o.tree)
363
inv_delta, unusual_modes = import_git_tree(repo.texts,
364
mapping, "", "", (base_tree, o.tree), base_inv,
365
None, rev.revision_id, [p.inventory for p in parent_trees],
366
lookup_object, (base_mode, stat.S_IFDIR), store_updater,
367
fileid_map.lookup_file_id,
308
inv_delta, unusual_modes, shamap = import_git_tree(repo.texts,
309
mapping, "", u"", (base_tree, o.tree), base_inv, base_inv_shamap,
310
None, rev.revision_id, parent_invs, lookup_object,
311
(base_mode, stat.S_IFDIR),
368
312
allow_submodules=getattr(repo._format, "supports_tree_reference", False))
314
for (oldpath, newpath, fileid, new_ie) in inv_delta:
316
entries.append((fileid, None, None, None))
318
if new_ie.kind in ("file", "symlink"):
319
entries.append((fileid, "blob", shamap[fileid], new_ie.revision))
320
elif new_ie.kind == "directory":
321
entries.append((fileid, "tree", shamap[fileid], rev.revision_id))
324
target_git_object_retriever._idmap.add_entries(rev.revision_id,
325
rev.parent_ids, head, o.tree, entries)
369
326
if unusual_modes != {}:
370
327
for path, mode in unusual_modes.iteritems():
371
328
warn_unusual_mode(rev.foreign_revid, path, mode)
376
333
basis_id = NULL_REVISION
378
335
rev.inventory_sha1, inv = repo.add_inventory_by_delta(basis_id,
379
inv_delta, rev.revision_id, rev.parent_ids, base_inv)
380
# FIXME: Check verifiers
381
testament = StrictTestament3(rev, inv)
382
calculated_verifiers = { "testament3-sha1": testament.as_sha1() }
383
if roundtrip_revid is not None:
384
original_revid = rev.revision_id
385
rev.revision_id = roundtrip_revid
386
if calculated_verifiers != verifiers:
387
trace.mutter("Testament SHA1 %r for %r did not match %r.",
388
calculated_verifiers["testament3-sha1"],
389
rev.revision_id, verifiers["testament3-sha1"])
390
rev.revision_id = original_revid
391
store_updater.add_object(o, calculated_verifiers, None)
392
store_updater.finish()
393
ret_tree = RevisionTree(repo, inv, rev.revision_id)
394
trees_cache.add(ret_tree)
336
inv_delta, rev.revision_id, rev.parent_ids,
338
parent_invs_cache.add(rev.revision_id, inv)
395
339
repo.add_revision(rev.revision_id, rev)
396
340
if "verify" in debug.debug_flags:
397
verify_commit_reconstruction(target_git_object_retriever,
398
lookup_object, o, rev, ret_tree, parent_trees, mapping,
341
new_unusual_modes = mapping.export_unusual_file_modes(rev)
342
if new_unusual_modes != unusual_modes:
343
raise AssertionError("unusual modes don't match: %r != %r" % (unusual_modes, new_unusual_modes))
344
objs = inventory_to_tree_and_blobs(inv, repo.texts, mapping, unusual_modes)
345
for newsha1, newobj, path in objs:
346
assert path is not None
350
(oldmode, oldsha1) = tree_lookup_path(lookup_object, o.tree, path)
351
if oldsha1 != newsha1:
352
raise AssertionError("%r != %r in %s" % (oldsha1, newsha1, path))
402
355
def import_git_objects(repo, mapping, object_iter,