def test_read_objects(self):
f = BytesIO()
entries = build_pack(f, [
(Blob.type_num, 'blob'),
(OFS_DELTA, (0, 'blob1')),
])
reader = PackStreamReader(f.read)
objects = list(reader.read_objects(compute_crc32=True))
self.assertEqual(2, len(objects))
unpacked_blob, unpacked_delta = objects
self.assertEqual(entries[0][0], unpacked_blob.offset)
self.assertEqual(Blob.type_num, unpacked_blob.pack_type_num)
self.assertEqual(Blob.type_num, unpacked_blob.obj_type_num)
self.assertEqual(None, unpacked_blob.delta_base)
self.assertEqual('blob', ''.join(unpacked_blob.decomp_chunks))
self.assertEqual(entries[0][4], unpacked_blob.crc32)
self.assertEqual(entries[1][0], unpacked_delta.offset)
self.assertEqual(OFS_DELTA, unpacked_delta.pack_type_num)
self.assertEqual(None, unpacked_delta.obj_type_num)
self.assertEqual(unpacked_delta.offset - unpacked_blob.offset,
unpacked_delta.delta_base)
delta = create_delta('blob', 'blob1')
self.assertEqual(delta, ''.join(unpacked_delta.decomp_chunks))
self.assertEqual(entries[1][4], unpacked_delta.crc32)
def test_read_objects(self):
f = BytesIO()
entries = build_pack(f, [
(Blob.type_num, 'blob'),
(OFS_DELTA, (0, 'blob1')),
])
reader = PackStreamReader(f.read)
objects = list(reader.read_objects(compute_crc32=True))
self.assertEqual(2, len(objects))
unpacked_blob, unpacked_delta = objects
self.assertEqual(entries[0][0], unpacked_blob.offset)
self.assertEqual(Blob.type_num, unpacked_blob.pack_type_num)
self.assertEqual(Blob.type_num, unpacked_blob.obj_type_num)
self.assertEqual(None, unpacked_blob.delta_base)
self.assertEqual('blob', ''.join(unpacked_blob.decomp_chunks))
self.assertEqual(entries[0][4], unpacked_blob.crc32)
self.assertEqual(entries[1][0], unpacked_delta.offset)
self.assertEqual(OFS_DELTA, unpacked_delta.pack_type_num)
self.assertEqual(None, unpacked_delta.obj_type_num)
self.assertEqual(unpacked_delta.offset - unpacked_blob.offset,
unpacked_delta.delta_base)
delta = create_delta('blob', 'blob1')
self.assertEqual(delta, ''.join(unpacked_delta.decomp_chunks))
self.assertEqual(entries[1][4], unpacked_delta.crc32)
def test_simple_delta(self):
b1 = Blob.from_string("a" * 101)
b2 = Blob.from_string("a" * 100)
delta = create_delta(b1.as_raw_string(), b2.as_raw_string())
self.assertEqual(
[(b1.type_num, b1.sha().digest(), None, b1.as_raw_string()),
(b2.type_num, b2.sha().digest(), b1.sha().digest(), delta)],
list(deltify_pack_objects([(b1, ""), (b2, "")])))
def test_simple_delta(self):
b1 = Blob.from_string("a" * 101)
b2 = Blob.from_string("a" * 100)
delta = create_delta(b1.as_raw_string(), b2.as_raw_string())
self.assertEqual([
(b1.type_num, b1.sha().digest(), None, b1.as_raw_string()),
(b2.type_num, b2.sha().digest(), b1.sha().digest(), delta)
],
list(deltify_pack_objects([(b1, ""), (b2, "")])))
def _test_roundtrip(self, base, target):
self.assertEquals(target, "".join(apply_delta(base, create_delta(base, target))))
def build_pack(f, objects_spec, store=None):
"""Write test pack data from a concise spec.
Args:
f: A file-like object to write the pack to.
objects_spec: A list of (type_num, obj). For non-delta types, obj
is the string of that object's data.
For delta types, obj is a tuple of (base, data), where:
* base can be either an index in objects_spec of the base for that
* delta; or for a ref delta, a SHA, in which case the resulting pack
* will be thin and the base will be an external ref.
* data is a string of the full, non-deltified data for that object.
Note that offsets/refs and deltas are computed within this function.
store: An optional ObjectStore for looking up external refs.
Returns: A list of tuples in the order specified by objects_spec:
(offset, type num, data, sha, CRC32)
"""
sf = SHA1Writer(f)
num_objects = len(objects_spec)
write_pack_header(sf, num_objects)
full_objects = {}
offsets = {}
crc32s = {}
while len(full_objects) < num_objects:
for i, (type_num, data) in enumerate(objects_spec):
if type_num not in DELTA_TYPES:
full_objects[i] = (type_num, data, obj_sha(type_num, [data]))
continue
base, data = data
if isinstance(base, int):
if base not in full_objects:
continue
base_type_num, _, _ = full_objects[base]
else:
base_type_num, _ = store.get_raw(base)
full_objects[i] = (
base_type_num,
data,
obj_sha(base_type_num, [data]),
)
for i, (type_num, obj) in enumerate(objects_spec):
offset = f.tell()
if type_num == OFS_DELTA:
base_index, data = obj
base = offset - offsets[base_index]
_, base_data, _ = full_objects[base_index]
obj = (base, create_delta(base_data, data))
elif type_num == REF_DELTA:
base_ref, data = obj
if isinstance(base_ref, int):
_, base_data, base = full_objects[base_ref]
else:
base_type_num, base_data = store.get_raw(base_ref)
base = obj_sha(base_type_num, base_data)
obj = (base, create_delta(base_data, data))
crc32 = write_pack_object(sf, type_num, obj)
offsets[i] = offset
crc32s[i] = crc32
expected = []
for i in range(num_objects):
type_num, data, sha = full_objects[i]
assert len(sha) == 20
expected.append((offsets[i], type_num, data, sha, crc32s[i]))
sf.write_sha()
f.seek(0)
return expected
def import_git_commit(self, commit):
self.ui.debug(_("importing: %s\n") % commit.id)
(strip_message, hg_renames,
hg_branch, extra) = self.extract_hg_metadata(commit.message)
# get a list of the changed, added, removed files
files = self.get_files_changed(commit)
date = (commit.author_time, -commit.author_timezone)
text = strip_message
origtext = text
try:
text.decode('utf-8')
except UnicodeDecodeError:
text = self.decode_guess(text, commit.encoding)
text = '\n'.join([l.rstrip() for l in text.splitlines()]).strip('\n')
if text + '\n' != origtext:
extra['message'] = create_delta(text +'\n', origtext)
author = commit.author
# convert extra data back to the end
if ' ext:' in commit.author:
regex = re.compile('^(.*?)\ ext:\((.*)\) <(.*)\>$')
m = regex.match(commit.author)
if m:
name = m.group(1)
ex = urllib.unquote(m.group(2))
email = m.group(3)
author = name + ' <' + email + '>' + ex
if ' <none@none>' in commit.author:
author = commit.author[:-12]
try:
author.decode('utf-8')
except UnicodeDecodeError:
origauthor = author
author = self.decode_guess(author, commit.encoding)
extra['author'] = create_delta(author, origauthor)
oldenc = self.swap_out_encoding()
def getfilectx(repo, memctx, f):
delete, mode, sha = files[f]
if delete:
raise IOError
data = self.git[sha].data
copied_path = hg_renames.get(f)
e = self.convert_git_int_mode(mode)
return context.memfilectx(f, data, 'l' in e, 'x' in e, copied_path)
gparents = map(self.map_hg_get, commit.parents)
p1, p2 = (nullid, nullid)
octopus = False
if len(gparents) > 1:
# merge, possibly octopus
def commit_octopus(p1, p2):
ctx = context.memctx(self.repo, (p1, p2), text, list(files), getfilectx,
author, date, {'hg-git': 'octopus'})
return hex(self.repo.commitctx(ctx))
octopus = len(gparents) > 2
p2 = gparents.pop()
p1 = gparents.pop()
while len(gparents) > 0:
p2 = commit_octopus(p1, p2)
p1 = gparents.pop()
else:
if gparents:
p1 = gparents.pop()
pa = None
if not (p2 == nullid):
node1 = self.repo.changectx(p1)
node2 = self.repo.changectx(p2)
pa = node1.ancestor(node2)
# if named branch, add to extra
if hg_branch:
extra['branch'] = hg_branch
# if committer is different than author, add it to extra
if commit.author != commit.committer \
or commit.author_time != commit.commit_time \
or commit.author_timezone != commit.commit_timezone:
extra['committer'] = "%s %d %d" % (
commit.committer, commit.commit_time, -commit.commit_timezone)
if commit.encoding:
extra['encoding'] = commit.encoding
if hg_branch:
extra['branch'] = hg_branch
if octopus:
extra['hg-git'] ='octopus-done'
# TODO use 'n in self.repo' when we require hg 1.5
def repo_contains(n):
try:
return bool(self.repo.lookup(n))
except error.RepoLookupError:
return False
if not (repo_contains(p1) and repo_contains(p2)):
raise hgutil.Abort(_('you appear to have run strip - '
'please run hg git-cleanup'))
ctx = context.memctx(self.repo, (p1, p2), text, list(files), getfilectx,
author, date, extra)
node = self.repo.commitctx(ctx)
self.swap_out_encoding(oldenc)
# save changeset to mapping file
cs = hex(node)
self.map_set(commit.id, cs)
def build_pack(f, objects_spec, store=None):
"""Write test pack data from a concise spec.
:param f: A file-like object to write the pack to.
:param objects_spec: A list of (type_num, obj). For non-delta types, obj
is the string of that object's data.
For delta types, obj is a tuple of (base, data), where:
* base can be either an index in objects_spec of the base for that
* delta; or for a ref delta, a SHA, in which case the resulting pack
* will be thin and the base will be an external ref.
* data is a string of the full, non-deltified data for that object.
Note that offsets/refs and deltas are computed within this function.
:param store: An optional ObjectStore for looking up external refs.
:return: A list of tuples in the order specified by objects_spec:
(offset, type num, data, sha, CRC32)
"""
sf = SHA1Writer(f)
num_objects = len(objects_spec)
write_pack_header(sf, num_objects)
full_objects = {}
offsets = {}
crc32s = {}
while len(full_objects) < num_objects:
for i, (type_num, data) in enumerate(objects_spec):
if type_num not in DELTA_TYPES:
full_objects[i] = (type_num, data,
obj_sha(type_num, [data]))
continue
base, data = data
if isinstance(base, int):
if base not in full_objects:
continue
base_type_num, _, _ = full_objects[base]
else:
base_type_num, _ = store.get_raw(base)
full_objects[i] = (base_type_num, data,
obj_sha(base_type_num, [data]))
for i, (type_num, obj) in enumerate(objects_spec):
offset = f.tell()
if type_num == OFS_DELTA:
base_index, data = obj
base = offset - offsets[base_index]
_, base_data, _ = full_objects[base_index]
obj = (base, create_delta(base_data, data))
elif type_num == REF_DELTA:
base_ref, data = obj
if isinstance(base_ref, int):
_, base_data, base = full_objects[base_ref]
else:
base_type_num, base_data = store.get_raw(base_ref)
base = obj_sha(base_type_num, base_data)
obj = (base, create_delta(base_data, data))
crc32 = write_pack_object(sf, type_num, obj)
offsets[i] = offset
crc32s[i] = crc32
expected = []
for i in range(num_objects):
type_num, data, sha = full_objects[i]
expected.append((offsets[i], type_num, data, sha, crc32s[i]))
sf.write_sha()
f.seek(0)
return expected
def _test_roundtrip(self, base, target):
self.assertEqual(target,
''.join(apply_delta(base, create_delta(base, target))))
def import_git_commit(self, commit):
self.ui.debug(_("importing: %s\n") % commit.id)
(strip_message, hg_renames,
hg_branch, extra) = self.extract_hg_metadata(commit.message)
# get a list of the changed, added, removed files
files = self.get_files_changed(commit)
date = (commit.author_time, -commit.author_timezone)
text = strip_message
origtext = text
try:
text.decode('utf-8')
except UnicodeDecodeError:
text = self.decode_guess(text, commit.encoding)
text = '\n'.join([l.rstrip() for l in text.splitlines()]).strip('\n')
if text + '\n' != origtext:
extra['message'] = create_delta(text +'\n', origtext)
author = commit.author
# convert extra data back to the end
if ' ext:' in commit.author:
regex = re.compile('^(.*?)\ ext:\((.*)\) <(.*)\>$')
m = regex.match(commit.author)
if m:
name = m.group(1)
ex = urllib.unquote(m.group(2))
email = m.group(3)
author = name + ' <' + email + '>' + ex
if ' <none@none>' in commit.author:
author = commit.author[:-12]
try:
author.decode('utf-8')
except UnicodeDecodeError:
origauthor = author
author = self.decode_guess(author, commit.encoding)
extra['author'] = create_delta(author, origauthor)
oldenc = self.swap_out_encoding()
def findconvergedfiles(p1, p2):
# If any files have the same contents in both parents of a merge
# (and are therefore not reported as changed by Git) but are at
# different file revisions in Mercurial (because they arrived at
# those contents in different ways), we need to include them in
# the list of changed files so that Mercurial can join up their
# filelog histories (same as if the merge was done in Mercurial to
# begin with).
if p2 == nullid:
return []
manifest1 = self.repo.changectx(p1).manifest()
manifest2 = self.repo.changectx(p2).manifest()
return [path for path, node1 in manifest1.iteritems()
if path not in files and manifest2.get(path, node1) != node1]
def getfilectx(repo, memctx, f):
info = files.get(f)
if info != None:
# it's a file reported as modified from Git
delete, mode, sha = info
if delete:
raise IOError
data = self.git[sha].data
copied_path = hg_renames.get(f)
e = self.convert_git_int_mode(mode)
else:
# it's a converged file
fc = context.filectx(self.repo, f, changeid=memctx.p1().rev())
data = fc.data()
e = fc.flags()
copied_path = fc.renamed()
return context.memfilectx(f, data, 'l' in e, 'x' in e, copied_path)
gparents = map(self.map_hg_get, commit.parents)
p1, p2 = (nullid, nullid)
octopus = False
if len(gparents) > 1:
# merge, possibly octopus
def commit_octopus(p1, p2):
ctx = context.memctx(self.repo, (p1, p2), text,
list(files) + findconvergedfiles(p1, p2),
getfilectx, author, date, {'hg-git': 'octopus'})
return hex(self.repo.commitctx(ctx))
octopus = len(gparents) > 2
p2 = gparents.pop()
p1 = gparents.pop()
while len(gparents) > 0:
p2 = commit_octopus(p1, p2)
p1 = gparents.pop()
else:
if gparents:
p1 = gparents.pop()
pa = None
if not (p2 == nullid):
node1 = self.repo.changectx(p1)
node2 = self.repo.changectx(p2)
pa = node1.ancestor(node2)
# if named branch, add to extra
if hg_branch:
extra['branch'] = hg_branch
# if committer is different than author, add it to extra
if commit.author != commit.committer \
or commit.author_time != commit.commit_time \
or commit.author_timezone != commit.commit_timezone:
extra['committer'] = "%s %d %d" % (
commit.committer, commit.commit_time, -commit.commit_timezone)
if commit.encoding:
extra['encoding'] = commit.encoding
if hg_branch:
extra['branch'] = hg_branch
if octopus:
extra['hg-git'] ='octopus-done'
# TODO use 'n in self.repo' when we require hg 1.5
def repo_contains(n):
try:
return bool(self.repo.lookup(n))
except error.RepoLookupError:
return False
if not (repo_contains(p1) and repo_contains(p2)):
raise hgutil.Abort(_('you appear to have run strip - '
'please run hg git-cleanup'))
ctx = context.memctx(self.repo, (p1, p2), text,
list(files) + findconvergedfiles(p1, p2),
getfilectx, author, date, extra)
node = self.repo.commitctx(ctx)
self.swap_out_encoding(oldenc)
# save changeset to mapping file
cs = hex(node)
self.map_set(commit.id, cs)
def _test_roundtrip(self, base, target):
self.assertEquals([target],
apply_delta(base, create_delta(base, target)))
def import_git_commit(self, commit):
self.ui.debug(_("importing: %s\n") % commit.id)
(strip_message, hg_renames,
hg_branch, extra) = self.extract_hg_metadata(commit.message)
# get a list of the changed, added, removed files
files = self.get_files_changed(commit)
date = (commit.author_time, -commit.author_timezone)
text = strip_message
origtext = text
try:
text.decode('utf-8')
except UnicodeDecodeError:
text = self.decode_guess(text, commit.encoding)
text = '\n'.join([l.rstrip() for l in text.splitlines()]).strip('\n')
if text + '\n' != origtext:
extra['message'] = create_delta(text +'\n', origtext)
author = commit.author
# convert extra data back to the end
if ' ext:' in commit.author:
regex = re.compile('^(.*?)\ ext:\((.*)\) <(.*)\>$')
m = regex.match(commit.author)
if m:
name = m.group(1)
ex = urllib.unquote(m.group(2))
email = m.group(3)
author = name + ' <' + email + '>' + ex
if ' <none@none>' in commit.author:
author = commit.author[:-12]
try:
author.decode('utf-8')
except UnicodeDecodeError:
origauthor = author
author = self.decode_guess(author, commit.encoding)
extra['author'] = create_delta(author, origauthor)
oldenc = self.swap_out_encoding()
def findconvergedfiles(p1, p2):
# If any files have the same contents in both parents of a merge
# (and are therefore not reported as changed by Git) but are at
# different file revisions in Mercurial (because they arrived at
# those contents in different ways), we need to include them in
# the list of changed files so that Mercurial can join up their
# filelog histories (same as if the merge was done in Mercurial to
# begin with).
if p2 == nullid:
return []
manifest1 = self.repo.changectx(p1).manifest()
manifest2 = self.repo.changectx(p2).manifest()
return [path for path, node1 in manifest1.iteritems()
if path not in files and manifest2.get(path, node1) != node1]
def getfilectx(repo, memctx, f):
info = files.get(f)
if info != None:
# it's a file reported as modified from Git
delete, mode, sha = info
if delete:
raise IOError
data = self.git[sha].data
copied_path = hg_renames.get(f)
e = self.convert_git_int_mode(mode)
else:
# it's a converged file
fc = context.filectx(self.repo, f, changeid=memctx.p1().rev())
data = fc.data()
e = fc.flags()
copied_path = fc.renamed()
return context.memfilectx(f, data, 'l' in e, 'x' in e, copied_path)
gparents = map(self.map_hg_get, commit.parents)
p1, p2 = (nullid, nullid)
octopus = False
if len(gparents) > 1:
# merge, possibly octopus
def commit_octopus(p1, p2):
ctx = context.memctx(self.repo, (p1, p2), text,
list(files) + findconvergedfiles(p1, p2),
getfilectx, author, date, {'hg-git': 'octopus'})
return hex(self.repo.commitctx(ctx))
octopus = len(gparents) > 2
p2 = gparents.pop()
p1 = gparents.pop()
while len(gparents) > 0:
p2 = commit_octopus(p1, p2)
p1 = gparents.pop()
else:
if gparents:
p1 = gparents.pop()
pa = None
if not (p2 == nullid):
node1 = self.repo.changectx(p1)
node2 = self.repo.changectx(p2)
pa = node1.ancestor(node2)
# if named branch, add to extra
if hg_branch:
extra['branch'] = hg_branch
# if committer is different than author, add it to extra
if commit.author != commit.committer \
or commit.author_time != commit.commit_time \
or commit.author_timezone != commit.commit_timezone:
extra['committer'] = "%s %d %d" % (
commit.committer, commit.commit_time, -commit.commit_timezone)
if commit.encoding:
extra['encoding'] = commit.encoding
if hg_branch:
extra['branch'] = hg_branch
if octopus:
extra['hg-git'] ='octopus-done'
# TODO use 'n in self.repo' when we require hg 1.5
def repo_contains(n):
try:
return bool(self.repo.lookup(n))
except error.RepoLookupError:
return False
if not (repo_contains(p1) and repo_contains(p2)):
raise hgutil.Abort(_('you appear to have run strip - '
'please run hg git-cleanup'))
ctx = context.memctx(self.repo, (p1, p2), text,
list(files) + findconvergedfiles(p1, p2),
getfilectx, author, date, extra)
node = self.repo.commitctx(ctx)
self.swap_out_encoding(oldenc)
# save changeset to mapping file
cs = hex(node)
self.map_set(commit.id, cs)
def import_git_commit(self, commit):
self.ui.debug(_("importing: %s\n") % commit.id)
(strip_message, hg_renames, hg_branch,
extra) = self.extract_hg_metadata(commit.message)
# get a list of the changed, added, removed files
files = self.get_files_changed(commit)
date = (commit.author_time, -commit.author_timezone)
text = strip_message
origtext = text
try:
text.decode('utf-8')
except UnicodeDecodeError:
text = self.decode_guess(text, commit.encoding)
text = '\n'.join([l.rstrip() for l in text.splitlines()]).strip('\n')
if text + '\n' != origtext:
extra['message'] = create_delta(text + '\n', origtext)
author = commit.author
# convert extra data back to the end
if ' ext:' in commit.author:
regex = re.compile('^(.*?)\ ext:\((.*)\) <(.*)\>$')
m = regex.match(commit.author)
if m:
name = m.group(1)
ex = urllib.unquote(m.group(2))
email = m.group(3)
author = name + ' <' + email + '>' + ex
if ' <none@none>' in commit.author:
author = commit.author[:-12]
try:
author.decode('utf-8')
except UnicodeDecodeError:
origauthor = author
author = self.decode_guess(author, commit.encoding)
extra['author'] = create_delta(author, origauthor)
oldenc = self.swap_out_encoding()
def getfilectx(repo, memctx, f):
delete, mode, sha = files[f]
if delete:
raise IOError
data = self.git[sha].data
copied_path = hg_renames.get(f)
e = self.convert_git_int_mode(mode)
return context.memfilectx(f, data, 'l' in e, 'x' in e, copied_path)
gparents = map(self.map_hg_get, commit.parents)
p1, p2 = (nullid, nullid)
octopus = False
if len(gparents) > 1:
# merge, possibly octopus
def commit_octopus(p1, p2):
ctx = context.memctx(self.repo, (p1, p2), text, list(files),
getfilectx, author, date,
{'hg-git': 'octopus'})
return hex(self.repo.commitctx(ctx))
octopus = len(gparents) > 2
p2 = gparents.pop()
p1 = gparents.pop()
while len(gparents) > 0:
p2 = commit_octopus(p1, p2)
p1 = gparents.pop()
else:
if gparents:
p1 = gparents.pop()
pa = None
if not (p2 == nullid):
node1 = self.repo.changectx(p1)
node2 = self.repo.changectx(p2)
pa = node1.ancestor(node2)
# if named branch, add to extra
if hg_branch:
extra['branch'] = hg_branch
# if committer is different than author, add it to extra
if commit.author != commit.committer \
or commit.author_time != commit.commit_time \
or commit.author_timezone != commit.commit_timezone:
extra['committer'] = "%s %d %d" % (
commit.committer, commit.commit_time, -commit.commit_timezone)
if commit.encoding:
extra['encoding'] = commit.encoding
if hg_branch:
extra['branch'] = hg_branch
if octopus:
extra['hg-git'] = 'octopus-done'
# TODO use 'n in self.repo' when we require hg 1.5
def repo_contains(n):
try:
return bool(self.repo.lookup(n))
except error.RepoLookupError:
return False
if not (repo_contains(p1) and repo_contains(p2)):
raise hgutil.Abort(
_('you appear to have run strip - '
'please run hg git-cleanup'))
ctx = context.memctx(self.repo, (p1, p2), text, list(files),
getfilectx, author, date, extra)
node = self.repo.commitctx(ctx)
self.swap_out_encoding(oldenc)
# save changeset to mapping file
cs = hex(node)
self.map_set(commit.id, cs)
def import_git_commit(self, commit):
self.ui.debug(_("importing: %s\n") % commit.id)
(strip_message, hg_renames,
hg_branch, extra) = self.extract_hg_metadata(commit.message)
# get a list of the changed, added, removed files
files = self.get_files_changed(commit)
# Handle gitlinks: collect
gitlinks = self.collect_gitlinks(commit.tree)
git_commit_tree = self.git[commit.tree]
# Analyze hgsubstate and build an updated version
# using SHAs from gitlinks
hgsubstate = None
if gitlinks:
hgsubstate = util.parse_hgsubstate(self.git_file_readlines(git_commit_tree, '.hgsubstate'))
for path, sha in gitlinks:
hgsubstate[path] = sha
# in case .hgsubstate wasn't among changed files
# force its inclusion
files['.hgsubstate'] = (False, 0100644, None)
# Analyze .hgsub and merge with .gitmodules
hgsub = None
gitmodules = self.parse_gitmodules(git_commit_tree)
if gitmodules or gitlinks:
hgsub = util.parse_hgsub(self.git_file_readlines(git_commit_tree, '.hgsub'))
for (sm_path, sm_url, sm_name) in gitmodules:
hgsub[sm_path] = '[git]' + sm_url
files['.hgsub'] = (False, 0100644, None)
elif commit.parents and '.gitmodules' in self.git[self.git[commit.parents[0]].tree]:
# no .gitmodules in this commit, however present in the parent
# mark its hg counterpart as deleted (assuming .hgsub is there
# due to the same import_git_commit process
files['.hgsub'] = (True, 0100644, None)
date = (commit.author_time, -commit.author_timezone)
text = strip_message
origtext = text
try:
text.decode('utf-8')
except UnicodeDecodeError:
text = self.decode_guess(text, commit.encoding)
text = '\n'.join([l.rstrip() for l in text.splitlines()]).strip('\n')
if text + '\n' != origtext:
extra['message'] = create_delta(text +'\n', origtext)
author = commit.author
# convert extra data back to the end
if ' ext:' in commit.author:
m = RE_GIT_AUTHOR_EXTRA.match(commit.author)
if m:
name = m.group(1)
ex = urllib.unquote(m.group(2))
email = m.group(3)
author = name + ' <' + email + '>' + ex
if ' <none@none>' in commit.author:
author = commit.author[:-12]
try:
author.decode('utf-8')
except UnicodeDecodeError:
origauthor = author
author = self.decode_guess(author, commit.encoding)
extra['author'] = create_delta(author, origauthor)
oldenc = self.swap_out_encoding()
def findconvergedfiles(p1, p2):
# If any files have the same contents in both parents of a merge
# (and are therefore not reported as changed by Git) but are at
# different file revisions in Mercurial (because they arrived at
# those contents in different ways), we need to include them in
# the list of changed files so that Mercurial can join up their
# filelog histories (same as if the merge was done in Mercurial to
# begin with).
if p2 == nullid:
return []
manifest1 = self.repo.changectx(p1).manifest()
manifest2 = self.repo.changectx(p2).manifest()
return [path for path, node1 in manifest1.iteritems()
if path not in files and manifest2.get(path, node1) != node1]
def getfilectx(repo, memctx, f):
info = files.get(f)
if info != None:
# it's a file reported as modified from Git
delete, mode, sha = info
if delete:
raise IOError
if not sha: # indicates there's no git counterpart
e = ''
copied_path = None
if '.hgsubstate' == f:
data = util.serialize_hgsubstate(hgsubstate)
elif '.hgsub' == f:
data = util.serialize_hgsub(hgsub)
else:
data = self.git[sha].data
copied_path = hg_renames.get(f)
e = self.convert_git_int_mode(mode)
else:
# it's a converged file
fc = context.filectx(self.repo, f, changeid=memctx.p1().rev())
data = fc.data()
e = fc.flags()
copied_path = fc.renamed()
return context.memfilectx(f, data, 'l' in e, 'x' in e, copied_path)
gparents = map(self.map_hg_get, commit.parents)
p1, p2 = (nullid, nullid)
octopus = False
if len(gparents) > 1:
# merge, possibly octopus
def commit_octopus(p1, p2):
ctx = context.memctx(self.repo, (p1, p2), text,
list(files) + findconvergedfiles(p1, p2),
getfilectx, author, date, {'hg-git': 'octopus'})
return hex(self.repo.commitctx(ctx))
octopus = len(gparents) > 2
p2 = gparents.pop()
p1 = gparents.pop()
while len(gparents) > 0:
p2 = commit_octopus(p1, p2)
p1 = gparents.pop()
else:
if gparents:
p1 = gparents.pop()
pa = None
if not (p2 == nullid):
node1 = self.repo.changectx(p1)
node2 = self.repo.changectx(p2)
pa = node1.ancestor(node2)
# if named branch, add to extra
if hg_branch:
extra['branch'] = hg_branch
# if committer is different than author, add it to extra
if commit.author != commit.committer \
or commit.author_time != commit.commit_time \
or commit.author_timezone != commit.commit_timezone:
extra['committer'] = "%s %d %d" % (
commit.committer, commit.commit_time, -commit.commit_timezone)
if commit.encoding:
extra['encoding'] = commit.encoding
if hg_branch:
extra['branch'] = hg_branch
if octopus:
extra['hg-git'] ='octopus-done'
# TODO use 'n in self.repo' when we require hg 1.5
def repo_contains(n):
try:
return bool(self.repo.lookup(n))
except error.RepoLookupError:
return False
if not (repo_contains(p1) and repo_contains(p2)):
raise hgutil.Abort(_('you appear to have run strip - '
'please run hg git-cleanup'))
ctx = context.memctx(self.repo, (p1, p2), text,
list(files) + findconvergedfiles(p1, p2),
getfilectx, author, date, extra)
node = self.repo.commitctx(ctx)
self.swap_out_encoding(oldenc)
# save changeset to mapping file
cs = hex(node)
self.map_set(commit.id, cs)