def test_symlink(self):
repo_dir = tempfile.mkdtemp()
self.addCleanup(shutil.rmtree, repo_dir)
with Repo.init(repo_dir) as repo:
# Populate repo
filed = Blob.from_string(b'file d')
filee = Blob.from_string(b'd')
tree = Tree()
tree[b'c/d'] = (stat.S_IFREG | 0o644, filed.id)
tree[b'c/e'] = (stat.S_IFLNK, filee.id) # symlink
repo.object_store.add_objects([(o, None)
for o in [filed, filee, tree]])
build_index_from_tree(repo.path, repo.index_path(),
repo.object_store, tree.id)
# Verify index entries
index = repo.open_index()
# symlink to d
epath = os.path.join(repo.path, 'c', 'e')
self.assertTrue(os.path.exists(epath))
self.assertReasonableIndexEntry(
index[b'c/e'], stat.S_IFLNK,
0 if sys.platform == 'win32' else 1, filee.id)
self.assertFileContents(epath, 'd', symlink=True)
def content(mode, hexsha):
if hexsha is None:
return Blob.from_string(b'')
elif S_ISGITLINK(mode):
return Blob.from_string(b"Submodule commit " + hexsha + b"\n")
else:
return store[hexsha]
def test_object_diff_bin_blob_force(self):
f = BytesIO()
# Prepare two slightly different PNG headers
b1 = Blob.from_string(b"\x89\x50\x4e\x47\x0d\x0a\x1a\x0a"
b"\x00\x00\x00\x0d\x49\x48\x44\x52"
b"\x00\x00\x01\xd5\x00\x00\x00\x9f"
b"\x08\x04\x00\x00\x00\x05\x04\x8b")
b2 = Blob.from_string(b"\x89\x50\x4e\x47\x0d\x0a\x1a\x0a"
b"\x00\x00\x00\x0d\x49\x48\x44\x52"
b"\x00\x00\x01\xd5\x00\x00\x00\x9f"
b"\x08\x03\x00\x00\x00\x98\xd3\xb3")
store = MemoryObjectStore()
store.add_objects([(b1, None), (b2, None)])
write_object_diff(f,
store, (b'foo.png', 0o644, b1.id),
(b'bar.png', 0o644, b2.id),
diff_binary=True)
self.assertEqual([
b'diff --git a/foo.png b/bar.png', b'index f73e47d..06364b7 644',
b'--- a/foo.png', b'+++ b/bar.png', b'@@ -1,4 +1,4 @@',
b' \x89PNG', b' \x1a', b' \x00\x00\x00',
b'-IHDR\x00\x00\x01\xd5\x00\x00\x00'
b'\x9f\x08\x04\x00\x00\x00\x05\x04\x8b',
b'\\ No newline at end of file',
b'+IHDR\x00\x00\x01\xd5\x00\x00\x00\x9f'
b'\x08\x03\x00\x00\x00\x98\xd3\xb3',
b'\\ No newline at end of file'
],
f.getvalue().splitlines())
def test_git_dir(self):
repo_dir = tempfile.mkdtemp()
self.addCleanup(shutil.rmtree, repo_dir)
with Repo.init(repo_dir) as repo:
# Populate repo
filea = Blob.from_string(b'file a')
filee = Blob.from_string(b'd')
tree = Tree()
tree[b'.git/a'] = (stat.S_IFREG | 0o644, filea.id)
tree[b'c/e'] = (stat.S_IFREG | 0o644, filee.id)
repo.object_store.add_objects([(o, None)
for o in [filea, filee, tree]])
build_index_from_tree(repo.path, repo.index_path(),
repo.object_store, tree.id)
# Verify index entries
index = repo.open_index()
self.assertEqual(len(index), 1)
# filea
apath = os.path.join(repo.path, '.git', 'a')
self.assertFalse(os.path.exists(apath))
# filee
epath = os.path.join(repo.path, 'c', 'e')
self.assertTrue(os.path.exists(epath))
self.assertReasonableIndexEntry(index[b'c/e'],
stat.S_IFREG | 0o644, 1, filee.id)
self.assertFileContents(epath, b'd')
def test_simple_delta(self):
b1 = Blob.from_string(b"a" * 101)
b2 = Blob.from_string(b"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, b""), (b2, b"")])))
def test_blob_diff(self):
f = BytesIO()
write_blob_diff(f,
(b"foo.txt", 0o644, Blob.from_string(b"old\nsame\n")),
(b"bar.txt", 0o644, Blob.from_string(b"new\nsame\n")))
self.assertEqual([
b"diff --git a/foo.txt b/bar.txt", b"index 3b0f961..a116b51 644",
b"--- a/foo.txt", b"+++ b/bar.txt", b"@@ -1,2 +1,2 @@", b"-old",
b"+new", b" same"
],
f.getvalue().splitlines())
def commit_diff(self, commit):
"""Return the list of changes introduced by `commit`."""
from my_klaus.utils import guess_is_binary
if commit.parents:
parent_tree = self[commit.parents[0]].tree
else:
parent_tree = None
summary = {'nfiles': 0, 'nadditions': 0, 'ndeletions': 0}
file_changes = [] # the changes in detail
dulwich_changes = self.object_store.tree_changes(
parent_tree, commit.tree)
for (oldpath, newpath), (oldmode,
newmode), (oldsha, newsha) in dulwich_changes:
summary['nfiles'] += 1
try:
oldblob = self.object_store[
oldsha] if oldsha else Blob.from_string(b'')
newblob = self.object_store[
newsha] if newsha else Blob.from_string(b'')
except KeyError:
# newsha/oldsha are probably related to submodules.
# Dulwich will handle that.
pass
# Check for binary files -- can't show diffs for these
if guess_is_binary(newblob) or \
guess_is_binary(oldblob):
file_changes.append({
'is_binary': True,
'old_filename': oldpath or '/dev/null',
'new_filename': newpath or '/dev/null',
'chunks': None
})
continue
additions, deletions, chunks = render_diff(oldblob.splitlines(),
newblob.splitlines())
change = {
'is_binary': False,
'old_filename': oldpath or '/dev/null',
'new_filename': newpath or '/dev/null',
'chunks': chunks,
'additions': additions,
'deletions': deletions,
}
summary['nadditions'] += additions
summary['ndeletions'] += deletions
file_changes.append(change)
return summary, file_changes
def test_nonempty(self):
repo_dir = tempfile.mkdtemp()
self.addCleanup(shutil.rmtree, repo_dir)
with Repo.init(repo_dir) as repo:
# Populate repo
filea = Blob.from_string(b'file a')
fileb = Blob.from_string(b'file b')
filed = Blob.from_string(b'file d')
tree = Tree()
tree[b'a'] = (stat.S_IFREG | 0o644, filea.id)
tree[b'b'] = (stat.S_IFREG | 0o644, fileb.id)
tree[b'c/d'] = (stat.S_IFREG | 0o644, filed.id)
repo.object_store.add_objects([
(o, None) for o in [filea, fileb, filed, tree]
])
build_index_from_tree(repo.path, repo.index_path(),
repo.object_store, tree.id)
# Verify index entries
index = repo.open_index()
self.assertEqual(len(index), 3)
# filea
apath = os.path.join(repo.path, 'a')
self.assertTrue(os.path.exists(apath))
self.assertReasonableIndexEntry(index[b'a'], stat.S_IFREG | 0o644,
6, filea.id)
self.assertFileContents(apath, b'file a')
# fileb
bpath = os.path.join(repo.path, 'b')
self.assertTrue(os.path.exists(bpath))
self.assertReasonableIndexEntry(index[b'b'], stat.S_IFREG | 0o644,
6, fileb.id)
self.assertFileContents(bpath, b'file b')
# filed
dpath = os.path.join(repo.path, 'c', 'd')
self.assertTrue(os.path.exists(dpath))
self.assertReasonableIndexEntry(index[b'c/d'],
stat.S_IFREG | 0o644, 6, filed.id)
self.assertFileContents(dpath, b'file d')
# Verify no extra files
self.assertEqual(['.git', 'a', 'b', 'c'],
sorted(os.listdir(repo.path)))
self.assertEqual(['d'],
sorted(os.listdir(os.path.join(repo.path, 'c'))))
def test_object_diff_blob(self):
f = BytesIO()
b1 = Blob.from_string(b"old\nsame\n")
b2 = Blob.from_string(b"new\nsame\n")
store = MemoryObjectStore()
store.add_objects([(b1, None), (b2, None)])
write_object_diff(f, store, (b"foo.txt", 0o644, b1.id),
(b"bar.txt", 0o644, b2.id))
self.assertEqual([
b"diff --git a/foo.txt b/bar.txt", b"index 3b0f961..a116b51 644",
b"--- a/foo.txt", b"+++ b/bar.txt", b"@@ -1,2 +1,2 @@", b"-old",
b"+new", b" same"
],
f.getvalue().splitlines())
def test_tree_diff(self):
f = BytesIO()
store = MemoryObjectStore()
added = Blob.from_string(b"add\n")
removed = Blob.from_string(b"removed\n")
changed1 = Blob.from_string(b"unchanged\nremoved\n")
changed2 = Blob.from_string(b"unchanged\nadded\n")
unchanged = Blob.from_string(b"unchanged\n")
tree1 = Tree()
tree1.add(b"removed.txt", 0o644, removed.id)
tree1.add(b"changed.txt", 0o644, changed1.id)
tree1.add(b"unchanged.txt", 0o644, changed1.id)
tree2 = Tree()
tree2.add(b"added.txt", 0o644, added.id)
tree2.add(b"changed.txt", 0o644, changed2.id)
tree2.add(b"unchanged.txt", 0o644, changed1.id)
store.add_objects([
(o, None) for o in
[tree1, tree2, added, removed, changed1, changed2, unchanged]
])
write_tree_diff(f, store, tree1.id, tree2.id)
self.assertEqual([
b'diff --git /dev/null b/added.txt',
b'new mode 644',
b'index 0000000..76d4bb8 644',
b'--- /dev/null',
b'+++ b/added.txt',
b'@@ -0,0 +1 @@',
b'+add',
b'diff --git a/changed.txt b/changed.txt',
b'index bf84e48..1be2436 644',
b'--- a/changed.txt',
b'+++ b/changed.txt',
b'@@ -1,2 +1,2 @@',
b' unchanged',
b'-removed',
b'+added',
b'diff --git a/removed.txt /dev/null',
b'deleted mode 644',
b'index 2c3f0b3..0000000',
b'--- a/removed.txt',
b'+++ /dev/null',
b'@@ -1 +0,0 @@',
b'-removed',
],
f.getvalue().splitlines())
def commit_handler(self, cmd):
"""Process a CommitCommand."""
commit = Commit()
if cmd.author is not None:
author = cmd.author
else:
author = cmd.committer
(author_name, author_email, author_timestamp, author_timezone) = author
(committer_name, committer_email, commit_timestamp,
commit_timezone) = cmd.committer
commit.author = author_name + b" <" + author_email + b">"
commit.author_timezone = author_timezone
commit.author_time = int(author_timestamp)
commit.committer = committer_name + b" <" + committer_email + b">"
commit.commit_timezone = commit_timezone
commit.commit_time = int(commit_timestamp)
commit.message = cmd.message
commit.parents = []
if cmd.from_:
cmd.from_ = self.lookup_object(cmd.from_)
self._reset_base(cmd.from_)
for filecmd in cmd.iter_files():
if filecmd.name == b"filemodify":
if filecmd.data is not None:
blob = Blob.from_string(filecmd.data)
self.repo.object_store.add(blob)
blob_id = blob.id
else:
blob_id = self.lookup_object(filecmd.dataref)
self._contents[filecmd.path] = (filecmd.mode, blob_id)
elif filecmd.name == b"filedelete":
del self._contents[filecmd.path]
elif filecmd.name == b"filecopy":
self._contents[filecmd.dest_path] = self._contents[
filecmd.src_path]
elif filecmd.name == b"filerename":
self._contents[filecmd.new_path] = self._contents[
filecmd.old_path]
del self._contents[filecmd.old_path]
elif filecmd.name == b"filedeleteall":
self._contents = {}
else:
raise Exception("Command %s not supported" % filecmd.name)
commit.tree = commit_tree(
self.repo.object_store,
((path, hexsha, mode)
for (path, (mode, hexsha)) in self._contents.items()))
if self.last_commit != ZERO_SHA:
commit.parents.append(self.last_commit)
for merge in cmd.merges:
commit.parents.append(self.lookup_object(merge))
self.repo.object_store.add_object(commit)
self.repo[cmd.ref] = commit.id
self.last_commit = commit.id
if cmd.mark:
self.markers[cmd.mark] = commit.id
def test_blob_add(self):
f = BytesIO()
write_blob_diff(f, (None, None, None),
(b"bar.txt", 0o644, Blob.from_string(b"new\nsame\n")))
self.assertEqual([
b'diff --git /dev/null b/bar.txt', b'new mode 644',
b'index 0000000..a116b51 644', b'--- /dev/null', b'+++ b/bar.txt',
b'@@ -0,0 +1,2 @@', b'+new', b'+same'
],
f.getvalue().splitlines())
def _get_example_tar_stream(self, *tar_stream_args, **tar_stream_kwargs):
store = MemoryObjectStore()
b1 = Blob.from_string(b"somedata")
store.add_object(b1)
t1 = Tree()
t1.add(b"somename", 0o100644, b1.id)
store.add_object(t1)
stream = b''.join(
tar_stream(store, t1, *tar_stream_args, **tar_stream_kwargs))
return BytesIO(stream)
def test_object_diff_bin_blob(self):
f = BytesIO()
# Prepare two slightly different PNG headers
b1 = Blob.from_string(b"\x89\x50\x4e\x47\x0d\x0a\x1a\x0a"
b"\x00\x00\x00\x0d\x49\x48\x44\x52"
b"\x00\x00\x01\xd5\x00\x00\x00\x9f"
b"\x08\x04\x00\x00\x00\x05\x04\x8b")
b2 = Blob.from_string(b"\x89\x50\x4e\x47\x0d\x0a\x1a\x0a"
b"\x00\x00\x00\x0d\x49\x48\x44\x52"
b"\x00\x00\x01\xd5\x00\x00\x00\x9f"
b"\x08\x03\x00\x00\x00\x98\xd3\xb3")
store = MemoryObjectStore()
store.add_objects([(b1, None), (b2, None)])
write_object_diff(f, store, (b'foo.png', 0o644, b1.id),
(b'bar.png', 0o644, b2.id))
self.assertEqual([
b'diff --git a/foo.png b/bar.png', b'index f73e47d..06364b7 644',
b'Binary files a/foo.png and b/bar.png differ'
],
f.getvalue().splitlines())
def test_blob_remove(self):
f = BytesIO()
write_blob_diff(f,
(b"bar.txt", 0o644, Blob.from_string(b"new\nsame\n")),
(None, None, None))
self.assertEqual([
b'diff --git a/bar.txt /dev/null', b'deleted mode 644',
b'index a116b51..0000000', b'--- a/bar.txt', b'+++ /dev/null',
b'@@ -1,2 +0,0 @@', b'-new', b'-same'
],
f.getvalue().splitlines())
def create_commit(marker=None):
blob = Blob.from_string(b'The blob content ' + marker)
tree = Tree()
tree.add(b"thefile " + marker, 0o100644, blob.id)
cmt = Commit()
cmt.tree = tree.id
cmt.author = cmt.committer = b"John Doe <*****@*****.**>"
cmt.message = marker
tz = parse_timezone(b'-0200')[0]
cmt.commit_time = cmt.author_time = int(time.time())
cmt.commit_timezone = cmt.author_timezone = tz
return cmt, tree, blob
def test_object_diff_add_blob(self):
f = BytesIO()
store = MemoryObjectStore()
b2 = Blob.from_string(b"new\nsame\n")
store.add_object(b2)
write_object_diff(f, store, (None, None, None),
(b"bar.txt", 0o644, b2.id))
self.assertEqual([
b'diff --git /dev/null b/bar.txt', b'new mode 644',
b'index 0000000..a116b51 644', b'--- /dev/null', b'+++ b/bar.txt',
b'@@ -0,0 +1,2 @@', b'+new', b'+same'
],
f.getvalue().splitlines())
def test_object_diff_remove_blob(self):
f = BytesIO()
b1 = Blob.from_string(b"new\nsame\n")
store = MemoryObjectStore()
store.add_object(b1)
write_object_diff(f, store, (b"bar.txt", 0o644, b1.id),
(None, None, None))
self.assertEqual([
b'diff --git a/bar.txt /dev/null', b'deleted mode 644',
b'index a116b51..0000000', b'--- a/bar.txt', b'+++ /dev/null',
b'@@ -1,2 +0,0 @@', b'-new', b'-same'
],
f.getvalue().splitlines())
def test_object_diff_remove_bin_blob(self):
f = BytesIO()
b1 = Blob.from_string(b'\x89\x50\x4e\x47\x0d\x0a\x1a\x0a'
b'\x00\x00\x00\x0d\x49\x48\x44\x52'
b'\x00\x00\x01\xd5\x00\x00\x00\x9f'
b'\x08\x04\x00\x00\x00\x05\x04\x8b')
store = MemoryObjectStore()
store.add_object(b1)
write_object_diff(f, store, (b'foo.png', 0o644, b1.id),
(None, None, None))
self.assertEqual([
b'diff --git a/foo.png /dev/null', b'deleted mode 644',
b'index f73e47d..0000000',
b'Binary files a/foo.png and /dev/null differ'
],
f.getvalue().splitlines())
def test_object_diff_add_bin_blob(self):
f = BytesIO()
b2 = Blob.from_string(b'\x89\x50\x4e\x47\x0d\x0a\x1a\x0a'
b'\x00\x00\x00\x0d\x49\x48\x44\x52'
b'\x00\x00\x01\xd5\x00\x00\x00\x9f'
b'\x08\x03\x00\x00\x00\x98\xd3\xb3')
store = MemoryObjectStore()
store.add_object(b2)
write_object_diff(f, store, (None, None, None),
(b'bar.png', 0o644, b2.id))
self.assertEqual([
b'diff --git /dev/null b/bar.png', b'new mode 644',
b'index 0000000..06364b7 644',
b'Binary files /dev/null and b/bar.png differ'
],
f.getvalue().splitlines())
def test_object_diff_kind_change(self):
f = BytesIO()
b1 = Blob.from_string(b"new\nsame\n")
store = MemoryObjectStore()
store.add_object(b1)
write_object_diff(f, store, (b"bar.txt", 0o644, b1.id),
(b"bar.txt", 0o160000,
b"06d0bdd9e2e20377b3180e4986b14c8549b393e4"))
self.assertEqual([
b'diff --git a/bar.txt b/bar.txt',
b'old mode 644',
b'new mode 160000',
b'index a116b51..06d0bdd 160000',
b'--- a/bar.txt',
b'+++ b/bar.txt',
b'@@ -1,2 +1 @@',
b'-new',
b'-same',
b'+Submodule commit 06d0bdd9e2e20377b3180e4986b14c8549b393e4',
],
f.getvalue().splitlines())
def test_git_submodule_exists(self):
repo_dir = tempfile.mkdtemp()
self.addCleanup(shutil.rmtree, repo_dir)
with Repo.init(repo_dir) as repo:
filea = Blob.from_string(b'file alalala')
subtree = Tree()
subtree[b'a'] = (stat.S_IFREG | 0o644, filea.id)
c = Commit()
c.tree = subtree.id
c.committer = c.author = b'Somebody <*****@*****.**>'
c.commit_time = c.author_time = 42342
c.commit_timezone = c.author_timezone = 0
c.parents = []
c.message = b'Subcommit'
tree = Tree()
tree[b'c'] = (S_IFGITLINK, c.id)
os.mkdir(os.path.join(repo_dir, 'c'))
repo.object_store.add_objects([(o, None) for o in [tree]])
build_index_from_tree(repo.path, repo.index_path(),
repo.object_store, tree.id)
# Verify index entries
index = repo.open_index()
self.assertEqual(len(index), 1)
# filea
apath = os.path.join(repo.path, 'c/a')
self.assertFalse(os.path.exists(apath))
# dir c
cpath = os.path.join(repo.path, 'c')
self.assertTrue(os.path.isdir(cpath))
self.assertEqual(index[b'c'][4], S_IFGITLINK) # mode
self.assertEqual(index[b'c'][8], c.id) # sha
def test_norewrite(self):
repo_dir = tempfile.mkdtemp()
self.addCleanup(shutil.rmtree, repo_dir)
with Repo.init(repo_dir) as repo:
# Populate repo
filea = Blob.from_string(b'file a')
filea_path = os.path.join(repo_dir, 'a')
tree = Tree()
tree[b'a'] = (stat.S_IFREG | 0o644, filea.id)
repo.object_store.add_objects([(o, None) for o in [filea, tree]])
# First Write
build_index_from_tree(repo.path, repo.index_path(),
repo.object_store, tree.id)
# Use sync as metadata can be cached on some FS
os.sync()
mtime = os.stat(filea_path).st_mtime
# Test Rewrite
build_index_from_tree(repo.path, repo.index_path(),
repo.object_store, tree.id)
os.sync()
self.assertEqual(mtime, os.stat(filea_path).st_mtime)
# Modify content
with open(filea_path, 'wb') as fh:
fh.write(b'test a')
os.sync()
mtime = os.stat(filea_path).st_mtime
# Test rewrite
build_index_from_tree(repo.path, repo.index_path(),
repo.object_store, tree.id)
os.sync()
with open(filea_path, 'rb') as fh:
self.assertEqual(b'file a', fh.read())
def test_no_decode_encode(self):
repo_dir = tempfile.mkdtemp()
repo_dir_bytes = repo_dir.encode(sys.getfilesystemencoding())
self.addCleanup(shutil.rmtree, repo_dir)
with Repo.init(repo_dir) as repo:
# Populate repo
file = Blob.from_string(b'foo')
tree = Tree()
latin1_name = u'À'.encode('latin1')
latin1_path = os.path.join(repo_dir_bytes, latin1_name)
utf8_name = u'À'.encode('utf8')
utf8_path = os.path.join(repo_dir_bytes, utf8_name)
tree[latin1_name] = (stat.S_IFREG | 0o644, file.id)
tree[utf8_name] = (stat.S_IFREG | 0o644, file.id)
repo.object_store.add_objects([(o, None) for o in [file, tree]])
try:
os.path.exists(latin1_path)
except UnicodeDecodeError:
# This happens e.g. with python3.6 on Windows.
# It implicitly decodes using utf8, which doesn't work.
self.skipTest('can not implicitly convert as utf8')
build_index_from_tree(repo.path, repo.index_path(),
repo.object_store, tree.id)
# Verify index entries
index = repo.open_index()
self.assertIn(latin1_name, index)
self.assertIn(utf8_name, index)
self.assertTrue(os.path.exists(latin1_path))
self.assertTrue(os.path.exists(utf8_path))
def test_blob_by_sha(self):
r = MemoryRepo()
b = Blob.from_string(b"Blah")
r.object_store.add_object(b)
self.assertEqual(b, parse_object(r, b.id))
def test_single(self):
b = Blob.from_string(b"foo")
self.assertEqual(
[(b.type_num, b.sha().digest(), None, b.as_raw_string())],
list(deltify_pack_objects([(b, b"")])))
def test_legacy_from_file(self):
b1 = Blob.from_string(b'foo')
b_raw = b1.as_legacy_object()
b2 = b1.from_file(BytesIO(b_raw))
self.assertEqual(b1, b2)
def test_chunks(self):
string = b'test 5\n'
b = Blob.from_string(string)
self.assertEqual([string], b.chunked)
def test_create_blob_from_string(self):
string = b'test 2\n'
b = Blob.from_string(string)
self.assertEqual(b.data, string)
self.assertEqual(b.sha().hexdigest().encode('ascii'), b_sha)
def blob_handler(self, cmd):
"""Process a BlobCommand."""
blob = Blob.from_string(cmd.data)
self.repo.object_store.add_object(blob)
if cmd.mark:
self.markers[cmd.mark] = blob.id