NotCommitError,
NotTreeError,
)
+from dulwich.file import GitFile
from dulwich.misc import (
make_sha,
)
-BLOB_ID = "blob"
-TAG_ID = "tag"
-TREE_ID = "tree"
-COMMIT_ID = "commit"
-PARENT_ID = "parent"
-AUTHOR_ID = "author"
-COMMITTER_ID = "committer"
-OBJECT_ID = "object"
-TYPE_ID = "type"
-TAGGER_ID = "tagger"
-ENCODING_ID = "encoding"
-
-S_IFGITLINK = 0160000
+
+# Header fields for commits
+_TREE_HEADER = "tree"
+_PARENT_HEADER = "parent"
+_AUTHOR_HEADER = "author"
+_COMMITTER_HEADER = "committer"
+_ENCODING_HEADER = "encoding"
+
+
+# Header fields for objects
+_OBJECT_HEADER = "object"
+_TYPE_HEADER = "type"
+_TAG_HEADER = "tag"
+_TAGGER_HEADER = "tagger"
+
+
+S_IFGITLINK = 0160000
+
def S_ISGITLINK(m):
return (stat.S_IFMT(m) == S_IFGITLINK)
+
def _decompress(string):
dcomp = zlib.decompressobj()
dcomped = dcomp.decompress(string)
return property(get, set, doc=docstring)
+def object_class(type):
+ """Get the object class corresponding to the given type.
+
+ :param type: Either a type name string or a numeric type.
+ :return: The ShaFile subclass corresponding to the given type.
+ """
+ return _TYPE_MAP[type]
+
+
class ShaFile(object):
"""A git SHA file."""
"""Parse a legacy object, creating it and setting object._text"""
text = _decompress(map)
object = None
- for posstype in type_map.keys():
- if text.startswith(posstype):
- object = type_map[posstype]()
- text = text[len(posstype):]
+ for cls in OBJECT_CLASSES:
+ if text.startswith(cls.type_name):
+ object = cls()
+ text = text[len(cls.type_name):]
break
assert object is not None, "%s is not a known object type" % text[:9]
assert text[0] == ' ', "%s is not a space" % text[0]
object.set_raw_string(text)
return object
+ def as_legacy_object_chunks(self):
+ compobj = zlib.compressobj()
+ yield compobj.compress(self._header())
+ for chunk in self.as_raw_chunks():
+ yield compobj.compress(chunk)
+ yield compobj.flush()
+
def as_legacy_object(self):
- text = self.as_raw_string()
- return zlib.compress("%s %d\0%s" % (self._type, len(text), text))
+ return "".join(self.as_legacy_object_chunks())
- def as_raw_string(self):
+ def as_raw_chunks(self):
if self._needs_serialization:
- self.serialize()
- return self._text
+ self._chunked_text = self._serialize()
+ self._needs_serialization = False
+ return self._chunked_text
+
+ def as_raw_string(self):
+ return "".join(self.as_raw_chunks())
def __str__(self):
return self.as_raw_string()
def _ensure_parsed(self):
if self._needs_parsing:
- self._parse_text()
+ self._deserialize(self._chunked_text)
+ self._needs_parsing = False
def set_raw_string(self, text):
if type(text) != str:
raise TypeError(text)
- self._text = text
+ self.set_raw_chunks([text])
+
+ def set_raw_chunks(self, chunks):
+ self._chunked_text = chunks
self._sha = None
self._needs_parsing = True
self._needs_serialization = False
used = 0
byte = ord(map[used])
used += 1
- num_type = (byte >> 4) & 7
+ type_num = (byte >> 4) & 7
try:
- object = num_type_map[num_type]()
+ object = object_class(type_num)()
except KeyError:
- raise AssertionError("Not a known type: %d" % num_type)
+ raise AssertionError("Not a known type: %d" % type_num)
while (byte & 0x80) != 0:
byte = ord(map[used])
used += 1
"""Don't call this directly"""
self._sha = None
- def _parse_text(self):
- """For subclasses to do initialisation time parsing"""
+ def _deserialize(self, chunks):
+ raise NotImplementedError(self._deserialize)
+
+ def _serialize(self):
+ raise NotImplementedError(self._serialize)
@classmethod
def from_file(cls, filename):
"""Get the contents of a SHA file on disk"""
size = os.path.getsize(filename)
- f = open(filename, 'rb')
+ f = GitFile(filename, 'rb')
try:
map = mmap.mmap(f.fileno(), size, access=mmap.ACCESS_READ)
shafile = cls._parse_file(map)
finally:
f.close()
- @classmethod
- def from_raw_string(cls, type, string):
+ @staticmethod
+ def from_raw_string(type_num, string):
"""Creates an object of the indicated type from the raw string given.
- Type is the numeric type of an object. String is the raw uncompressed
- contents.
+ :param type_num: The numeric type of the object.
+ :param string: The raw uncompressed contents.
"""
- real_class = num_type_map[type]
- obj = real_class()
- obj.type = type
+ obj = object_class(type_num)()
obj.set_raw_string(string)
return obj
+ @staticmethod
+ def from_raw_chunks(type_num, chunks):
+ """Creates an object of the indicated type from the raw chunks given.
+
+ :param type_num: The numeric type of the object.
+ :param chunks: An iterable of the raw uncompressed contents.
+ """
+ obj = object_class(type_num)()
+ obj.set_raw_chunks(chunks)
+ return obj
+
@classmethod
def from_string(cls, string):
"""Create a blob from a string."""
- shafile = cls()
- shafile.set_raw_string(string)
- return shafile
+ obj = cls()
+ obj.set_raw_string(string)
+ return obj
def _header(self):
- return "%s %lu\0" % (self._type, len(self.as_raw_string()))
+ return "%s %lu\0" % (self.type_name, self.raw_length())
+
+ def raw_length(self):
+ """Returns the length of the raw string of this object."""
+ ret = 0
+ for chunk in self.as_raw_chunks():
+ ret += len(chunk)
+ return ret
+
+ def _make_sha(self):
+ ret = make_sha()
+ ret.update(self._header())
+ for chunk in self.as_raw_chunks():
+ ret.update(chunk)
+ return ret
def sha(self):
"""The SHA1 object that is the name of this object."""
if self._needs_serialization or self._sha is None:
- self._sha = make_sha()
- self._sha.update(self._header())
- self._sha.update(self.as_raw_string())
+ self._sha = self._make_sha()
return self._sha
@property
return self.sha().hexdigest()
def get_type(self):
- return self._num_type
+ return self.type_num
def set_type(self, type):
- self._num_type = type
+ self.type_num = type
+ # DEPRECATED: use type_num or type_name as needed.
type = property(get_type, set_type)
def __repr__(self):
return self.id != other.id
def __eq__(self, other):
- """Return true id the sha of the two objects match.
+ """Return true if the sha of the two objects match.
The __le__ etc methods aren't overriden as they make no sense,
certainly at this level.
class Blob(ShaFile):
"""A Git Blob object."""
- _type = BLOB_ID
- _num_type = 3
- _needs_serialization = False
- _needs_parsing = False
+ type_name = 'blob'
+ type_num = 3
+
+ def __init__(self):
+ super(Blob, self).__init__()
+ self._chunked_text = []
+ self._needs_parsing = False
+ self._needs_serialization = False
- def get_data(self):
- return self._text
+ def _get_data(self):
+ return self.as_raw_string()
- def set_data(self, data):
- self._text = data
+ def _set_data(self, data):
+ self.set_raw_string(data)
- data = property(get_data, set_data,
+ data = property(_get_data, _set_data,
"The text contained within the blob object.")
+ def _get_chunked(self):
+ return self._chunked_text
+
+ def _set_chunked(self, chunks):
+ self._chunked_text = chunks
+
+ chunked = property(_get_chunked, _set_chunked,
+ "The text within the blob object, as chunks (not necessarily lines).")
+
@classmethod
def from_file(cls, filename):
blob = ShaFile.from_file(filename)
- if blob._type != cls._type:
+ if not isinstance(blob, cls):
raise NotBlobError(filename)
return blob
class Tag(ShaFile):
"""A Git Tag object."""
- _type = TAG_ID
- _num_type = 4
+ type_name = 'tag'
+ type_num = 4
def __init__(self):
super(Tag, self).__init__()
@classmethod
def from_file(cls, filename):
- blob = ShaFile.from_file(filename)
- if blob._type != cls._type:
- raise NotBlobError(filename)
- return blob
+ tag = ShaFile.from_file(filename)
+ if not isinstance(tag, cls):
+ raise NotTagError(filename)
+ return tag
@classmethod
def from_string(cls, string):
shafile.set_raw_string(string)
return shafile
- def serialize(self):
- f = StringIO()
- f.write("%s %s\n" % (OBJECT_ID, self._object_sha))
- f.write("%s %s\n" % (TYPE_ID, num_type_map[self._object_type]._type))
- f.write("%s %s\n" % (TAG_ID, self._name))
+ def _serialize(self):
+ chunks = []
+ chunks.append("%s %s\n" % (_OBJECT_HEADER, self._object_sha))
+ chunks.append("%s %s\n" % (_TYPE_HEADER, self._object_class.type_name))
+ chunks.append("%s %s\n" % (_TAG_HEADER, self._name))
if self._tagger:
if self._tag_time is None:
- f.write("%s %s\n" % (TAGGER_ID, self._tagger))
+ chunks.append("%s %s\n" % (_TAGGER_HEADER, self._tagger))
else:
- f.write("%s %s %d %s\n" % (TAGGER_ID, self._tagger, self._tag_time, format_timezone(self._tag_timezone)))
- f.write("\n") # To close headers
- f.write(self._message)
- self._text = f.getvalue()
- self._needs_serialization = False
-
- def _parse_text(self):
+ chunks.append("%s %s %d %s\n" % (
+ _TAGGER_HEADER, self._tagger, self._tag_time,
+ format_timezone(self._tag_timezone)))
+ chunks.append("\n") # To close headers
+ chunks.append(self._message)
+ return chunks
+
+ def _deserialize(self, chunks):
"""Grab the metadata attached to the tag"""
self._tagger = None
- f = StringIO(self._text)
+ f = StringIO("".join(chunks))
for l in f:
l = l.rstrip("\n")
if l == "":
break # empty line indicates end of headers
(field, value) = l.split(" ", 1)
- if field == OBJECT_ID:
+ if field == _OBJECT_HEADER:
self._object_sha = value
- elif field == TYPE_ID:
- self._object_type = type_map[value]
- elif field == TAG_ID:
+ elif field == _TYPE_HEADER:
+ self._object_class = object_class(value)
+ elif field == _TAG_HEADER:
self._name = value
- elif field == TAGGER_ID:
+ elif field == _TAGGER_HEADER:
try:
sep = value.index("> ")
except ValueError:
else:
raise AssertionError("Unknown field %s" % field)
self._message = f.read()
- self._needs_parsing = False
- def get_object(self):
- """Returns the object pointed by this tag, represented as a tuple(type, sha)"""
+ def _get_object(self):
+ """Get the object pointed to by this tag.
+
+ :return: tuple of (object class, sha).
+ """
self._ensure_parsed()
- return (self._object_type, self._object_sha)
+ return (self._object_class, self._object_sha)
- def set_object(self, value):
+ def _set_object(self, value):
self._ensure_parsed()
- (self._object_type, self._object_sha) = value
+ (self._object_class, self._object_sha) = value
self._needs_serialization = True
- object = property(get_object, set_object)
+ object = property(_get_object, _set_object)
name = serializable_property("name", "The name of this tag")
tagger = serializable_property("tagger",
def parse_tree(text):
- ret = {}
+ """Parse a tree text.
+
+ :param text: Serialized text to parse
+ :yields: tuples of (name, mode, sha)
+ """
count = 0
l = len(text)
while count < l:
mode_end = text.index(' ', count)
mode = int(text[count:mode_end], 8)
-
name_end = text.index('\0', mode_end)
name = text[mode_end+1:name_end]
-
count = name_end+21
-
sha = text[name_end+1:count]
+ yield (name, mode, sha_to_hex(sha))
+
+
+def serialize_tree(items):
+ """Serialize the items in a tree to a text.
+
+ :param items: Sorted iterable over (name, mode, sha) tuples
+ :return: Serialized tree text as chunks
+ """
+ for name, mode, hexsha in items:
+ yield "%04o %s\0%s" % (mode, name, hex_to_sha(hexsha))
+
- ret[name] = (mode, sha_to_hex(sha))
+def sorted_tree_items(entries):
+ """Iterate over a tree entries dictionary in the order in which
+ the items would be serialized.
- return ret
+ :param entries: Dictionary mapping names to (mode, sha) tuples
+ :return: Iterator over (name, mode, sha)
+ """
+ def cmp_entry((name1, value1), (name2, value2)):
+ if stat.S_ISDIR(value1[0]):
+ name1 += "/"
+ if stat.S_ISDIR(value2[0]):
+ name2 += "/"
+ return cmp(name1, name2)
+ for name, entry in sorted(entries.iteritems(), cmp=cmp_entry):
+ yield name, entry[0], entry[1]
class Tree(ShaFile):
"""A Git tree object"""
- _type = TREE_ID
- _num_type = 2
+ type_name = 'tree'
+ type_num = 2
def __init__(self):
super(Tree, self).__init__()
@classmethod
def from_file(cls, filename):
tree = ShaFile.from_file(filename)
- if tree._type != cls._type:
+ if not isinstance(tree, cls):
raise NotTreeError(filename)
return tree
def entries(self):
"""Return a list of tuples describing the tree entries"""
self._ensure_parsed()
- # The order of this is different from iteritems() for historical reasons
- return [(mode, name, hexsha) for (name, mode, hexsha) in self.iteritems()]
+ # The order of this is different from iteritems() for historical
+ # reasons
+ return [
+ (mode, name, hexsha) for (name, mode, hexsha) in self.iteritems()]
def iteritems(self):
- def cmp_entry((name1, value1), (name2, value2)):
- if stat.S_ISDIR(value1[0]):
- name1 += "/"
- if stat.S_ISDIR(value2[0]):
- name2 += "/"
- return cmp(name1, name2)
+ """Iterate over all entries in the order in which they would be
+ serialized.
+
+ :return: Iterator over (name, mode, sha) tuples
+ """
self._ensure_parsed()
- for name, entry in sorted(self._entries.iteritems(), cmp=cmp_entry):
- yield name, entry[0], entry[1]
+ return sorted_tree_items(self._entries)
- def _parse_text(self):
+ def _deserialize(self, chunks):
"""Grab the entries in the tree"""
- self._entries = parse_tree(self._text)
+ parsed_entries = parse_tree("".join(chunks))
+ # TODO: list comprehension is for efficiency in the common (small) case;
+ # if memory efficiency in the large case is a concern, use a genexp.
+ self._entries = dict([(n, (m, s)) for n, m, s in parsed_entries])
self._needs_parsing = False
- def serialize(self):
- f = StringIO()
- for name, mode, hexsha in self.iteritems():
- f.write("%04o %s\0%s" % (mode, name, hex_to_sha(hexsha)))
- self._text = f.getvalue()
- self._needs_serialization = False
+ def _serialize(self):
+ return list(serialize_tree(self.iteritems()))
def as_pretty_string(self):
- text = ""
+ text = []
for name, mode, hexsha in self.iteritems():
if mode & stat.S_IFDIR:
kind = "tree"
else:
kind = "blob"
- text += "%04o %s %s\t%s\n" % (mode, kind, hexsha, name)
- return text
+ text.append("%04o %s %s\t%s\n" % (mode, kind, hexsha, name))
+ return "".join(text)
def parse_timezone(text):
class Commit(ShaFile):
"""A git commit object"""
- _type = COMMIT_ID
- _num_type = 1
+ type_name = 'commit'
+ type_num = 1
def __init__(self):
super(Commit, self).__init__()
self._encoding = None
self._needs_parsing = False
self._needs_serialization = True
+ self._extra = {}
@classmethod
def from_file(cls, filename):
commit = ShaFile.from_file(filename)
- if commit._type != cls._type:
+ if not isinstance(commit, cls):
raise NotCommitError(filename)
return commit
- def _parse_text(self):
+ def _deserialize(self, chunks):
self._parents = []
+ self._extra = []
self._author = None
- f = StringIO(self._text)
+ f = StringIO("".join(chunks))
for l in f:
l = l.rstrip("\n")
if l == "":
# Empty line indicates end of headers
break
(field, value) = l.split(" ", 1)
- if field == TREE_ID:
+ if field == _TREE_HEADER:
self._tree = value
- elif field == PARENT_ID:
+ elif field == _PARENT_HEADER:
self._parents.append(value)
- elif field == AUTHOR_ID:
+ elif field == _AUTHOR_HEADER:
self._author, timetext, timezonetext = value.rsplit(" ", 2)
self._author_time = int(timetext)
self._author_timezone = parse_timezone(timezonetext)
- elif field == COMMITTER_ID:
+ elif field == _COMMITTER_HEADER:
self._committer, timetext, timezonetext = value.rsplit(" ", 2)
self._commit_time = int(timetext)
self._commit_timezone = parse_timezone(timezonetext)
- elif field == ENCODING_ID:
+ elif field == _ENCODING_HEADER:
self._encoding = value
else:
- raise AssertionError("Unknown field %s" % field)
+ self._extra.append((field, value))
self._message = f.read()
- self._needs_parsing = False
- def serialize(self):
- f = StringIO()
- f.write("%s %s\n" % (TREE_ID, self._tree))
+ def _serialize(self):
+ chunks = []
+ chunks.append("%s %s\n" % (_TREE_HEADER, self._tree))
for p in self._parents:
- f.write("%s %s\n" % (PARENT_ID, p))
- f.write("%s %s %s %s\n" % (AUTHOR_ID, self._author, str(self._author_time), format_timezone(self._author_timezone)))
- f.write("%s %s %s %s\n" % (COMMITTER_ID, self._committer, str(self._commit_time), format_timezone(self._commit_timezone)))
+ chunks.append("%s %s\n" % (_PARENT_HEADER, p))
+ chunks.append("%s %s %s %s\n" % (
+ _AUTHOR_HEADER, self._author, str(self._author_time),
+ format_timezone(self._author_timezone)))
+ chunks.append("%s %s %s %s\n" % (
+ _COMMITTER_HEADER, self._committer, str(self._commit_time),
+ format_timezone(self._commit_timezone)))
if self.encoding:
- f.write("%s %s\n" % (ENCODING_ID, self.encoding))
- f.write("\n") # There must be a new line after the headers
- f.write(self._message)
- self._text = f.getvalue()
- self._needs_serialization = False
+ chunks.append("%s %s\n" % (_ENCODING_HEADER, self.encoding))
+ for k, v in self.extra:
+ if "\n" in k or "\n" in v:
+ raise AssertionError("newline in extra data: %r -> %r" % (k, v))
+ chunks.append("%s %s\n" % (k, v))
+ chunks.append("\n") # There must be a new line after the headers
+ chunks.append(self._message)
+ return chunks
tree = serializable_property("tree", "Tree that is the state of this commit")
- def get_parents(self):
+ def _get_parents(self):
"""Return a list of parents of this commit."""
self._ensure_parsed()
return self._parents
- def set_parents(self, value):
- """Return a list of parents of this commit."""
+ def _set_parents(self, value):
+ """Set a list of parents of this commit."""
self._ensure_parsed()
self._needs_serialization = True
self._parents = value
- parents = property(get_parents, set_parents)
+ parents = property(_get_parents, _set_parents)
+
+ def _get_extra(self):
+ """Return extra settings of this commit."""
+ self._ensure_parsed()
+ return self._extra
+
+ extra = property(_get_extra)
author = serializable_property("author",
"The name of the author of the commit")
"Encoding of the commit message.")
-type_map = {
- BLOB_ID : Blob,
- TREE_ID : Tree,
- COMMIT_ID : Commit,
- TAG_ID: Tag,
-}
+OBJECT_CLASSES = (
+ Commit,
+ Tree,
+ Blob,
+ Tag,
+ )
+
+_TYPE_MAP = {}
+
+for cls in OBJECT_CLASSES:
+ _TYPE_MAP[cls.type_name] = cls
+ _TYPE_MAP[cls.type_num] = cls
-num_type_map = {
- 0: None,
- 1: Commit,
- 2: Tree,
- 3: Blob,
- 4: Tag,
- # 5 Is reserved for further expansion
-}
+
+# Hold on to the pure-python implementations for testing
+_parse_tree_py = parse_tree
+_sorted_tree_items_py = sorted_tree_items
try:
# Try to import C versions
- from dulwich._objects import parse_tree
+ from dulwich._objects import parse_tree, sorted_tree_items
except ImportError:
pass
-
git.samba.org / jelmer / dulwich-libgit2.git / blobdiff