def _parse_legacy_object(self, map):
"""Parse a legacy object, setting the raw string."""
text = _decompress(map)
header_end = text.find(b'\0')
if header_end < 0:
raise ObjectFormatException("Invalid object header, no \\0")
self.set_raw_string(text[header_end + 1:])
def _parse_object_header(magic, f):
"""Parse a new style object, creating it but not reading the file."""
num_type = (ord(magic[0:1]) >> 4) & 7
obj_class = object_class(num_type)
if not obj_class:
raise ObjectFormatException("Not a known type %d" % num_type)
return obj_class()
def from_file(cls, f):
"""Get the contents of a SHA file on disk."""
try:
obj = cls._parse_file(f)
obj._sha = None
return obj
except (IndexError, ValueError):
raise ObjectFormatException("invalid object header")
def check_hexsha(hex, error_msg):
"""Check if a string is a valid hex sha string.
:param hex: Hex string to check
:param error_msg: Error message to use in exception
:raise ObjectFormatException: Raised when the string is not valid
"""
if not valid_hexsha(hex):
raise ObjectFormatException("%s %s" % (error_msg, hex))
def _deserialize(self, chunks):
"""Grab the entries in the tree"""
try:
parsed_entries = parse_tree(b''.join(chunks))
except ValueError as e:
raise ObjectFormatException(e)
# 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])
def _check_has_member(self, member, error_msg):
"""Check that the object has a given member variable.
:param member: the member variable to check for
:param error_msg: the message for an error if the member is missing
:raise ObjectFormatException: with the given error_msg if member is
missing or is None
"""
if getattr(self, member, None) is None:
raise ObjectFormatException(error_msg)
def check(self):
"""Check this object for internal consistency.
Raises:
ObjectFormatException: if the object is malformed in some way
"""
super(Commit, self).check()
self._check_has_member("_tree", "missing tree")
self._check_has_member("_author", "missing author")
self._check_has_member("_committer", "missing committer")
self._check_has_member("_author_time", "missing author time")
self._check_has_member("_commit_time", "missing commit time")
for parent in self._parents:
check_hexsha(parent, "invalid parent sha")
check_hexsha(self._tree, "invalid tree sha")
check_identity(self._author, "invalid author")
check_identity(self._committer, "invalid committer")
check_time(self._author_time)
check_time(self._commit_time)
last = None
for field, _ in _parse_message(self._chunked_text):
if field == _TREE_HEADER and last is not None:
raise ObjectFormatException("unexpected tree")
elif field == _PARENT_HEADER and last not in (
_PARENT_HEADER,
_TREE_HEADER,
):
raise ObjectFormatException("unexpected parent")
elif field == _AUTHOR_HEADER and last not in (
_TREE_HEADER,
_PARENT_HEADER,
):
raise ObjectFormatException("unexpected author")
elif field == _COMMITTER_HEADER and last != _AUTHOR_HEADER:
raise ObjectFormatException("unexpected committer")
elif field == _ENCODING_HEADER and last != _COMMITTER_HEADER:
raise ObjectFormatException("unexpected encoding")
last = field
def check_hexsha(hex, error_msg):
"""Check if a string is a valid hex sha string.
:param hex: Hex string to check
:param error_msg: Error message to use in exception
:raise ObjectFormatException: Raised when the string is not valid
"""
try:
hex_to_sha(hex)
except (TypeError, AssertionError, ValueError):
raise ObjectFormatException("%s %s" % (error_msg, hex))
def from_file(cls, f):
"""Get the contents of a SHA file on disk."""
try:
obj = cls._parse_file_header(f)
obj._sha = None
obj._needs_parsing = True
obj._needs_serialization = True
obj._file = f
return obj
except (IndexError, ValueError), e:
raise ObjectFormatException("invalid object header")
def check_time(time_seconds):
"""Check if the specified time is not prone to overflow error.
This will raise an exception if the time is not valid.
:param time_info: author/committer/tagger info
"""
# Prevent overflow error
if time_seconds > MAX_TIME:
raise ObjectFormatException(
'Date field should not exceed %s' % MAX_TIME)
def check_time(time_seconds):
"""Check if the specified time is not prone to overflow error.
This will raise an exception if the time is not valid.
Args:
time_seconds: time in seconds
"""
# Prevent overflow error
if time_seconds > MAX_TIME:
raise ObjectFormatException("Date field should not exceed %s" % MAX_TIME)
def _deserialize(self, chunks):
"""Grab the metadata attached to the tag"""
self._tagger = None
self._tag_time = None
self._tag_timezone = None
self._tag_timezone_neg_utc = False
for field, value in _parse_message(chunks):
if field == _OBJECT_HEADER:
self._object_sha = value
elif field == _TYPE_HEADER:
obj_class = object_class(value)
if not obj_class:
raise ObjectFormatException("Not a known type: %s" % value)
self._object_class = obj_class
elif field == _TAG_HEADER:
self._name = value
elif field == _TAGGER_HEADER:
try:
sep = value.index(b'> ')
except ValueError:
self._tagger = value
self._tag_time = None
self._tag_timezone = None
self._tag_timezone_neg_utc = False
else:
self._tagger = value[0:sep + 1]
try:
(timetext,
timezonetext) = (value[sep + 2:].rsplit(b' ', 1))
self._tag_time = int(timetext)
self._tag_timezone, self._tag_timezone_neg_utc = (
parse_timezone(timezonetext))
except ValueError as e:
raise ObjectFormatException(e)
elif field is None:
self._message = value
else:
raise ObjectFormatException("Unknown field %s" % field)
def _parse_legacy_object_header(magic, f):
"""Parse a legacy object, creating it but not reading the file."""
bufsize = 1024
decomp = zlib.decompressobj()
header = decomp.decompress(magic)
start = 0
end = -1
while end < 0:
extra = f.read(bufsize)
header += decomp.decompress(extra)
magic += extra
end = header.find(b'\0', start)
start = len(header)
header = header[:end]
type_name, size = header.split(b' ', 1)
try:
int(size) # sanity check
except ValueError as e:
raise ObjectFormatException("Object size not an integer: %s" % e)
obj_class = object_class(type_name)
if not obj_class:
raise ObjectFormatException("Not a known type: %s" % type_name)
return obj_class()
def _deserialize(self, chunks):
"""Grab the metadata attached to the tag"""
self._tagger = None
self._tag_time = None
self._tag_timezone = None
self._tag_timezone_neg_utc = False
for field, value in _parse_message(chunks):
if field == _OBJECT_HEADER:
self._object_sha = value
elif field == _TYPE_HEADER:
obj_class = object_class(value)
if not obj_class:
raise ObjectFormatException("Not a known type: %s" % value)
self._object_class = obj_class
elif field == _TAG_HEADER:
self._name = value
elif field == _TAGGER_HEADER:
(
self._tagger,
self._tag_time,
(self._tag_timezone, self._tag_timezone_neg_utc),
) = parse_time_entry(value)
elif field is None:
if value is None:
self._message = None
self._signature = None
else:
try:
sig_idx = value.index(BEGIN_PGP_SIGNATURE)
except ValueError:
self._message = value
self._signature = None
else:
self._message = value[:sig_idx]
self._signature = value[sig_idx:]
else:
raise ObjectFormatException("Unknown field %s" % field)
def check_identity(identity, error_msg):
"""Check if the specified identity is valid.
This will raise an exception if the identity is not valid.
:param identity: Identity string
:param error_msg: Error message to use in exception
"""
email_start = identity.find("<")
email_end = identity.find(">")
if (email_start < 0 or email_end < 0 or email_end <= email_start
or identity.find("<", email_start + 1) >= 0
or identity.find(">", email_end + 1) >= 0
or not identity.endswith(">")):
raise ObjectFormatException(error_msg)
def _deserialize(self, chunks):
"""Grab the metadata attached to the tag"""
self._tagger = None
self._tag_time = None
self._tag_timezone = None
self._tag_timezone_neg_utc = False
for field, value in _parse_message(chunks):
if field == _OBJECT_HEADER:
self._object_sha = value
elif field == _TYPE_HEADER:
obj_class = object_class(value)
if not obj_class:
raise ObjectFormatException("Not a known type: %s" % value)
self._object_class = obj_class
elif field == _TAG_HEADER:
self._name = value
elif field == _TAGGER_HEADER:
(self._tagger, self._tag_time,
(self._tag_timezone,
self._tag_timezone_neg_utc)) = parse_time_entry(value)
elif field is None:
self._message = value
else:
raise ObjectFormatException("Unknown field %s" % field)
def check(self):
"""Check this object for internal consistency.
:raise ObjectFormatException: if the object is malformed in some way
:raise ChecksumMismatch: if the object was created with a SHA that does
not match its contents
"""
# TODO: if we find that error-checking during object parsing is a
# performance bottleneck, those checks should be moved to the class's
# check() method during optimization so we can still check the object
# when necessary.
old_sha = self.id
try:
self._deserialize(self.as_raw_chunks())
self._sha = None
new_sha = self.id
except Exception, e:
raise ObjectFormatException(e)
def parse_time_entry(value):
"""Parse time entry behavior
:param value: Bytes representing a git commit/tag line
:raise: ObjectFormatException in case of parsing error (malformed
field date)
:return: Tuple of (author, time, (timezone, timezone_neg_utc))
"""
try:
sep = value.rindex(b'> ')
except ValueError:
return (value, None, (None, False))
try:
person = value[0:sep+1]
rest = value[sep+2:]
timetext, timezonetext = rest.rsplit(b' ', 1)
time = int(timetext)
timezone, timezone_neg_utc = parse_timezone(timezonetext)
except ValueError as e:
raise ObjectFormatException(e)
return person, time, (timezone, timezone_neg_utc)
def _parse_legacy_object_header(magic, f):
"""Parse a legacy object, creating it but not reading the file."""
bufsize = 1024
decomp = zlib.decompressobj()
header = decomp.decompress(magic)
start = 0
end = -1
while end < 0:
extra = f.read(bufsize)
header += decomp.decompress(extra)
magic += extra
end = header.find("\0", start)
start = len(header)
header = header[:end]
type_name, size = header.split(" ", 1)
size = int(size) # sanity check
obj_class = object_class(type_name)
if not obj_class:
raise ObjectFormatException("Not a known type: %s" % type_name)
ret = obj_class()
ret._magic = magic
return ret
self._tag_timezone = None
self._tag_timezone_neg_utc = False
else:
self._tagger = value[0:sep + 1]
try:
(timetext,
timezonetext) = value[sep + 2:].rsplit(" ", 1)
self._tag_time = int(timetext)
self._tag_timezone, self._tag_timezone_neg_utc = \
parse_timezone(timezonetext)
except ValueError, e:
raise ObjectFormatException(e)
elif field is None:
self._message = value
else:
raise ObjectFormatException("Unknown field %s" % field)
def _get_object(self):
"""Get the object pointed to by this tag.
:return: tuple of (object class, sha).
"""
self._ensure_parsed()
return (self._object_class, self._object_sha)
def _set_object(self, value):
self._ensure_parsed()
(self._object_class, self._object_sha) = value
self._needs_serialization = True
object = property(_get_object, _set_object)
def _deserialize(self, chunks):
"""Grab the entries in the tree"""
try:
parsed_entries = parse_tree("".join(chunks))
except ValueError, e:
raise ObjectFormatException(e)
def check_hexsha(hex, error_msg):
try:
hex_to_sha(hex)
except (TypeError, AssertionError):
raise ObjectFormatException("%s %s" % (error_msg, hex))
