def from_string(cls, s):
hbyte = ord(s[0:1])
if hbyte < 2:
st = cls.struct
# Freq, Doc freq, min len, max len, max w, max WOL, min ID, max ID
f, df, ml, xl, xw, xwol, mid, xid = st.unpack(s[1:st.size + 1])
mid = None if mid == NO_ID else mid
xid = None if xid == NO_ID else xid
# Postings
pstr = s[st.size + 1:]
if hbyte == 0:
p = unpack_long(pstr)[0]
else:
p = loads(pstr + b("."))
else:
# Old format was encoded as a variable length pickled tuple
v = loads(s + b("."))
if len(v) == 1:
f = df = 1
p = v[0]
elif len(v) == 2:
f = df = v[1]
p = v[0]
else:
f, p, df = v
# Fake values for stats which weren't stored before
ml = 1
xl = 106374
xw = 999999999
xwol = 999999999
mid = -1
xid = -1
return cls(f, df, ml, xl, xw, xwol, mid, xid, p)
def from_bytes(cls, s):
st = cls._struct
vals = st.unpack(s[:st.size])
terminfo = cls()
flags = vals[0]
terminfo._weight = vals[1]
terminfo._df = vals[2]
terminfo._minlength = byte_to_length(vals[3])
terminfo._maxlength = byte_to_length(vals[4])
terminfo._maxweight = vals[5]
terminfo._minid = None if vals[6] == 0xffffffff else vals[6]
terminfo._maxid = None if vals[7] == 0xffffffff else vals[7]
if flags:
# Postings are stored inline
terminfo._inlined = loads(s[st.size:])
else:
# Last bytes are pointer into posting file and length
offpos = st.size
lenpos = st.size + _LONG_SIZE
terminfo._offset = unpack_long(s[offpos:lenpos])[0]
terminfo._length = unpack_int(s[lenpos:lenpos + _INT_SIZE])
return terminfo
def test_pickleability():
# Ignore base classes
ignore = (columns.Column, columns.WrappedColumn, columns.ListColumn)
# Required arguments
init_args = {
"ClampedNumericColumn": (columns.NumericColumn("B"), ),
"FixedBytesColumn": (5, ),
"FixedBytesListColumn": (5, ),
"NumericColumn": ("i", ),
"PickleColumn": (columns.VarBytesColumn(), ),
"StructColumn": ("=if", (0, 0.0)),
}
coltypes = [
c for _, c in inspect.getmembers(columns, inspect.isclass)
if issubclass(c, columns.Column) and not c in ignore
]
for coltype in coltypes:
args = init_args.get(coltype.__name__, ())
try:
inst = coltype(*args)
except TypeError:
e = sys.exc_info()[1]
raise TypeError("Error instantiating %r: %s" % (coltype, e))
_ = loads(dumps(inst, -1))
def __getitem__(self, num):
if num > self.length - 1:
raise IndexError("Tried to get document %s, file has %s"
% (num, self.length))
dbfile = self.dbfile
start = self.directory_offset + num * stored_pointer_size
dbfile.seek(start)
ptr = dbfile.read(stored_pointer_size)
if len(ptr) != stored_pointer_size:
raise Exception("Error reading %r @%s %s < %s"
% (dbfile, start, len(ptr), stored_pointer_size))
position, length = unpack_stored_pointer(ptr)
vlist = loads(dbfile.map[position:position + length] + b("."))
names = self.names
# Recreate a dictionary by putting the field names and values back
# together by position. We can't just use dict(zip(...)) because we
# want to filter out the None values.
values = dict((names[i], vlist[i]) for i in xrange(len(names))
if vlist[i] is not None)
# Pull any extra stored dynamic field values off the end of the list
if len(vlist) > len(names):
values.update(dict(vlist[len(names):]))
return values
def from_bytes(cls, s):
st = cls._struct
vals = st.unpack(s[:st.size])
terminfo = cls()
flags = vals[0]
terminfo._weight = vals[1]
terminfo._df = vals[2]
terminfo._minlength = byte_to_length(vals[3])
terminfo._maxlength = byte_to_length(vals[4])
terminfo._maxweight = vals[5]
terminfo._minid = None if vals[6] == 0xffffffff else vals[6]
terminfo._maxid = None if vals[7] == 0xffffffff else vals[7]
if flags:
# Postings are stored inline
terminfo._inlined = loads(s[st.size:])
else:
# Last bytes are pointer into posting file and length
offpos = st.size
lenpos = st.size + _LONG_SIZE
terminfo._offset = unpack_long(s[offpos:lenpos])[0]
terminfo._length = unpack_int(s[lenpos:lenpos + _INT_SIZE])
return terminfo
def decode_positions(self, valuestring):
codes = loads(valuestring[_INT_SIZE + _FLOAT_SIZE:] + b("."))
position = 0
posns = []
for code in codes:
position = code[0] + position
posns.append(position)
return posns
def decode_positions(self, valuestring):
codes = loads(valuestring[_INT_SIZE + _FLOAT_SIZE:] + b("."))
position = 0
posns = []
for code in codes:
position = code[0] + position
posns.append(position)
return posns
def _read_stored_fields(self):
sfs = {}
c = self._find_line(2, "DOCFIELD")
while c is not None:
v = c.get("v")
if v is not None:
v = loads(v)
sfs[c["fn"]] = v
c = self._find_line(2, "DOCFIELD")
return sfs
def decode_positions(self, valuestring):
if not valuestring.endswith(b(".")):
valuestring += b(".")
codes = loads(valuestring[_INT_SIZE + _FLOAT_SIZE:])
position = 0
posns = []
for code in codes:
position = code[0] + position
posns.append(position)
return posns
def from_string(cls, s):
assert isinstance(s, bytes_type)
if isinstance(s, string_type):
hbyte = ord(s[0]) # Python 2.x - str
else:
hbyte = s[0] # Python 3 - bytes
if hbyte < 2:
st = cls.struct
# Weight, Doc freq, min len, max len, max w, unused, min ID, max ID
w, df, ml, xl, xw, _, mid, xid = st.unpack(s[1:st.size + 1])
mid = None if mid == NO_ID else mid
xid = None if xid == NO_ID else xid
# Postings
pstr = s[st.size + 1:]
if hbyte == 0:
p = unpack_long(pstr)[0]
else:
p = loads(pstr + b("."))
else:
# Old format was encoded as a variable length pickled tuple
v = loads(s + b("."))
if len(v) == 1:
w = df = 1
p = v[0]
elif len(v) == 2:
w = df = v[1]
p = v[0]
else:
w, p, df = v
# Fake values for stats which weren't stored before
ml = 1
xl = 255
xw = 999999999
mid = -1
xid = -1
ml = byte_to_length(ml)
xl = byte_to_length(xl)
obj = cls(w, df, ml, xl, xw, mid, xid)
obj.postings = p
return obj
def _read_stored_fields(self):
sfs = {}
c = self._find_line(2, "DOCFIELD")
while c is not None:
v = c.get("v")
if v is not None:
v = loads(v)
sfs[c["fn"]] = v
c = self._find_line(2, "DOCFIELD")
return sfs
def decode_positions(self, valuestring):
if not valuestring.endswith(b(".")):
valuestring += b(".")
codes = loads(valuestring[_INT_SIZE + _FLOAT_SIZE:])
position = 0
posns = []
for code in codes:
position = code[0] + position
posns.append(position)
return posns
def decode_characters(self, valuestring):
codes = loads(valuestring[_INT_SIZE:] + b("."))
position = 0
endchar = 0
posns_chars = []
for code in codes:
position = code[0] + position
startchar = code[1] + endchar
endchar = code[2] + startchar
posns_chars.append((position, startchar, endchar))
return posns_chars
def deminimize_ids(typecode, count, string, compression=0):
if compression:
string = decompress(string)
if typecode == '':
return loads(string)
else:
arry = array(typecode)
arry.fromstring(string)
if not IS_LITTLE:
arry.byteswap()
return arry
def decode_character_boosts(self, valuestring):
codes = loads(valuestring[_INT_SIZE + _FLOAT_SIZE:] + b("."))
position = 0
endchar = 0
posn_char_boosts = []
for code in codes:
position = position + code[0]
startchar = endchar + code[1]
endchar = startchar + code[2]
posn_char_boosts.append((position, startchar, endchar, code[3]))
return posn_char_boosts
def decode_character_boosts(self, valuestring):
codes = loads(valuestring[_INT_SIZE + _FLOAT_SIZE:] + b("."))
position = 0
endchar = 0
posn_char_boosts = []
for code in codes:
position = position + code[0]
startchar = endchar + code[1]
endchar = startchar + code[2]
posn_char_boosts.append((position, startchar, endchar, code[3]))
return posn_char_boosts
def deminimize_values(postingsize, count, string, compression=0):
if compression:
string = decompress(string)
if postingsize < 0:
return loads(string)
elif postingsize == 0:
return [None] * count
else:
return [string[i:i + postingsize] for i
in xrange(0, len(string), postingsize)]
def decode_characters(self, valuestring):
codes = loads(valuestring[_INT_SIZE:] + b("."))
position = 0
endchar = 0
posns_chars = []
for code in codes:
position = code[0] + position
startchar = code[1] + endchar
endchar = code[2] + startchar
posns_chars.append((position, startchar, endchar))
return posns_chars
def _read_data(self):
# Load block data tuple from disk
datalen = self._nextoffset - self._dataoffset
b = self._postfile.get(self._dataoffset, datalen)
# Decompress the pickled data if necessary
if self._compression:
b = zlib.decompress(b)
# Unpickle the data tuple and save it in an attribute
self._data = loads(b)
def _read_data(self):
# Load block data tuple from disk
datalen = self._nextoffset - self._dataoffset
b = self._postfile.get(self._dataoffset, datalen)
# Decompress the pickled data if necessary
if self._compression:
b = zlib.decompress(b)
# Unpickle the data tuple and save it in an attribute
self._data = loads(b)
def read_ids(self):
dataoffset = self.dataoffset
ids_string = self.postfile.map[dataoffset:dataoffset + self.idslen]
if self.compression:
ids_string = decompress(ids_string)
if self.stringids:
ids = loads(ids_string)
else:
ids = array(self.typecode)
ids.fromstring(ids_string)
if not IS_LITTLE:
ids.byteswap()
self.ids = ids
return ids
def __iter__(self):
dbfile = self.dbfile
names = self.names
lengths = array("I")
dbfile.seek(self.directory_offset)
for i in xrange(self.length):
dbfile.seek(_LONG_SIZE, 1)
lengths.append(dbfile.read_uint())
dbfile.seek(self.basepos)
for length in lengths:
vlist = loads(dbfile.read(length) + b("."))
vdict = dict((names[i], vlist[i]) for i in xrange(len(vlist))
if vlist[i] is not None)
yield vdict
def __iter__(self):
dbfile = self.dbfile
names = self.names
lengths = array("I")
dbfile.seek(self.directory_offset)
for i in xrange(self.length):
dbfile.seek(_LONG_SIZE, 1)
lengths.append(dbfile.read_uint())
dbfile.seek(self.basepos)
for length in lengths:
vlist = loads(dbfile.read(length) + b("."))
vdict = dict((names[i], vlist[i]) for i in xrange(len(vlist))
if vlist[i] is not None)
yield vdict
def read_ids(self):
dataoffset = self.dataoffset
ids_string = self.postfile.map[dataoffset:dataoffset + self.idslen]
if self.compression:
ids_string = decompress(ids_string)
if self.stringids:
ids = loads(ids_string)
else:
ids = array(self.typecode)
ids.fromstring(ids_string)
if not IS_LITTLE:
ids.byteswap()
self.ids = ids
return ids
def read_values(self):
postingsize = self.postingsize
if postingsize == 0:
values = [None] * self.postcount
else:
offset = self.dataoffset + self.idslen + self.weightslen
values_string = self.postfile.map[offset:self.nextoffset]
if self.compression:
values_string = decompress(values_string)
if postingsize < 0:
values = loads(values_string)
else:
values = [values_string[i:i + postingsize]
for i in xrange(0, len(values_string), postingsize)]
self.values = values
return values
def read_values(self):
postingsize = self.postingsize
if postingsize == 0:
values = [None] * self.postcount
else:
offset = self.dataoffset + self.idslen + self.weightslen
values_string = self.postfile.map[offset:self.nextoffset]
if self.compression:
values_string = decompress(values_string)
if postingsize < 0:
values = loads(values_string)
else:
values = [
values_string[i:i + postingsize]
for i in xrange(0, len(values_string), postingsize)
]
self.values = values
return values
def test_pickleability():
# Ignore base classes
ignore = (columns.Column, columns.WrappedColumn, columns.ListColumn)
# Required arguments
init_args = {"ClampedNumericColumn": (columns.NumericColumn("B"),),
"FixedBytesColumn": (5,),
"FixedBytesListColumn": (5,),
"NumericColumn": ("i",),
"PickleColumn": (columns.VarBytesColumn(),),
"StructColumn": ("=if", (0, 0.0)),
}
coltypes = [c for _, c in inspect.getmembers(columns, inspect.isclass)
if issubclass(c, columns.Column) and not c in ignore]
for coltype in coltypes:
args = init_args.get(coltype.__name__, ())
try:
inst = coltype(*args)
except TypeError:
e = sys.exc_info()[1]
raise TypeError("Error instantiating %r: %s" % (coltype, e))
_ = loads(dumps(inst, -1))
def __iter__(self):
for v in self._child:
if not v:
yield None
else:
yield loads(v)
def __iter__(self):
for v in self._child:
if not v:
yield None
else:
yield loads(v)
def __getitem__(self, docnum):
v = self._child[docnum]
if not v:
return None
else:
return loads(v)
def __getitem__(self, docnum):
v = self._child[docnum]
if not v:
return None
else:
return loads(v)
