def translate(self, data):
"""
:param data bytes: raw MARC21 field data content, in either utf8 or marc8 encoding
:rtype: str
:return: A NFC normalized unicode str
"""
if self.rec.marc8():
data = mnemonics.read(data)
return marc8.translate(data)
return normalize('NFC', data.decode('utf8'))
def translate(bytes_in, leader_says_marc8=False):
"""
Converts MARC8 to unicode
"""
marc8 = MARC8ToUnicode(quiet=True)
if leader_says_marc8:
data = marc8.translate(mnemonics.read(bytes_in))
else:
data = bytes_in.decode('utf-8')
return normalize('NFC', data)
def translate(self, data):
utf8 = self.rec.leader()[9] == 'a'
if utf8:
try:
data = data.decode('utf-8')
except:
utf8 = False
if not utf8:
data = mnemonics.read(data)
data = marc8.translate(data)
data = normalize('NFC', data)
return data
def translate(self, data):
utf8 = self.rec.leader()[9] == 'a'
if utf8:
try:
data = data.decode('utf-8')
except:
utf8 = False
if not utf8:
data = mnemonics.read(data)
data = marc8.translate(data)
data = normalize('NFC', data)
return data
def translate(data, bad_ia_charset=False, leader_says_marc8=False):
if leader_says_marc8:
data = marc8.translate(data)
data = mnemonics.read(data)
return data
if bad_ia_charset:
#data = data.decode('utf-8')
data = marc8.translate(data)
return normalize('NFC', data)
data = mnemonics.read(data)
if type(data) == unicode:
return normalize('NFC', data)
try:
data = data.decode('utf-8')
is_utf8 = True
except UnicodeDecodeError:
is_utf8 = False
if not is_utf8:
data = marc8.translate(data)
if type(data) == unicode:
return normalize('NFC', data)
else:
return data
def translate(bytes_in, leader_says_marc8=False):
"""
Converts a binary MARC field value to unicode str,
from either MARC8 or UTF8 encoded bytes.
:param bytes_in bytes:
:rtype: str
"""
assert isinstance(bytes_in, bytes)
marc8 = MARC8ToUnicode(quiet=True)
if leader_says_marc8:
data = marc8.translate(mnemonics.read(bytes_in))
else:
data = bytes_in.decode('utf-8')
return normalize('NFC', data)
def translate(data, bad_ia_charset=False):
if bad_ia_charset:
data = data.decode("utf-8")
data = marc8.translate(data)
return normalize("NFC", data)
data = mnemonics.read(data)
if type(data) == unicode:
return normalize("NFC", data)
try:
data = data.decode("utf8")
is_utf8 = True
except UnicodeDecodeError:
is_utf8 = False
if not is_utf8:
data = marc8.translate(data)
if type(data) == unicode:
return normalize("NFC", data)
else:
return data
def test_read_no_change():
input_ = b'El Ing.{eniero} Federico E. Capurro y el nacimiento de la profesi\xe2on bibliotecaria en el Uruguay.'
assert read(input_) == input_
def test_read_conversion_to_marc8():
input_ = b'Tha{mllhring}{macr}alib{macr}i, {mllhring}Abd al-Malik ibn Mu{dotb}hammad,'
output = b'Tha\xb0\xe5alib\xe5i, \xb0Abd al-Malik ibn Mu\xf2hammad,'
assert read(input_) == output