def Object_load(typ,inFile,format='auto',overrideHashPercent={}):
def typeChecked(obj,type):
if type==None: return obj
if not isinstance(obj,typ): raise TypeError('Loaded object of type '+obj.__class__.__name__+' is not a '+typ.__name__)
return obj
validFormats=('auto','boost::serialization','expr','pickle','json')
if format not in validFormats: raise ValueError('format must be one of '+', '.join(validFormats)+'.')
if format=='auto':
format=None
## DO NOT use extensions to determine type, that is evil
# check for compression first
head=open(inFile,'rb').read(100)
# we might want to pass this data to ObjectIO, which currently determines format by looking at extension
if head[:2]==b'\x1f\x8b':
import gzip
head=gzip.open(inFile,'rb').read(100)
elif head[:2]==b'BZ':
import bz2
head=bz2.BZ2File(inFile,'rb').read(100)
# detect real format here (head is uncompressed already)
# the string between nulls is 'serialization::archive'
# see http://stackoverflow.com/questions/10614215/magic-for-detecting-boostserialization-file
# newer versions (1.51 and perhaps greater) put '\n' (\x0a) after serialization::archive instead of null,
# so let's just not test the byte after "archive"
if head.startswith(b'\x16\x00\x00\x00\x00\x00\x00\x00\x73\x65\x72\x69\x61\x6c\x69\x7a\x61\x74\x69\x6f\x6e\x3a\x3a\x61\x72\x63\x68\x69\x76\x65'):
format='boost::serialization'
elif head.startswith(b'<?xml version="1.0"'):
format='boost::serialization'
elif head.startswith(b'##woo-expression##'):
format='expr'
else:
# test pickling by trying to load
try: return typeChecked(pickle.load(open(inFile,'rb')),typ) # open again to seek to the beginning
except (IOError,KeyError,pickle.UnpicklingError,EOFError): pass
try: return typeChecked(WooJSONDecoder().decode(codecs.open(inFile,'rb','utf-8').read()),typ)
except (IOError,ValueError): pass
if not format: raise RuntimeError('File format detection failed on %s (head: %s, bin: %s)'%(inFile,''.join(["\\x%02x"%(x if py3k else ord(x)) for x in head]),str(head))) # in py3k, bytes contain integers rather than chars
if format not in validFormats: raise RuntimeError("format='%s'??"%format)
assert format in validFormats
if overrideHashPercent and format!='expr': raise ValueError("overrideHashPercent only applicable with the 'expr' format (not '%s')"%format)
if format==None:
raise IOError('Input file format not detected')
elif format=='boost::serialization':
# ObjectIO takes care of detecting binary, xml, compression independently
return typeChecked(Object._boostLoad(str(inFile)),typ) # convert unicode to str, if necessary, as the c++ type is std::string
elif format=='expr':
buf=codecs.open(inFile,'rb','utf-8').read()
return typeChecked(wooExprEval(buf,inFile,__overrideHashPercent=overrideHashPercent),typ)
elif format=='pickle':
return typeChecked(pickle.load(open(inFile,'rb')),typ)
elif format=='json':
return typeChecked(WooJSONDecoder().decode(codecs.open(inFile,'rb','utf-8').read()),typ)
assert False
def Object_load(typ,inFile,format='auto',overrideHashColon={}):
def typeChecked(obj,type):
if type==None: return obj
if not isinstance(obj,typ): raise TypeError('Loaded object of type '+obj.__class__.__name__+' is not a '+typ.__name__)
return obj
validFormats=('auto','boost::serialization','expr','pickle','json')
if format not in validFormats: raise ValueError('format must be one of '+', '.join(validFormats)+'.')
if format=='auto':
format=None
## DO NOT use extensions to determine type, that is evil
# check for compression first
head=open(inFile,'rb').read(100)
# we might want to pass this data to ObjectIO, which currently determines format by looking at extension
if head[:2]==b'\x1f\x8b':
import gzip
head=gzip.open(inFile,'rb').read(100)
elif head[:2]==b'BZ':
import bz2
head=bz2.BZ2File(inFile,'rb').read(100)
# detect real format here (head is uncompressed already)
# the string between nulls is 'serialization::archive'
# see http://stackoverflow.com/questions/10614215/magic-for-detecting-boostserialization-file
# newer versions (1.51 and perhaps greater) put '\n' (\x0a) after serialization::archive instead of null,
# so let's just not test the byte after "archive"
if head.startswith(b'\x16\x00\x00\x00\x00\x00\x00\x00\x73\x65\x72\x69\x61\x6c\x69\x7a\x61\x74\x69\x6f\x6e\x3a\x3a\x61\x72\x63\x68\x69\x76\x65'):
format='boost::serialization'
elif head.startswith(b'<?xml version="1.0"'):
format='boost::serialization'
elif head.startswith(b'##woo-expression##'):
format='expr'
else:
# test pickling by trying to load
try: return typeChecked(pickle.load(open(inFile,'rb')),typ) # open again to seek to the beginning
except (IOError,KeyError,pickle.UnpicklingError,EOFError): pass
try: return typeChecked(WooJSONDecoder().decode(codecs.open(inFile,'rb','utf-8').read()),typ)
except (IOError,ValueError): pass
if not format: raise RuntimeError('File format detection failed on %s (head: %s, bin: %s)'%(inFile,''.join(["\\x%02x"%(x if py3k else ord(x)) for x in head]),str(head))) # in py3k, bytes contain integers rather than chars
if format not in validFormats: raise RuntimeError("format='%s'??"%format)
assert format in validFormats
if overrideHashColon and format!='expr': raise ValueError("overrideHashColon only applicable with the 'expr' format (not '%s')"%format)
if format==None:
raise IOError('Input file format not detected')
elif format=='boost::serialization':
# ObjectIO takes care of detecting binary, xml, compression independently
return typeChecked(Object._boostLoad(str(inFile)),typ) # convert unicode to str, if necessary, as the c++ type is std::string
elif format=='expr':
buf=codecs.open(inFile,'rb','utf-8').read()
return typeChecked(wooExprEval(buf,inFile,__overrideHashColon=overrideHashColon),typ)
elif format=='pickle':
return typeChecked(pickle.load(open(inFile,'rb')),typ)
elif format=='json':
return typeChecked(WooJSONDecoder().decode(codecs.open(inFile,'rb','utf-8').read()),typ)
assert False