liblinear.patch

*** liblinear.py	Wed Apr 25 07:32:21 2012
--- liblinear.py	Thu Aug 30 10:59:42 2012
***************
*** 42,55 ****
  	_types = [c_int, c_double]
  	_fields_ = genFields(_names, _types)
  
  def gen_feature_nodearray(xi, feature_max=None, issparse=True):
  	if isinstance(xi, dict):
  		index_range = xi.keys()
  	elif isinstance(xi, (list, tuple)):
  		xi = [0] + xi  # idx should start from 1
  		index_range = range(1, len(xi))
  	else:
! 		raise TypeError('xi should be a dictionary, list or tuple')
  
  	if feature_max:
  		assert(isinstance(feature_max, int))
--- 42,77 ----
  	_types = [c_int, c_double]
  	_fields_ = genFields(_names, _types)
  
+ def gen_feature_nodearray_from_array(xi, feature_max=None, issparse=True):
+ 	values = xi
+ 	if feature_max:
+ 		assert(isinstance(feature_max, int))
+ 		values = filter(lambda x: x[0] <= feature_max, values)
+ 	if issparse: 
+ 		values = filter(lambda x: x[1] != 0, values)
+ 
+ 	values = sorted(values, key=lambda x: x[0])
+ 	ret = (feature_node * (len(values)+2))()
+ 	ret[-1].index = -1 # for bias term
+ 	ret[-2].index = -1
+ 	for idx, (feature, value) in enumerate(values):
+ 		ret[idx].index = feature
+ 		ret[idx].value = value
+ 	max_idx = 0
+ 	if values: 
+ 		max_idx = values[-1][0]
+ 	return ret, max_idx
+ 
  def gen_feature_nodearray(xi, feature_max=None, issparse=True):
  	if isinstance(xi, dict):
  		index_range = xi.keys()
  	elif isinstance(xi, (list, tuple)):
  		xi = [0] + xi  # idx should start from 1
  		index_range = range(1, len(xi))
+ 	elif isinstance(xi, Array):
+ 		return gen_feature_nodearray_from_array(xi, feature_max, issparse)
  	else:
! 		raise TypeError('xi should be a dictionary, list, tuple or ctypes.Array')
  
  	if feature_max:
  		assert(isinstance(feature_max, int))