Merge pull request #4214 from abergeron/fix_cudnn_mac

theano/compile/ops.py

@@ -11,7 +11,7 @@ import warnings
 import theano
 from theano import gof
 from theano.compat import OrderedDict
-from six import iteritems
+from six import iteritems, integer_types
 from six.moves import xrange
 
 
@@ -647,7 +647,7 @@ class Rebroadcast(gof.Op):
         items = sorted(axis)
         self.axis = OrderedDict(items)
         for axis, broad in iteritems(self.axis):
-            if not isinstance(axis, (numpy.integer, int)):
+            if not isinstance(axis, (numpy.integer, integer_types)):
                 raise TypeError("Rebroadcast needs integer axes. "
                                 "Got {}".format(axis))
 

theano/d3viz/tests/models.py

 import numpy as np
 
+from six import integer_types
 import theano as th
 import theano.tensor as T
 
@@ -9,7 +10,7 @@ class Mlp(object):
     def __init__(self, nfeatures=100, noutputs=10, nhiddens=50, rng=None):
         if rng is None:
             rng = 0
-        if isinstance(rng, int):
+        if isinstance(rng, integer_types):
             rng = np.random.RandomState(rng)
         self.rng = rng
         self.nfeatures = nfeatures

theano/gof/graph.py

@@ -383,7 +383,7 @@ class Variable(Node):
         if owner is not None and not isinstance(owner, Apply):
             raise TypeError("owner must be an Apply instance", owner)
         self.owner = owner
-        if index is not None and not isinstance(index, int):
+        if index is not None and not isinstance(index, integer_types):
             raise TypeError("index must be an int", index)
         self.index = index
         if name is not None and not isinstance(name, string_types):

theano/gof/opt.py

@@ -19,7 +19,7 @@ import numpy
 import theano
 from theano import config
 from theano.compat import izip, OrderedDict
-from six import string_types, iteritems, itervalues
+from six import string_types, iteritems, itervalues, integer_types
 from six.moves import reduce
 from theano.gof import graph, op, utils, unify, toolbox
 from theano.gof.fg import InconsistencyError
@@ -1507,7 +1507,7 @@ class PatternSub(LocalOptimizer):
                     return retry_with_equiv()
                 else:
                     u = u.merge(expr, v)
-            elif (isinstance(pattern, (int, float)) and
+            elif (isinstance(pattern, (integer_types, float)) and
                     isinstance(expr, graph.Constant)):
                 if numpy.all(
                         theano.tensor.constant(pattern).value == expr.value):
@@ -1534,7 +1534,7 @@ class PatternSub(LocalOptimizer):
                     return pattern[0](*args)
                 elif isinstance(pattern, string_types):
                     return u[unify.Var(pattern)]
-                elif isinstance(pattern, (int, float)):
+                elif isinstance(pattern, (integer_types, float)):
                     return pattern
                 else:
                     return pattern.clone()

theano/gof/optdb.py

@@ -4,7 +4,7 @@ import sys
 
 from theano.compat import DefaultOrderedDict
 from theano.misc.ordered_set import OrderedSet
-from six import StringIO
+from six import StringIO, integer_types
 from theano.gof import opt
 from theano import config
 
@@ -310,7 +310,7 @@ class SequenceDB(DB):
 
     def register(self, name, obj, position, *tags):
         super(SequenceDB, self).register(name, obj, *tags)
-        assert isinstance(position, (int, float))
+        assert isinstance(position, (integer_types, float))
         self.__position__[name] = position
 
     def query(self, *tags, **kwtags):

theano/gof/utils.py

@@ -3,7 +3,7 @@ import linecache
 import sys
 
 import numpy
-from six import iteritems
+from six import iteritems, integer_types, string_types
 
 from theano import config
 from theano.compat import OrderedDict, PY3
@@ -326,7 +326,7 @@ RETRY = Keyword("RETRY", False)
 FAILURE = Keyword("FAILURE", False)
 
 
-simple_types = (int, float, str, bool, None.__class__, Keyword)
+simple_types = integer_types + string_types + (float, bool, None.__class__, Keyword)
 
 
 ANY_TYPE = Keyword("ANY_TYPE")
@@ -521,7 +521,7 @@ def hash_from_dict(d):
     first_part = [k for k, v in items]
     second_part = []
     for k, v in items:
-        assert isinstance(k, (str, int, float))
+        assert isinstance(k, (string_types, integer_types, float))
         if isinstance(v, (tuple, list)):
             second_part += [tuple(v)]
         else:

theano/printing.py

@@ -13,11 +13,11 @@ import hashlib
 
 import numpy as np
 from six import string_types, integer_types, iteritems
+from six.moves import StringIO, reduce
 
 import theano
 from theano import gof
 from theano import config
-from six.moves import StringIO, reduce
 from theano.gof import Op, Apply
 from theano.compile import Function, debugmode, SharedVariable
 from theano.compile.profilemode import ProfileMode
@@ -88,7 +88,7 @@ def debugprint(obj, depth=-1, print_type=False,
     to the Apply's identifier, to indicate which output a line corresponds to.
 
     """
-    if not isinstance(depth, int):
+    if not isinstance(depth, integer_types):
         raise Exception("depth parameter must be an int")
     if file == 'str':
         _file = StringIO()

theano/sandbox/cuda/blas.py

@@ -4,10 +4,12 @@ import os
 import logging
 _logger = logging.getLogger(__name__)
 
+from six import integer_types
+from six.moves import StringIO, reduce
+
 import theano
 from theano import Apply
 from theano import tensor
-from six.moves import StringIO, reduce
 from theano.sandbox.cuda.type import CudaNdarrayType
 from theano.sandbox.cuda import GpuOp
 from theano.sandbox.cuda.basic_ops import (as_cuda_ndarray_variable,
@@ -874,7 +876,7 @@ class BaseGpuCorrMM(GpuOp):
             if border_mode != "valid":
                 raise ValueError("border_mode must be 'valid' if pad is given")
             border_mode = pad
-        if isinstance(border_mode, int):
+        if isinstance(border_mode, integer_types):
             border_mode = (border_mode, border_mode)
         if isinstance(border_mode, tuple):
             pad_h, pad_w = map(int, border_mode)

theano/sandbox/cuda/dnn.py

@@ -2,6 +2,8 @@ import os
 import numpy
 import warnings
 
+from six import integer_types
+
 import theano
 from theano import Apply, tensor, config, Variable
 from theano.scalar import as_scalar, constant, Log
@@ -127,7 +129,7 @@ class GpuDnnConvDesc(GpuOp):
 
     def __init__(self, border_mode, subsample=(1, 1), conv_mode='conv',
                  precision="float32"):
-        if isinstance(border_mode, int):
+        if isinstance(border_mode, integer_types):
             border_mode = (border_mode,) * len(subsample)
         if isinstance(border_mode, tuple):
             assert len(border_mode) == len(subsample)

theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py

@@ -217,19 +217,29 @@ def _params_allgood(ishape, kshape, mode, subsample=(1, 1), img_stride=(1, 1),
     assert [(sh == 1) is br for
             sh, br in zip(cpuval.shape[:2], op.type.broadcastable[:2])]
 
-    if (t2 is not None):
+    if (t2 is not None and verbose > 0):
         if mode == 'valid':
             approx_fp = cpuval.size * ishape[1] * kshape[2] * kshape[3] * 2
         else:
             approx_fp = (ishape[0] * kshape[0] * kshape[1] * kshape[2] *
                          kshape[3] * ishape[2] * ishape[3] * 2)
         approx_fp /= 1e6
-        cpu_mflops = approx_fp / (t1 - t0)
-        gpu_mflops = approx_fp / (t3 - t2)
-        if verbose > 0:
-            print('%15s' % str(ishape), '%15s' % str(kshape), end=' ', file=sys.stdout)
-            print('%12.5f  %7.2f %7.2f %7.1f' % (approx_fp,
-                    cpu_mflops, gpu_mflops, (t1 - t0) / (t2 - t1)), file=sys.stdout)
+        if t1 - t0 != 0:
+            cpu_mflops = approx_fp / (t1 - t0)
+        else:
+            cpu_mflops = float('inf')
+        if t3 - t2 != 0:
+            gpu_mflops = approx_fp / (t3 - t2)
+        else:
+            gpu_mflops = float('inf')
+
+        if t2 - t1 != 0:
+            div = (t1 - t0) / (t2 - t1)
+        else:
+            div = float('inf')
+        print('%15s' % str(ishape), '%15s' % str(kshape), end=' ')
+        print('%12.5f  %7.2f %7.2f %7.1f' % (
+                approx_fp, cpu_mflops, gpu_mflops, div))
 
 
 def exec_conv(version, shapes, verbose, random, mode,

theano/sandbox/gpuarray/dnn.py

 import os
-import numpy
 import warnings
 
+import numpy
+from six import integer_types
+
 import theano
 from theano import Op, Apply, tensor, config, Variable
 from theano.scalar import as_scalar, constant, Log
@@ -281,7 +283,7 @@ class GpuDnnConvDesc(COp):
                  precision="float32"):
         COp.__init__(self, ["conv_desc.c"], "APPLY_SPECIFIC(conv_desc)")
 
-        if isinstance(border_mode, int):
+        if isinstance(border_mode, integer_types):
             border_mode = (border_mode,) * len(subsample)
         if isinstance(border_mode, tuple):
             assert len(border_mode) == len(subsample)

theano/sandbox/gpuarray/subtensor.py

@@ -4,10 +4,11 @@ import os
 import copy
 
 import numpy
+from six import integer_types
+from six.moves import StringIO
 
 import theano
 from theano import tensor, gof
-from six.moves import StringIO
 from theano.tensor.subtensor import IncSubtensor, Subtensor, get_idx_list
 import theano.tensor.inplace
 
@@ -116,7 +117,7 @@ class GpuSubtensor(HideC, Subtensor):
         def fix_idx(idx):
             if idx is None:
                 return "0", 1
-            elif isinstance(idx, (numpy.integer, int)):
+            elif isinstance(idx, (numpy.integer, integer_types)):
                 return str(idx), 0
             elif isinstance(idx, gof.Type):
                 return indices.pop(0), 0
@@ -143,7 +144,7 @@ class GpuSubtensor(HideC, Subtensor):
             else:
                 if isinstance(idx, gof.Type):
                     start = indices.pop(0)
-                elif isinstance(idx, (numpy.integer, int)):
+                elif isinstance(idx, (numpy.integer, integer_types)):
                     start = idx
                 else:
                     assert 0, idx

theano/sandbox/linalg/ops.py

@@ -3,7 +3,7 @@ import logging
 logger = logging.getLogger(__name__)
 import numpy
 
-from six import iteritems
+from six import iteritems, integer_types
 from six.moves import xrange
 
 from theano.gof import Op, Apply
@@ -412,7 +412,7 @@ def spectral_radius_bound(X, log2_exponent):
     """
     if X.type.ndim != 2:
         raise TypeError('spectral_radius_bound requires a matrix argument', X)
-    if not isinstance(log2_exponent, int):
+    if not isinstance(log2_exponent, integer_types):
         raise TypeError('spectral_radius_bound requires an integer exponent',
                         log2_exponent)
     if log2_exponent <= 0:

theano/sandbox/rng_mrg.py

@@ -9,6 +9,7 @@ from __future__ import print_function
 import warnings
 
 import numpy
+from six import integer_types
 from six.moves import xrange
 
 from theano import Op, Apply, shared, config, Variable
@@ -396,7 +397,8 @@ class mrg_uniform(mrg_uniform_base):
             NORM = '4.656612873077392578125e-10'
         return """
         //////// <code generated by mrg_uniform>
-        npy_int64 odims[%(ndim)s];
+        // The +1 is to avoid odims[0] which fails on windows
+        npy_int64 odims[%(ndim)s+1];
         npy_int64 n_elements = 1;
         int n_streams = 0;
         int must_alloc_sample = ((NULL == %(o_sample)s)
@@ -667,7 +669,8 @@ class GPU_mrg_uniform(mrg_uniform_base, GpuOp):
         return """
         //////// <code generated by mrg_uniform>
         npy_int64 M1 = 2147483647;      //2^31 - 1
-        npy_int64 odims[%(ndim)s];
+        // The +1 is to avoid odims[0] which fails on windows
+        npy_int64 odims[%(ndim)s+1];
         npy_int64 n_elements = 1;
         int n_streams, n_streams_used_in_this_call;
         int must_alloc_sample = ((NULL == %(o_sample)s)
@@ -931,7 +934,8 @@ class GPUA_mrg_uniform(GpuKernelBase, mrg_uniform_base):
 
         return """
         npy_int64 M1 = 2147483647;      //2^31 - 1
-        size_t odims[%(ndim)s];
+        // The +1 is to avoid odims[0] which fails on windows
+        size_t odims[%(ndim)s+1];
         size_t n_elements = 1;
         unsigned int n_streams;
         int must_alloc_sample = ((NULL == %(o_sample)s)
@@ -1060,7 +1064,7 @@ def guess_n_streams(size, warn=False):
     # Note that this code was moved out of `MRG_RandomStreams` so that it can
     # be easily accessed from tests, where we want to disable the warning.
     if (isinstance(size, (tuple, list)) and
-            all([isinstance(i, int) for i in size])):
+            all([isinstance(i, integer_types) for i in size])):
         # We can make a guess.
         r = 1
         for s in size:
@@ -1127,7 +1131,7 @@ class MRG_RandomStreams(object):
 
     def set_rstate(self, seed):
         # TODO : need description for method, parameter
-        if isinstance(seed, int):
+        if isinstance(seed, integer_types):
             if seed == 0:
                 raise ValueError('seed should not be 0', seed)
             elif seed >= M2:
@@ -1289,9 +1293,9 @@ class MRG_RandomStreams(object):
 
         if isinstance(size, tuple):
             msg = "size must be a tuple of int or a Theano variable"
-            assert all([isinstance(i, (numpy.integer, int, Variable))
+            assert all([isinstance(i, (numpy.integer, integer_types, Variable))
                         for i in size]), msg
-            if any([isinstance(i, (numpy.integer, int)) and i <= 0
+            if any([isinstance(i, (numpy.integer, integer_types)) and i <= 0
                     for i in size]):
                 raise ValueError(
                     "The specified size contains a dimension with value <= 0",
@@ -1377,7 +1381,7 @@ class MRG_RandomStreams(object):
             raise TypeError("You have to specify pvals")
         pvals = as_tensor_variable(pvals)
         if size is not None:
-            if any([isinstance(i, int) and i <= 0 for i in size]):
+            if any([isinstance(i, integer_types) and i <= 0 for i in size]):
                 raise ValueError(
                     "The specified size contains a dimension with value <= 0",
                     size)
@@ -1483,7 +1487,7 @@ class MRG_RandomStreams(object):
         evened = False
         constant = False
         if (isinstance(size, tuple) and
-                all([isinstance(i, (numpy.integer, int)) for i in size])):
+                all([isinstance(i, (numpy.integer, integer_types)) for i in size])):
             constant = True
             # Force dtype because it defaults to float when size is empty
             n_samples = numpy.prod(size, dtype='int64')

theano/scalar/sharedvar.py

@@ -17,6 +17,8 @@ way (as scan does) to create a shared variable of this kind.
 """
 
 import numpy
+from six import integer_types
+
 from theano.compile import SharedVariable
 from .basic import Scalar, _scalar_py_operators
 
@@ -46,7 +48,7 @@ def shared(value, name=None, strict=False, allow_downcast=None):
     We implement this using 0-d tensors for now.
 
     """
-    if not isinstance(value, (numpy.number, float, int, complex)):
+    if not isinstance(value, (numpy.number, float, integer_types, complex)):
         raise TypeError()
     try:
         dtype = value.dtype

theano/scan_module/scan.py

@@ -48,7 +48,7 @@ import numpy
 import warnings
 
 from theano.compat import ifilter, izip
-from six import iteritems
+from six import iteritems, integer_types
 from six.moves import xrange
 from theano.compile import SharedVariable, function
 from theano import compile
@@ -372,7 +372,7 @@ def scan(fn,
     # To do that we check here to see the nature of n_steps
     n_fixed_steps = None
 
-    if isinstance(n_steps, (float, int)):
+    if isinstance(n_steps, (float, integer_types)):
         n_fixed_steps = int(n_steps)
     else:
         try:

theano/scan_module/scan_op.py

@@ -62,7 +62,7 @@ import logging
 import time
 
 import numpy
-from six import iteritems
+from six import iteritems, integer_types
 from six.moves import xrange
 
 import theano
@@ -837,7 +837,7 @@ class Scan(PureOp):
 
         profile = None
         if (theano.config.profile or
-            (isinstance(self.profile, (string_types, bool, int))
+            (isinstance(self.profile, (string_types, bool, integer_types))
                                       and self.profile)):
             if isinstance(self.profile, string_types):
                 profile = ScanProfileStats(name=self.profile)

theano/sparse/basic.py

@@ -14,6 +14,7 @@ import sys
 
 import numpy
 from numpy.lib.stride_tricks import as_strided
+from six import integer_types
 from six.moves import xrange
 import scipy.sparse
 
@@ -1402,7 +1403,7 @@ class GetItemScalar(gof.op.Op):
                 raise Exception("GetItemScalar called with a slice as index!")
 
             # in case of indexing using int instead of theano variable
-            elif isinstance(ind, int):
+            elif isinstance(ind, integer_types):
                 ind = theano.tensor.constant(ind)
                 input_op += [ind]
 

theano/tensor/basic.py

@@ -4,6 +4,7 @@ import sys
 import warnings
 
 import numpy
+from six import integer_types
 from six.moves import xrange
 import numbers
 
@@ -21,7 +22,6 @@ from theano.tensor.type import TensorType, values_eq_approx_always_true
 from theano.tensor.type_other import NoneConst
 from theano import scalar as scal
 from functools import partial
-from six import integer_types
 from theano import compile, printing
 from theano.printing import pprint, min_informative_str
 # For history
@@ -606,7 +606,7 @@ def get_scalar_constant_value(orig_v, elemwise=True,
             # to depend on passing it None)
             raise NotScalarConstantError()
 
-        if isinstance(v, (numpy.integer, int, float)):
+        if isinstance(v, (numpy.integer, integer_types, float)):
             return numpy.asarray(v)
 
         if isinstance(v, numpy.ndarray):
@@ -786,7 +786,7 @@ def tensor(*args, **kwargs):
 
 def _multi(*fns):
     def f2(f, *names):
-        if names and isinstance(names[0], int):
+        if names and isinstance(names[0], integer_types):
             if names == 1:
                 return f()
             else:
@@ -1290,7 +1290,7 @@ class MaxAndArgmax(Op):
     def make_node(self, x, axis=None):
         x = _as_tensor_variable(x)
 
-        if isinstance(axis, (int, numpy.integer)):
+        if isinstance(axis, (integer_types, numpy.integer)):
             axis = [int(axis)]
         elif isinstance(axis, numpy.ndarray) and axis.ndim == 0:
             axis = [int(axis)]
@@ -1307,7 +1307,7 @@ class MaxAndArgmax(Op):
             else:
                 assert (axis.dtype.startswith("int") or
                         axis.dtype.startswith("uint"))
-                if isinstance(axis.data, (int, numpy.integer)) or \
+                if isinstance(axis.data, (integer_types, numpy.integer)) or \
                    (isinstance(axis.data, numpy.ndarray) and
                         axis.data.ndim == 0):
                     axis = [int(axis.data)]
@@ -1536,7 +1536,7 @@ def makeKeepDims(x, y, axis):
 
     if axis is None:
         axis = list(range(x.type.ndim))
-    elif isinstance(axis, (int, numpy.integer)):
+    elif isinstance(axis, (integer_types, numpy.integer)):
         axis = [axis]
     elif isinstance(axis, numpy.ndarray) and axis.ndim == 0:
         axis = [int(axis)]
@@ -1544,7 +1544,7 @@ def makeKeepDims(x, y, axis):
         axis = [int(a) for a in axis]
     newaxis = []
     for a in axis:
-        if not isinstance(a, int):
+        if not isinstance(a, integer_types):
             raise ValueError(
                 "keepdims option can be used only with constant axis")
         if a < 0:
@@ -3082,7 +3082,7 @@ def mean(input, axis=None, dtype=None, op=False, keepdims=False,
 
     if axis is None:
         axis = list(range(input.ndim))
-    elif isinstance(axis, (int, numpy.integer)):
+    elif isinstance(axis, (integer_types, numpy.integer)):
         axis = [axis]
     elif isinstance(axis, numpy.ndarray) and axis.ndim == 0:
         axis = [int(axis)]
@@ -3126,7 +3126,7 @@ def var(input, axis=None, keepdims=False):
     input_ndim = input.type.ndim
     if axis is None:
         axis = list(range(input_ndim))
-    elif isinstance(axis, (int, numpy.integer)):
+    elif isinstance(axis, (integer_types, numpy.integer)):
         axis = [axis]
     elif isinstance(axis, numpy.ndarray) and axis.ndim == 0:
         axis = [int(axis)]
@@ -3769,7 +3769,7 @@ class Join(Op):
                 as_tensor_variable_args[0].type.broadcastable)
             ndim = len(bcastable)
             # Axis can also be a constant
-            if not isinstance(axis, int):
+            if not isinstance(axis, integer_types):
                 try:
                     # Note : `get_scalar_constant_value` returns a ndarray not
                     # an int
@@ -3777,7 +3777,7 @@ class Join(Op):
 
                 except NotScalarConstantError:
                     pass
-            if isinstance(axis, int):
+            if isinstance(axis, integer_types):
                 # Basically, broadcastable -> length 1, but the
                 # converse does not hold. So we permit e.g. T/F/T
                 # joins, and if they fail at runtime they fail, but if

theano/tensor/elemwise.py

@@ -3,12 +3,12 @@ import sys
 from copy import copy
 
 import numpy
+from six import iteritems, integer_types
+from six.moves import xrange
 
 import theano
 from theano import gof
 from theano.compat import izip
-from six import iteritems
-from six.moves import xrange
 from theano.gof import Apply, Op, OpenMPOp
 from theano import scalar
 from theano.scalar import get_scalar_type
@@ -135,10 +135,11 @@ class DimShuffle(Op):
 
         for i, j in enumerate(new_order):
             if j != 'x':
-                # There is a bug in numpy that results in isinstance(x, int)
-                # returning False for numpy integers.
-                # See <http://projects.scipy.org/numpy/ticket/2235>.
-                if not isinstance(j, (int, numpy.integer)):
+                # There is a bug in numpy that results in
+                # isinstance(x, integer_types) returning False for
+                # numpy integers.  See
+                # <http://projects.scipy.org/numpy/ticket/2235>.
+                if not isinstance(j, (integer_types, numpy.integer)):
                     raise TypeError("DimShuffle indices must be python ints.")
                 if j >= len(input_broadcastable):
                     raise ValueError(("new_order[%d] is %d, but the input "
@@ -1325,10 +1326,10 @@ class CAReduce(Op):
 
         if axis is None:
             self.axis = axis
-        # There is a bug in numpy that results in isinstance(x, int) returning
-        # False for numpy integers.
-        # See <http://projects.scipy.org/numpy/ticket/2235>.
-        elif isinstance(axis, (int, numpy.integer)):
+        # There is a bug in numpy that results in isinstance(x,
+        # integer_types) returning False for numpy integers.  See
+        # <http://projects.scipy.org/numpy/ticket/2235>.
+        elif isinstance(axis, (integer_types, numpy.integer)):
             self.axis = (axis,)
         elif isinstance(axis, numpy.ndarray) and axis.ndim == 0:
             self.axis = (int(axis),)

theano/tensor/nnet/abstract_conv.py

@@ -4,7 +4,7 @@ Abstract conv interface
 
 import numpy as np
 import logging
-from six import reraise
+from six import reraise, integer_types
 import sys
 
 import theano
@@ -238,17 +238,17 @@ def conv2d_grad_wrt_inputs(output_grad,
     # checking the type of input_shape
     for dim in [0, 1]:
         assert isinstance(input_shape[dim], (theano.tensor.TensorConstant,
-                                             int, type(None)))
+                                             integer_types, type(None)))
     for dim in [2, 3]:
         assert isinstance(input_shape[dim], (theano.tensor.TensorVariable,
                                              theano.tensor.TensorConstant,
-                                             int))
+                                             integer_types))
 
     # checking the type of filter_shape
     if filter_shape is not None:
         for dim in [0, 1, 2, 3]:
             assert isinstance(filter_shape[dim], (theano.tensor.TensorConstant,
-                                                  int, type(None)))
+                                                  integer_types, type(None)))
 
     # setting the last two dimensions of input_shape to None, if
     # the type of these dimensions is TensorVariable.
@@ -361,17 +361,17 @@ def conv2d_grad_wrt_weights(input,
     # checking the type of filter_shape
     for dim in [0, 1]:
         assert isinstance(filter_shape[dim], (theano.tensor.TensorConstant,
-                                              int, type(None)))
+                                              integer_types, type(None)))
     for dim in [2, 3]:
         assert isinstance(filter_shape[dim], (theano.tensor.TensorVariable,
                                               theano.tensor.TensorConstant,
-                                              int))
+                                              integer_types))
 
     # checking the type of input_shape
     if input_shape is not None:
         for dim in [0, 1, 2, 3]:
             assert isinstance(input_shape[dim], (theano.tensor.TensorConstant,
-                                                 int, type(None)))
+                                                 integer_types, type(None)))
 
     # setting the last two dimensions of filter_shape to None, if
     # the type of these dimensions is TensorVariable.
@@ -625,7 +625,7 @@ class BaseAbstractConv2d(Op):
                  border_mode="valid", subsample=(1, 1),
                  filter_flip=True):
 
-        if isinstance(border_mode, int):
+        if isinstance(border_mode, integer_types):
             border_mode = (border_mode, border_mode)
         if isinstance(border_mode, tuple):
             pad_h, pad_w = map(int, border_mode)

theano/tensor/nnet/corr.py

 import os
 import logging
 
+from six import integer_types
+
 import theano
 from theano import Apply
 from theano import gof
@@ -30,7 +32,7 @@ class BaseCorrMM(gof.Op):
     __props__ = ('border_mode', 'subsample')
 
     def __init__(self, border_mode="valid", subsample=(1, 1)):
-        if isinstance(border_mode, int):
+        if isinstance(border_mode, integer_types):
             if border_mode < 0:
                 raise ValueError(
                     'invalid border_mode {}, which must be a '

theano/tensor/nnet/tests/test_corr.py

 from nose.plugins.skip import SkipTest
 from nose.plugins.attrib import attr
 import numpy
+from six import integer_types
 
 import theano
 import theano.tensor as T
@@ -92,7 +93,7 @@ class TestCorr2D(utt.InferShapeTester):
             padHW = numpy.floor(fil_shape2d / 2).astype('int32')
         elif isinstance(border_mode, tuple):
             padHW = numpy.array(border_mode)
-        elif isinstance(border_mode, int):
+        elif isinstance(border_mode, integer_types):
             padHW = numpy.array([border_mode, border_mode])
         else:
             raise NotImplementedError('Unsupported border_mode {}'.format(border_mode))

theano/tensor/opt.py

@@ -14,14 +14,12 @@ import traceback
 import warnings
 
 import numpy
-import numpy as N  # guys... please don't do this in the library :(
-from six.moves import xrange
+from six import integer_types, iteritems
+from six.moves import reduce, xrange
 
 import theano
 from theano import gof
 from theano.compat import izip
-from six import integer_types, iteritems
-from six.moves import reduce
 from theano.gof import opt, InconsistencyError, TopoOptimizer, graph
 from theano.gof import Variable, Constant
 from theano.gof.utils import MethodNotDefined
@@ -1169,11 +1167,11 @@ class ShapeFeature(object):
                 #  - Shape_i(i)(other_r);
                 #  - Shape_i(i)(r).
                 merged_shape.append(r_shape[i])
-            elif isinstance(r_shape[i], (Constant, int)):
+            elif isinstance(r_shape[i], (Constant, integer_types)):
                 # We do this to call less often ancestors and make
                 # sure we have the simplest shape possible.
                 merged_shape.append(r_shape[i])
-            elif isinstance(other_shape[i], (Constant, int)):
+            elif isinstance(other_shape[i], (Constant, integer_types)):
                 # We do this to call less often ancestors and make
                 # sure we have the simplest shape possible.
                 merged_shape.append(other_shape[i])
@@ -1826,7 +1824,7 @@ def local_subtensor_make_vector(node):
     else:
         return
 
-    if isinstance(idx, (int, numpy.integer)):
+    if isinstance(idx, (integer_types, numpy.integer)):
         # We don't need to copy over any stack traces here
         return [x.owner.inputs[idx]]
     elif isinstance(idx, Variable):
@@ -2452,7 +2450,7 @@ def local_useless_subtensor(node):
 
             length_pos = shape_of[node.inputs[0]][pos]
 
-            if isinstance(idx.stop, (int, numpy.integer)):
+            if isinstance(idx.stop, (integer_types, numpy.integer)):
                 length_pos_data = sys.maxsize
                 try:
                     length_pos_data = get_scalar_constant_value(length_pos)
@@ -4497,7 +4495,7 @@ class Canonizer(gof.LocalOptimizer):
         num, denum = self.simplify(list(orig_num), list(orig_denum), out.type)
 
         def same(x, y):
-            return len(x) == len(y) and all(N.all(xe == ye) for xe, ye in
+            return len(x) == len(y) and all(numpy.all(xe == ye) for xe, ye in
                                             zip(x, y))
 
         if same(orig_num, num) and same(orig_denum, denum):
@@ -4538,7 +4536,7 @@ def mul_calculate(num, denum, aslist=False, out_type=None):
         if aslist:
             return []
         else:
-            return N.int8(1)
+            return numpy.int8(1)
 
     # Make sure we do not accidently upcast data types.
     if out_type is None:
@@ -4547,9 +4545,9 @@ def mul_calculate(num, denum, aslist=False, out_type=None):
         out_dtype = out_type.dtype
     one = theano._asarray(1, dtype=out_dtype)
 
-    v = reduce(N.multiply, num, one) / reduce(N.multiply, denum, one)
+    v = reduce(numpy.multiply, num, one) / reduce(numpy.multiply, denum, one)
     if aslist:
-        if N.all(v == 1):
+        if numpy.all(v == 1):
             return []
         else:
             return [v]
@@ -5205,7 +5203,7 @@ register_canonicalize(local_mul_zero)
 
 @gof.local_optimizer([T.true_div])
 def local_div_to_inv(node):
-    if node.op == T.true_div and N.all(
+    if node.op == T.true_div and numpy.all(
             local_mul_canonizer.get_constant(node.inputs[0]) == 1.0):
         out = node.outputs[0]
         new_out = T.inv(local_mul_canonizer.merge_num_denum(node.inputs[1:],
@@ -5286,19 +5284,19 @@ def local_pow_specialize(node):
                                               ysym.type.broadcastable):
             rval = None
 
-            if N.all(y == 2):
+            if numpy.all(y == 2):
                 rval = [T.sqr(xsym)]
-            if N.all(y == 1):
+            if numpy.all(y == 1):
                 rval = [xsym]
-            if N.all(y == 0):
+            if numpy.all(y == 0):
                 rval = [T.fill(xsym, numpy.asarray(1, dtype=odtype))]
-            if N.all(y == 0.5):
+            if numpy.all(y == 0.5):
                 rval = [T.sqrt(xsym)]
-            if N.all(y == -0.5):
+            if numpy.all(y == -0.5):
                 rval = [T.inv(T.sqrt(xsym))]
-            if N.all(y == -1):
+            if numpy.all(y == -1):
                 rval = [T.inv(xsym)]
-            if N.all(y == -2):
+            if numpy.all(y == -2):
                 rval = [T.inv(T.sqr(xsym))]
             if rval:
                 rval[0] = T.cast(rval[0], odtype)
@@ -5637,9 +5635,9 @@ def add_calculate(num, denum, aslist=False, out_type=None):
         zero = theano._asarray(0, dtype=out_type.dtype)
     # zero = 0.0 if out_type is None else theano._asarray(0,
     # dtype=out_type.dtype)
-    v = reduce(N.add, num, zero) - reduce(N.add, denum, zero)
+    v = reduce(numpy.add, num, zero) - reduce(numpy.add, denum, zero)
     if aslist:
-        if N.all(v == 0):
+        if numpy.all(v == 0):
             return []
         else:
             return [v]

theano/tensor/raw_random.py

@@ -5,14 +5,14 @@ import sys
 from copy import copy
 
 import numpy
+from six import string_types
+from six.moves import reduce, xrange
 
 # local imports
 import theano
-from six.moves import reduce, xrange
 from theano import tensor
 from theano.tensor import opt
 from theano import gof
-from six import string_types
 from theano.compile import optdb
 
 __docformat__ = "restructuredtext en"

theano/tensor/sharedvar.py

 import traceback
 
 import numpy
+from six import integer_types
 
 import theano.tensor.basic
 from theano.tensor.basic import TensorType, _tensor_py_operators
@@ -84,7 +85,7 @@ def scalar_constructor(value, name=None, strict=False, allow_downcast=None,
     if target != 'cpu':
         raise TypeError('not for cpu')
 
-    if not isinstance(value, (numpy.number, float, int, complex)):
+    if not isinstance(value, (numpy.number, float, integer_types, complex)):
         raise TypeError()
     try:
         dtype = value.dtype

theano/tensor/signal/pool.py

@@ -6,11 +6,12 @@ Pool, DownsampleAvg, DownsampleSoftmax.
 """
 from __future__ import print_function
 # This file should move along with conv.py
-from six.moves import xrange
-import six.moves.builtins as builtins
 import warnings
 
 import numpy
+from six import integer_types
+from six.moves import xrange
+import six.moves.builtins as builtins
 
 import theano
 from theano import gof, Op, tensor, Variable, Apply
@@ -233,7 +234,7 @@ class Pool(Op):
     def __init__(self, ds, ignore_border=False, st=None, padding=(0, 0),
                  mode='max'):
         self.ds = tuple(ds)
-        if not all([isinstance(d, int) for d in ds]):
+        if not all([isinstance(d, integer_types) for d in ds]):
             raise ValueError(
                 "Pool downsample parameters must be ints."
                 " Got %s" % str(ds))
@@ -890,7 +891,7 @@ class DownsampleFactorMaxGradGrad(Op):
 
     def __init__(self, ds, ignore_border, st=None, padding=(0, 0), mode='max'):
         self.ds = tuple(ds)
-        if not all([isinstance(d, int) for d in ds]):
+        if not all([isinstance(d, integer_types) for d in ds]):
             raise ValueError(
                 "Pool downsample parameters must be ints."
                 " Got %s" % str(ds))

theano/tensor/subtensor.py

@@ -5,11 +5,11 @@ import warnings
 import logging
 
 import numpy
+from six import integer_types
 from six.moves import xrange
 
 import theano
 from theano.compat import izip
-from six import integer_types
 from theano.gradient import DisconnectedType
 from theano import gof
 from theano.gof import Apply, Constant, hashtype, Op, Type, MethodNotDefined
@@ -513,7 +513,7 @@ class Subtensor(Op):
                         if start is None:
                             start = 0
                         if (p.stop is None or
-                            (isinstance(p.stop, (int, numpy.integer,
+                            (isinstance(p.stop, (integer_types, numpy.integer,
                                                  numpy.ndarray)) and
                              p.stop > start)):
                             broadcastable.append(True)
@@ -680,7 +680,7 @@ class Subtensor(Op):
             return pos[1]
 
         def init_entry(entry, depth=0):
-            if isinstance(entry, (numpy.integer, int)):
+            if isinstance(entry, (numpy.integer, integer_types)):
                 init_cmds.append(
                     "subtensor_spec[%i] = %i;" % (spec_pos(),
                                                   entry))
@@ -972,7 +972,7 @@ class SubtensorPrinter:
             sidxs = []
             inbrack_pstate = pstate.clone(precedence=-1000)
             for entry in idxs:
-                if isinstance(entry, int):
+                if isinstance(entry, integer_types):
                     sidxs.append(str(entry))
                 elif isinstance(entry, scal.Scalar):
                     sidxs.append(inbrack_pstate.pprinter.process(inputs.pop()))

theano/tensor/tests/test_keepdims.py

 import unittest
 
 import numpy
+from six import integer_types
 
 import theano
 from theano import tensor, function
@@ -14,7 +15,7 @@ class TestKeepDims(unittest.TestCase):
     def makeKeepDims_local(self, x, y, axis):
         if axis is None:
             newaxis = list(range(x.ndim))
-        elif isinstance(axis, int):
+        elif isinstance(axis, integer_types):
             if axis < 0:
                 newaxis = [axis + x.type.ndim]
             else:

theano/tensor/var.py

@@ -3,6 +3,7 @@ import traceback as tb
 import warnings
 
 import numpy
+from six import integer_types
 from six.moves import xrange
 
 import theano
@@ -319,7 +320,7 @@ class _tensor_py_operators(object):
         """
 
         if ndim is not None:
-            if not isinstance(ndim, int):
+            if not isinstance(ndim, integer_types):
                 raise ValueError("Expected ndim to be an integer, is " +
                                  str(type(ndim)))
 

theano/tests/unittest_tools.py

@@ -2,10 +2,12 @@ from __future__ import print_function
 from copy import copy, deepcopy
 from functools import wraps
 import logging
-from six.moves import StringIO
 import sys
 import unittest
 
+from six import integer_types
+from six.moves import StringIO
+
 try:
     from nose.plugins.attrib import attr
 except ImportError:
@@ -204,7 +206,7 @@ class InferShapeTester(unittest.TestCase):
             mode = mode.excluding(*excluding)
         if warn:
             for var, inp in zip(inputs, numeric_inputs):
-                if isinstance(inp, (int, float, list, tuple)):
+                if isinstance(inp, (integer_types, float, list, tuple)):
                     inp = var.type.filter(inp)
                 if not hasattr(inp, "shape"):
                     continue