Merge pull request #60 from goodfeli/Rop_defense

theano/gof/op.py

@@ -567,3 +567,23 @@ def get_test_value(v):
     v_tensor = theano.tensor.as_tensor_variable(v)
     return PureOp._get_test_value(v_tensor)
 
+def missing_test_message(msg):
+    """ Displays msg, a message saying that some test_value is missing,
+    in the appropriate form based on config.compute_test_value:
+
+        off: the interactive debugger is off, so we do nothing
+        ignore: the interactive debugger is set to ignore missing inputs,
+                so do nothing
+        warn: display msg as a warning
+        raise: raise an AttributeError with msg as the exception text
+    """
+
+
+    action = config.compute_test_value
+
+    if action == 'raise':
+        raise AttributeError(msg)
+    elif action == 'warn':
+        warnings.warn(msg, stacklevel = 2)
+    else:
+        assert action in [ 'ignore', 'off' ]

theano/tensor/basic.py

@@ -20,6 +20,7 @@ from theano.gof import Apply, Constant, Op, Type, Value, Variable
 import elemwise
 from theano import scalar as scal
 from theano.gof.python25 import partial, any, all
+from theano.gof.op import get_test_value, missing_test_message
 from theano import compile, printing
 from theano.printing import pprint, min_informative_str
 
@@ -5216,8 +5217,50 @@ class Dot(Op):
         # simply c \dot b + a \dot d
         if None in eval_points:
             return [None]
-        t1 = self.make_node(eval_points[0], inputs[1]).outputs[0]
-        t2 = self.make_node(inputs[0], eval_points[1]).outputs[0]
+
+        assert len(inputs) == 2
+        assert len(eval_points) == 2
+
+        debugger_available = config.compute_test_value != 'off'
+
+        if debugger_available:
+            try:
+                iv0 = get_test_value(inputs[0])
+            except AttributeError:
+                missing_test_message('first input passed to Dot.R_op has no test value')
+                debugger_available = False
+
+            try:
+                iv1 = get_test_value(inputs[1])
+            except AttributeError:
+                missing_test_message('second input passed to Dot.R_op has no test value')
+                debugger_available = False
+
+            try:
+                ev0 = get_test_value(eval_points[0])
+            except AttributeError:
+                missing_test_message('first eval point passed to Dot.R_op has no test value')
+                debugger_available = False
+            try:
+                ev1 = get_test_value(eval_points[1])
+            except AttributeError:
+                missing_test_message('second eval point passed to Dot.R_op has no test value')
+                debugger_available = False
+
+        if debugger_available:
+            input_values = [ iv0, iv1]
+            eval_point_values = [ ev0, ev1 ]
+
+            for i in xrange(2):
+                if input_values[i].shape != eval_point_values[i].shape:
+                    raise ValueError('input '+str(i)+' and eval_point '+str(i)+' to Dot.R_op '
+                            'should have the '
+                        'same shape, but their shapes are %s and %s, respectively' % ( \
+                                str(input_values[i].shape), str(eval_point_values[i].shape) ) )
+
+        t1 = self(eval_points[0], inputs[1])
+        t2 = self(inputs[0], eval_points[1])
+
         return [t1+t2]
 
 

theano/tensor/tensor_grad.py

@@ -64,6 +64,20 @@ def Rop(f, wrt, eval_points):
 
     assert len(wrt) == len(eval_points)
 
+    for pack in enumerate(zip(wrt, eval_points)):
+        i = pack[0]
+        wrt_elem, eval_point = pack[1]
+
+        wrt_elem = as_tensor_variable(wrt_elem)
+        eval_point = as_tensor_variable(eval_point)
+
+        wrt_dim = len(wrt_elem.type.broadcastable)
+        eval_dim = len(eval_point.type.broadcastable)
+
+        if wrt_dim != eval_dim:
+            raise ValueError('Element '+str(i)+' of wrt/eval_point have mismatched '
+                    'dimensionality: '+str(wrt_dim)+' versus '+str(eval_dim))
+
     seen_nodes = {}
 
     def _traverse(node):

theano/tensor/tests/test_rop.py

@@ -261,3 +261,14 @@ class test_RopLop(unittest.TestCase):
                 self.mat_in_shape[0], self.mat_in_shape[1], self.in_shape[0])
         self.check_rop_lop(out3d.flatten(),
                 self.mat_in_shape[0] * self.mat_in_shape[1] * self.in_shape[0])
+
+    def test_invalid_input(self):
+        success = False
+
+        try:
+            tensor.Rop(0., [ tensor.matrix() ], [ tensor.vector() ] )
+            success = True
+        except ValueError:
+            pass
+
+        assert not success