Add c_code to GpuFromHost too.

theano/sandbox/gpuarray/basic_ops.py

@@ -44,10 +44,7 @@ class HostFromGpu(Op):
         z[0] = numpy.asarray(x)
 
     def c_code(self, node, name, inputs, outputs, sub):
-        inp = inputs[0]
-        out = outputs[0]
-        fail = sub['fail']
-        return """{
+        return """
         GpuArray %(name)s_ga_s;
         GpuArray *%(name)s_ga;
         int %(name)serr;
@@ -80,7 +77,8 @@ class HostFromGpu(Op):
         if (%(name)serr != GA_NO_ERROR) {
             %(fail)s
         }
-        }""" % {'name': name, 'fail': sub['fail'], 'inp': inp, 'out': out}
+        """ % {'name': name, 'fail': sub['fail'], 'inp': inputs[0],
+                'out': outputs[0]}
 
     def grad(self, inputs, grads):
         gz, = grads
@@ -133,6 +131,40 @@ class GpuFromHost(Op):
     def infer_shape(self, node, xshp):
         return xshp
 
+    def c_code(self, node, name, inputs, outputs, sub):
+        type = node.outputs[0].type
+        return """
+        PyArrayObject *%(name)s_tmp;
+        int %(name)serr;
+        %(name)s_tmp = PyArray_GETCONTIGUOUS(%(inp)s);
+        if (%(name)s_tmp == NULL) {
+            %(fail)s
+        }
+        Py_DECREF(%(inp)s);
+        %(inp)s = %(name)s_tmp;
+        %(out)s = new_GpuArray((PyObject *)&GpuArrayType);
+        if (%(out)s == NULL) {
+            %(fail)s
+        }
+        %(name)serr = GpuArray_empty(&%(out)s->ga, compyte_get_ops("%(kind)s"),
+                                     (void *)%(ctx)s, %(typecode)s,
+                                     PyArray_NDIM(%(inp)s),
+                                     (size_t *)PyArray_DIMS(%(inp)s),
+                                     GA_C_ORDER);
+        if (%(name)serr != GA_NO_ERROR) {
+            Py_DECREF(%(out)s);
+            %(fail)s
+        }
+        %(name)serr = GpuArray_write(&%(out)s->ga, PyArray_DATA(%(inp)s),
+                                     PyArray_NBYTES(%(inp)s));
+        if (%(name)serr != GA_NO_ERROR) {
+            Py_DECREF(%(out)s);
+            %(fail)s
+        }
+        """ % {'name': name, 'kind': type.kind, 'ctx': hex(type.context),
+               'inp': inputs[0], 'out': outputs[0], 'fail': sub['fail'],
+               'typecode': type.typecode}
+
 
 gpu_from_host = GpuFromHost()
 
@@ -157,6 +189,8 @@ class GpuFromCuda(Op):
             base = x
             while hasattr(base, 'base') and base.base is not None:
                 base = base.base
+            # TODO: I know how to do this in C, but I don't know about python.
+            #       Is perform() actually required to work?
             raise NotImplementedError("How are we going to get a gpudata pointer from here")
             x[0] = gpuarray.from_gpudata(b, 0, x.dtype, x.shape,
                                          base=base, kind=globals.kind,