use latest error handling conventions

nnvm/python/nnvm/frontend/__init__.py

@@ -1,16 +1,12 @@
 """NNVM frontends."""
 from __future__ import absolute_import
+from tvm.error import OpError, OpNotImplemented, OpAttributeRequired, \
+                      OpAttributeInvalid, OpAttributeUnimplemented
 from .mxnet import from_mxnet
 from .onnx import from_onnx
 from .coreml import from_coreml
 from .keras import from_keras
 from .darknet import from_darknet
 from .tensorflow import from_tensorflow
 from .caffe2 import from_caffe2
-from .common import raise_not_supported, get_nnvm_op, required_attr, \
-                    warn_not_used, parse_tshape, parse_bool_str
-from tvm.error_handling import raise_attribute_required, \
-                               raise_attribute_invalid, \
-                               raise_operator_unimplemented, \
-                               raise_attribute_unimplemented, \
-                               warn_not_used
+from .common import get_nnvm_op, required_attr, parse_tshape, parse_bool_str

nnvm/python/nnvm/frontend/caffe2.py

@@ -73,7 +73,8 @@ def get_converter(cls):
 
         if hasattr(cls, '_impl'):
             return getattr(cls, '_impl')
-        raise_operator_unimplemented(cls.__name__)
+        raise OpNotImplemented(
+            'Operator {} is not implemented in frontend Caffe2.'.format(cls.__name__))
 
 
 _caffe2_internal_args = {
@@ -175,7 +176,7 @@ def _get_axis_from_order_str(order):
                 return 1
             if order == 'NHWC':
                 return 3
-            raise_attribute_invalid(order, 'storage order', 'concat')
+            raise OpAttributeInvalid('Value {} in attribute {} of operator {} is not valid.'.format(order, 'order', 'Concat'))
 
         return AttrCvt(
             op_name='concatenate',
@@ -425,7 +426,8 @@ def _convert_operator(self,
             # Add a sanitizing step to convert all byte strings in args to strings
             sym = convert_map[op_type](inputs, args, self._params)
         else:
-            raise_operator_unimplemented(op_type)
+            raise OpNotImplemented(
+                'Operator {} is not supported in frontend Caffe2.'.format(op_type))
         return sym
 
 

nnvm/python/nnvm/frontend/common.py

@@ -7,13 +7,15 @@
 def get_nnvm_op(op_name):
     op = getattr(_sym, op_name)
     if not op:
-        raise_operator_unimplemented(op_name)
+        raise OpNotImplemented(
+            'Operator {} is not supported.'.format(op))
     return op
 
 def required_attr(attr, key, op_name):
     assert isinstance(attr, dict)
     if key not in attr:
-        raise_attribute_required(key, op_name)
+        raise OpAttributeRequired(
+            'Required attribute {} not found in operator {}'.format(key, op_name))
     return attr[key]
 
 def parse_tshape(tshape):

nnvm/python/nnvm/frontend/coreml.py

@@ -83,7 +83,8 @@ def BatchnormLayerParams(op, insym, symtab):
     """Get layer of batchnorm parameter"""
     # this changes the symbol
     if op.instanceNormalization:
-        raise_operator_unimplemented('instance normalization')
+        msg = 'Operator "instance normalization" is not supported in frontend CoreML.'
+        raise OpNotImplemented(msg)
     else:
         params = {'gamma':symtab.new_const(list(op.gamma.floatValue)),
                   'beta':symtab.new_const(list(op.beta.floatValue)),
@@ -136,7 +137,8 @@ def ActivationParams(op, insym, symtab):
         betasym = symtab.new_const(beta)
         return _sym.broadcast_mul(_sym.log(_sym.broadcast_add(
             _sym.exp(insym), betasym)), alphasym)
-    raise_operator_unimplemented(whichActivation)
+    raise OpNotImplemented(
+        'Operator {} is not supported in frontend CoreML.'.format(whichActivation))
 
 def ScaleLayerParams(op, insym, symtab):
     """Scale layer params."""
@@ -158,7 +160,8 @@ def PoolingLayerParams(op, insym, symtab):
             return _sym.global_max_pool2d(insym)
         if op.type == 1:
             return _sym.global_avg_pool2d(insym)
-        raise_operator_unimplemented('pooling (not max or average)')
+        raise OpNotImplemented(
+            'Operator pooling (not max or average) is not supported in frontend CoreML.')
 
     else:
         params = {'pool_size':list(op.kernelSize),
@@ -178,8 +181,8 @@ def PoolingLayerParams(op, insym, symtab):
             params['padding'] = padding
             params['ceil_mode'] = True
         else:
-            raise_attribute_invalid(op.WhichOneof('PoolingPaddingType'),
-                                    'PoolingPaddingType', 'pooling')
+            msg = 'Value {} in attribute PoolingPaddingType of operator Pooling is not valid.'
+            raise OpAttributeInvalid(msg.format(op.WhichOneof('PoolingPaddingType')))
 
         # consume padding layer
         if symtab.in_padding:
@@ -191,7 +194,8 @@ def PoolingLayerParams(op, insym, symtab):
             return _sym.max_pool2d(insym, **params)
         if op.type == 1:
             return _sym.avg_pool2d(insym, **params)
-        raise_operator_unimplemented('pooling (not max or average)')
+        msg = 'Operator pooling (not max or average) is not supported in frontend CoreML.'
+        raise OpNotImplemented(msg)
 
 def SoftmaxLayerParams(op, insym, symtab):
     return _sym.softmax(_sym.flatten(insym))
@@ -230,7 +234,8 @@ def ConcatLayerParams(op, insyms, symtab):
     if not isinstance(insyms, list):
         insyms = [insyms]
     if op.sequenceConcat:
-        raise_operator_unimplemented('sequence concat')
+        raise OpNotImplemented(
+            'Operator Sequence Concat is not supported in frontend CoreML.')
     ret = _sym.concatenate(*insyms, axis=1)
     return ret
 
@@ -244,14 +249,16 @@ def PaddingLayerParams(op, insym, symtab):
     if op.WhichOneof('PaddingType') == 'constant':
         constant = op.constant
         if constant.value != 0:
-            raise_attribute_invalid(constant.value, 'padding value', 'padding')
+            msg = 'Value {} in attribute "padding value" of operator Padding is not valid.'
+            raise OpAttributeInvalid(msg.format(constant.value))
         padding = [b.startEdgeSize for b in op.paddingAmounts.borderAmounts]
         padding2 = [b.endEdgeSize for b in op.paddingAmounts.borderAmounts]
         for i, j in zip(padding, padding2):
             assert i == j
         symtab.set_padding(padding)
     else:
-        raise_operator_unimplemented('non-constant padding')
+        raise OpNotImplemented(
+            'Operator "non-constant padding" is not supported in frontend CoreML.')
     return insym
 
 def PermuteLayerParams(op, insym, symtab):
@@ -260,8 +267,8 @@ def PermuteLayerParams(op, insym, symtab):
 
 def UpsampleLayerParams(op, insym, symtab):
     if op.scalingFactor[0] != op.scalingFactor[1]:
-        raise_attribute_invalid(op.scalingFactor, 'scaling factors',
-                                'upsample')
+        raise OpAttributeInvalid(
+            'Height and width scaling factors of Upsample operator must be equal.')
     interpolationMode = 'NEAREST_NEIGHBOR' if op.mode == 0 else 'BILINEAR'
     return _sym.upsampling(insym, scale=op.scalingFactor[0], method=interpolationMode)
 
@@ -342,7 +349,8 @@ def coreml_op_to_nnvm(op, inname, outname, symtab):
     """
     classname = type(op).__name__
     if classname not in _convert_map:
-        raise_operator_unimplemented(classname)
+        raise OpNotImplemented(
+            'Operator {} is not supported in frontend CoreML.'.format(classname))
     if isinstance(inname, string_types):
         insym = symtab.get_var(inname)
     else:

nnvm/python/nnvm/frontend/darknet.py

@@ -61,7 +61,8 @@ def _darknet_maxpooling(inputs, attrs):
     """Process the max pool 2d operation."""
     kernel = parse_tshape(required_attr(attrs, 'kernel', 'maxpool'))
     if len(kernel) != 1:
-        raise_attribute_unimplemented('non-2d kernel', 'pool_2d')
+        raise OpAttributeUnimplemented(
+            'Non-2D kernels for Max Pooling are not supported in frontend Darknet.')
 
     op_name, new_attrs = 'max_pool2d', {}
     strides = int(attrs.get('stride', (1, 1)))
@@ -79,7 +80,8 @@ def _darknet_avgpooling(inputs, attrs):
     """Process the average pool 2d operation."""
     kernel = parse_tshape(required_attr(attrs, 'kernel', 'avgpool'))
     if len(kernel) != 1:
-        raise_attribute_unimplemented('non-2d kernel', 'pool_2d')
+        raise OpAttributeUnimplemented(
+            'Non-2D kernels for Average Pooling are not supported in frontend Darknet.')
 
     op_name, new_attrs = 'avg_pool2d', {}
     strides = int(attrs.get('stride', (1, 1)))
@@ -103,10 +105,12 @@ def _darknet_conv2d(inputs, attrs):
     """Process the convolution 2d operation."""
     kernel = parse_tshape(required_attr(attrs, 'kernel', 'conv2d'))
     if len(kernel) != 1:
-        raise_attribute_unimplemented('non 2d kernel', 'conv2d')
+        raise OpAttributeUnimplemented('Non-2D kernels for Conv2D are unsupported in frontend '
+                                       'Darknet.')
     layout = attrs.get('layout', 'NCHW')
     if layout not in ['NCHW', 'NHWC']:
-        raise_attribute_invalid(layout, 'layout', 'conv2d')
+        raise OpAttributeInvalid(
+            'Value {} in attribute "layout" of operator Conv2D is not valid.'.format(layout))
     strides = int(attrs.get('stride', (1, 1)))
     pads = int(attrs.get('pad', (0, 0)))
 
@@ -142,13 +146,16 @@ def _darknet_conv2d(inputs, attrs):
 def _darknet_conv2d_transpose(inputs, attrs):
     """Process the convolution 2d transpose operation."""
     if 'target_shape' in attrs:
-        raise_attribute_unimplemented('target_shape', 'conv2d_transpose')
+        raise OpAttributeUnimplemented(
+            'Attribute "target_shape" is not supported in operator Conv2D-transpose.')
     kernel = parse_tshape(required_attr(attrs, 'kernel', 'conv2d_transpose'))
     if len(kernel) != 2:
-        raise_attribute_unimplemented('non-2d kernel', 'conv2d_transpose')
+        raise OpAttributeUnimplemented(
+            'Non-2D kernels are not supported in operator Conv2D-transpose.')
     layout = attrs.get('layout', 'NCHW')
     if layout not in ['NCHW', 'NHWC']:
-        raise_attribute_invalid(layout, 'layout', 'conv2d_transpose')
+        msg = 'Value {} in attribute "layout" of operator Conv2D-transpose is not valid.'
+        raise OpAttributeInvalid(msg.format(layout))
     op_name, new_attrs = 'conv2d_transpose', {}
     new_attrs['channels'] = required_attr(attrs, 'num_filter', 'conv2d_transpose')
     new_attrs['kernel_size'] = kernel
@@ -222,7 +229,8 @@ def _darknet_dropout(inputs, attrs):
 def _darknet_reshape(inputs, attrs):
     """Process the reshape operation."""
     if parse_bool_str(attrs, 'reverse'):
-        raise_attribute_unimplemented('reverse', 'reshape')
+        raise OpAttributeUnimplemented(
+            'Attribute "reverse" is not supported in operator Reshape.')
     op_name, new_attrs = 'reshape', {}
     new_attrs['shape'] = required_attr(attrs, 'shape', 'reshape')
     return get_nnvm_op(op_name)(*inputs, **new_attrs), None
@@ -324,7 +332,8 @@ def _darknet_activations(inputs, attrs):
     elif ACTIVATION.ELU == act:
         act_type = 'elu'
     else:
-        raise_operator_unimplemented('act: ' + act)
+        raise OpNotImplemented(
+            'Operator act: {} is not supported in framework Darknet.'.format(act))
 
     if act_type in ['relu', 'tanh']:
         op_name, new_attrs = act_type, {}
@@ -339,7 +348,8 @@ def _darknet_activations(inputs, attrs):
         op_name, new_attrs = act_type, {}
         sym = get_nnvm_op(op_name)(*inputs, **new_attrs)
     else:
-        raise_operator_unimplemented('act_type: ' + act_type)
+        raise OpNotImplemented(
+            'Operator act: {} is not supported in framework Darknet.'.format(act))
     return sym, None
 
 def _darknet_op_not_support(inputs, attrs):
@@ -402,7 +412,8 @@ def _darknet_convert_symbol(op_name, inputs, attrs):
     if op_name in _DARKNET_CONVERT_MAP:
         sym, out_name = _DARKNET_CONVERT_MAP[op_name](inputs, attrs)
     else:
-        raise_operator_unimplemented(op_name)
+        raise OpNotImplemented(
+            'Operator {} is not supported in frontend Darknet.'.format(op_name))
     if out_name is  None:
         out_name = sym.list_output_names()[0].replace('_output', '')
     return out_name, sym
@@ -448,9 +459,10 @@ def _get_convolution_weights(self, layer, opname):
         if layer.nweights == 0:
             return
 
-        if (layer.n * layer.c * layer.size * layer.size) != layer.nweights:
-            raise_attribute_invalid(layer.n * layer.c * layer.size * layer.size,
-                                    'layer weights size', 'conv2d')
+        if layer.n * layer.c * layer.size * layer.size != layer.nweights:
+            msg = 'nweights ({}) != n * c * h * w ({}) in operator {}'
+            msg = msg.format(layer.nweights, layer.n * layer.c * layer.size ** 2, opname)
+            raise OpAttributeInvalid(msg)
 
         shape = (layer.n, layer.c, layer.size, layer.size)
         weights = self._read_memory_buffer(shape, layer.weights)
@@ -630,7 +642,8 @@ def _get_darknet_attrs(self, layer, layer_num):
             pass
 
         else:
-            raise_operator_unimplemented(layer.type)
+            raise OpNotImplemented(
+                'Operator {} is not supported in frontend Darknet.'.format(layer.type))
 
         return attr
 
@@ -763,7 +776,8 @@ def _handle_darknet_rnn_layers(self, layer_num, sym):
 
         elif LAYERTYPE.LSTM == layer.type:
             if layer.steps > 1:
-                raise_attribute_invalid(layer.steps, 'number of steps', 'RNN')
+                raise OpAttributeInvalid(
+                    'Number of steps {} of RNN is not valid.'.format(layer.steps))
 
             op_name_add = 'elemwise_add'
             op_name_mul = 'elemwise_mul'
@@ -829,7 +843,8 @@ def _handle_darknet_rnn_layers(self, layer_num, sym):
 
         elif LAYERTYPE.GRU == layer.type:
             if layer.steps > 1:
-                raise_attribute_invalid(layer.steps, 'number of steps', 'RNN')
+                raise OpAttributeInvalid(
+                    'Number of steps {} is not valid in RNN.'.format(layer.steps))
 
             op_name_add = 'elemwise_add'
             op_name_mul = 'elemwise_mul'