Merge pull request #94 from novas0x2a/unicode-input

Unicode input

capnp/lib/capnp.pyx

@@ -679,7 +679,7 @@ cdef _setBytes(_DynamicSetterClasses thisptr, field, value):
     thisptr.set(field, temp)
 
 cdef _setBaseString(_DynamicSetterClasses thisptr, field, value):
-    encoded_value = value.encode()
+    encoded_value = value.encode('utf-8')
     cdef capnp.StringPtr temp_string = capnp.StringPtr(<char*>encoded_value, len(encoded_value))
     cdef C_DynamicValue.Reader temp = C_DynamicValue.Reader(temp_string)
     thisptr.set(field, temp)
@@ -690,7 +690,7 @@ cdef _setBytesField(DynamicStruct_Builder thisptr, _StructSchemaField field, val
     thisptr.setByField(field.thisptr, temp)
 
 cdef _setBaseStringField(DynamicStruct_Builder thisptr, _StructSchemaField field, value):
-    encoded_value = value.encode()
+    encoded_value = value.encode('utf-8')
     cdef capnp.StringPtr temp_string = capnp.StringPtr(<char*>encoded_value, len(encoded_value))
     cdef C_DynamicValue.Reader temp = C_DynamicValue.Reader(temp_string)
     thisptr.setByField(field.thisptr, temp)
@@ -1025,7 +1025,7 @@ cdef class _DynamicStructReader:
         try:
             return self._get(field)
         except KjException as e:
-            raise e._to_python()
+            raise e._to_python(), None, _sys.exc_info()[2]
 
     cpdef _get_by_field(self, _StructSchemaField field):
         return to_python_reader(self.thisptr.getByField(field.thisptr), self._parent)
@@ -1225,7 +1225,7 @@ cdef class _DynamicStructBuilder:
         try:
             return self._get(field)
         except KjException as e:
-            raise e._to_python()
+            raise e._to_python(), None, _sys.exc_info()[2]
 
     cpdef _set(self, field, value):
         _setDynamicField(self.thisptr, field, value, self._parent)
@@ -1237,7 +1237,7 @@ cdef class _DynamicStructBuilder:
         try:
             self._set(field, value)
         except KjException as e:
-            raise e._to_python()
+            raise e._to_python(), None, _sys.exc_info()[2]
 
     cpdef _has(self, field):
         return self.thisptr.has(field)
@@ -1456,7 +1456,7 @@ cdef class _DynamicStructPipeline:
         try:
             return self._get(field)
         except KjException as e:
-            raise e._to_python()
+            raise e._to_python(), None, _sys.exc_info()[2]
 
     property schema:
         """A property that returns the _StructSchema object matching this reader"""
@@ -1936,7 +1936,7 @@ cdef class _RemotePromise:
         try:
             return self._get(field)
         except KjException as e:
-            raise e._to_python()
+            raise e._to_python(), None, _sys.exc_info()[2]
 
     property schema:
         """A property that returns the _StructSchema object matching this reader"""
@@ -2039,7 +2039,7 @@ cdef class _DynamicCapabilityServer:
         try:
             return getattr(self.server, field)
         except KjException as e:
-            raise e._to_python()
+            raise e._to_python(), None, _sys.exc_info()[2]
 
 cdef class _DynamicCapabilityClient:
     cdef C_DynamicCapability.Client thisptr
@@ -2121,7 +2121,7 @@ cdef class _DynamicCapabilityClient:
                 raise AttributeError('Method named %s not found' % name)
             return _partial(self._send, name)
         except KjException as e:
-            raise e._to_python()
+            raise e._to_python(), None, _sys.exc_info()[2]
 
     cpdef upcast(self, schema) except +reraise_kj_exception:
         cdef _InterfaceSchema s

capnp/templates/module.pyx

@@ -90,7 +90,7 @@ cpdef _set_{{field.name}}(self, value):
     if type(value) is bytes:
         temp_string = StringPtr(<char*>value, len(value))
     else:
-        encoded_value = value.encode()
+        encoded_value = value.encode('utf-8')
         temp_string = StringPtr(<char*>encoded_value, len(encoded_value))
     self.thisptr_child.set{{field.c_name}}(temp_string)
     {% elif 'data' == field['type'] -%}
@@ -99,7 +99,7 @@ cpdef _set_{{field.name}}(self, value):
     if type(value) is bytes:
         temp_string = StringPtr(<char*>value, len(value))
     else:
-        encoded_value = value.encode()
+        encoded_value = value.encode('utf-8')
         temp_string = StringPtr(<char*>encoded_value, len(encoded_value))
     self.thisptr_child.set{{field.c_name}}(ArrayPtr[byte](<byte *>temp_string.begin(), temp_string.size()))
     {% else -%}
@@ -133,40 +133,6 @@ cdef _from_list(_DynamicListBuilder msg, list d):
         msg._set(count, val)
         count += 1
 
-cdef DynamicValue.Reader to_dynamic_value(value):
-    cdef DynamicValue.Reader temp
-    cdef StringPtr temp_string
-    value_type = type(value)
-
-    if value_type is int or value_type is long:
-        if value < 0:
-           temp = DynamicValue.Reader(<long long>value)
-        else:
-           temp = DynamicValue.Reader(<unsigned long long>value)
-    elif value_type is float:
-        temp = DynamicValue.Reader(<double>value)
-    elif value_type is bool:
-        temp = DynamicValue.Reader(<cbool>value)
-    elif value_type is bytes:
-        temp_string = StringPtr(<char*>value, len(value))
-        temp = DynamicValue.Reader(temp_string)
-    elif isinstance(value, basestring):
-        encoded_value = value.encode()
-        temp_string = StringPtr(<char*>encoded_value, len(encoded_value))
-        temp = DynamicValue.Reader(temp_string)
-    elif value is None:
-        temp = DynamicValue.Reader(VOID)
-    elif value_type is _DynamicStructBuilder:
-        temp = _extract_dynamic_struct_builder(value)
-    elif value_type is _DynamicStructReader:
-        temp = _extract_dynamic_struct_reader(value)
-    elif value_type is _DynamicEnum:
-        temp = _extract_dynamic_enum(value)
-    else:
-        raise ValueError("Tried to convert value of: '{}' which is an unsupported type: '{}'".format(str(value), str(type(value))))
-
-    return temp
-
 
 cdef extern from "{{file.filename}}.h":
     {%- for node in code.nodes %}

test/all-types.txt

@@ -25,7 +25,7 @@
     uInt64Field = 345678901234567890,
     float32Field = -1.25e-10,
     float64Field = 345,
-    textField = "baz",
+    textField = "\xe2\x98\x83",
     dataField = "qux",
     structField = (
       voidField = void,

test/test_regression.py

@@ -1,10 +1,19 @@
+ # -*- coding: utf-8 -*-
+
 import pytest
 import capnp
 import os
 import math
+import sys
 
 this_dir = os.path.dirname(__file__)
 
+if sys.version_info[0] < 3:
+    EXPECT_BYTES = True
+else:
+    EXPECT_BYTES = False
+
+
 @pytest.fixture
 def addressbook():
      return capnp.load(os.path.join(this_dir, 'addressbook.capnp'))
@@ -300,7 +309,7 @@ def init_all_types(builder):
     subBuilder.uInt64Field = 345678901234567890
     subBuilder.float32Field = -1.25e-10
     subBuilder.float64Field = 345
-    subBuilder.textField = "baz"
+    subBuilder.textField = b"\xe2\x98\x83".decode('utf-8') # This is u"☃", but py3.2 doesn't support u
     subBuilder.dataField = b"qux"
     subSubBuilder = subBuilder.structField
     subSubBuilder.textField = "nested"
@@ -390,7 +399,15 @@ def check_all_types(reader):
     assert subReader.uInt64Field == 345678901234567890
     assert_almost(subReader.float32Field, -1.25e-10)
     assert subReader.float64Field == 345
-    assert subReader.textField == "baz"
+
+    assert subReader.textField == "☃"
+    # This assertion highlights the encoding we expect to see here, since
+    # otherwise this appears a bit magical...
+    if EXPECT_BYTES:
+        assert len(subReader.textField) == 3
+    else:
+        assert len(subReader.textField) == 1
+
     assert subReader.dataField == b"qux"
 
     subSubReader = subReader.structField