Support empty string values in NumpyProtocolHandler

Mark Florisson · Mark Florisson · commit 6d1af98a6488 · 2016-04-27T12:06:45.000+01:00
diff --git a/cassandra/deserializers.pyx b/cassandra/deserializers.pyx
@@ -44,13 +44,17 @@ cdef class Deserializer:
 
 cdef class DesBytesType(Deserializer):
     cdef deserialize(self, Buffer *buf, int protocol_version):
+        if buf.size == 0:
+            return ""
         return to_bytes(buf)
 
 # this is to facilitate cqlsh integration, which requires bytearrays for BytesType
 # It is switched in by simply overwriting DesBytesType:
 # deserializers.DesBytesType = deserializers.DesBytesTypeByteArray
 cdef class DesBytesTypeByteArray(Deserializer):
     cdef deserialize(self, Buffer *buf, int protocol_version):
+        if buf.size == 0:
+            return bytearray()
         return bytearray(buf.ptr[:buf.size])
 
 # TODO: Use libmpdec: http://www.bytereef.org/mpdecimal/index.html
@@ -84,6 +88,8 @@ cdef class DesByteType(Deserializer):
 
 cdef class DesAsciiType(Deserializer):
     cdef deserialize(self, Buffer *buf, int protocol_version):
+        if buf.size == 0:
+            return ""
         if PY2:
             return to_bytes(buf)
         return to_bytes(buf).decode('ascii')
@@ -169,6 +175,8 @@ cdef class DesTimeType(Deserializer):
 
 cdef class DesUTF8Type(Deserializer):
     cdef deserialize(self, Buffer *buf, int protocol_version):
+        if buf.size == 0:
+            return ""
         cdef val = to_bytes(buf)
         return val.decode('utf8')
 
diff --git a/cassandra/numpy_parser.pyx b/cassandra/numpy_parser.pyx
@@ -52,12 +52,14 @@ ctypedef struct ArrDesc:
     Py_uintptr_t buf_ptr
     int stride # should be large enough as we allocate contiguous arrays
     int is_object
+    int empty_binary_ok
 
 arrDescDtype = np.dtype(
     [ ('buf_ptr', np.uintp)
     , ('stride', np.dtype('i'))
     , ('is_object', np.dtype('i'))
-    ])
+    , ('empty_binary_ok', np.dtype('i'))
+    ], align=True)
 
 _cqltype_to_numpy = {
     cqltypes.LongType:          np.dtype('>i8'),
@@ -117,6 +119,7 @@ def make_arrays(ParseDesc desc, array_size):
         array_descs[i]['buf_ptr'] = arr.ctypes.data
         array_descs[i]['stride'] = arr.strides[0]
         array_descs[i]['is_object'] = coltype not in _cqltype_to_numpy
+        array_descs[i]['empty_binary_ok'] = coltype.empty_binary_ok
         arrays.append(arr)
 
     return array_descs, arrays
@@ -145,8 +148,9 @@ cdef inline int unpack_row(
         get_buf(reader, &buf)
         arr = arrays[i]
 
-        if buf.size == 0:
+        if buf.size < 0 or (buf.size == 0 and not arr.empty_binary_ok):
             raise ValueError("Cannot handle NULL value")
+
         if arr.is_object:
             deserializer = desc.deserializers[i]
             val = from_binary(deserializer, &buf, desc.protocol_version)
diff --git a/tests/integration/standard/utils.py b/tests/integration/standard/utils.py
@@ -45,7 +45,11 @@ def get_all_primitive_params(key):
     """
     params = [key]
     for datatype in PRIMITIVE_DATATYPES:
-        params.append(get_sample(datatype))
+        # Also test for empty strings
+        if key == 1 and datatype == 'ascii':
+            params.append('')
+        else:
+            params.append(get_sample(datatype))
     return params
 
 
