Fix crash when dealing with surrogate pairs

This addresses the "core dump" part of issue #81, but it's not yet clear what should
happen: simplejson handles them without problems, while ujson rejects them... Given that
AFAICT rapidjson cannot decode them (it raises `kParseErrorStringUnicodeSurrogateInvalid`)
it seems reasonable to reject them at encode time as well.

rapidjson.cpp

@@ -1675,6 +1675,8 @@ dumps_internal(
         else if (PyUnicode_Check(object)) {
             Py_ssize_t l;
             char* s = PyUnicode_AsUTF8AndSize(object, &l);
+            if (s == NULL)
+                return NULL;
             writer->String(s, l);
         }
         else if (PyList_Check(object)) {
@@ -1712,6 +1714,8 @@ dumps_internal(
                     if (PyUnicode_Check(key)) {
                         Py_ssize_t l;
                         char* key_str = PyUnicode_AsUTF8AndSize(key, &l);
+                        if (key_str == NULL)
+                            goto error;
                         stack.push_back(WriterContext(NULL, 0, item, false, nextLevel));
                         stack.push_back(WriterContext(key_str, l, NULL, false, nextLevel));
                     }
@@ -1728,6 +1732,8 @@ dumps_internal(
                     if (PyUnicode_Check(key)) {
                         Py_ssize_t l;
                         char* key_str = PyUnicode_AsUTF8AndSize(key, &l);
+                        if (key_str == NULL)
+                            goto error;
                         items.push_back(DictItem(key_str, l, item));
                     }
                     else if (!skipKeys) {

tests/test_unicode.py

@@ -24,3 +24,25 @@ def test_unicode(u, dumps, loads):
     assert ju == js
     assert ju.lower() == json.dumps(u).lower()
     assert dumps(u, ensure_ascii=False) == json.dumps(u, ensure_ascii=False)
+
+
+@pytest.mark.unit
+@pytest.mark.parametrize('o', [
+    "\ud80d",
+    {"foo": "\ud80d"},
+    {"\ud80d": "foo"},
+])
+def test_dump_surrogate(o, dumps):
+    with pytest.raises(UnicodeEncodeError, match="surrogates not allowed"):
+        dumps(o)
+
+
+@pytest.mark.unit
+@pytest.mark.parametrize('j', [
+    '"\\ud80d"',
+    '{"foo": "\\ud80d"}',
+    '{"\\ud80d": "foo"}',
+])
+def test_load_surrogate(j, loads):
+    with pytest.raises(ValueError, match="surrogate pair in string is invalid"):
+        loads(j)