BUG28280321: Fix segmentation fault when using unicode characters in tables

johannes · nmariz · commit bb054a9a34d8 · 2018-11-28T16:36:03.000Z
With Python 2.7 and use_pure=False, a segmentation fault is raised when using
unicode characters in table objects operations.

This patch fixes this issue by handling unicode string casts in Python 2.

Tests were added for both tables and collections.
diff --git a/src/mysqlxpb/mysqlxpb.cc b/src/mysqlxpb/mysqlxpb.cc
@@ -423,8 +423,12 @@ static void AddPyListToMessageRepeatedMessage(
   if (list_size > 0) {
     mutable_field->Reserve(list_size);
     for (Py_ssize_t idx = 0; idx < list_size; ++idx) {
-      mutable_field->AddAllocated(
-        CreateMessage(PyList_GetItem(list, idx), factory));
+      google::protobuf::Message* msg = CreateMessage(PyList_GetItem(list, idx), factory);
+      if (!msg) {
+        // CreateMessage already reported an error, we can leave quietly
+        return;
+      }
+      mutable_field->AddAllocated(msg);
     }
   }
 }
diff --git a/src/mysqlxpb/python_cast.h b/src/mysqlxpb/python_cast.h
@@ -133,6 +133,15 @@ std::string python_cast<std::string>(PyObject* obj) {
   } else if (PyBytes_CheckExact(obj)) {
     return std::string(PyBytes_AsString(obj), PyBytes_Size(obj));
   }
+#else
+  else if (PyUnicode_CheckExact(obj)) {
+    PyObject* pystring = PyUnicode_AsUTF8String(obj);
+    if (pystring) {
+      std::string result(PyString_AsString(pystring), PyString_Size(pystring));
+      Py_DECREF(pystring);
+      return result;
+    }
+  }
 #endif
 
   THROW_BAD_PYTHON_CAST(std::string);
diff --git a/tests/test_mysqlx_crud.py b/tests/test_mysqlx_crud.py
@@ -599,6 +599,11 @@ def test_add(self):
         result = collection.find().execute()
         self.assertEqual(7, len(result.fetch_all()))
 
+        # test unicode
+        result = collection.add({"age": 1, "name": u"😀"}).execute()
+        self.assertEqual(result.get_affected_items_count(), 1)
+        self.assertEqual(8, collection.count())
+
         if tests.MYSQL_VERSION > (8, 0, 4):
             # Following test are only possible on servers with id generetion.
             # Ensure _id is created at the server side
@@ -2110,6 +2115,12 @@ def test_insert(self):
         rows = result.fetch_all()
         self.assertEqual(6, len(rows))
 
+        # test unicode
+        table.insert("age", "name").values(1, u"😀").execute()
+        result = table.select().execute()
+        rows = result.fetch_all()
+        self.assertEqual(7, len(rows))
+
         drop_table(self.schema, "test")
 
     def test_update(self):

Original file line number	Diff line number	Diff line change
`@@ -423,8 +423,12 @@ static void AddPyListToMessageRepeatedMessage(`
`423`	`423`	`if (list_size > 0) {`
`424`	`424`	`mutable_field->Reserve(list_size);`
`425`	`425`	`for (Py_ssize_t idx = 0; idx < list_size; ++idx) {`
`426`		`- mutable_field->AddAllocated(`
`427`		`- CreateMessage(PyList_GetItem(list, idx), factory));`
	`426`	`+ google::protobuf::Message* msg = CreateMessage(PyList_GetItem(list, idx), factory);`
	`427`	`+ if (!msg) {`
	`428`	`+ // CreateMessage already reported an error, we can leave quietly`
	`429`	`+ return;`
	`430`	`+ }`
	`431`	`+ mutable_field->AddAllocated(msg);`
`428`	`432`	`}`
`429`	`433`	`}`
`430`	`434`	`}`