Merge pull request #59 from maxmind/greg/fix-segfaults

Fix segfault on invalid UTF-8. Closes #58.

HISTORY.rst

@@ -3,6 +3,12 @@
 History
 -------
 
+1.5.3 (2020-05-04)
+++++++++++++++++++
+
+* Fix a segfault when decoding a database with a corrupt data section.
+  Reported by Robert Scott. GitHub #58.
+
 1.5.2 (2019-12-20)
 ++++++++++++++++++
 

extension/maxminddb.c

@@ -204,6 +204,11 @@ static int get_record(PyObject *self, PyObject *args, PyObject **record) {
     *record = from_entry_data_list(&entry_data_list);
     MMDB_free_entry_data_list(original_entry_data_list);
 
+    // from_entry_data_list will return NULL on errors.
+    if (*record == NULL) {
+        return -1;
+    }
+
     return prefix_len;
 }
 
@@ -526,6 +531,11 @@ static PyObject *from_map(MMDB_entry_data_list_s **entry_data_list) {
         PyObject *key = PyUnicode_FromStringAndSize(
             (char *)(*entry_data_list)->entry_data.utf8_string,
             (*entry_data_list)->entry_data.data_size);
+        if (!key) {
+            // PyUnicode_FromStringAndSize will set an appropriate exception
+            // in this case.
+            return NULL;
+        }
 
         *entry_data_list = (*entry_data_list)->next;
 

tests/reader_test.py

@@ -286,6 +286,15 @@ def test_nondatabase(self):
         ):
             open_database("README.rst", self.mode)
 
+    # This is from https://github.com/maxmind/MaxMind-DB-Reader-python/issues/58
+    def test_database_with_invalid_utf8_key(self):
+        reader = open_database(
+            "tests/data/bad-data/maxminddb-python/bad-unicode-in-map-key.mmdb",
+            self.mode,
+        )
+        with self.assertRaises(UnicodeDecodeError):
+            reader.get_with_prefix_len("163.254.149.39")
+
     def test_too_many_constructor_args(self):
         with self.assertRaises(TypeError):
             self.readerClass[0]("README.md", self.mode, 1)