fix segmentation fault in process.cdist

CHANGELOG.md

@@ -1,5 +1,10 @@
 ## Changelog
 
+
+### [2.11.1] - 2022-10-05
+#### Fixed
+- Fix segmentation fault in `process.cdist` when used with an empty query sequence
+
 ### [2.11.0] - 2022-10-02
 #### Changes
 - move jarowinkler dependency into rapidfuzz to simplify maintenance

docs/conf.py

@@ -22,7 +22,7 @@ copyright = "2021, Max Bachmann"
 author = "Max Bachmann"
 
 # The full version, including alpha/beta/rc tags
-release = "2.11.0"
+release = "2.11.1"
 
 
 # -- General configuration ---------------------------------------------------

setup.py

@@ -13,7 +13,7 @@ with open("README.md", encoding="utf8") as f:
 
 setup_args = {
     "name": "rapidfuzz",
-    "version": "2.11.0",
+    "version": "2.11.1",
     "extras_require": {"full": ["numpy"]},
     "url": "https://github.com/maxbachmann/RapidFuzz",
     "author": "Max Bachmann",

src/rapidfuzz/__init__.py

@@ -3,6 +3,6 @@ rapid string matching library
 """
 __author__: str = "Max Bachmann"
 __license__: str = "MIT"
-__version__: str = "2.11.0"
+__version__: str = "2.11.1"
 
 from rapidfuzz import distance, fuzz, process, string_metric, utils

src/rapidfuzz/process_cpp.hpp

@@ -483,6 +483,9 @@ static Matrix cdist_two_lists_impl(const RF_ScorerFlags* scorer_flags, const RF_
     Matrix matrix(dtype, static_cast<size_t>(rows), static_cast<size_t>(cols));
     bool multiStringInit = scorer_flags->flags & RF_SCORER_FLAG_MULTI_STRING_INIT;
 
+    if (queries.empty() || choices.empty())
+        return matrix;
+
     if (multiStringInit) {
         std::vector<size_t> row_idx(rows);
         std::iota(row_idx.begin(), row_idx.end(), 0);

tests/test_process.py

@@ -338,5 +338,14 @@ def test_extractOne_use_first_match(scorer):
     )
 
 
+@pytest.mark.parametrize("scorer", [fuzz.ratio, fuzz.WRatio, custom_scorer])
+def test_cdist_empty_seq(scorer):
+    pytest.importorskip("numpy")
+    assert process_cpp.cdist([], ["a", "b"], scorer=scorer).shape == (0, 2)
+    assert process_cpp.cdist(["a", "b"], [], scorer=scorer).shape == (2, 0)
+    assert process_py.cdist([], ["a", "b"], scorer=scorer).shape == (0, 2)
+    assert process_py.cdist(["a", "b"], [], scorer=scorer).shape == (2, 0)
+
+
 if __name__ == "__main__":
     unittest.main()