Only require numpy for the vectorized UDF path

src/python_udf.cpp

@@ -530,26 +530,26 @@ struct PythonUDFData {
 	ScalarFunction GetFunction(const nb::callable &udf, PythonExceptionHandling exception_handling, bool side_effects,
 	                           const ClientProperties &client_properties) {
 
-		// Import this module, because importing this from a non-main thread causes a segfault
-
-		auto &import_cache = *DuckDBPyConnection::ImportCache();
-		nb::handle core;
-		auto numpy = import_cache.numpy();
-		if (!numpy) {
-			throw InvalidInputException("'numpy' is required for this operation, but it wasn't installed");
-		}
-		// numpy.__version__ is a string; nb::cast<nb::tuple> rejects a non-tuple, so convert it explicitly.
-		nb::object numpy_version_str = numpy.attr("__version__");
-		auto numpy_version = nb::tuple(numpy_version_str);
-		if (NumpyDeprecatesAccessToCore(numpy_version)) {
-			core = numpy.attr("_core");
-		} else {
-			core = numpy.attr("core");
-		}
-		(void)core.attr("multiarray");
-
 		scalar_function_t func;
 		if (vectorized) {
+			// Only the vectorized (pyarrow) path needs numpy; import it here rather than before
+			// the branch. Importing off the main thread causes a segfault.
+			auto &import_cache = *DuckDBPyConnection::ImportCache();
+			nb::handle core;
+			auto numpy = import_cache.numpy();
+			if (!numpy) {
+				throw InvalidInputException("'numpy' is required for this operation, but it wasn't installed");
+			}
+			// numpy.__version__ is a string; nb::cast<nb::tuple> rejects a non-tuple, so convert it explicitly.
+			nb::object numpy_version_str = numpy.attr("__version__");
+			auto numpy_version = nb::tuple(numpy_version_str);
+			if (NumpyDeprecatesAccessToCore(numpy_version)) {
+				core = numpy.attr("_core");
+			} else {
+				core = numpy.attr("core");
+			}
+			(void)core.attr("multiarray");
+
 			func = CreateVectorizedFunction(udf.ptr(), exception_handling, null_handling);
 		} else {
 			func = CreateNativeFunction(udf.ptr(), exception_handling, client_properties, null_handling);

tests/fast/udf/test_scalar_arrow.py

@@ -1,3 +1,6 @@
+import subprocess
+import sys
+
 import pytest
 
 import duckdb
@@ -214,3 +217,21 @@ def func(data):
 
         res = duckdb_cursor.sql("select func(1).y").fetchone()
         assert res == ("this is not an inlined string",)
+
+    def test_arrow_udf_requires_numpy(self):
+        # Subprocess isolation, same reasoning as test_native_udf_without_numpy in
+        # test_scalar_native.py (numpy's import cache is a process-lifetime singleton).
+        script = """
+import sys
+sys.modules["numpy"] = None
+import duckdb
+from duckdb.sqltypes import VARCHAR
+
+try:
+    duckdb.create_function("arrow_no_numpy", lambda: "x", [], VARCHAR, type="arrow")
+    raise SystemExit("expected InvalidInputException, but create_function succeeded")
+except duckdb.InvalidInputException as e:
+    assert "numpy" in str(e), f"unexpected error message: {e}"
+"""
+        result = subprocess.run([sys.executable, "-c", script], capture_output=True, text=True)
+        assert result.returncode == 0, result.stderr

tests/fast/udf/test_scalar_native.py

@@ -1,3 +1,6 @@
+import subprocess
+import sys
+
 import pytest
 
 import duckdb
@@ -254,3 +257,23 @@ def example():
         (res,) = duckdb_cursor.sql("SELECT example()").fetchone()
         for key, val in res.items():
             assert key == val
+
+    def test_native_udf_without_numpy(self):
+        # Runs in a subprocess because the numpy import cache is a process-lifetime singleton -
+        # once something else in this interpreter has imported numpy, sys.modules trickery here
+        # can't make it forget that.
+        script = """
+import sys
+sys.modules["numpy"] = None
+import duckdb
+from duckdb.sqltypes import VARCHAR
+
+def generate_name():
+    return "hello-world"
+
+duckdb.create_function("native_no_numpy", generate_name, [], VARCHAR)
+res = duckdb.sql("select native_no_numpy()").fetchall()
+assert res == [("hello-world",)]
+"""
+        result = subprocess.run([sys.executable, "-c", script], capture_output=True, text=True)
+        assert result.returncode == 0, result.stderr