Enforce valid UTF-8 encoding for sqlite collation names (#6015)

ever0de · web-flow · commit 57bdf35ee685 · 2025-07-21T23:42:47.000+09:00
* Make public `PyStr::ensure_valid_utf8`

* Enforce valid UTF-8 encoding for sqlite collation names
diff --git a/Lib/test/test_sqlite3/test_regression.py b/Lib/test/test_sqlite3/test_regression.py
@@ -263,8 +263,6 @@ def test_connection_call(self):
         """
         self.assertRaises(TypeError, self.con, b"select 1")
 
-    # TODO: RUSTPYTHON
-    @unittest.expectedFailure
     def test_collation(self):
         def collation_cb(a, b):
             return 1
diff --git a/stdlib/src/sqlite.rs b/stdlib/src/sqlite.rs
@@ -1163,6 +1163,7 @@ mod _sqlite {
             callable: PyObjectRef,
             vm: &VirtualMachine,
         ) -> PyResult<()> {
+            name.ensure_valid_utf8(vm)?;
             let name = name.to_cstring(vm)?;
             let db = self.db_lock(vm)?;
             let Some(data) = CallbackData::new(callable.clone(), vm) else {
diff --git a/vm/src/builtins/str.rs b/vm/src/builtins/str.rs
@@ -457,7 +457,7 @@ impl PyStr {
         self.data.as_str()
     }
 
-    fn ensure_valid_utf8(&self, vm: &VirtualMachine) -> PyResult<()> {
+    pub fn ensure_valid_utf8(&self, vm: &VirtualMachine) -> PyResult<()> {
         if self.is_utf8() {
             Ok(())
         } else {

Original file line number	Diff line number	Diff line change
`@@ -457,7 +457,7 @@ impl PyStr {`
`457`	`457`	`self.data.as_str()`
`458`	`458`	`}`
`459`	`459`
`460`		`- fn ensure_valid_utf8(&self, vm: &VirtualMachine) -> PyResult<()> {`
	`460`	`+ pub fn ensure_valid_utf8(&self, vm: &VirtualMachine) -> PyResult<()> {`
`461`	`461`	`if self.is_utf8() {`
`462`	`462`	`Ok(())`
`463`	`463`	`} else {`