Avoid break string inside unicode char when formatting (#767)

tespent · web-flow · commit ad727ff3e1ea · 2025-07-13T17:55:49.000-07:00
diff --git a/src/python_data_access.rs b/src/python_data_access.rs
@@ -409,8 +409,8 @@ where
         let value = copy_string(addr as *const I::StringObject, process)?
             .replace('\'', "\\\"")
             .replace('\n', "\\n");
-        if value.len() as isize >= max_length - 5 {
-            format!("\"{}...\"", &value[..(max_length - 5) as usize])
+        if let Some((offset, _)) = value.char_indices().nth((max_length - 5) as usize) {
+            format!("\"{}...\"", &value[..offset])
         } else {
             format!("\"{}\"", value)
         }
diff --git a/tests/integration_test.rs b/tests/integration_test.rs
@@ -280,7 +280,7 @@ fn test_local_vars() {
     let frame = &trace.frames[0];
     let locals = frame.locals.as_ref().unwrap();
 
-    assert_eq!(locals.len(), 27);
+    assert_eq!(locals.len(), 28);
 
     let arg1 = &locals[0];
     assert_eq!(arg1.name, "arg1");
@@ -415,6 +415,13 @@ fn test_local_vars() {
     assert_eq!(local24.name, "local24");
     test_repr_prefix(local24, "<numpy.clongdouble at");
 
+    // https://github.com/benfred/py-spy/issues/766
+    let local25 = &locals[27];
+    assert_eq!(local25.name, "local25");
+    let unicode_val = local25.repr.as_ref().unwrap();
+    let end = unicode_val.char_indices().map(|(i, _)| i).nth(4).unwrap();
+    assert_eq!(unicode_val[0..end], *"\"测试1");
+
     // we only support dictionary lookup on python 3.6+ right now
     if runner.spy.version.major == 3 && runner.spy.version.minor >= 6 {
         assert_eq!(
diff --git a/tests/scripts/local_vars.py b/tests/scripts/local_vars.py
@@ -39,6 +39,8 @@ def local_variable_lookup(arg1="foo", arg2=None, arg3=True):
     local23 = np.complex128(0.3+5j)
     local24 = np.clongdouble(0.3+5j)
 
+    # https://github.com/benfred/py-spy/issues/766
+    local25 = "测试1" * 500
 
     time.sleep(100000)
 
