fix u style unicode strings in python (#110)

crates/jiter-python/tests/test_jiter.py

@@ -1,3 +1,4 @@
+import json
 from decimal import Decimal
 
 import jiter
@@ -239,3 +240,17 @@ def test_lossless_floats_int():
     v = jiter.from_json(b'123', lossless_floats=True)
     assert isinstance(v, int)
     assert v == 123
+
+
+def test_unicode_roundtrip():
+    original = ['中文']
+    json_data = json.dumps(original).encode()
+    assert jiter.from_json(json_data) == original
+    assert json.loads(json_data) == original
+
+
+def test_unicode_roundtrip_ensure_ascii():
+    original = {'name': '中文'}
+    json_data = json.dumps(original, ensure_ascii=False).encode()
+    assert jiter.from_json(json_data, cache_mode=False) == original
+    assert json.loads(json_data) == original

crates/jiter/src/string_decoder.rs

@@ -118,6 +118,7 @@ fn decode_to_tape<'t, 'j>(
                 b't' => tape.push(b'\t'),
                 b'u' => {
                     let (c, new_index) = parse_escape(data, index)?;
+                    ascii_only = false;
                     index = new_index;
                     tape.extend_from_slice(c.encode_utf8(&mut [0_u8; 4]).as_bytes());
                 }

crates/jiter/tests/main.rs

@@ -1594,3 +1594,16 @@ fn jiter_partial_string() {
         JiterErrorType::JsonError(JsonErrorType::EofWhileParsingList)
     );
 }
+
+#[test]
+fn test_unicode_roundtrip() {
+    // '"中文"'
+    let json_bytes = b"\"\\u4e2d\\u6587\"";
+    let value = JsonValue::parse(json_bytes, false).unwrap();
+    let cow = match value {
+        JsonValue::Str(s) => s,
+        _ => panic!("expected string"),
+    };
+    assert_eq!(cow, "中文");
+    assert!(matches!(cow, Cow::Owned(_)));
+}