fix a bug of `str`

src/bindings/py_str.c

@@ -228,7 +228,8 @@ static bool str__getitem__(int argc, py_Ref argv) {
     py_Ref _1 = py_arg(1);
     if(_1->type == tp_int) {
         int index = py_toint(py_arg(1));
-        if(!pk__normalize_index(&index, self.size)) return false;
+        int u8_len = c11_sv__u8_length(self);
+        if(!pk__normalize_index(&index, u8_len)) return false;
         c11_sv res = c11_sv__u8_getitem(self, index);
         py_newstrv(py_retval(), res);
         return true;

src/common/str.c

@@ -104,8 +104,10 @@ c11_sv c11_sv__slice(c11_sv sv, int start) { return c11_sv__slice2(sv, start, sv
 
 c11_sv c11_sv__slice2(c11_sv sv, int start, int stop) {
     if(start < 0) start = 0;
-    if(stop < start) stop = start;
+    if(start > sv.size) start = sv.size;
+    if(stop < 0) stop = 0;
     if(stop > sv.size) stop = sv.size;
+    if(stop < start) stop = start;
     return (c11_sv){sv.data + start, stop - start};
 }
 
@@ -997,7 +999,6 @@ const static c11_u32_range kLoRanges[] = {
 // clang-format on
 
 bool c11__is_unicode_Lo_char(int c) {
-    if(c == 0x1f955) return true;
     const char* data =
         c11__search_u32_ranges(c, kLoRanges, sizeof(kLoRanges) / sizeof(c11_u32_range));
     return data != NULL;

src/modules/pickle.c

@@ -501,10 +501,10 @@ bool py_pickle_loads_body(const unsigned char* p, int memo_length, c11_smallmap_
 bool py_pickle_loads(const unsigned char* data, int size) {
     const unsigned char* p = data;
 
-    // \xf0\x9f\xa5\x95
-    if(size < 4 || p[0] != 240 || p[1] != 159 || p[2] != 165 || p[3] != 149)
+    // PK
+    if(size < 2 || p[0] != 'P' || p[1] != 'K')
         return ValueError("invalid pickle data");
-    p += 4;
+    p += 2;
 
     c11_smallmap_d2d type_mapping;
     c11_smallmap_d2d__ctor(&type_mapping);
@@ -780,7 +780,7 @@ bool py_pickle_loads_body(const unsigned char* p, int memo_length, c11_smallmap_
 static bool PickleObject__py_submit(PickleObject* self, py_OutRef out) {
     c11_sbuf cleartext;
     c11_sbuf__ctor(&cleartext);
-    c11_sbuf__write_cstr(&cleartext, "\xf0\x9f\xa5\x95");
+    c11_sbuf__write_cstr(&cleartext, "PK");
     // line 1: type mapping
     for(py_Type type = 0; type < self->used_types_length; type++) {
         if(self->used_types[type]) {

tests/041_str.py

@@ -221,6 +221,11 @@ assert chr(0x1f955) == '🥕'
 assert ord('测') == 27979
 assert chr(27979) == '测'
 
+assert '测试'[0] == '测'
+assert '测试'[1] == '试'
+assert '测试'[-1] == '试'
+assert '测试'[-2] == '测'
+
 # test format()
 assert "Hello, {}!".format("World") == "Hello, World!"
 assert "{} {} {}".format("I", "love", "Python") == "I love Python"