improve lastof

3 files changed, 39 insertions, 25 deletions

diff --git a/core/src/xmake/utf8/lastof.c b/core/src/xmake/utf8/lastof.c
index 8e22b16ba..165da3b69 100644
--- a/core/src/xmake/utf8/lastof.c
+++ b/core/src/xmake/utf8/lastof.c
@@ -36,42 +36,30 @@ tb_int_t xm_utf8_lastof(lua_State *lua) {
     size_t sublen;
     tb_char_t const* sub = luaL_checklstring(lua, 2, &sublen);
     tb_int_t plain = lua_toboolean(lua, 3);
+    tb_long_t byte_pos = 0;
 
     if (plain) {
-        tb_long_t char_pos = xm_utf8_lastof_impl(s, len, sub, sublen);
-        if (char_pos > 0) {
-            lua_pushinteger(lua, char_pos);
-        } else {
-            lua_pushnil(lua);
-        }
-        return 1;
+        byte_pos = xm_utf8_lastof_impl(s, len, sub, sublen);
     } else {
         lua_getglobal(lua, "string");
         lua_getfield(lua, -1, "lastof");
         lua_pushvalue(lua, 1); // s
         lua_pushvalue(lua, 2); // pattern
         lua_pushboolean(lua, 0); // plain = false
-
         lua_call(lua, 3, 1);
-
-        // Stack: [args, string_table, result]
-        if (lua_isnil(lua, -1)) {
-            return 1;
+        if (!lua_isnil(lua, -1)) {
+            byte_pos = (tb_long_t)lua_tointeger(lua, -1);
         }
+    }
 
-        lua_Integer byte_pos = lua_tointeger(lua, -1);
-        if (byte_pos > 0) {
-            tb_long_t count = xm_utf8_len_impl(s, len, 1, byte_pos - 1, tb_true, tb_null);
-            if (count >= 0) {
-                if (xm_utf8_iscont(s[byte_pos - 1])) {
-                    lua_pushinteger(lua, count);
-                } else {
-                    lua_pushinteger(lua, count + 1);
-                }
-                return 1;
-            }
+    if (byte_pos > 0) {
+        tb_long_t char_pos = xm_utf8_charpos(s, len, byte_pos);
+        if (char_pos > 0) {
+            lua_pushinteger(lua, char_pos);
+            return 1;
         }
-        lua_pushnil(lua);
-        return 1;
     }
+
+    lua_pushnil(lua);
+    return 1;
 }
diff --git a/core/src/xmake/utf8/utf8.c b/core/src/xmake/utf8/utf8.c
index 3b23532c5..3748e0975 100644
--- a/core/src/xmake/utf8/utf8.c
+++ b/core/src/xmake/utf8/utf8.c
@@ -105,6 +105,27 @@ tb_size_t xm_utf8_encode(tb_char_t* s, xm_utf8_int_t val) {
     return 0;
 }
 
+tb_long_t xm_utf8_charpos(tb_char_t const* s, tb_size_t len, tb_long_t byte_pos) {
+    if (byte_pos <= 0) return 0;
+    if (byte_pos > len + 1) byte_pos = len + 1;
+
+    // adjust byte_pos to the start of the character
+    // 
+    // performance: 
+    // 0(1) complexity, because utf8 sequence is max 4 bytes
+    while (byte_pos > 1 && xm_utf8_iscont(s[byte_pos - 1])) {
+        byte_pos--;
+    }
+    
+    // get character position
+    tb_long_t count = xm_utf8_len_impl(s, len, 1, byte_pos - 1, tb_true, tb_null);
+    return count >= 0? count + 1 : -1;
+}
+
+/* //////////////////////////////////////////////////////////////////////////////////////
+ * implementation interfaces
+ */
+
 tb_long_t xm_utf8_len_impl(tb_char_t const* s, tb_size_t len, tb_long_t posi, tb_long_t posj, tb_bool_t strict, tb_size_t* errpos) {
     tb_assert_and_check_return_val(s, -1);
 
@@ -313,3 +334,4 @@ tb_char_t const* xm_utf8_sub_impl(tb_char_t const* s, tb_size_t len, tb_long_t i
     *psublen = end_byte - start_byte;
     return s + start_byte - 1;
 }
+
diff --git a/core/src/xmake/utf8/utf8.h b/core/src/xmake/utf8/utf8.h
index f50b21e44..f1d374af3 100644
--- a/core/src/xmake/utf8/utf8.h
+++ b/core/src/xmake/utf8/utf8.h
@@ -41,7 +41,11 @@ static __tb_inline__ tb_long_t xm_utf8_posrelat(tb_long_t pos, tb_size_t len) {
 
 tb_char_t const*    xm_utf8_decode(tb_char_t const* s, xm_utf8_int_t* val, tb_bool_t strict);
 tb_size_t           xm_utf8_encode(tb_char_t* s, xm_utf8_int_t val);
+tb_long_t           xm_utf8_charpos(tb_char_t const* s, tb_size_t len, tb_long_t byte_pos);
 
+/* //////////////////////////////////////////////////////////////////////////////////////
+ * implementation interfaces
+ */
 tb_long_t           xm_utf8_len_impl(tb_char_t const* s, tb_size_t len, tb_long_t posi, tb_long_t posj, tb_bool_t strict, tb_size_t* errpos);
 tb_long_t           xm_utf8_offset_impl(tb_char_t const* s, tb_size_t len, tb_long_t n, tb_long_t posi);
 tb_bool_t           xm_utf8_codepoint_impl(tb_char_t const* s, tb_size_t len, tb_long_t posi, tb_long_t posj, tb_bool_t strict, xm_utf8_codepoint_func_t func, tb_cpointer_t udata);