improve utf8.lastof

6 files changed, 73 insertions, 24 deletions

diff --git a/core/src/xmake/utf8/find.c b/core/src/xmake/utf8/find.c
index 8f8a5a97f..cc57f1e69 100644
--- a/core/src/xmake/utf8/find.c
+++ b/core/src/xmake/utf8/find.c
@@ -25,7 +25,7 @@
 #include "utf8.h"
 
 /* //////////////////////////////////////////////////////////////////////////////////////
- * implementation
+ * private implementation
  */
 
 static tb_int_t xm_utf8_find_impl_plain(lua_State* lua, tb_char_t const* s, size_t len, tb_char_t const* sub, size_t sublen, lua_Integer init) {
@@ -110,6 +110,10 @@ static tb_int_t xm_utf8_find_impl_pattern(lua_State* lua, tb_char_t const* s, si
     return nres;
 }
 
+/* //////////////////////////////////////////////////////////////////////////////////////
+ * implementation
+ */
+
 /* utf8.find(s, target [, init [, plain]])
  */
 tb_int_t xm_utf8_find(lua_State *lua) {
diff --git a/core/src/xmake/utf8/lastof.c b/core/src/xmake/utf8/lastof.c
index 8c5ec7dd3..8e22b16ba 100644
--- a/core/src/xmake/utf8/lastof.c
+++ b/core/src/xmake/utf8/lastof.c
@@ -28,38 +28,50 @@
  * implementation
  */
 
-/* utf8.lastof(s, substr)
+/* utf8.lastof(s, pattern, plain)
  */
 tb_int_t xm_utf8_lastof(lua_State *lua) {
     size_t len;
     tb_char_t const* s = luaL_checklstring(lua, 1, &len);
     size_t sublen;
     tb_char_t const* sub = luaL_checklstring(lua, 2, &sublen);
+    tb_int_t plain = lua_toboolean(lua, 3);
 
-    if (sublen == 0) {
-        lua_pushnil(lua);
+    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;
-    }
+    } 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
 
-    tb_char_t const* p = s;
-    tb_char_t const* last = tb_null;
-    
-    while (1) {
-        p = tb_strstr(p, sub);
-        if (!p) break;
-        last = p;
-        p += 1; 
-    }
+        lua_call(lua, 3, 1);
 
-    if (last) {
-        tb_long_t count = xm_utf8_len_impl(s, len, 1, last - s, tb_true, tb_null);
-        if (count < 0) {
-             lua_pushnil(lua);
-             return 1;
+        // Stack: [args, string_table, result]
+        if (lua_isnil(lua, -1)) {
+            return 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;
+            }
         }
-        lua_pushinteger(lua, count + 1);
-    } else {
         lua_pushnil(lua);
+        return 1;
     }
-    return 1;
 }
diff --git a/core/src/xmake/utf8/utf8.c b/core/src/xmake/utf8/utf8.c
index 3b54a0f37..3b23532c5 100644
--- a/core/src/xmake/utf8/utf8.c
+++ b/core/src/xmake/utf8/utf8.c
@@ -239,6 +239,29 @@ tb_long_t xm_utf8_find_impl(tb_char_t const* s, tb_size_t len, tb_char_t const*
     return char_start;
 }
 
+tb_long_t xm_utf8_lastof_impl(tb_char_t const* s, tb_size_t len, tb_char_t const* sub, tb_size_t sublen) {
+    tb_assert_and_check_return_val(s && sub, 0);
+
+    if (sublen == 0) return 0;
+
+    tb_char_t const* p = s;
+    tb_char_t const* last = tb_null;
+    
+    while (1) {
+        p = tb_strstr(p, sub);
+        if (!p) break;
+        last = p;
+        p += 1; 
+    }
+
+    if (last) {
+        tb_long_t count = xm_utf8_len_impl(s, len, 1, last - s, tb_true, tb_null);
+        if (count < 0) return 0;
+        return count + 1;
+    }
+    return 0;
+}
+
 tb_char_t const* xm_utf8_sub_impl(tb_char_t const* s, tb_size_t len, tb_long_t i, tb_long_t j, tb_size_t* psublen) {
     tb_assert_and_check_return_val(s && psublen, tb_null);
     *psublen = 0;
diff --git a/core/src/xmake/utf8/utf8.h b/core/src/xmake/utf8/utf8.h
index 8a61afe10..f50b21e44 100644
--- a/core/src/xmake/utf8/utf8.h
+++ b/core/src/xmake/utf8/utf8.h
@@ -46,6 +46,7 @@ tb_long_t           xm_utf8_len_impl(tb_char_t const* s, tb_size_t len, tb_long_
 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);
 tb_long_t           xm_utf8_find_impl(tb_char_t const* s, tb_size_t len, tb_char_t const* sub, tb_size_t sublen, tb_long_t init, tb_long_t* pchar_end);
+tb_long_t           xm_utf8_lastof_impl(tb_char_t const* s, tb_size_t len, tb_char_t const* sub, tb_size_t sublen);
 tb_char_t const*    xm_utf8_sub_impl(tb_char_t const* s, tb_size_t len, tb_long_t i, tb_long_t j, tb_size_t* psublen);
 
 #endif
diff --git a/tests/modules/utf8/test.lua b/tests/modules/utf8/test.lua
index 51b21d1c3..af0c98997 100644
--- a/tests/modules/utf8/test.lua
+++ b/tests/modules/utf8/test.lua
@@ -105,6 +105,15 @@ function test_lastof(t)
     t:are_equal(utf8.lastof("A你好A", "好"), 3)
 
     t:are_equal(utf8.lastof("ABC", "D"), nil)
+    
+    -- plain
+    t:are_equal(utf8.lastof("ABC", "A", true), 1)
+    t:are_equal(utf8.lastof("ABC", "B", true), 2)
+    t:are_equal(utf8.lastof("ABC", ".", true), nil)
+
+    -- pattern
+    t:are_equal(utf8.lastof("ABC", "."), 3)
+    t:are_equal(utf8.lastof("你好", "."), 2)
 end
 
 function test_find(t)
diff --git a/xmake/core/base/utf8.lua b/xmake/core/base/utf8.lua
index 2878404fb..dfcfdfbf0 100644
--- a/xmake/core/base/utf8.lua
+++ b/xmake/core/base/utf8.lua
@@ -30,8 +30,8 @@ local utf8 = utf8 or {}
 -- @interface   utf8.offset(s, n [, i])
 -- @interface   utf8.codes(s [, lax])
 -- @interface   utf8.sub(s, i [, j])
--- @interface   utf8.lastof(s, substr)
--- @interface   utf8.find(s, target [, init])
+-- @interface   utf8.lastof(s, pattern [, plain])
+-- @interface   utf8.find(s, pattern [, init [, plain]])
 --
 
 -- the char pattern