From 09d117debf1a1bde32073445f2d9dcb7e5f09695 Mon Sep 17 00:00:00 2001
From: Kittycannon <alejandro.cervera@sintekanalytics.com>
Date: Sun, 19 Jul 2026 14:45:49 -0600
Subject: [PATCH] begin of testing

---
 src/spider/compiler/Compiler.cpp  | 15 +++++++++++++++
 src/spider/compiler/text/utf8.hpp |  8 +++++---
 2 files changed, 20 insertions(+), 3 deletions(-)

diff --git a/src/spider/compiler/Compiler.cpp b/src/spider/compiler/Compiler.cpp
index 6654941..ed21b53 100644
--- a/src/spider/compiler/Compiler.cpp
+++ b/src/spider/compiler/Compiler.cpp
@@ -1,5 +1,8 @@
 #include <iostream>
 
+#include <spider/compiler/common.hpp>
+#include <spider/compiler/text/utf8.hpp>
+
 namespace spider {
 
 
@@ -7,6 +10,18 @@ namespace spider {
 }
 
 int main() {
+    std::string test = "UTF-STR WITH EMOJIS 😀🚀";
+    // Extracted UTF-8 chars:
+    std::u32string out;
+    bool r = spider::utf8::toUTF32(test, out);
+
+    std::cout << "INPUT: " << test << std::endl;
+    std::cout << "RESULT: " << int(r) << std::endl;
+    for(spider::u32 ch : out) {
+        std::cout << ch << " ";
+    }
+    std::cout << std::endl;
+
     std::cout << "Happy Day!" << std::endl;
     return 0;
 }
diff --git a/src/spider/compiler/text/utf8.hpp b/src/spider/compiler/text/utf8.hpp
index 7e1a896..d230442 100644
--- a/src/spider/compiler/text/utf8.hpp
+++ b/src/spider/compiler/text/utf8.hpp
@@ -86,10 +86,11 @@ namespace spider {
             return out;
         }
 
-        inline bool charAt(const std::string& str, isize& index, u32& out) {
-            isize chlen = isValidSeq(str.c_str(), str.size());
+        inline bool charAt(std::string_view str, isize& index, u32& out) {
+            isize chlen = isValidSeq(str.begin(), str.size());
             if(chlen == 0) return false;
-            out = decodeArr(str.c_str(), chlen);
+            out = decodeArr(str.begin(), chlen);
+            index += chlen;
             return true;
         }
 
@@ -103,6 +104,7 @@ namespace spider {
                 _size = isValidSeq(cptr + _i, csize - _i);
                 if(_size == 0) return false;
                 out += decodeArr(cptr + _i, _size);
+                _i += _size;
             }
             
             return _i == csize;