From 6e13571f607068e94b61704ae10949c9aa093bac Mon Sep 17 00:00:00 2001
From: Kittcannon <alejandro.cervera@sintekanalytics.com>
Date: Sat, 26 Sep 2026 22:21:12 -0600
Subject: [PATCH] unicode

---
 src/spider/compiler/Compiler.cpp        |  4 ++--
 src/spider/compiler/text/TextReader.cpp |  8 ++++----
 src/spider/compiler/text/Token.cpp      | 10 +++++-----
 3 files changed, 11 insertions(+), 11 deletions(-)

diff --git a/src/spider/compiler/Compiler.cpp b/src/spider/compiler/Compiler.cpp
index 2879e91..b86fa33 100644
--- a/src/spider/compiler/Compiler.cpp
+++ b/src/spider/compiler/Compiler.cpp
@@ -20,7 +20,7 @@ static bool run_test_case(const Token* token, const std::string& input, bool exp
     if (status_ok && match_ok) {
         std::cout << "[PASS] Input: \"" << input << "\" -> "
             << (res.success ? "SUCCESS" : "FAILURE")
-            << (res.success ? (" (Matched: \"" + utf8::toUTF8(res.flatMatch()) + "\")") : "")
+            << (res.success ? (" (Matched: \"" + unicode::toUTF8(res.flatMatch()) + "\")") : "")
             << "\n";
         return true;
     }
@@ -28,7 +28,7 @@ static bool run_test_case(const Token* token, const std::string& input, bool exp
     std::cerr << "[FAIL] Input: \"" << input << "\"\n"
         << "       Expected Success: " << (expectedSuccess ? "true" : "false") << ", Got: " << (res.success ? "true" : "false") << "\n";
     if (expectedSuccess && !expectedMatch.empty()) {
-        std::cerr << "       Expected Match:   \"" << utf8::toUTF8(expectedMatch) << "\", Got: \"" << utf8::toUTF8(res.flatMatch()) << "\"\n";
+        std::cerr << "       Expected Match:   \"" << unicode::toUTF8(expectedMatch) << "\", Got: \"" << unicode::toUTF8(res.flatMatch()) << "\"\n";
     }
     return false;
 }
diff --git a/src/spider/compiler/text/TextReader.cpp b/src/spider/compiler/text/TextReader.cpp
index 6f157ca..13c0542 100644
--- a/src/spider/compiler/text/TextReader.cpp
+++ b/src/spider/compiler/text/TextReader.cpp
@@ -1,6 +1,6 @@
 #include "TextReader.hpp"
 
-#include <spider/compiler/text/utf8.hpp>
+#include <spider/compiler/text/unicode.hpp>
 
 #include <stdexcept>
 
@@ -28,7 +28,7 @@ namespace spider {
         // instead of whatever that was, convert to UTF-32
         // and then do an easy compare!
         std::u32string str;
-        if(!utf8::toUTF32(chars, str)) throw std::runtime_error("Specified invalid UTF-8 string!");
+        if(!unicode::toUTF32(chars, str)) throw std::runtime_error("Specified invalid UTF-8 string!");
         return eat(str);
     }
 
@@ -143,7 +143,7 @@ namespace spider {
         if (err) return false;
         if (eof) return false;
 
-        isize chsize = utf8::seqlen(u8(bytes[bindex]));
+        isize chsize = unicode::seqlen(u8(bytes[bindex]));
         if(chsize == 0) {
             err = true;
             errmsg = "Invalid start of UTF-8 sequence.";
@@ -158,7 +158,7 @@ namespace spider {
         }
 
         u32 decodedChar;
-        if(!utf8::decodeArr(bytes, chsize, decodedChar)) {
+        if(!unicode::decodeArr(bytes, chsize, decodedChar)) {
             err = true;
             errmsg = "Invalid UTF-8 sequence.";
             return false;
diff --git a/src/spider/compiler/text/Token.cpp b/src/spider/compiler/text/Token.cpp
index 472316c..4f8d53c 100644
--- a/src/spider/compiler/text/Token.cpp
+++ b/src/spider/compiler/text/Token.cpp
@@ -87,7 +87,7 @@ namespace spider {
     // ============================================================================
 
     LitToken::LitToken(std::string_view lit) {
-        if (!utf8::toUTF32(lit, literal)) throw std::runtime_error("Illegal UTF8 literal!");
+        if (!unicode::toUTF32(lit, literal)) throw std::runtime_error("Illegal UTF8 literal!");
     }
 
     LitToken::LitToken(std::u32string lit) : literal(std::move(lit)) {}
@@ -116,7 +116,7 @@ namespace spider {
         }
 
         r.success = !r.match.empty();
-        if(r.success) std::cout << "[fn] matched: " << utf8::toUTF8(r.flatMatch()) << std::endl;
+        if(r.success) std::cout << "[fn] matched: " << unicode::toUTF8(r.flatMatch()) << std::endl;
         return r;
     }
 
@@ -166,7 +166,7 @@ namespace spider {
             // Short-circuit branch: return immediately on first valid choice match
             TokenResult res = token_ref->test(ctx);
             if (res.success) {
-                std::cout << "[or] matched: " << utf8::toUTF8(res.flatMatch()) << std::endl;
+                std::cout << "[or] matched: " << unicode::toUTF8(res.flatMatch()) << std::endl;
                 return res;
             }
 
@@ -188,7 +188,7 @@ namespace spider {
         TokenResult res = target->test(ctx);
 
         if (res.success) {
-            std::cout << "[~] matched: " << utf8::toUTF8(res.flatMatch()) << std::endl;
+            std::cout << "[~] matched: " << unicode::toUTF8(res.flatMatch()) << std::endl;
             return res; // Option matched exactly 1 instance successfully
         }
 
@@ -225,7 +225,7 @@ namespace spider {
         // Repetition rules (* token) always evaluate to successful
         // completion state, even with 0 matches.
         r.success = true;
-        std::cout << "[*] matched: " << utf8::toUTF8(r.flatMatch()) << std::endl;
+        std::cout << "[*] matched: " << unicode::toUTF8(r.flatMatch()) << std::endl;
         return r;
     }