From 29e03ef74a814cb31a0ae53192e25cc75b638256 Mon Sep 17 00:00:00 2001
From: mattkae <mattkae@protonmail.com>
Date: Thu, 22 Jun 2023 13:48:11 -0400
Subject: Handling less than signs in plain text

---
 CMakeLists.txt     |  6 ++++++
 examples/1.html    |  2 ++
 src/html_token.cpp |  2 +-
 src/html_token.hpp |  2 +-
 src/tokenizer.cpp  | 19 +++++++++++++------
 src/tokenizer.hpp  |  2 +-
 6 files changed, 24 insertions(+), 9 deletions(-)

diff --git a/CMakeLists.txt b/CMakeLists.txt
index a65ed8b..56e8ff8 100644
--- a/CMakeLists.txt
+++ b/CMakeLists.txt
@@ -7,6 +7,12 @@ set(CMAKE_CXX_STANDARD_REQUIRED ON)
 
 add_executable(html_parser src/main.cpp src/html_token.cpp src/tokenizer.cpp)
 
+if(MSVC)
+  target_compile_options(html_parser PRIVATE /W4 /WX)
+else()
+  target_compile_options(html_parser PRIVATE -Wall -Wextra -Wpedantic -Wswitch-enum)
+endif()
+
 target_link_libraries(html_parser -lmatte)
 
 install(TARGETS html_parser
diff --git a/examples/1.html b/examples/1.html
index 8193bf0..f83a60a 100644
--- a/examples/1.html
+++ b/examples/1.html
@@ -1,3 +1,5 @@
+Meow is < bark
+
 <div>
   <h1>Hello World</h1>
   <p>
diff --git a/src/html_token.cpp b/src/html_token.cpp
index 8589ba8..1d0952d 100644
--- a/src/html_token.cpp
+++ b/src/html_token.cpp
@@ -18,7 +18,7 @@ void HtmlToken::print() {
         break;
     case HtmlTokenType_StartTag:
     case HtmlTokenType_EndTag:
-        logger_info("%s, %s", name, tag_name.c_str());
+        logger_info("%s, %S", name, tag_name.c_str());
         break;
     default:
         logger_info("%s", name);
diff --git a/src/html_token.hpp b/src/html_token.hpp
index 3d848d9..e691d21 100644
--- a/src/html_token.hpp
+++ b/src/html_token.hpp
@@ -37,7 +37,7 @@ struct HtmlToken {
         code_point_t value = 0x0000;
         if (is_hex) {
             int multiplier = 1;
-            for (size_t i = code_entity.size() - 1; i >= 0; i--) {
+            for (size_t i = code_entity.size() - 1; i > 0; i--) {
                 auto c = code_entity[i];
                 if (c >= CodePoints::LOWERCASE_A) { // [a, z]
                     c = 10 + c - CodePoints::LOWERCASE_A;
diff --git a/src/tokenizer.cpp b/src/tokenizer.cpp
index 9931d59..dc0b8d7 100644
--- a/src/tokenizer.cpp
+++ b/src/tokenizer.cpp
@@ -28,6 +28,12 @@ inline void unconsume_previous(Tokenizer* tokenizer) {
     tokenizer->ptr--;
 }
 
+inline void emit_character(Tokenizer* tokenizer, code_point_t c) {
+    tokenizer->last.type = HtmlTokenType_Character;
+    tokenizer->last.character_token = c;
+    tokenizer->flag |= TokenizerFlag_Emit;
+}
+
 
 /// https://dev.w3.org/html5/spec-LC/tokenization.html#data-state
 inline void data_state(Tokenizer* tokenizer) {
@@ -45,9 +51,7 @@ inline void data_state(Tokenizer* tokenizer) {
         break;
     default:
         // TODO: @Error If null, throw an error
-        tokenizer->last.type = HtmlTokenType_Character;
-        tokenizer->last.character_token = *tokenizer->ptr;
-        tokenizer->flag |= TokenizerFlag_Emit;
+        emit_character(tokenizer, *tokenizer->ptr);
         break;
     }
 }
@@ -80,8 +84,11 @@ inline void tag_open_state(Tokenizer* tokenizer) {
             tokenizer->state = TokenizerState_BogusComment; // TODO:
         }
         else {
+            emit_character(tokenizer, CodePoints::LESS_THAN_SIGN);
             tokenizer->state = TokenizerState_Data;
+            tokenizer->flag = tokenizer->flag | TokenizerFlag_NoIncrement;
         }
+        break;
     }
 }
 
@@ -114,7 +121,7 @@ inline void tag_name_state(Tokenizer* tokenizer) {
         else if (c == EOF) {
             // TODO: @Error
             tokenizer->state = TokenizerState_Data;
-            tokenizer->flag = tokenizer->flag & TokenizerFlag_DecrementPtr;
+            tokenizer->flag = tokenizer->flag | TokenizerFlag_NoIncrement;
         }
         else {
             tokenizer->last.append_to_tag_name(c);
@@ -151,7 +158,7 @@ inline void end_tag_open_state(Tokenizer* tokenizer) {
         tokenizer->last.type = HtmlTokenType_Character;
         tokenizer->last.character_token = CodePoints::SOLIDUS;
         
-        tokenizer->flag = tokenizer->flag & TokenizerFlag_DecrementPtr;
+        tokenizer->flag = tokenizer->flag | TokenizerFlag_NoIncrement;
     }
     else if (c == CodePoints::NULL_CHAR) {
         // TODO: @Error
@@ -286,7 +293,7 @@ HtmlToken read_next(Tokenizer* tokenizer) {
             exit(1);
         }
         
-        if ((tokenizer->flag & TokenizerFlag_DecrementPtr) == 0) {
+        if ((tokenizer->flag & TokenizerFlag_NoIncrement) == 0) {
             tokenizer->ptr++;
         }
 
diff --git a/src/tokenizer.hpp b/src/tokenizer.hpp
index 4cd9245..8b844cd 100644
--- a/src/tokenizer.hpp
+++ b/src/tokenizer.hpp
@@ -7,7 +7,7 @@
 enum TokenizerFlag {
     TokenizerFlag_None = 0,
     TokenizerFlag_Emit = 1,
-    TokenizerFlag_DecrementPtr = 2
+    TokenizerFlag_NoIncrement = 2
 };
 
 enum TokenizerState {
-- 
cgit v1.2.1