From d53f2e7107cf63669b705c3abf08c129eeb0315e Mon Sep 17 00:00:00 2001
From: mattkae <mattkae@protonmail.com>
Date: Thu, 22 Jun 2023 13:29:49 -0400
Subject: Parsing most html entities except for the last category

---
 src/html_token.hpp | 46 ++++++++++++++++++++++++++++++++++++++++++++--
 1 file changed, 44 insertions(+), 2 deletions(-)

(limited to 'src/html_token.hpp')

diff --git a/src/html_token.hpp b/src/html_token.hpp
index ee385ce..3d848d9 100644
--- a/src/html_token.hpp
+++ b/src/html_token.hpp
@@ -17,14 +17,56 @@ enum HtmlTokenType {
 struct HtmlToken {
     HtmlTokenType type;
 
-    // TODO: Performance
+    // TODO: @Performance
     char character_token;
-    std::string tag_name;
+    std::wstring tag_name;
+    std::wstring code_entity;
 
     void append_to_tag_name(code_point_t c) {
         tag_name += c;
     }
 
+    void append_to_code_entity(code_point_t c) {
+        code_entity += c;
+    }
+
+    /// Transforms the code_entity into a usable value.
+    /// Note that we are assuming that the code_entity is
+    /// valid at this point in time.
+    code_point_t code_entity_to_value(bool is_hex) {
+        code_point_t value = 0x0000;
+        if (is_hex) {
+            int multiplier = 1;
+            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;
+                }
+                else if (c >= CodePoints::UPPERCASE_A) { // [A, Z]
+                    c = 10 + c - CodePoints::UPPERCASE_A;
+                }
+                else { // [0, 9]
+                    c = c - CodePoints::DIGIT_ZERO; // Now it is between 0 and 9
+                }
+
+                // Now we have c in decimal, let's convert it to the final value.
+                c = c * multiplier;
+                value += c;
+                multiplier *= 16;
+            }
+        }
+        else {
+            int multiplier = 1;
+            for (int i = code_entity.size() - 1; i >= 0; i--) {
+                auto c = code_entity[i];
+                c = c - CodePoints::DIGIT_ZERO; // Now it is between 0 and 9
+                value += c * multiplier;
+                multiplier *= 10;
+            }
+        }
+        return value;
+    }
+
     void print();
     void reset();
 };
-- 
cgit v1.2.1