More parsing implementation

src/html_parse.cpp

@@ -1,5 +1,6 @@
 #include "html_parse.hpp"
 
+#include <cassert>
 #include <iostream>
 
 namespace html {
@@ -10,11 +11,44 @@ Tokenizer::Tokenizer(std::u32string source)
 {
 }
 
+static bool is_alpha(char32_t c)
+{
+    return (c >= 65 && c <= 90) || (c >= 97 && c <= 122);
+}
+
+static bool is_num(char32_t c)
+{
+    return c >= 48 && c <= 57;
+}
+
+static bool is_alpha_num(char32_t c)
+{
+    return is_alpha(c) || is_num(c);
+}
+
+static bool is_space(char32_t c)
+{
+    return c == 32;
+}
+
+static bool is_newline(char32_t c)
+{
+    return c == 10;
+}
+
 std::vector<Token> Tokenizer::tokenize()
 {
+    std::u32string buffer;
     std::vector<Token> tokens;
     while (peek().has_value()) {
-        if (peek().value() == '<') {
+        if (is_alpha(peek().value())) {
+            do {
+                buffer.push_back(consume());
+            } while (peek().has_value() && is_alpha_num(peek().value()));
+            tokens.push_back({ .type = TokenType::ident, .value = buffer });
+            buffer.clear();
+        }
+        else if (peek().value() == '<') {
             consume();
             tokens.push_back({ .type = TokenType::lt });
         }
@@ -22,13 +56,35 @@ std::vector<Token> Tokenizer::tokenize()
             consume();
             tokens.push_back({ .type = TokenType::gt });
         }
+        else if (peek().value() == '!') {
+            consume();
+            tokens.push_back({ .type = TokenType::exclaim });
+        }
+        else if (peek().value() == '/') {
+            consume();
+            tokens.push_back({ .type = TokenType::fslash });
+        }
+        else if (peek().value() == '=') {
+            consume();
+            tokens.push_back({ .type = TokenType::eq });
+        }
+        else if (peek().value() == '"') {
+            do {
+                buffer.push_back(consume());
+            } while (peek().has_value() && peek().value() != '"');
+            tokens.push_back({ .type = TokenType::str, .value = buffer });
+            buffer.clear();
+        }
+        else if (is_space(peek().value()) || is_newline(peek().value())) {
+            consume();
+        }
         else {
             std::cout << "[WARN] Unexpected token: " << std::to_string(peek().value()) << std::endl;
             consume();
         }
     }
 
-    return {};
+    return tokens;
 }
 std::optional<char32_t> Tokenizer::peek(size_t ahead)
 {
@@ -43,4 +99,35 @@ char32_t Tokenizer::consume()
     return m_source.at(m_index++);
 }
 
+std::u32string Token::to_string()
+{
+    std::u32string buffer;
+    switch (type) {
+    case TokenType::lt:
+        buffer.push_back('<');
+        return buffer;
+    case TokenType::gt:
+        buffer.push_back('>');
+        return buffer;
+    case TokenType::ident:
+        assert(value.has_value());
+        return value.value();
+    case TokenType::fslash:
+        buffer.push_back('/');
+        return buffer;
+    case TokenType::str:
+        assert(value.has_value());
+        return value.value();
+    case TokenType::exclaim:
+        buffer.push_back('!');
+        return buffer;
+    case TokenType::eq:
+        buffer.push_back('=');
+        return buffer;
+    default:
+        assert(false && "Unimplemented");
+        return buffer;
+    }
 }
+
+}

src/html_parse.hpp

@@ -6,17 +6,13 @@
 
 namespace html {
 
-enum class TokenType {
-    lt,
-    gt,
-    ident,
-    fslash,
-    quote,
-};
+enum class TokenType { lt, gt, ident, fslash, str, exclaim, eq };
 
 struct Token {
     TokenType type;
-    std::optional<std::string> value = {};
+    std::optional<std::u32string> value = {};
+
+    std::u32string to_string();
 };
 
 enum class NodeType {

src/main.cpp

@@ -1,13 +1,15 @@
 #include "fetch.hpp"
 
+#include <codecvt>
+#include <iostream>
 #include <optional>
 
-#include "html_parse.hpp"
-
 #define RAYGUI_IMPLEMENTATION
 #include <raygui.h>
 #include <raylib-cpp.hpp>
 
+#include "html_parse.hpp"
+
 int main()
 {
     init_curl();
@@ -32,7 +34,11 @@ int main()
     if (page_data.has_value()) {
         std::u32string page_data_unicode(page_data.value().begin(), page_data.value().end());
         html::Tokenizer tokenizer(page_data_unicode);
-        tokenizer.tokenize();
+        std::vector<html::Token> tokens = tokenizer.tokenize();
+        std::wstring_convert<std::codecvt_utf8<char32_t>, char32_t> converter;
+        for (html::Token token : tokens) {
+            std::cout << converter.to_bytes(token.to_string()) << std::endl;
+        }
     }
 
     while (!window.ShouldClose()) {
@@ -54,7 +60,7 @@ int main()
         scroll_pos += GetMouseWheelMove();
 
         if (page_data.has_value()) {
-            DrawTextEx(sans_font, page_data.value(), { 0, 20 + scroll_pos * 20 }, 24, 1.0f, BLACK);
+            DrawTextEx(sans_font, page_data.value(), { 0, 20 + scroll_pos * 40 }, 24, 1.0f, BLACK);
         }
 
         EndDrawing();