From 614dcb875411fcca63f9b8de50256c8b53889355 Mon Sep 17 00:00:00 2001 From: Richard Qin Date: Sat, 11 Apr 2026 02:48:24 +0800 Subject: feat: implement HTML lexer and integrate tokenization into browser output --- src/core/browser.cc | 50 ++++++++++++++++++++++++++++++++++++-------------- 1 file changed, 36 insertions(+), 14 deletions(-) (limited to 'src/core/browser.cc') diff --git a/src/core/browser.cc b/src/core/browser.cc index ccda9a1..1e9d044 100644 --- a/src/core/browser.cc +++ b/src/core/browser.cc @@ -18,26 +18,48 @@ #include "ricochet/core/browser.hh" #include "ricochet/net/http_client.hh" +#include "ricochet/parser/html_lexer.hh" #include +namespace { +struct TokenPrinter { + void operator()(const ricochet::parser::TextToken& t) const { + std::cout << "[TEXT] " << t.content << "\n"; + } + void operator()(const ricochet::parser::TagOpenToken& t) const { + std::cout << "[OPEN] <" << t.name << ">\n"; + } + void operator()(const ricochet::parser::TagCloseToken& t) const { + std::cout << "[CLOSE] \n"; + } +}; +} // namespace namespace ricochet::core { int Browser::run( std::string_view initial_url ) { - std::cout << "Browser: Navigating to " << initial_url << "...\n"; - - const net::HttpClient client; - auto response_result = client.fetch( initial_url ); - if ( !response_result.has_value() ) { - std::cerr << "[!] Ricochet failed to load page: " << response_result.error() << "\n"; - return 1; - } - - const auto& response = response_result.value(); - std::cout << "\n[=== HTTP Status: " << response.status_code << " ===]\n"; - std::cout << "[=== HTML Content Below ===]\n\n"; - std::cout << response.body << "\n"; - return 0; + std::cout << "-> Fetching: " << initial_url << "...\n"; + + const net::HttpClient client; + auto response_result = client.fetch(initial_url); + + if (!response_result.has_value()) { + std::cerr << "[!] Ricochet failed to load page: " << response_result.error() << "\n"; + return 1; + } + + const auto& response = response_result.value(); + std::cout << "[=== HTTP Status: " << response.status_code << " ===]\n"; + + std::cout << "\n[=== Tokenizing HTML ===]\n"; + const parser::HtmlLexer lexer; + auto tokens = lexer.tokenize(response.body); + + for (const auto& token : tokens) { + std::visit(TokenPrinter{}, token); + } + + return 0; } } // namespace ricochet::core -- cgit v1.2.3