/* * Copyright (c) 2023, Tim Flynn * Copyright (c) 2023, Cameron Youell * * SPDX-License-Identifier: BSD-2-Clause */ #include #include #include #include #if defined(ENABLE_PUBLIC_SUFFIX) # include #endif namespace WebView { static Optional query_public_suffix_list(StringView url_string) { auto out = MUST(String::from_utf8(url_string)); if (!out.contains("://"sv)) out = MUST(String::formatted("https://{}"sv, out)); auto url = URL::create_with_url_or_path(out.to_deprecated_string()); if (!url.is_valid()) return {}; if (url.host().has() || url.host().has()) return url; if (url.scheme() != "http"sv && url.scheme() != "https"sv) return url; if (url.host().has()) { auto const& host = url.host().get(); if (auto public_suffix = get_public_suffix(host); public_suffix.has_value()) return url; if (host.ends_with_bytes(".local"sv) || host.ends_with_bytes("localhost"sv)) return url; } return {}; } bool is_public_suffix([[maybe_unused]] StringView host) { #if defined(ENABLE_PUBLIC_SUFFIX) return PublicSuffixData::the()->is_public_suffix(host); #else return false; #endif } Optional get_public_suffix([[maybe_unused]] StringView host) { #if defined(ENABLE_PUBLIC_SUFFIX) return MUST(PublicSuffixData::the()->get_public_suffix(host)); #else return {}; #endif } Optional sanitize_url(StringView url, Optional search_engine, AppendTLD append_tld) { if (FileSystem::exists(url)) { auto path = FileSystem::real_path(url); if (path.is_error()) return {}; return URL::create_with_file_scheme(path.value().to_deprecated_string()); } auto format_search_engine = [&]() -> Optional { if (!search_engine.has_value()) return {}; return MUST(String::formatted(*search_engine, URL::percent_decode(url))); }; String url_buffer; if (append_tld == AppendTLD::Yes) { // FIXME: Expand the list of top level domains. if (!url.ends_with(".com"sv) && !url.ends_with(".net"sv) && !url.ends_with(".org"sv)) { url_buffer = MUST(String::formatted("{}.com", url)); url = url_buffer; } } auto result = query_public_suffix_list(url); if (!result.has_value()) return format_search_engine(); return result.release_value(); } static URLParts break_file_url_into_parts(URL const& url, StringView url_string) { auto scheme = url_string.substring_view(0, url.scheme().bytes_as_string_view().length() + "://"sv.length()); auto path = url_string.substring_view(scheme.length()); return URLParts { scheme, path, {} }; } static URLParts break_web_url_into_parts(URL const& url, StringView url_string) { auto scheme = url_string.substring_view(0, url.scheme().bytes_as_string_view().length() + "://"sv.length()); auto url_without_scheme = url_string.substring_view(scheme.length()); StringView domain; StringView remainder; if (auto index = url_without_scheme.find_any_of("/?#"sv); index.has_value()) { domain = url_without_scheme.substring_view(0, *index); remainder = url_without_scheme.substring_view(*index); } else { domain = url_without_scheme; } auto public_suffix = get_public_suffix(domain); if (!public_suffix.has_value() || !domain.ends_with(*public_suffix)) return { scheme, domain, remainder }; auto subdomain = domain.substring_view(0, domain.length() - public_suffix->bytes_as_string_view().length()); subdomain = subdomain.trim("."sv, TrimMode::Right); if (auto index = subdomain.find_last('.'); index.has_value()) { subdomain = subdomain.substring_view(0, *index + 1); domain = domain.substring_view(subdomain.length()); } else { subdomain = {}; } auto scheme_and_subdomain = url_string.substring_view(0, scheme.length() + subdomain.length()); return { scheme_and_subdomain, domain, remainder }; } Optional break_url_into_parts(StringView url_string) { auto url = URL::create_with_url_or_path(url_string); if (!url.is_valid()) return {}; auto scheme_length = url.scheme().bytes_as_string_view().length(); if (!url_string.substring_view(scheme_length).starts_with("://"sv)) return {}; if (url.scheme() == "file"sv) return break_file_url_into_parts(url, url_string); if (url.scheme().is_one_of("http"sv, "https"sv, "gemini"sv)) return break_web_url_into_parts(url, url_string); return {}; } }