URL.hpp

// Copyright Take Vos 2019-2022.
// Distributed under the Boost Software License, Version 1.0.
// (See accompanying file LICENSE_1_0.txt or copy at https://www.boost.org/LICENSE_1_0.txt)
 
#pragma once
 
#include "utility.hpp"
#include "assert.hpp"
#include "URI.hpp"
#include <string>
#include <string_view>
#include <optional>
#include <vector>
#include <unordered_map>
#include <memory>
#include <ostream>
#include <mutex>
#include <filesystem>
 
namespace hi { inline namespace v1 {
class resource_view;
 
class URL : public URI {
public:
    constexpr URL() noexcept = default;
    constexpr URL(URL const&) noexcept = default;
    constexpr URL(URL&&) noexcept = default;
    constexpr URL& operator=(URL const&) noexcept = default;
    constexpr URL& operator=(URL&&) noexcept = default;
 
    constexpr explicit URL(URI const& other) noexcept : URI(other) {}
    constexpr explicit URL(URI&& other) noexcept : URI(std::move(other)){};
 
    static std::string make_file_url_string(std::filesystem::path const& path)
    {
        auto r = std::string{};
 
        hilet root_name = path.root_name().generic_string();
        if (root_name.empty()) {
            // No root-name.
            if (not path.root_directory().empty()) {
                // An absolute path should start with the file: scheme.
                r += "file:" + path.root_directory().generic_string();
            } else {
                // A relative path should not be prefixed with a scheme.
                ;
            }
 
        } else if (hilet i = root_name.find(':'); i != std::string::npos) {
            if (i == 1) {
                // Root name is a drive-letter, followed by potentially a relative path.
                r += "file:///" + root_name + path.root_directory().generic_string();
            } else {
                throw url_error("Paths containing a device are not allowed to be converted to a URL.");
            }
        } else {
            // Root name is a server.
            r += "file://" + root_name + path.root_directory().generic_string();
            if (not path.root_directory().empty()) {
                throw url_error("Invalid path contains server name without a root directory.");
            }
        }
 
        return r + path.relative_path().generic_string();
    }
 
    explicit URL(std::filesystem::path const& path) : URI(make_file_url_string(path)) {}
 
    constexpr void static validate_file_segment(std::string_view segment)
    {
        for (auto c : segment) {
            if (c == '/' or c == '\\') {
                throw url_error("Filename server name may not contain slash or back-slash.");
            }
        }
    }
 
    constexpr void static validate_file_server(std::string_view server)
    {
        for (auto c : server) {
            if (c == '/' or c == '\\') {
                throw url_error("Filename segments may not contain slash or back-slash.");
            }
        }
    }
 
    [[nodiscard]] std::string generic_path() const
    {
        if (not(not scheme() or scheme() == "file")) {
            throw url_error("URL::generic_path() is only valid on a file: scheme URL");
        }
 
        auto r = std::string{};
        hilet& p = path();
        hilet first = p.begin();
        hilet last = p.end();
        auto it = first;
 
        auto has_root_name = false;
        if (authority()) {
            // file://server/filename is valid.
            auto server = to_string(*authority());
            if (not server.empty() and server != "localhost") {
                validate_file_server(server);
                has_root_name = true;
                r += '/';
                r += '/';
                r += server;
                r += '/';
            }
        }
 
        // If a server was found than the path must be absolute.
        hi_axiom(has_root_name == false or p.absolute());
 
        if (p.double_absolute()) {
            // file:////server/filename is valid.
            if (has_root_name) {
                // file://server//server/filename is invalid.
                throw url_error("file URL has two server names.");
            }
 
            has_root_name = true;
            r += '/';
            r += '/';
            it += 2;
            validate_file_server(*it);
            r += *it++;
            r += '/';
        }
 
        // Find the drive letter.
        auto empty_segment = false;
        while (it != last) {
            validate_file_segment(*it);
            empty_segment = it->empty();
 
            if (it == first and empty_segment) {
                // Skip leading '/' in front of drive letter.
                ++it;
 
            } else if (auto i = it->find(':'); i != std::string::npos) {
                // Found a drive letter.
                if (i != 1) {
                    throw url_error("file URL contains a device name which is a security issue.");
                }
 
                if (has_root_name or p.absolute()) {
                    r += it->front();
                    // Use $ when the drive letter is on a server.
                    r += has_root_name ? '$' : ':';
                    // Add potentially a relative segment after the driver letter.
                    r += it->substr(2);
 
                } else {
                    // Take the drive letter and optional relative directory directly on relative paths.
                    // C:dirname valid
                    // C:/dirname valid
                    // file:C:dirname valid
                    // file:C:/dirname valid
                    r += *it;
                }
 
                has_root_name = true;
                r += '/';
                ++it;
                break;
 
            } else {
                // Found a normal directory or filename.
                break;
            }
        }
 
        // The rest are directories followed by a single (optionally empty) filename
        for (; it != last; ++it) {
            validate_file_segment(*it);
            empty_segment = it->empty();
            r += *it;
            r += '/';
        }
 
        if (not empty_segment) {
            // Remove trailing backslash if the last segment was a filename.
            r.pop_back();
        }
 
        return r;
    }
 
    [[nodiscard]] std::filesystem::path filesystem_path() const
    {
        if (auto scheme_ = scheme()) {
            if (scheme_ == "resource") {
                return URL::url_from_resource_directory() / *this;
            } else if (scheme_ == "file") {
                return {generic_path()};
            } else {
                throw url_error("URL can not be converted to a filesystem path.");
            }
        } else {
            return {generic_path()};
        }
    }
 
    operator std::filesystem::path() const
    {
        return filesystem_path();
    }
 
    [[nodiscard]] std::unique_ptr<resource_view> loadView() const;
 
    [[nodiscard]] constexpr friend URL operator/(URL const& base, URI const& ref) noexcept
    {
        return URL{up_cast<URI const&>(base) / ref};
    }
 
    [[nodiscard]] constexpr friend URL operator/(URL const& base, std::string_view ref) noexcept
    {
        return URL{up_cast<URI const&>(base) / ref};
    }
 
    [[nodiscard]] static URL url_from_current_working_directory() noexcept;
    [[nodiscard]] static URL url_from_resource_directory() noexcept;
    [[nodiscard]] static URL url_from_executable_directory() noexcept;
    [[nodiscard]] static URL url_from_executable_file() noexcept;
    [[nodiscard]] static URL url_from_application_data_directory() noexcept;
    [[nodiscard]] static URL url_from_application_log_directory() noexcept;
    [[nodiscard]] static URL url_from_system_font_directory() noexcept;
    [[nodiscard]] static URL url_from_application_preferences_file() noexcept;
};
}} // namespace hi::v1
 
template<>
struct std::hash<hi::URL> {
    [[nodiscard]] size_t operator()(hi::URL const& rhs) const noexcept
    {
        return std::hash<hi::URI>{}(rhs);
    }
};
 
template<typename CharT>
struct std::formatter<hi::URL, CharT> : std::formatter<hi::URI, CharT> {
    auto format(hi::URL const& t, auto& fc)
    {
        return std::formatter<hi::URI, CharT>::format(t, fc);
    }
};