Program Listing for File parser.hpp

Return to documentation for file (zeem/parser.hpp)

// Copyright (c) 2024-2026 Maarten L. Hekkelman
// SPDX-License-Identifier: BSD-2-Clause

#pragma once

#ifndef ZEEM_CXX_MODULE
# include "zeem/error.hpp"

# include <functional>
# include <istream>
# include <memory>
# include <string>
# include <string_view>
# include <utility>
# include <vector>
#endif

namespace zeem
{

enum class encoding_type;
struct version_type;


ZEEM_EXPORT class invalid_exception : public exception
{
  public:
    explicit invalid_exception(std::string msg)
        : exception(std::move(msg))
    {
    }
};


ZEEM_EXPORT class not_wf_exception : public exception
{
  public:
    explicit not_wf_exception(std::string msg)
        : exception(std::move(msg))
    {
    }
};

ZEEM_EXPORT class parser
{
  public:
    struct attr
    {
        std::string m_ns;
        std::string m_name;
        std::string m_value;
        bool m_id{};
    };

    using attr_list_type = std::vector<attr>;

    explicit parser(std::istream &is);

    // Avoid default constructor
    parser() = delete;

    // Avoid copy
    parser(const parser &) = delete;

    // Move constructor
    parser(parser &&rhs) noexcept;

    // Assignment operators
    parser &operator=(const parser &) = delete;
    parser &operator=(parser &&rhs) noexcept;

    virtual ~parser();

    std::function<void(encoding_type encoding, bool standalone, version_type version)> xml_decl_handler;
    std::function<void(std::string name, std::string uri, const attr_list_type &atts)> start_element_handler;
    std::function<void(std::string name, std::string uri)> end_element_handler;
    std::function<void(std::string data)> character_data_handler;
    std::function<void(std::string target, std::string data)> processing_instruction_handler;
    std::function<void(std::string data)> comment_handler;
    std::function<void()> start_cdata_section_handler;
    std::function<void()> end_cdata_section_handler;
    std::function<void(std::string prefix, std::string uri)> start_namespace_decl_handler;
    std::function<void(std::string prefix)> end_namespace_decl_handler;
    std::function<void(std::string root, std::string publicId, std::string uri)> doctype_decl_handler;
    std::function<void(std::string name, std::string systemId, std::string publicId)> notation_decl_handler;
    std::function<std::unique_ptr<std::istream>(std::string_view base, std::string_view pubid, std::string_view uri)> external_entity_ref_handler;
    std::function<void(std::string msg)> report_invalidation_handler;

    void parse(bool validate, bool validate_ns);

  protected:
    friend struct parser_imp;

    virtual void xml_decl(encoding_type encoding, bool standalone, version_type version);

    virtual void doctype_decl(std::string root, std::string publicId, std::string uri);

    virtual void start_element(std::string name, std::string uri, const attr_list_type &atts);

    virtual void end_element(std::string name, std::string uri);

    virtual void character_data(std::string data);

    virtual void processing_instruction(std::string target, std::string data);

    virtual void comment(std::string data);

    virtual void start_cdata_section();

    virtual void end_cdata_section();

    virtual void start_namespace_decl(std::string prefix, std::string uri);

    virtual void end_namespace_decl(std::string prefix);

    virtual void notation_decl(std::string name,
        std::string systemId, std::string publicId);

    virtual void report_invalidation(std::string msg);

    virtual std::unique_ptr<std::istream> external_entity_ref(std::string_view base,
        std::string_view pubid, std::string_view uri);

  private:
    std::unique_ptr<struct parser_imp> m_impl;

};

} // namespace zeem