sentence.hpp

/*
 * CoNLL-U - https://github.com/tugrulgungor/CoNLL-U
 *
 * Copyright (c) 2026. All rights reserved.
 * Tuğrul Güngör - https://conll-u.tugrulgungor.me
 *
 * Distributed under the MIT License.
 * https://opensource.org/license/mit/
 */

 
#ifndef TG_CONLLU_SENTENCE_HPP
#define TG_CONLLU_SENTENCE_HPP 1
 
#include "conllu/error.hpp"
#include "conllu/internal/string_hash.hpp"
#include "conllu/word.hpp"
 
#include <expected>
#include <string>
#include <unordered_map>
#include <vector>
 
namespace tg::conllu {
    class Treebank;

    class Sentence final {
        std::string sent_id_{};
        std::string text_{};
 
        std::unordered_map<std::string, std::string, internal::StringHash, std::equal_to<>> metadata_{};
 
        std::vector<std::unique_ptr<Token>> tokens_{};
        std::vector<std::unique_ptr<Word>> words_{};
 
        Treebank* treebank_;
 
        static constexpr void split_metadata_line_(std::string_view& line, std::string_view& key,
                                                   std::string_view& value) noexcept;
 
        [[nodiscard]] constexpr std::optional<std::error_code> set_known_metadata_(std::string_view key,
                                                                                   std::string_view value) noexcept;
 
        [[nodiscard]] constexpr std::error_code set_sent_id_(std::string_view sent_id) noexcept;
 
        constexpr std::expected<Word*, std::error_code> add_word_unsafe_(Word::Range range) noexcept;
        constexpr void remove_empty_words_() noexcept;
 
        explicit(true) constexpr Sentence(Treebank* treebank) noexcept;
 
        friend class Parser;
        friend class Token;
        friend class Treebank;
        friend class Word;
        friend class Writer;
 
    public:
        constexpr Sentence() noexcept = delete;
        [[nodiscard]] constexpr Treebank* treebank() const noexcept;
        [[nodiscard]] constexpr std::string_view get_sent_id() const noexcept;
        constexpr std::error_code set_sent_id(std::string_view sent_id) noexcept;
        [[nodiscard]] constexpr std::string_view get_text() const noexcept;
        [[nodiscard]] constexpr std::error_code set_text(std::string_view text) noexcept;
        [[nodiscard]] constexpr auto tokens() const noexcept;
        [[nodiscard]] constexpr std::size_t num_tokens() const noexcept;
        [[nodiscard]] constexpr std::optional<Token*> get_token_by_id(TokenId id) const noexcept;
        [[nodiscard]] constexpr std::expected<Token*, std::error_code> add_token() noexcept;
        constexpr std::error_code remove_token(TokenId id, bool skip_dependency_validation = false) noexcept;
        constexpr std::error_code remove_token(Token* token, bool skip_dependency_validation = false) noexcept;
        constexpr void remove_all_tokens() noexcept;
        [[nodiscard]] constexpr auto words() const noexcept;
        [[nodiscard]] constexpr std::size_t num_words() const noexcept;
        [[nodiscard]] constexpr std::expected<Word*, std::error_code> add_word(TokenId first, TokenId last) noexcept;
        [[nodiscard]] constexpr std::expected<Word*, std::error_code> add_word(Word::Range range) noexcept;
        constexpr std::error_code remove_word(const Word* word, bool remove_tokens = false) noexcept;
        constexpr void remove_all_words(bool remove_tokens = false) noexcept;
        [[nodiscard]] constexpr auto metadata() const noexcept;
        [[nodiscard]] constexpr std::size_t num_metadata_entries() const noexcept;
        [[nodiscard]] constexpr std::optional<std::string_view> get_metadata(std::string_view key) const noexcept;
        [[nodiscard]] constexpr std::error_code set_metadata(std::string_view key,
                                                             std::string_view value = std::string_view{}) noexcept;
        [[nodiscard]] constexpr std::error_code set_metadata_from_comment_line(std::string_view line) noexcept;
        constexpr std::error_code unset_metadata(std::string_view key) noexcept;
        constexpr void remove_all_metadata() noexcept;
        [[nodiscard]] constexpr bool empty() const noexcept;
        constexpr void clear() noexcept;
        [[nodiscard]] constexpr std::error_code validate() const noexcept;
        constexpr ~Sentence() noexcept = default;
    };
} // namespace tg::conllu
 
#endif // TG_CONLLU_SENTENCE_HPP