format_parse.hpp

Go to the documentation of this file.

// -----------------------------------------------------------------------------------------------------
// Copyright (c) 2006-2021, Knut Reinert & Freie Universität Berlin
// Copyright (c) 2016-2021, Knut Reinert & MPI für molekulare Genetik
// This file may be used, modified and/or redistributed under the terms of the 3-clause BSD-License
// shipped with this file and also available at: https://github.com/seqan/seqan3/blob/master/LICENSE.md
// -----------------------------------------------------------------------------------------------------
 
#pragma once
 
#include <seqan3/std/charconv>
#include <seqan3/std/concepts>
#include <sstream>
#include <string>
#include <vector>
 
#include <seqan3/argument_parser/detail/format_base.hpp>
#include <seqan3/utility/char_operations/predicate.hpp>
#include <seqan3/utility/detail/type_name_as_string.hpp>
 
namespace seqan3::detail
{
 
class format_parse : public format_base
{
public:
    format_parse() = delete;                                     
    format_parse(format_parse const & pf) = default;             
    format_parse & operator=(format_parse const & pf) = default; 
    format_parse(format_parse &&) = default;                     
    format_parse & operator=(format_parse &&) = default;         
    ~format_parse() = default;                                   
 
    format_parse(int const argc_, std::vector<std::string> argv_) :
        argc{argc_ - 1}, argv{std::move(argv_)}
    {}
 
    template <typename option_type, typename validator_type>
    void add_option(option_type & value,
                    char const short_id,
                    std::string const & long_id,
                    std::string const & SEQAN3_DOXYGEN_ONLY(desc),
                    option_spec const spec,
                    validator_type && option_validator)
    {
        option_calls.push_back([this, &value, short_id, long_id, spec, option_validator]()
        {
            get_option(value, short_id, long_id, spec, option_validator);
        });
    }
 
    void add_flag(bool & value,
                  char const short_id,
                  std::string const & long_id,
                  std::string const & SEQAN3_DOXYGEN_ONLY(desc),
                  option_spec const & SEQAN3_DOXYGEN_ONLY(spec))
    {
        flag_calls.push_back([this, &value, short_id, long_id]()
        {
            get_flag(value, short_id, long_id);
        });
    }
 
    template <typename option_type, typename validator_type>
    void add_positional_option(option_type & value,
                               std::string const & SEQAN3_DOXYGEN_ONLY(desc),
                               validator_type && option_validator)
    {
        positional_option_calls.push_back([this, &value, option_validator]()
        {
            get_positional_option(value, option_validator);
        });
    }
 
    void parse(argument_parser_meta_data const & /*meta*/)
    {
        end_of_options_it = std::find(argv.begin(), argv.end(), "--");
 
        // parse options first, because we need to rule out -keyValue pairs
        // (e.g. -AnoSpaceAfterIdentifierA) before parsing flags
        for (auto && f : option_calls)
            f();
 
        for (auto && f : flag_calls)
            f();
 
        check_for_unknown_ids();
 
        if (end_of_options_it != argv.end())
            *end_of_options_it = ""; // remove -- before parsing positional arguments
 
        for (auto && f : positional_option_calls)
            f();
 
        check_for_left_over_args();
    }
 
    // functions are not needed for command line parsing but are part of the format interface.
    void add_section(std::string const &, option_spec const) {}
    void add_subsection(std::string const &, option_spec const) {}
    void add_line(std::string const &, bool, option_spec const) {}
    void add_list_item(std::string const &, std::string const &, option_spec const) {}
 
    template <typename id_type>
    static bool is_empty_id(id_type const & id)
    {
        if constexpr (std::same_as<std::remove_cvref_t<id_type>, std::string>)
            return id.empty();
        else // char
            return is_char<'\0'>(id);
    }
 
    template <typename iterator_type, typename id_type>
    static iterator_type find_option_id(iterator_type begin_it, iterator_type end_it, id_type const & id)
    {
        if (is_empty_id(id))
            return end_it;
 
        return (std::find_if(begin_it, end_it,
            [&] (std::string const & current_arg)
            {
                std::string full_id = prepend_dash(id);
 
                if constexpr (std::same_as<id_type, char>) // short id
                {
                    // check if current_arg starts with "-o", i.e. it correctly identifies all short notations:
                    // "-ovalue", "-o=value", and "-o value".
                    return current_arg.substr(0, full_id.size()) == full_id;
                }
                else
                {
                    // only "--opt Value" or "--opt=Value" are valid
                    return current_arg.substr(0, full_id.size()) == full_id && // prefix is the same
                           (current_arg.size() == full_id.size() || current_arg[full_id.size()] == '='); // space or `=`
                }
            }));
    }
 
private:
    enum class option_parse_result
    {
        success, 
        error, 
        overflow_error 
    };
 
    static std::string prepend_dash(std::string const & long_id)
    {
        return {"--" + long_id};
    }
 
    static std::string prepend_dash(char const short_id)
    {
        return {"-" + std::string{short_id}};
    }
 
    std::string combine_option_names(char const short_id, std::string const & long_id)
    {
        if (short_id == '\0')
            return prepend_dash(long_id);
        else if (long_id.empty())
            return prepend_dash(short_id);
        else // both are set (note: both cannot be empty, this is caught before)
            return prepend_dash(short_id) + "/" + prepend_dash(long_id);
    }
 
    bool flag_is_set(std::string const & long_id)
    {
        auto it = std::find(argv.begin(), end_of_options_it, prepend_dash(long_id));
 
        if (it != end_of_options_it)
            *it = ""; // remove seen flag
 
        return(it != end_of_options_it);
    }
 
    bool flag_is_set(char const short_id)
    {
        // short flags need special attention, since they could be grouped (-rGv <=> -r -G -v)
        for (std::string & arg : argv)
        {
            if (arg[0] == '-' && arg.size() > 1 && arg[1] != '-') // is option && not dash && no long option
            {
                auto pos = arg.find(short_id);
 
                if (pos != std::string::npos)
                {
                    arg.erase(pos, 1); // remove seen bool
 
                    if (arg == "-") // if flag is empty now
                        arg = "";
 
                    return true;
                }
            }
        }
        return false;
    }
 
    template <typename option_t>
        requires input_stream_over<std::istringstream, option_t>
    option_parse_result parse_option_value(option_t & value, std::string const & in)
    {
        std::istringstream stream{in};
        stream >> value;
 
        if (stream.fail() || !stream.eof())
            return option_parse_result::error;
 
        return option_parse_result::success;
    }
 
    template <named_enumeration option_t>
    option_parse_result parse_option_value(option_t & value, std::string const & in)
    {
        auto map = seqan3::enumeration_names<option_t>;
 
        if (auto it = map.find(in); it == map.end())
        {
            std::vector<std::pair<std::string_view, option_t>> key_value_pairs(map.begin(), map.end());
            std::ranges::sort(key_value_pairs, [] (auto pair1, auto pair2)
                {
                    if constexpr (std::totally_ordered<option_t>)
                    {
                        if (pair1.second != pair2.second)
                            return pair1.second < pair2.second;
                    }
                    return pair1.first < pair2.first;
                });
 
            throw user_input_error{detail::to_string("You have chosen an invalid input value: ", in,
                                                     ". Please use one of: ", key_value_pairs | std::views::keys)};
        }
        else
        {
            value = it->second;
        }
 
        return option_parse_result::success;
    }
 
    option_parse_result parse_option_value(std::string & value, std::string const & in)
    {
        value = in;
        return option_parse_result::success;
    }
 
    template <sequence_container container_option_t>
        requires requires (format_parse fp, typename container_option_t::value_type & container_value, std::string const & in)
        {
            SEQAN3_RETURN_TYPE_CONSTRAINT(fp.parse_option_value(container_value, in), std::same_as, option_parse_result);
        }
    option_parse_result parse_option_value(container_option_t & value, std::string const & in)
    {
        typename container_option_t::value_type tmp{};
 
        auto res = parse_option_value(tmp, in);
 
        if (res == option_parse_result::success)
            value.push_back(tmp);
 
        return res;
    }
 
    template <arithmetic option_t>
        requires input_stream_over<std::istringstream, option_t>
    option_parse_result parse_option_value(option_t & value, std::string const & in)
    {
        auto res = std::from_chars(&in[0], &in[in.size()], value);
 
        if (res.ec == std::errc::result_out_of_range)
            return option_parse_result::overflow_error;
        else if (res.ec == std::errc::invalid_argument || res.ptr != &in[in.size()])
            return option_parse_result::error;
 
        return option_parse_result::success;
    }
 
    option_parse_result parse_option_value(bool & value, std::string const & in)
    {
        if (in == "0")
            value = false;
        else if (in == "1")
            value = true;
        else if (in == "true")
            value = true;
        else if (in == "false")
            value = false;
        else
            return option_parse_result::error;
 
        return option_parse_result::success;
    }
 
    template <typename option_type>
    void throw_on_input_error(option_parse_result const res,
                              std::string const & option_name,
                              std::string const & input_value)
    {
        std::string msg{"Value parse failed for " + option_name + ": "};
 
        if (res == option_parse_result::error)
        {
            throw user_input_error{msg + "Argument " + input_value + " could not be parsed as type " +
                                   get_type_name_as_string(option_type{}) + "."};
        }
 
        if constexpr (arithmetic<option_type>)
        {
            if (res == option_parse_result::overflow_error)
            {
                throw user_input_error{msg + "Numeric argument " + input_value + " is not in the valid range [" +
                                       std::to_string(std::numeric_limits<option_type>::min()) + "," +
                                       std::to_string(std::numeric_limits<option_type>::max()) + "]."};
            }
        }
 
        assert(res == option_parse_result::success); // if nothing was thrown, the result must have been a success
    }
 
    template <typename option_type, typename id_type>
    bool identify_and_retrieve_option_value(option_type & value,
                                            std::vector<std::string>::iterator & option_it,
                                            id_type const & id)
    {
        if (option_it != end_of_options_it)
        {
            std::string input_value;
            size_t id_size = (prepend_dash(id)).size();
 
            if ((*option_it).size() > id_size) // identifier includes value (-keyValue or -key=value)
            {
                if ((*option_it)[id_size] == '=') // -key=value
                {
                    if ((*option_it).size() == id_size + 1) // malformed because no value follows '-i='
                        throw too_few_arguments("Missing value for option " + prepend_dash(id));
                    input_value = (*option_it).substr(id_size + 1);
                }
                else // -kevValue
                {
                    input_value = (*option_it).substr(id_size);
                }
 
                *option_it = ""; // remove used identifier-value pair
            }
            else // -key value
            {
                *option_it = ""; // remove used identifier
                ++option_it;
                if (option_it == end_of_options_it) // should not happen
                    throw too_few_arguments("Missing value for option " + prepend_dash(id));
                input_value = *option_it;
                *option_it = ""; // remove value
            }
 
            auto res = parse_option_value(value, input_value);
            throw_on_input_error<option_type>(res, prepend_dash(id), input_value);
 
            return true;
        }
        return false;
    }
 
    template <typename option_type, typename id_type>
    bool get_option_by_id(option_type & value, id_type const & id)
    {
        auto it = find_option_id(argv.begin(), end_of_options_it, id);
 
        if (it != end_of_options_it)
            identify_and_retrieve_option_value(value, it, id);
 
        if (find_option_id(it, end_of_options_it, id) != end_of_options_it) // should not be found again
           throw option_declared_multiple_times("Option " + prepend_dash(id) +
                                                " is no list/container but declared multiple times.");
 
       return (it != end_of_options_it); // first search was successful or not
    }
 
    template <sequence_container option_type, typename id_type>
        requires (!std::is_same_v<option_type, std::string>)
    bool get_option_by_id(option_type & value, id_type const & id)
    {
        auto it = find_option_id(argv.begin(), end_of_options_it, id);
        bool seen_at_least_once{it != end_of_options_it};
 
        if (seen_at_least_once)
            value.clear();
 
        while (it != end_of_options_it)
        {
            identify_and_retrieve_option_value(value, it, id);
            it = find_option_id(it, end_of_options_it, id);
        }
 
        return seen_at_least_once;
    }
 
    void check_for_unknown_ids()
    {
        for (auto it = argv.begin(); it != end_of_options_it; ++it)
        {
            std::string arg{*it};
            if (!arg.empty() && arg[0] == '-') // may be an identifier
            {
                if (arg == "-")
                {
                    continue; // positional option
                }
                else if (arg[1] != '-' && arg.size() > 2) // one dash, but more than one character (-> multiple flags)
                {
                    throw unknown_option("Unknown flags " + expand_multiple_flags(arg) +
                                         ". In case this is meant to be a non-option/argument/parameter, " +
                                         "please specify the start of arguments with '--'. " +
                                         "See -h/--help for program information.");
                }
                else // unknown short or long option
                {
                    throw unknown_option("Unknown option " + arg +
                                         ". In case this is meant to be a non-option/argument/parameter, " +
                                         "please specify the start of non-options with '--'. " +
                                         "See -h/--help for program information.");
                }
            }
        }
    }
 
    void check_for_left_over_args()
    {
        if (std::find_if(argv.begin(), argv.end(), [](std::string const & s){return (s != "");}) != argv.end())
            throw too_many_arguments("Too many arguments provided. Please see -h/--help for more information.");
    }
 
    template <typename option_type, typename validator_type>
    void get_option(option_type & value,
                     char const short_id,
                     std::string const & long_id,
                     option_spec const spec,
                     validator_type && validator)
    {
        bool short_id_is_set{get_option_by_id(value, short_id)};
        bool long_id_is_set{get_option_by_id(value, long_id)};
 
        // if value is no container we need to check for multiple declarations
        if (short_id_is_set && long_id_is_set &&
            !(sequence_container<option_type> && !std::is_same_v<option_type, std::string>))
            throw option_declared_multiple_times("Option " + combine_option_names(short_id, long_id) +
                                                 " is no list/container but specified multiple times");
 
        if (short_id_is_set || long_id_is_set)
        {
            try
            {
                validator(value);
            }
            catch (std::exception & ex)
            {
                throw validation_error(std::string("Validation failed for option ") +
                                        combine_option_names(short_id, long_id) + ": " + ex.what());
            }
        }
        else // option is not set
        {
            // check if option is required
            if (spec & option_spec::required)
                throw required_option_missing("Option " + combine_option_names(short_id, long_id) +
                                              " is required but not set.");
        }
    }
 
    void get_flag(bool & value,
                  char const short_id,
                  std::string const & long_id)
    {
        value = flag_is_set(short_id) || flag_is_set(long_id);
    }
 
    template <typename option_type, typename validator_type>
    void get_positional_option(option_type & value,
                               validator_type && validator)
    {
        ++positional_option_count;
        auto it = std::find_if(argv.begin(), argv.end(), [](std::string const & s){return (s != "");});
 
        if (it == argv.end())
            throw too_few_arguments("Not enough positional arguments provided (Need at least " +
                                    std::to_string(positional_option_calls.size()) +
                                    "). See -h/--help for more information.");
 
        if constexpr (sequence_container<option_type> && !std::is_same_v<option_type, std::string>) // vector/list will be filled with all remaining arguments
        {
            assert(positional_option_count == positional_option_calls.size()); // checked on set up.
 
            value.clear();
 
            while (it != argv.end())
            {
                auto res = parse_option_value(value, *it);
                std::string id = "positional option" + std::to_string(positional_option_count);
                throw_on_input_error<option_type>(res, id, *it);
 
                *it = ""; // remove arg from argv
                it = std::find_if(it, argv.end(), [](std::string const & s){return (s != "");});
                ++positional_option_count;
            }
        }
        else
        {
            auto res = parse_option_value(value, *it);
            std::string id = "positional option" + std::to_string(positional_option_count);
            throw_on_input_error<option_type>(res, id, *it);
 
            *it = ""; // remove arg from argv
        }
 
        try
        {
            validator(value);
        }
        catch (std::exception & ex)
        {
            throw validation_error("Validation failed for positional option " +
                                    std::to_string(positional_option_count) + ": " + ex.what());
        }
    }
 
    std::vector<std::function<void()>> option_calls;
    std::vector<std::function<void()>> flag_calls;
    std::vector<std::function<void()>> positional_option_calls;
    unsigned positional_option_count{0};
    int argc;
    std::vector<std::string> argv;
    std::vector<std::string>::iterator end_of_options_it;
};
 
} // namespace seqan3