#ifndef regex_hh_INCLUDED #define regex_hh_INCLUDED #include "string.hh" #include "exception.hh" #include "utf8_iterator.hh" #include namespace Kakoune { struct regex_error : runtime_error { regex_error(StringView desc) : runtime_error{format("regex error: '{}'", desc)} {} }; using RegexBase = boost::basic_regex>; // Regex that keeps track of its string representation struct Regex : RegexBase { Regex() = default; explicit Regex(StringView re, flag_type flags = ECMAScript); bool empty() const { return m_str.empty(); } bool operator==(const Regex& other) const { return m_str == other.m_str; } bool operator!=(const Regex& other) const { return m_str != other.m_str; } const String& str() const { return m_str; } static constexpr const char* option_type_name = "regex"; private: String m_str; }; template using RegexUtf8It = utf8::iterator; template using RegexIteratorBase = boost::regex_iterator, wchar_t, boost::c_regex_traits>; namespace RegexConstant = boost::regex_constants; template struct MatchResults : boost::match_results> { using ParentType = boost::match_results>; struct SubMatch : std::pair { SubMatch() = default; SubMatch(const boost::sub_match>& m) : std::pair{m.first.base(), m.second.base()}, matched{m.matched} {} bool matched = false; }; struct iterator : boost::match_results>::iterator { using ParentType = typename boost::match_results>::iterator; iterator(const ParentType& it) : ParentType(it) {} SubMatch operator*() const { return {ParentType::operator*()}; } }; iterator begin() const { return {ParentType::begin()}; } iterator cbegin() const { return {ParentType::cbegin()}; } iterator end() const { return {ParentType::end()}; } iterator cend() const { return {ParentType::cend()}; } SubMatch operator[](size_t s) const { return {ParentType::operator[](s)}; } }; template struct RegexIterator : RegexIteratorBase { using Utf8It = RegexUtf8It; using ValueType = MatchResults; RegexIterator() = default; RegexIterator(Iterator begin, Iterator end, const Regex& re, RegexConstant::match_flag_type flags = RegexConstant::match_default) : RegexIteratorBase{Utf8It{begin, begin, end}, Utf8It{end, begin, end}, re, flags} {} const ValueType& operator*() const { return *reinterpret_cast(&RegexIteratorBase::operator*()); } const ValueType* operator->() const { return reinterpret_cast(RegexIteratorBase::operator->()); } }; inline RegexConstant::match_flag_type match_flags(bool bol, bool eol, bool bow, bool eow) { return (bol ? RegexConstant::match_default : RegexConstant::match_not_bol) | (eol ? RegexConstant::match_default : RegexConstant::match_not_eol) | (bow ? RegexConstant::match_default : RegexConstant::match_not_bow) | (eow ? RegexConstant::match_default : RegexConstant::match_not_eow); } template bool regex_match(It begin, It end, const Regex& re) { using Utf8It = RegexUtf8It; return boost::regex_match(Utf8It{begin, begin, end}, Utf8It{end, begin, end}, re); } template bool regex_match(It begin, It end, MatchResults& res, const Regex& re) { using Utf8It = RegexUtf8It; return boost::regex_match(Utf8It{begin, begin, end}, Utf8It{end, begin, end}, res, re); } template bool regex_search(It begin, It end, const Regex& re, RegexConstant::match_flag_type flags = RegexConstant::match_default) { using Utf8It = RegexUtf8It; return boost::regex_search(Utf8It{begin, begin, end}, Utf8It{end, begin, end}, re); } template bool regex_search(It begin, It end, MatchResults& res, const Regex& re, RegexConstant::match_flag_type flags = RegexConstant::match_default) { using Utf8It = RegexUtf8It; return boost::regex_search(Utf8It{begin, begin, end}, Utf8It{end, begin, end}, res, re); } String option_to_string(const Regex& re); void option_from_string(StringView str, Regex& re); } #endif // regex_hh_INCLUDED