1049 lines
36 KiB
C++
1049 lines
36 KiB
C++
|
|
#pragma once
|
|
|
|
#if !defined(MIJIN_UTIL_STRING_HPP_INCLUDED)
|
|
#define MIJIN_UTIL_STRING_HPP_INCLUDED 1
|
|
|
|
#include <algorithm>
|
|
#include <array>
|
|
#include <charconv>
|
|
#include <climits>
|
|
#include <cstdlib>
|
|
#include <cstring>
|
|
#include <iterator>
|
|
#include <limits>
|
|
#include <locale>
|
|
#include <sstream>
|
|
#include <string>
|
|
#include <string_view>
|
|
#include <vector>
|
|
|
|
#include "./iterators.hpp"
|
|
#include "../internal/common.hpp"
|
|
#include "../util/traits.hpp"
|
|
|
|
namespace mijin
|
|
{
|
|
|
|
//
|
|
// public defines
|
|
//
|
|
|
|
//
|
|
// public constants
|
|
//
|
|
|
|
namespace detail
|
|
{
|
|
template<typename TChar>
|
|
static constexpr std::array DEFAULT_TRIM_CHARS_DATA = {TChar(' '), TChar('\t'), TChar('\r'), TChar('\n')};
|
|
}
|
|
|
|
template<typename TChar>
|
|
static const std::basic_string_view<TChar, std::char_traits<TChar>> DEFAULT_TRIM_CHARS
|
|
= {detail::DEFAULT_TRIM_CHARS_DATA<TChar>.begin(), detail::DEFAULT_TRIM_CHARS_DATA<TChar>.end()};
|
|
|
|
//
|
|
// public traits
|
|
//
|
|
|
|
template<typename T>
|
|
inline constexpr bool is_string_v = is_template_instance_v<std::basic_string, std::remove_cvref_t<T>>;
|
|
|
|
template<typename T>
|
|
concept std_string_type = is_string_v<T>;
|
|
|
|
template<typename T>
|
|
inline constexpr bool is_string_view_v = is_template_instance_v<std::basic_string_view, std::remove_cvref_t<T>>;
|
|
|
|
template<typename T>
|
|
concept std_string_view_type = is_string_view_v<T>;
|
|
|
|
template<typename T>
|
|
inline constexpr bool is_char_v = is_any_type_v<std::remove_cvref_t<T>, char, wchar_t, char8_t, char16_t, char32_t>;
|
|
|
|
template<typename T>
|
|
concept char_type = is_char_v<T>;
|
|
|
|
template<typename T>
|
|
inline constexpr bool is_cstring_v = std::is_pointer_v<T> && is_char_v<std::remove_pointer_t<T>>;
|
|
|
|
template<typename T>
|
|
concept cstring_type = is_cstring_v<T>;
|
|
|
|
template<typename T>
|
|
struct str_char_type
|
|
{
|
|
using type = void;
|
|
};
|
|
|
|
template<char_type T>
|
|
struct str_char_type<T*>
|
|
{
|
|
using type = std::remove_cvref_t<T>;
|
|
};
|
|
|
|
template<std_string_view_type T>
|
|
struct str_char_type<T>
|
|
{
|
|
using type = typename std::remove_cvref_t<T>::value_type;
|
|
};
|
|
|
|
template<std_string_type T>
|
|
struct str_char_type<T>
|
|
{
|
|
using type = typename std::remove_cvref_t<T>::value_type;
|
|
};
|
|
|
|
template<typename T>
|
|
using str_char_type_t = str_char_type<T>::type;
|
|
|
|
//
|
|
// public types
|
|
//
|
|
|
|
struct SplitOptions
|
|
{
|
|
std::size_t limitParts = std::numeric_limits<std::size_t>::max();
|
|
bool ignoreEmpty = true;
|
|
};
|
|
|
|
struct [[nodiscard]] ConvertCharTypeResult
|
|
{
|
|
unsigned numRead = 0;
|
|
unsigned numWritten = 0;
|
|
|
|
constexpr operator bool() const MIJIN_NOEXCEPT
|
|
{
|
|
return numRead != 0 || numWritten != 0;
|
|
}
|
|
constexpr bool operator !() const MIJIN_NOEXCEPT
|
|
{
|
|
return !static_cast<bool>(*this);
|
|
}
|
|
};
|
|
|
|
struct SplitViewOptions
|
|
{
|
|
bool ignoreEmpty = true;
|
|
bool trim = false;
|
|
};
|
|
|
|
template<typename TChar, TChar splitAt, SplitViewOptions options = SplitViewOptions(), typename TCharTraits = std::char_traits<TChar>>
|
|
struct SplitStringTraitsCT
|
|
{
|
|
using char_t = TChar;
|
|
using string_view_t = std::basic_string_view<TChar, TCharTraits>;
|
|
|
|
static constexpr char_t getSplitAt() MIJIN_NOEXCEPT { return splitAt; }
|
|
static constexpr bool getIgnoreEmpty() MIJIN_NOEXCEPT { return options.ignoreEmpty; }
|
|
static constexpr bool getTrim() MIJIN_NOEXCEPT { return options.trim; }
|
|
static constexpr auto getTrimChars() MIJIN_NOEXCEPT { return DEFAULT_TRIM_CHARS<char_t>; }
|
|
};
|
|
|
|
template<typename TChar, typename TCharTraits = std::char_traits<TChar>>
|
|
struct SplitStringTraitsRT
|
|
{
|
|
using char_t = TChar;
|
|
using string_view_t = std::basic_string_view<TChar, TCharTraits>;
|
|
|
|
char_t splitAt;
|
|
bool ignoreEmpty;
|
|
string_view_t trimChars = {};
|
|
|
|
constexpr char_t getSplitAt() const MIJIN_NOEXCEPT { return splitAt; }
|
|
constexpr bool getIgnoreEmpty() const MIJIN_NOEXCEPT { return ignoreEmpty; }
|
|
constexpr bool getTrim() const MIJIN_NOEXCEPT { return !trimChars.empty(); }
|
|
constexpr string_view_t getTrimChars() const MIJIN_NOEXCEPT { return trimChars; }
|
|
};
|
|
|
|
template<typename T, typename TChar>
|
|
concept SplitStringTraitsType = std::is_copy_constructible_v<T> && requires(const T& object)
|
|
{
|
|
typename T::char_t;
|
|
typename T::string_view_t;
|
|
{ object.getSplitAt() } -> std::convertible_to<TChar>;
|
|
{ object.getIgnoreEmpty() } -> std::convertible_to<bool>;
|
|
{ object.getTrim() } -> std::convertible_to<bool>;
|
|
{ object.getTrimChars() } -> std::convertible_to<typename T::string_view_t>;
|
|
};
|
|
static_assert(SplitStringTraitsType<SplitStringTraitsCT<char, ' '>, char>);
|
|
static_assert(SplitStringTraitsType<SplitStringTraitsRT<char>, char>);
|
|
|
|
template<typename TChar, typename TLine, SplitViewOptions options = SplitViewOptions(), typename TCharTraits = std::char_traits<TChar>>
|
|
struct SplitLineTraitsCT : SplitStringTraitsCT<TChar, '\n', options, TCharTraits>
|
|
{
|
|
using base_t = SplitStringTraitsCT<TChar, '\n', options, TCharTraits>;
|
|
using char_t = TChar;
|
|
using line_t = TLine;
|
|
|
|
line_t line = 1;
|
|
|
|
using base_t::getSplitAt;
|
|
using base_t::getIgnoreEmpty;
|
|
using base_t::getTrim;
|
|
using base_t::getTrimChars;
|
|
constexpr void onNext() MIJIN_NOEXCEPT {
|
|
++line;
|
|
}
|
|
constexpr line_t getLine() const MIJIN_NOEXCEPT { return line; }
|
|
};
|
|
|
|
template<typename TChar, typename TLine, typename TCharTraits = std::char_traits<TChar>>
|
|
struct SplitLineTraitsRT
|
|
{
|
|
using char_t = TChar;
|
|
using line_t = TLine;
|
|
using string_view_t = std::basic_string_view<TChar, TCharTraits>;
|
|
|
|
line_t line = 1;
|
|
bool ignoreEmpty;
|
|
string_view_t trimChars = {};
|
|
|
|
constexpr char_t getSplitAt() const MIJIN_NOEXCEPT { return '\n'; }
|
|
constexpr bool getIgnoreEmpty() const MIJIN_NOEXCEPT { return ignoreEmpty; }
|
|
constexpr bool getTrim() const MIJIN_NOEXCEPT { return !trimChars.empty(); }
|
|
constexpr string_view_t getTrimChars() const MIJIN_NOEXCEPT { return trimChars; }
|
|
constexpr line_t getLine() const MIJIN_NOEXCEPT { return line; }
|
|
constexpr void onNext() MIJIN_NOEXCEPT {
|
|
++line;
|
|
}
|
|
};
|
|
|
|
template<typename T, typename TChar, typename TLine>
|
|
concept SplitLineTraitsType = SplitStringTraitsType<T, TChar> && requires (const T& object)
|
|
{
|
|
{ object.getLine() } -> std::convertible_to<TLine>;
|
|
};
|
|
static_assert(SplitLineTraitsType<SplitLineTraitsCT<char, unsigned>, char, unsigned>);
|
|
static_assert(SplitLineTraitsType<SplitLineTraitsRT<char, unsigned>, char, unsigned>);
|
|
|
|
template<typename TString, typename TChars>
|
|
[[nodiscard]]
|
|
auto trim(TString&& string, TChars&& chars);
|
|
|
|
template<typename TChar, SplitStringTraitsType<TChar> TTraits>
|
|
class SplitStringIterator
|
|
{
|
|
public:
|
|
using char_t = TChar;
|
|
using traits_t = TTraits;
|
|
using string_view_t = traits_t::string_view_t;
|
|
using base_t = string_view_t::iterator;
|
|
using value_type = string_view_t;
|
|
private:
|
|
[[no_unique_address]] traits_t traits_;
|
|
|
|
string_view_t full_;
|
|
string_view_t::iterator pos_;
|
|
string_view_t::iterator next_;
|
|
|
|
public:
|
|
constexpr SplitStringIterator(string_view_t full, base_t pos, traits_t traits = {}) MIJIN_NOEXCEPT_IF(std::is_nothrow_move_constructible_v<traits_t>)
|
|
: full_(full), pos_(pos), traits_(std::move(traits))
|
|
{
|
|
findNext();
|
|
}
|
|
constexpr explicit SplitStringIterator(traits_t traits = {}) MIJIN_NOEXCEPT_IF(std::is_nothrow_move_constructible_v<traits_t>)
|
|
: traits_(std::move(traits)) {}
|
|
constexpr SplitStringIterator(const SplitStringIterator&) noexcept(std::is_nothrow_copy_constructible_v<traits_t>) = default;
|
|
constexpr SplitStringIterator(SplitStringIterator&&) noexcept(std::is_nothrow_move_constructible_v<traits_t>) = default;
|
|
constexpr SplitStringIterator& operator=(const SplitStringIterator&) noexcept(std::is_nothrow_copy_assignable_v<traits_t>) = default;
|
|
constexpr SplitStringIterator& operator=(SplitStringIterator&&) noexcept(std::is_nothrow_move_assignable_v<traits_t>) = default;
|
|
|
|
constexpr bool operator==(const SplitStringIterator& other) const MIJIN_NOEXCEPT { MIJIN_ASSERT(full_ == other.full_, "Comparing unrelated iterators."); return pos_ == other.pos_; }
|
|
constexpr bool operator!=(const SplitStringIterator& other) const MIJIN_NOEXCEPT { MIJIN_ASSERT(full_ == other.full_, "Comparing unrelated iterators."); return pos_ != other.pos_; }
|
|
constexpr bool operator<(const SplitStringIterator& other) const MIJIN_NOEXCEPT { MIJIN_ASSERT(full_ == other.full_, "Comparing unrelated iterators."); return pos_ < other.pos_; }
|
|
constexpr bool operator<=(const SplitStringIterator& other) const MIJIN_NOEXCEPT { MIJIN_ASSERT(full_ == other.full_, "Comparing unrelated iterators."); return pos_ <= other.pos_; }
|
|
constexpr bool operator>(const SplitStringIterator& other) const MIJIN_NOEXCEPT { MIJIN_ASSERT(full_ == other.full_, "Comparing unrelated iterators."); return pos_ > other.pos_; }
|
|
constexpr bool operator>=(const SplitStringIterator& other) const MIJIN_NOEXCEPT { MIJIN_ASSERT(full_ == other.full_, "Comparing unrelated iterators."); return pos_ >= other.pos_; }
|
|
|
|
[[nodiscard]]
|
|
traits_t& getTraits() MIJIN_NOEXCEPT
|
|
{
|
|
return traits_;
|
|
}
|
|
|
|
[[nodiscard]]
|
|
const traits_t& getTraits() const MIJIN_NOEXCEPT
|
|
{
|
|
return traits_;
|
|
}
|
|
|
|
constexpr value_type operator*() const MIJIN_NOEXCEPT
|
|
{
|
|
MIJIN_ASSERT(pos_ != full_.end(), "Dereferencing an invalid iterator.");
|
|
string_view_t result{pos_, next_};
|
|
if (traits_.getTrim())
|
|
{
|
|
result = trim(result, traits_.getTrimChars());
|
|
}
|
|
return result;
|
|
}
|
|
|
|
constexpr SplitStringIterator& operator++() MIJIN_NOEXCEPT
|
|
{
|
|
MIJIN_ASSERT(pos_ != full_.end(), "Iterating past end.");
|
|
if (next_ == full_.end()) {
|
|
pos_ = full_.end();
|
|
}
|
|
else
|
|
{
|
|
pos_ = std::next(next_);
|
|
findNext();
|
|
}
|
|
return *this;
|
|
}
|
|
|
|
constexpr SplitStringIterator operator++(int) const MIJIN_NOEXCEPT
|
|
{
|
|
SplitStringIterator copy(*this);
|
|
++copy;
|
|
return copy;
|
|
}
|
|
|
|
// TODO
|
|
// SplitStringIterator& operator--() MIJIN_NOEXCEPT
|
|
// {
|
|
// MIJIN_ASSERT(pos_ != full_.begin(), "Iterating past begin.");
|
|
// next_ = std::prev(pos_);
|
|
// pos_ = std::find(std::reverse_iterator(next_), std::reverse_iterator(full_.begin()), separator).base();
|
|
// }
|
|
private:
|
|
constexpr void findNext()
|
|
{
|
|
while (true)
|
|
{
|
|
if constexpr (requires{{ traits_.onNext() };}) {
|
|
traits_.onNext();
|
|
}
|
|
next_ = std::find(pos_, full_.end(), traits_.getSplitAt());
|
|
|
|
if (!traits_.getIgnoreEmpty() || pos_ == full_.end()) {
|
|
break;
|
|
}
|
|
if (traits_.getTrim())
|
|
{
|
|
const string_view_t trimChars = traits_.getTrimChars();
|
|
typename string_view_t::iterator trimmedPos = std::find_if(pos_, next_, [&](char_t chr)
|
|
{
|
|
return !trimChars.contains(chr);
|
|
});
|
|
if (trimmedPos == next_)
|
|
{
|
|
pos_ = next_; // skip this part
|
|
}
|
|
}
|
|
|
|
if (pos_ != next_) {
|
|
break;
|
|
}
|
|
pos_ = std::next(pos_);
|
|
}
|
|
}
|
|
};
|
|
|
|
template<typename TChar, SplitStringTraitsType<TChar> TTraits>
|
|
class SplitStringRange
|
|
{
|
|
public:
|
|
using char_t = TChar;
|
|
using traits_t = TTraits;
|
|
using string_view_t = traits_t::string_view_t;
|
|
using iterator = SplitStringIterator<char_t, traits_t>;
|
|
private:
|
|
[[no_unique_address]] traits_t traits_;
|
|
string_view_t stringView_;
|
|
public:
|
|
constexpr explicit SplitStringRange(string_view_t stringView, traits_t traits = {}) MIJIN_NOEXCEPT_IF(std::is_nothrow_move_constructible_v<traits_t>)
|
|
: stringView_(stringView), traits_(std::move(traits)) {}
|
|
constexpr SplitStringRange(const SplitStringRange&) noexcept(std::is_nothrow_copy_constructible_v<traits_t>) = default;
|
|
constexpr SplitStringRange(SplitStringRange&&) noexcept(std::is_nothrow_move_constructible_v<traits_t>) = default;
|
|
constexpr SplitStringRange& operator=(const SplitStringRange&) noexcept(std::is_nothrow_copy_assignable_v<traits_t>) = default;
|
|
constexpr SplitStringRange& operator=(SplitStringRange&&) noexcept(std::is_nothrow_move_assignable_v<traits_t>) = default;
|
|
constexpr auto operator<=>(const SplitStringRange&) const noexcept = default;
|
|
|
|
constexpr iterator begin() const MIJIN_NOEXCEPT_IF(std::is_nothrow_copy_constructible_v<traits_t>)
|
|
{
|
|
return iterator(stringView_, stringView_.begin(), traits_);
|
|
}
|
|
|
|
constexpr iterator end() const MIJIN_NOEXCEPT_IF(std::is_nothrow_copy_constructible_v<traits_t>)
|
|
{
|
|
return iterator(stringView_, stringView_.end(), traits_);
|
|
}
|
|
};
|
|
|
|
template<typename TChar, typename TLine = unsigned, SplitLineTraitsType<TChar, TLine> TTraits = SplitLineTraitsCT<TChar, TLine>>
|
|
class LineIterator
|
|
{
|
|
public:
|
|
using char_t = TChar;
|
|
using line_t = TLine;
|
|
using traits_t = TTraits;
|
|
using base_t = SplitStringIterator<TChar, traits_t>;
|
|
using string_view_t = base_t::string_view_t;
|
|
using value_type = std::pair<string_view_t, line_t>;
|
|
private:
|
|
base_t base_ = {};
|
|
public:
|
|
constexpr LineIterator(string_view_t full, string_view_t::iterator pos, traits_t traits = {}) MIJIN_NOEXCEPT_IF(std::is_nothrow_move_constructible_v<traits_t>)
|
|
: base_(full, pos, std::move(traits)) {}
|
|
constexpr explicit LineIterator(traits_t traits = {}) MIJIN_NOEXCEPT_IF(std::is_nothrow_move_constructible_v<traits_t>)
|
|
: base_(std::move(traits)) {}
|
|
LineIterator(const LineIterator&) noexcept = default;
|
|
LineIterator(LineIterator&&) noexcept = default;
|
|
|
|
LineIterator& operator=(const LineIterator&) noexcept = default;
|
|
LineIterator& operator=(LineIterator&&) noexcept = default;
|
|
|
|
bool operator==(const LineIterator& other) const MIJIN_NOEXCEPT { return base_ == other.base_; }
|
|
bool operator!=(const LineIterator& other) const MIJIN_NOEXCEPT { return base_ != other.base_; }
|
|
bool operator<(const LineIterator& other) const MIJIN_NOEXCEPT { return base_ < other.base_; }
|
|
bool operator>(const LineIterator& other) const MIJIN_NOEXCEPT { return base_ > other.base_; }
|
|
bool operator<=(const LineIterator& other) const MIJIN_NOEXCEPT { return base_ <= other.base_; }
|
|
bool operator>=(const LineIterator& other) const MIJIN_NOEXCEPT { return base_ >= other.base_; }
|
|
|
|
constexpr value_type operator*() const MIJIN_NOEXCEPT
|
|
{
|
|
string_view_t stringView = *base_;
|
|
if (!base_.getTraits().getTrim())
|
|
{
|
|
// always split \r, even if not trimming other whitespace
|
|
if (stringView.ends_with('\r')) {
|
|
stringView = stringView.substr(0, stringView.size() - 1);
|
|
}
|
|
}
|
|
return {stringView, base_.getTraits().line};
|
|
}
|
|
|
|
constexpr LineIterator& operator++() MIJIN_NOEXCEPT
|
|
{
|
|
++base_;
|
|
return *this;
|
|
}
|
|
|
|
constexpr LineIterator operator++(int) const MIJIN_NOEXCEPT
|
|
{
|
|
SplitStringIterator copy(*this);
|
|
++copy;
|
|
return copy;
|
|
}
|
|
};
|
|
|
|
template<typename TChar, typename TLine = unsigned, SplitLineTraitsType<TChar, TLine> TTraits = SplitLineTraitsCT<TChar, TLine>>
|
|
class LineRange
|
|
{
|
|
public:
|
|
using char_t = TChar;
|
|
using line_t = TLine;
|
|
using traits_t = TTraits;
|
|
using iterator = LineIterator<char_t, line_t, traits_t>;
|
|
using string_view_t = iterator::string_view_t;
|
|
private:
|
|
[[no_unique_address]] traits_t traits_;
|
|
string_view_t stringView_;
|
|
public:
|
|
constexpr explicit LineRange(string_view_t stringView, traits_t traits = {}) MIJIN_NOEXCEPT_IF(std::is_nothrow_move_constructible_v<traits_t>)
|
|
: traits_(std::move(traits)), stringView_(stringView) {}
|
|
constexpr LineRange(const LineRange&) noexcept(std::is_nothrow_copy_constructible_v<traits_t>) = default;
|
|
constexpr LineRange(LineRange&&) noexcept(std::is_nothrow_move_constructible_v<traits_t>) = default;
|
|
constexpr LineRange& operator=(const LineRange&) noexcept(std::is_nothrow_copy_assignable_v<traits_t>) = default;
|
|
constexpr LineRange& operator=(LineRange&&) noexcept(std::is_nothrow_move_assignable_v<traits_t>) = default;
|
|
constexpr auto operator<=>(const LineRange&) const noexcept = default;
|
|
|
|
constexpr iterator begin() const MIJIN_NOEXCEPT
|
|
{
|
|
return iterator(stringView_, stringView_.begin(), traits_);
|
|
}
|
|
|
|
constexpr iterator end() const MIJIN_NOEXCEPT
|
|
{
|
|
return iterator(stringView_, stringView_.end(), traits_);
|
|
}
|
|
};
|
|
|
|
//
|
|
// public functions
|
|
//
|
|
|
|
template <typename TRange, typename TValue = typename TRange::value_type>
|
|
[[nodiscard]] std::string join(const TRange& elements, const char* const delimiter)
|
|
{
|
|
std::ostringstream oss;
|
|
auto first = std::begin(elements);
|
|
auto last = std::end(elements);
|
|
|
|
if (first != last)
|
|
{
|
|
std::copy(first, std::prev(last), std::ostream_iterator<TValue>(oss, delimiter));
|
|
first = std::prev(last);
|
|
}
|
|
if (first != last)
|
|
{
|
|
oss << *first;
|
|
}
|
|
|
|
return oss.str();
|
|
}
|
|
|
|
namespace detail
|
|
{
|
|
template<typename TChar, typename TTraits, typename TOutIterator>
|
|
void splitImpl(std::basic_string_view<TChar, TTraits> stringView,
|
|
std::basic_string_view<TChar, TTraits> separator,
|
|
const SplitOptions& options,
|
|
TOutIterator outIterator,
|
|
std::size_t& numEmitted)
|
|
{
|
|
using sv_t = std::basic_string_view<TChar, TTraits>;
|
|
|
|
MIJIN_ASSERT(options.limitParts > 0, "Cannot split to zero parts.");
|
|
MIJIN_ASSERT(!separator.empty(), "Separator cannot be empty.");
|
|
|
|
if (separator.empty())
|
|
{
|
|
return;
|
|
}
|
|
|
|
auto emit = [&](std::string_view result)
|
|
{
|
|
*outIterator = result;
|
|
++outIterator;
|
|
++numEmitted;
|
|
};
|
|
|
|
if (options.limitParts <= 1)
|
|
{
|
|
emit(stringView);
|
|
return;
|
|
}
|
|
auto start = stringView.begin();
|
|
auto pos = start;
|
|
const auto end = stringView.end();
|
|
|
|
auto seperatorFound = [&]()
|
|
{
|
|
return pos <= (end - separator.size()) && sv_t(pos, pos + separator.size()) == separator;
|
|
};
|
|
|
|
while (pos != end)
|
|
{
|
|
if (seperatorFound())
|
|
{
|
|
if (!options.ignoreEmpty || pos != start)
|
|
{
|
|
emit(std::string_view(start, pos));
|
|
}
|
|
start = pos = (pos + separator.size());
|
|
if (numEmitted == options.limitParts - 1)
|
|
{
|
|
if (options.ignoreEmpty)
|
|
{
|
|
while (seperatorFound())
|
|
{
|
|
pos += separator.size();
|
|
}
|
|
}
|
|
emit(std::string_view(pos, end));
|
|
return;
|
|
}
|
|
if (!options.ignoreEmpty && pos == end)
|
|
{
|
|
// skipped a separator at the very end, add an empty entry
|
|
emit("");
|
|
return;
|
|
}
|
|
}
|
|
else
|
|
{
|
|
++pos;
|
|
}
|
|
}
|
|
|
|
if (start != end)
|
|
{
|
|
emit(std::string_view(start, end));
|
|
}
|
|
}
|
|
|
|
template<typename TChar, typename TTraits>
|
|
std::vector<std::basic_string_view<TChar, TTraits>> splitImpl(std::basic_string_view<TChar, TTraits> stringView,
|
|
std::basic_string_view<TChar, TTraits> separator,
|
|
const SplitOptions& options)
|
|
{
|
|
std::vector<std::basic_string_view<TChar, TTraits>> result;
|
|
std::size_t numEmitted = 0;
|
|
splitImpl(stringView, separator, options, std::back_inserter(result), numEmitted);
|
|
return result;
|
|
}
|
|
|
|
|
|
template<std::size_t count, typename TChar, typename TTraits>
|
|
std::array<std::basic_string_view<TChar, TTraits>, count> splitFixedImpl(std::basic_string_view<TChar, TTraits> stringView,
|
|
std::basic_string_view<TChar, TTraits> separator,
|
|
SplitOptions options,
|
|
std::size_t* outNumResults)
|
|
{
|
|
options.limitParts = count;
|
|
|
|
std::array<std::basic_string_view<TChar, TTraits>, count> result;
|
|
std::size_t numEmitted = 0;
|
|
splitImpl(stringView, separator, options, result.begin(), numEmitted);
|
|
if (outNumResults != nullptr)
|
|
{
|
|
*outNumResults = numEmitted;
|
|
}
|
|
return result;
|
|
}
|
|
|
|
template<typename TChar, typename TTraitsA, typename TTraitsB>
|
|
bool equalsIgnoreCaseImpl(std::basic_string_view<TChar, TTraitsA> stringA, std::basic_string_view<TChar, TTraitsB> stringB) MIJIN_NOEXCEPT
|
|
{
|
|
if (stringA.size() != stringB.size())
|
|
{
|
|
return false;
|
|
}
|
|
|
|
for (const auto [charA, charB] : zip(stringA, stringB))
|
|
{
|
|
if (std::tolower(charA) != std::tolower(charB))
|
|
{
|
|
return false;
|
|
}
|
|
}
|
|
|
|
return true;
|
|
}
|
|
|
|
template<typename TChar, typename TTraits>
|
|
std::basic_string_view<TChar, TTraits> trimPrefixImpl(std::basic_string_view<TChar, TTraits> stringView,
|
|
std::basic_string_view<TChar, TTraits> charsToTrim)
|
|
{
|
|
stringView.remove_prefix(std::min(stringView.find_first_not_of(charsToTrim), stringView.size()));
|
|
return stringView;
|
|
}
|
|
|
|
template<typename TChar, typename TTraits>
|
|
std::basic_string_view<TChar, TTraits> trimSuffixImpl(std::basic_string_view<TChar, TTraits> stringView,
|
|
std::basic_string_view<TChar, TTraits> charsToTrim)
|
|
{
|
|
stringView.remove_suffix(stringView.size() - std::min(stringView.find_last_not_of(charsToTrim) + 1, stringView.size()));
|
|
return stringView;
|
|
}
|
|
|
|
template<typename TChar, typename TTraits>
|
|
std::basic_string_view<TChar, TTraits> trimImpl(std::basic_string_view<TChar, TTraits> stringView,
|
|
std::basic_string_view<TChar, TTraits> charsToTrim)
|
|
{
|
|
return trimPrefixImpl(trimSuffixImpl(stringView, charsToTrim), charsToTrim);
|
|
}
|
|
}
|
|
|
|
template<typename TLeft, typename TRight>
|
|
[[nodiscard]] auto split(TLeft&& stringView, TRight&& separator, const SplitOptions& options = {})
|
|
{
|
|
return detail::splitImpl(std::basic_string_view(std::forward<TLeft>(stringView)),
|
|
std::basic_string_view(std::forward<TRight>(separator)), options);
|
|
}
|
|
|
|
template<std::size_t count, typename TLeft, typename TRight>
|
|
[[nodiscard]] auto splitFixed(TLeft&& stringView, TRight&& separator, SplitOptions options = {}, std::size_t* outNumResults = nullptr)
|
|
{
|
|
return detail::splitFixedImpl<count>(std::basic_string_view(std::forward<TLeft>(stringView)),
|
|
std::basic_string_view(std::forward<TRight>(separator)), options, outNumResults);
|
|
}
|
|
|
|
template<typename TTraits, typename TChar> requires (SplitStringTraitsType<TTraits, TChar>)
|
|
[[nodiscard]] SplitStringRange<TChar, TTraits> splitView(typename TTraits::string_view_t stringView, TTraits traits = {}) MIJIN_NOEXCEPT_IF(std::is_nothrow_move_constructible_v<TTraits>)
|
|
{
|
|
return SplitStringRange<TChar, TTraits>(stringView, std::move(traits));
|
|
}
|
|
|
|
template<auto splitAt, SplitViewOptions options = SplitViewOptions(), typename TStringView, typename TChar = decltype(splitAt), typename TCharTraits = std::char_traits<TChar>>
|
|
[[nodiscard]] SplitStringRange<TChar, SplitStringTraitsCT<TChar, splitAt, options, TCharTraits>> splitView(TStringView&& stringView) MIJIN_NOEXCEPT
|
|
{
|
|
return splitView<SplitStringTraitsCT<TChar, splitAt, options, TCharTraits>>(std::basic_string_view<TChar, TCharTraits>(std::forward<TStringView>(stringView)));
|
|
}
|
|
|
|
template<typename TStringView, typename TChar, typename TCharTraits = std::char_traits<TChar>, typename TTrimChars = std::basic_string_view<TChar, TCharTraits>>
|
|
[[nodiscard]] auto splitView(TStringView&& stringView, TChar splitAt, bool ignoreEmpty = true, TTrimChars trimChars = {}) MIJIN_NOEXCEPT
|
|
{
|
|
return splitView(std::basic_string_view<TChar, TCharTraits>(std::forward<TStringView>(stringView)), SplitStringTraitsRT<TChar, TCharTraits>{
|
|
.splitAt = splitAt,
|
|
.ignoreEmpty = ignoreEmpty,
|
|
.trimChars = std::basic_string_view<TChar, TCharTraits>(std::forward<TTrimChars>(trimChars))
|
|
});
|
|
}
|
|
|
|
template<typename TTraits, typename TChar, typename TLine = unsigned> requires(SplitLineTraitsType<TTraits, TChar, TLine>)
|
|
[[nodiscard]] LineRange<TChar, TLine, TTraits> splitLines(typename TTraits::string_view_t stringView, TTraits traits = {}) MIJIN_NOEXCEPT_IF(std::is_nothrow_move_constructible_v<TTraits>)
|
|
{
|
|
return LineRange<TChar, TLine, TTraits>(stringView, std::move(traits));
|
|
}
|
|
|
|
template<typename TLine = unsigned, SplitViewOptions options = SplitViewOptions{.ignoreEmpty=false},
|
|
typename TParam,
|
|
typename TStringView = decltype(std::basic_string_view(std::declval<TParam&&>())),
|
|
typename TChar = typename TStringView::value_type,
|
|
typename TCharTraits = typename TStringView::traits_type,
|
|
typename TTraits = SplitLineTraitsCT<TChar, TLine, options, TCharTraits>>
|
|
[[nodiscard]]
|
|
auto splitLines(TParam&& stringView) MIJIN_NOEXCEPT -> LineRange<TChar, TLine, TTraits>
|
|
{
|
|
return LineRange<TChar, TLine, TTraits>(std::basic_string_view(std::forward<TParam>(stringView)));
|
|
}
|
|
|
|
template<typename TParam, typename TLine = unsigned,
|
|
typename TStringView = decltype(std::basic_string_view(std::declval<TParam&&>())),
|
|
typename TChar = typename TStringView::value_type,
|
|
typename TCharTraits = typename TStringView::traits_type,
|
|
typename TTraits = SplitLineTraitsRT<TChar, TLine, TCharTraits>,
|
|
typename TTrimChars = TStringView>
|
|
[[nodiscard]]
|
|
auto splitLines(TParam&& stringView, bool ignoreEmpty, TTrimChars&& trimChars = {}) MIJIN_NOEXCEPT -> LineRange<TChar, TLine, TTraits>
|
|
{
|
|
return LineRange<TChar, TLine, TTraits>(std::basic_string_view(std::forward<TParam>(stringView)), TTraits{
|
|
.ignoreEmpty = ignoreEmpty,
|
|
.trimChars = std::basic_string_view<TChar, TCharTraits>(std::forward<TTrimChars>(trimChars))
|
|
});
|
|
}
|
|
|
|
template<typename TString, typename TChars>
|
|
[[nodiscard]]
|
|
auto trimPrefix(TString&& string, TChars&& chars)
|
|
{
|
|
return detail::trimPrefixImpl(std::string_view(std::forward<TString>(string)), std::string_view(std::forward<TChars>(chars)));
|
|
}
|
|
|
|
template<typename TString>
|
|
[[nodiscard]]
|
|
auto trimPrefix(TString&& string)
|
|
{
|
|
return trimPrefix(string, DEFAULT_TRIM_CHARS<str_char_type_t<TString>>);
|
|
}
|
|
|
|
template<typename TString, typename TChars>
|
|
[[nodiscard]]
|
|
auto trimSuffix(TString&& string, TChars&& chars)
|
|
{
|
|
return detail::trimSuffixImpl(std::string_view(std::forward<TString>(string)), std::string_view(std::forward<TChars>(chars)));
|
|
}
|
|
|
|
template<typename TString>
|
|
[[nodiscard]]
|
|
auto trimSuffix(TString&& string)
|
|
{
|
|
return trimSuffix(string, DEFAULT_TRIM_CHARS<str_char_type_t<TString>>);
|
|
}
|
|
|
|
template<typename TString, typename TChars>
|
|
[[nodiscard]]
|
|
auto trim(TString&& string, TChars&& chars)
|
|
{
|
|
return detail::trimImpl(std::string_view(std::forward<TString>(string)), std::string_view(std::forward<TChars>(chars)));
|
|
}
|
|
|
|
template<typename TString>
|
|
[[nodiscard]]
|
|
auto trim(TString&& string)
|
|
{
|
|
return trim(string, DEFAULT_TRIM_CHARS<str_char_type_t<TString>>);
|
|
}
|
|
|
|
template<typename TLeft, typename TRight>
|
|
[[nodiscard]] bool equalsIgnoreCase(TLeft&& left, TRight&& right) MIJIN_NOEXCEPT
|
|
{
|
|
return detail::equalsIgnoreCaseImpl(std::string_view(left), std::string_view(right));
|
|
}
|
|
|
|
template<typename TChar, typename TTraits, typename TAllocator>
|
|
constexpr void makeLower(std::basic_string<TChar, TTraits, TAllocator>& string)
|
|
{
|
|
std::transform(string.begin(), string.end(), string.begin(), [locale = std::locale()](TChar chr)
|
|
{
|
|
return std::tolower<TChar>(chr, locale);
|
|
});
|
|
}
|
|
|
|
template<typename TChar, typename TTraits, typename TAllocator>
|
|
constexpr void makeUpper(std::basic_string<TChar, TTraits, TAllocator>& string)
|
|
{
|
|
std::transform(string.begin(), string.end(), string.begin(), [locale = std::locale()](TChar chr)
|
|
{
|
|
return std::toupper<TChar>(chr, locale);
|
|
});
|
|
}
|
|
|
|
template<typename... TArgs>
|
|
[[nodiscard]]
|
|
constexpr auto toLower(TArgs&&... args)
|
|
{
|
|
std::basic_string string(std::forward<TArgs>(args)...);
|
|
makeLower(string);
|
|
return string;
|
|
}
|
|
|
|
template<typename... TArgs>
|
|
[[nodiscard]]
|
|
constexpr auto toUpper(TArgs&&... args)
|
|
{
|
|
std::basic_string string(std::forward<TArgs>(args)...);
|
|
makeUpper(string);
|
|
return string;
|
|
}
|
|
|
|
|
|
template<std::integral TNumber>
|
|
[[nodiscard]]
|
|
constexpr bool toNumber(std::string_view stringView, TNumber& outNumber, int base = 10) MIJIN_NOEXCEPT
|
|
{
|
|
const char* start = &*stringView.begin();
|
|
const char* end = start + stringView.size();
|
|
const std::from_chars_result res = std::from_chars(start, end, outNumber, base);
|
|
return res.ec == std::errc{} && res.ptr == end;
|
|
}
|
|
|
|
template<typename TChar, typename TTraits, std::integral TNumber>
|
|
[[nodiscard]]
|
|
constexpr bool toNumber(std::basic_string_view<TChar, TTraits> stringView, TNumber& outNumber, int base = 10) MIJIN_NOEXCEPT requires (!std::is_same_v<TChar, char>)
|
|
{
|
|
std::string asString;
|
|
asString.resize(stringView.size());
|
|
// should only contain number symbols, so just cast down to char
|
|
std::transform(stringView.begin(), stringView.end(), asString.begin(), [](TChar chr) { return static_cast<char>(chr); });
|
|
return toNumber(asString, outNumber, base);
|
|
}
|
|
|
|
template<std::floating_point TNumber>
|
|
[[nodiscard]]
|
|
constexpr bool toNumber(std::string_view stringView, TNumber& outNumber, std::chars_format fmt = std::chars_format::general) MIJIN_NOEXCEPT
|
|
{
|
|
const char* start = &*stringView.begin();
|
|
const char* end = start + stringView.size();
|
|
const std::from_chars_result res = std::from_chars(start, end, outNumber, fmt);
|
|
return res.ec == std::errc{} && res.ptr == end;
|
|
}
|
|
|
|
template<typename TChar, typename TTraits, std::floating_point TNumber>
|
|
[[nodiscard]]
|
|
constexpr bool toNumber(std::basic_string_view<TChar, TTraits> stringView, TNumber& outNumber, std::chars_format fmt = std::chars_format::general) MIJIN_NOEXCEPT requires (!std::is_same_v<TChar, char>)
|
|
{
|
|
std::string asString;
|
|
asString.resize(stringView.size());
|
|
// should only contain number symbols, so just cast down to char
|
|
std::transform(stringView.begin(), stringView.end(), asString.begin(), [](TChar chr) { return static_cast<char>(chr); });
|
|
return toNumber(asString, outNumber, fmt);
|
|
}
|
|
|
|
template<typename TChar>
|
|
[[nodiscard]]
|
|
constexpr bool isDecimalChar(TChar chr) noexcept
|
|
{
|
|
return (chr >= TChar('0') && chr <= TChar('9'));
|
|
}
|
|
|
|
template<typename TChar>
|
|
[[nodiscard]]
|
|
constexpr bool isHexadecimalChar(TChar chr) noexcept
|
|
{
|
|
return isDecimalChar(chr)
|
|
|| (chr >= TChar('A') && chr <= TChar('F'))
|
|
|| (chr >= TChar('a') && chr <= TChar('f'));
|
|
}
|
|
|
|
template<typename TChar>
|
|
bool compareIgnoreCase(TChar left, TChar right) MIJIN_NOEXCEPT
|
|
{
|
|
return std::tolower(left) == std::tolower(right);
|
|
}
|
|
|
|
[[nodiscard]]
|
|
inline auto findIgnoreCase(std::string_view haystack, std::string_view needle)
|
|
{
|
|
return std::ranges::search(haystack, needle, &compareIgnoreCase<char>);
|
|
}
|
|
|
|
[[nodiscard]]
|
|
inline bool startsWithIgnoreCase(std::string_view string, std::string_view part)
|
|
{
|
|
if (part.size() > string.size())
|
|
{
|
|
return false;
|
|
}
|
|
return std::ranges::equal(string.substr(0, part.size()), part, &compareIgnoreCase<char>);
|
|
}
|
|
|
|
[[nodiscard]]
|
|
inline bool endsWithIgnoreCase(std::string_view string, std::string_view part)
|
|
{
|
|
if (part.size() > string.size())
|
|
{
|
|
return false;
|
|
}
|
|
return std::ranges::equal(string.substr(string.size() - part.size()), part, &compareIgnoreCase<char>);
|
|
}
|
|
|
|
namespace pipe
|
|
{
|
|
struct Join
|
|
{
|
|
const char* delimiter;
|
|
|
|
explicit Join(const char* delimiter_) MIJIN_NOEXCEPT: delimiter(delimiter_) {}
|
|
};
|
|
|
|
template<typename TIterable>
|
|
auto operator|(TIterable&& iterable, const Join& joiner)
|
|
{
|
|
return join(std::forward<TIterable>(iterable), joiner.delimiter);
|
|
}
|
|
} // namespace pipe
|
|
|
|
template<typename TFrom, typename TTo>
|
|
ConvertCharTypeResult convertCharType(const TFrom* chrFrom, std::size_t numFrom, TTo* outTo, std::size_t numTo, std::mbstate_t& mbstate) MIJIN_NOEXCEPT
|
|
{
|
|
if constexpr (std::is_same_v<TFrom, char>)
|
|
{
|
|
if constexpr (std::is_same_v<TTo, wchar_t>)
|
|
{
|
|
const std::size_t result = std::mbrtowc(outTo, chrFrom, numFrom, &mbstate);
|
|
if (result == static_cast<std::size_t>(-1))
|
|
{
|
|
return {};
|
|
}
|
|
return {
|
|
.numRead = static_cast<unsigned>(result),
|
|
.numWritten = 1
|
|
};
|
|
}
|
|
if constexpr (std::is_same_v<TTo, char8_t>)
|
|
{
|
|
|
|
}
|
|
}
|
|
if constexpr (std::is_same_v<TFrom, wchar_t>)
|
|
{
|
|
if constexpr (std::is_same_v<TTo, char>)
|
|
{
|
|
if (numTo < MB_CUR_MAX)
|
|
{
|
|
char tmpBuf[MB_LEN_MAX];
|
|
const ConvertCharTypeResult result = convertCharType(chrFrom, numFrom, tmpBuf, MB_LEN_MAX, mbstate);
|
|
if (result && result.numWritten <= numTo)
|
|
{
|
|
std::memcpy(outTo, tmpBuf, result.numWritten);
|
|
}
|
|
return result;
|
|
}
|
|
const std::size_t result = std::wcrtomb(outTo, *chrFrom, &mbstate);
|
|
if (result == static_cast<std::size_t>(-1))
|
|
{
|
|
return {};
|
|
}
|
|
return {
|
|
.numRead = 1,
|
|
.numWritten = static_cast<unsigned>(result)
|
|
};
|
|
}
|
|
if constexpr (std::is_same_v<TTo, char8_t>)
|
|
{
|
|
|
|
}
|
|
}
|
|
if constexpr (std::is_same_v<TFrom, char8_t>)
|
|
{
|
|
|
|
}
|
|
}
|
|
template<typename TFrom, typename TTo>
|
|
ConvertCharTypeResult convertCharType(const TFrom* chrFrom, std::size_t numFrom, TTo* outTo, std::size_t numTo) MIJIN_NOEXCEPT
|
|
{
|
|
std::mbstate_t mbstate;
|
|
return convertCharType(chrFrom, numFrom, outTo, numTo, mbstate);
|
|
}
|
|
|
|
template<typename TIterator>
|
|
struct [[nodiscard]] ConvertStringTypeResult
|
|
{
|
|
TIterator iterator;
|
|
bool success;
|
|
};
|
|
|
|
template<typename TTo, typename TFrom, typename TFromTraits, std::output_iterator<TTo> TIterator>
|
|
ConvertStringTypeResult<TIterator> convertStringType(std::basic_string_view<TFrom, TFromTraits> strFrom, TIterator outIterator)
|
|
{
|
|
TTo outBuffer[MB_LEN_MAX];
|
|
|
|
std::mbstate_t mbstate = {};
|
|
for (auto it = strFrom.begin(); it != strFrom.end();)
|
|
{
|
|
const std::size_t remaining = std::distance(it, strFrom.end());
|
|
const ConvertCharTypeResult result = convertCharType(&*it, remaining, outBuffer, MB_LEN_MAX, mbstate);
|
|
if (!result)
|
|
{
|
|
return {
|
|
.iterator = outIterator,
|
|
.success = false
|
|
};
|
|
}
|
|
for (unsigned pos = 0; pos < result.numWritten; ++pos)
|
|
{
|
|
*outIterator = outBuffer[pos];
|
|
++outIterator;
|
|
}
|
|
it = std::next(it, result.numRead);
|
|
}
|
|
|
|
return {
|
|
.iterator = outIterator,
|
|
.success = true
|
|
};
|
|
}
|
|
|
|
template<typename TFrom, typename TFromTraits, typename TTo, typename TToTraits, typename TToAllocator>
|
|
bool convertStringType(std::basic_string_view<TFrom, TFromTraits> strFrom, std::basic_string<TTo, TToTraits, TToAllocator>& outString)
|
|
{
|
|
if constexpr (std::is_same_v<TTo, TFrom>)
|
|
{
|
|
outString += strFrom;
|
|
return true;
|
|
}
|
|
else
|
|
{
|
|
return convertStringType<TTo>(strFrom, std::back_inserter(outString)).success;
|
|
}
|
|
}
|
|
|
|
template<typename TFrom, typename TTo, typename TToTraits, typename TToAllocator>
|
|
bool convertStringType(const TFrom* strFrom, std::basic_string<TTo, TToTraits, TToAllocator>& outString)
|
|
{
|
|
return convertStringType(std::basic_string_view<TFrom>(strFrom), outString);
|
|
}
|
|
|
|
template<typename TChar, typename TTraits, typename TAlloc = MIJIN_DEFAULT_ALLOCATOR<TChar>>
|
|
std::basic_string<TChar, TTraits, TAlloc> quoted(std::basic_string_view<TChar, TTraits> input)
|
|
{
|
|
std::basic_string<TChar, TTraits> result;
|
|
result.reserve(input.size() + 2);
|
|
result.push_back(TChar('"'));
|
|
for (const TChar chr : input)
|
|
{
|
|
if (chr == TChar('"') || chr == TChar('\\')) {
|
|
result.push_back(TChar('\\'));
|
|
}
|
|
result.push_back(chr);
|
|
}
|
|
result.push_back(TChar('"'));
|
|
return result;
|
|
}
|
|
|
|
template<typename TChar, typename TTraits, typename TAlloc>
|
|
std::basic_string<TChar, TTraits, TAlloc> quoted(const std::basic_string<TChar, TTraits, TAlloc>& input)
|
|
{
|
|
return quoted<TChar, TTraits, TAlloc>(std::basic_string_view(input));
|
|
}
|
|
} // namespace mijin
|
|
|
|
#endif // !defined(MIJIN_UTIL_STRING_HPP_INCLUDED)
|