json-validator/include/jvalidate/detail/string.h

/**
 * Utility functions for managing strings, specifically because C++'s
 * std::string/std::regex is not well suited for UTF8 comprehensions.
 */
#pragma once
#include <jvalidate/_config.h>

#include <cstring>
#include <memory>
#include <string_view>

#if JVALIDATE_HAS_ICU
#include <unicode/brkiter.h>
#include <unicode/unistr.h>
#endif

#if JVALIDATE_HAS_IDNA
#include <ada/idna/unicode_transcoding.h>
#endif

#include <jvalidate/detail/expect.h>

namespace jvalidate::detail {
inline std::string_view to_u8(std::string_view arg) { return arg; }
inline std::u32string_view to_u32(std::u32string_view arg) { return arg; }
}

namespace jvalidate::detail {
#if JVALIDATE_HAS_IDNA
/**
 * @brief Calculates the string-length of the argument, treating multi-byte
 * characters as their individual bytes (as if the string was a std::string).
 *
 * @param arg Any UTF8 compatible string (including a standard ASCII string)
 *
 * @returns A number no greater than arg.length(), depending on the number of
 * graphemes/codepoints in the string.
 */
inline size_t length(std::string_view arg) {
  return ada::idna::utf32_length_from_utf8(arg.data(), arg.length());
}

inline std::string to_u8(std::u32string_view str) {
  auto data = std::make_unique_for_overwrite<char[]>(4 * str.length());
  size_t bytes = ada::idna::utf32_to_utf8(str.data(), str.length(), data.get());
  return std::string(data.get(), data.get() + bytes);
}

inline std::u32string to_u32(std::string_view str) {
  auto data = std::make_unique_for_overwrite<char32_t[]>(str.length());
  size_t bytes = ada::idna::utf8_to_utf32(str.data(), str.length(), data.get());
  return std::u32string(data.get(), data.get() + bytes);
}
#elif JVALIDATE_HAS_ICU
/**
 * @brief Calculates the string-length of the argument, treating multi-byte
 * characters and unicode graphemes as single characters (which std::string
 * cannot do).
 *
 * @param arg Any UTF8 compatible string (including a standard ASCII string)
 *
 * @returns A number no greater than arg.length(), depending on the number of
 * graphemes/codepoints in the string.
 */
inline size_t length(std::string_view arg) {
  icu::UnicodeString ucs = icu::UnicodeString::fromUTF8(icu::StringPiece(arg));
  return ucs.countChar32();
}
#else
/**
 * @brief Calculates the string-length of the argument, without attempting to
 * parse out graphemes. This method is provided so that it is possible to use
 * maxLength and minLength constraints even when building without IDNA or ICU
 * in the toolchain.
 *
 * @param arg Any UTF8 compatible string (including a standard ASCII string)
 *
 * @returns arg.length()
 */
inline size_t length(std::string_view arg) { return arg.length(); }
#endif
}