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 <ostream>
#include <string>
#include <string_view>
#if __has_include(<unicode/std_string.h>)
#define JVALIDATE_HAS_ICU
#include <unicode/brkiter.h>
#include <unicode/unistr.h>
#endif

#include <jvalidate/detail/expect.h>

#ifdef JVALIDATE_HAS_ICU
namespace jvalidate::detail {
/**
 * @brief Calclates the string-length of the argument, treating multi-byte
 * characters an 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.size(), 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();
}

inline std::string_view to_u8(std::string_view arg) { return arg; }

inline std::string to_u8(std::u32string_view arg) {
  icu::UnicodeString const ucs =
      icu::UnicodeString::fromUTF32(reinterpret_cast<int const *>(arg.data()), arg.size());
  std::string out;
  return ucs.toUTF8String(out);
}

inline std::u32string_view to_u32(std::u32string_view arg) { return arg; }

inline std::u32string to_u32(std::string_view arg) {
  icu::UnicodeString const ucs = icu::UnicodeString::fromUTF8(icu::StringPiece(arg));

  std::u32string rval;
  size_t const capacity = ucs.countChar32();
  rval.resize(capacity);

  UErrorCode status = U_ZERO_ERROR;
  ucs.toUTF32(reinterpret_cast<int *>(rval.data()), capacity, status);
  // This should never occur - unless there's like an alloc error
  if (U_FAILURE(status)) {
    JVALIDATE_THROW(std::runtime_error, "UTF-32 Translation Error");
  }

  return rval;
}
}

namespace std {
inline std::ostream & operator<<(std::ostream & os, std::u32string_view str) {
  return os << jvalidate::detail::to_u8(str);
}
}
#else
#endif