ꦫꦣꦺꦤ꧀ꦄꦤ꧀ꦠꦱꦺꦤ

PATH: opt / alt / alt-nodejs20 / root / usr / include / unicode

// © 2025 and later: Unicode, Inc. and others.
// License & terms of use: https://www.unicode.org/copyright.html

// utfstring.h
// created: 2025jul18 Markus W. Scherer

#ifndef __UTFSTRING_H__
#define __UTFSTRING_H__

#include "unicode/utypes.h"

#if U_SHOW_CPLUSPLUS_API || U_SHOW_CPLUSPLUS_HEADER_API || !defined(UTYPES_H)

#include "unicode/utf16.h"

/**
 * \file
 * \brief C++ header-only API: C++ string helper functions.
 */

#ifndef U_HIDE_DRAFT_API

namespace U_HEADER_ONLY_NAMESPACE {
namespace utfstring {

// Write code points to strings -------------------------------------------- ***

#ifndef U_IN_DOXYGEN
namespace prv {

// This function, and the public wrappers,
// want to be U_FORCE_INLINE but the gcc-debug-build-and-test CI check failed with
// error: ‘always_inline’ function might not be inlinable [-Werror=attributes]
template<typename StringClass, bool validate>
inline StringClass &appendCodePoint(StringClass &s, uint32_t c) {
    using Unit = typename StringClass::value_type;
    if constexpr (sizeof(Unit) == 1) {
        // UTF-8: Similar to U8_APPEND().
        if (c <= 0x7f) {
            s.push_back(static_cast<Unit>(c));
        } else {
            Unit buf[4];
            uint8_t len;
            if (c <= 0x7ff) {
                len = 2;
                buf[2] = (c >> 6) | 0xc0;
            } else {
                if (validate ?
                        c < 0xd800 ||
                            (c < 0xe000 || c > 0x10ffff ? (c = 0xfffd, true) : c <= 0xffff) :
                        c <= 0xffff) {
                    len = 3;
                    buf[1] = (c >> 12) | 0xe0;
                } else {
                    len = 4;
                    buf[0] = (c >> 18) | 0xf0;
                    buf[1] = ((c >> 12) & 0x3f) | 0x80;
                }
                buf[2] = ((c >> 6) & 0x3f) | 0x80;
            }
            buf[3] = (c & 0x3f) | 0x80;
            s.append(buf + 4 - len, len);
        }
    } else if constexpr (sizeof(Unit) == 2) {
        // UTF-16: Similar to U16_APPEND().
        if (validate ?
                c < 0xd800 || (c < 0xe000 || c > 0x10ffff ? (c = 0xfffd, true) : c <= 0xffff) :
                c <= 0xffff) {
            s.push_back(static_cast<Unit>(c));
        } else {
            Unit buf[2] = { U16_LEAD(c), U16_TRAIL(c) };
            s.append(buf, 2);
        }
    } else {
        // UTF-32
        s.push_back(!validate || U_IS_SCALAR_VALUE(c) ? c : 0xfffd);
    }
    return s;
}

}  // namespace prv
#endif  // U_IN_DOXYGEN

#ifndef U_HIDE_DRAFT_API
/**
 * Appends the code point to the string.
 * Appends the U+FFFD replacement character instead if c is not a scalar value.
 * See https://www.unicode.org/glossary/#unicode_scalar_value
 *
 * @tparam StringClass A version of std::basic_string (or a compatible type)
 * @param s The string to append to
 * @param c The code point to append
 * @return s
 * @draft ICU 78
 * @see U_IS_SCALAR_VALUE
 */
template<typename StringClass>
inline StringClass &appendOrFFFD(StringClass &s, UChar32 c) {
    return prv::appendCodePoint<StringClass, true>(s, c);
}

/**
 * Appends the code point to the string.
 * The code point must be a scalar value; otherwise the behavior is undefined.
 * See https://www.unicode.org/glossary/#unicode_scalar_value
 *
 * @tparam StringClass A version of std::basic_string (or a compatible type)
 * @param s The string to append to
 * @param c The code point to append (must be a scalar value)
 * @return s
 * @draft ICU 78
 * @see U_IS_SCALAR_VALUE
 */
template<typename StringClass>
inline StringClass &appendUnsafe(StringClass &s, UChar32 c) {
    return prv::appendCodePoint<StringClass, false>(s, c);
}

/**
 * Returns the code point as a string of code units.
 * Returns the U+FFFD replacement character instead if c is not a scalar value.
 * See https://www.unicode.org/glossary/#unicode_scalar_value
 *
 * @tparam StringClass A version of std::basic_string (or a compatible type)
 * @param c The code point
 * @return the string of c's code units
 * @draft ICU 78
 * @see U_IS_SCALAR_VALUE
 */
template<typename StringClass>
inline StringClass encodeOrFFFD(UChar32 c) {
    StringClass s;
    prv::appendCodePoint<StringClass, true>(s, c);
    return s;
}

/**
 * Returns the code point as a string of code units.
 * The code point must be a scalar value; otherwise the behavior is undefined.
 * See https://www.unicode.org/glossary/#unicode_scalar_value
 *
 * @tparam StringClass A version of std::basic_string (or a compatible type)
 * @param c The code point
 * @return the string of c's code units
 * @draft ICU 78
 * @see U_IS_SCALAR_VALUE
 */
template<typename StringClass>
inline StringClass encodeUnsafe(UChar32 c) {
    StringClass s;
    prv::appendCodePoint<StringClass, false>(s, c);
    return s;
}
#endif  // U_HIDE_DRAFT_API

}  // namespace utfstring
}  // namespace U_HEADER_ONLY_NAMESPACE

#endif  // U_HIDE_DRAFT_API
#endif  // U_SHOW_CPLUSPLUS_API || U_SHOW_CPLUSPLUS_HEADER_API
#endif  // __UTFSTRING_H__

[+] ..
[-] localpointer.h [edit]
[-] stringpiece.h [edit]
[-] uenum.h [edit]
[-] ureldatefmt.h [edit]
[-] measure.h [edit]
[-] numberrangeformatter.h [edit]
[-] ucharstrie.h [edit]
[-] rep.h [edit]
[-] numfmt.h [edit]
[-] ulistformatter.h [edit]
[-] umachine.h [edit]
[-] unum.h [edit]
[-] simplenumberformatter.h [edit]
[-] filteredbrk.h [edit]
[-] umutablecptrie.h [edit]
[-] unumberoptions.h [edit]
[-] messageformat2_data_model_names.h [edit]
[-] nounit.h [edit]
[-] gender.h [edit]
[-] stsearch.h [edit]
[-] ufieldpositer.h [edit]
[-] dtptngen.h [edit]
[-] messageformat2.h [edit]
[-] ucat.h [edit]
[-] ustdio.h [edit]
[-] rbtz.h [edit]
[-] udata.h [edit]
[-] unumberformatter.h [edit]
[-] utf.h [edit]
[-] tznames.h [edit]
[-] timezone.h [edit]
[-] region.h [edit]
[-] formattednumber.h [edit]
[-] curramt.h [edit]
[-] bytestriebuilder.h [edit]
[-] utfstring.h [edit]
[-] tblcoll.h [edit]
[-] dtrule.h [edit]
[-] unifunct.h [edit]
[-] uformattednumber.h [edit]
[-] parseerr.h [edit]
[-] std_string.h [edit]
[-] messageformat2_data_model.h [edit]
[-] ucol.h [edit]
[-] platform.h [edit]
[-] measunit.h [edit]
[-] uscript.h [edit]
[-] ucnvsel.h [edit]
[-] unifilt.h [edit]
[-] uformattable.h [edit]
[-] uloc.h [edit]
[-] udatpg.h [edit]
[-] listformatter.h [edit]
[-] tmutamt.h [edit]
[-] numsys.h [edit]
[-] utf8.h [edit]
[-] ubidi.h [edit]
[-] selfmt.h [edit]
[-] uspoof.h [edit]
[-] fieldpos.h [edit]
[-] translit.h [edit]
[-] rbbi.h [edit]
[-] appendable.h [edit]
[-] localematcher.h [edit]
[-] basictz.h [edit]
[-] usetiter.h [edit]
[-] tzrule.h [edit]
[-] currpinf.h [edit]
[-] uregion.h [edit]
[-] scientificnumberformatter.h [edit]
[-] ucnv.h [edit]
[-] ucasemap.h [edit]
[-] unorm.h [edit]
[-] utext.h [edit]
[-] urename.h [edit]
[-] dbbi.h [edit]
[-] icudataver.h [edit]
[-] idna.h [edit]
[-] unistr.h [edit]
[-] umisc.h [edit]
[-] uidna.h [edit]
[-] utypes.h [edit]
[-] stringtriebuilder.h [edit]
[-] ures.h [edit]
[-] reldatefmt.h [edit]
[-] messagepattern.h [edit]
[-] utmscale.h [edit]
[-] simpleformatter.h [edit]
[-] uset.h [edit]
[-] utf16.h [edit]
[-] ustringtrie.h [edit]
[-] udisplaycontext.h [edit]
[-] resbund.h [edit]
[-] uregex.h [edit]
[-] ugender.h [edit]
[-] locid.h [edit]
[-] umsg.h [edit]
[-] casemap.h [edit]
[-] normalizer2.h [edit]
[-] putil.h [edit]
[-] normlzr.h [edit]
[-] uldnames.h [edit]
[-] ubiditransform.h [edit]
[-] utfiterator.h [edit]
[-] choicfmt.h [edit]
[-] dtfmtsym.h [edit]
[-] tmutfmt.h [edit]
[-] utrace.h [edit]
[-] ubrk.h [edit]
[-] ulocdata.h [edit]
[-] utf32.h [edit]
[-] usimplenumberformatter.h [edit]
[-] unimatch.h [edit]
[-] tzfmt.h [edit]
[-] edits.h [edit]
[-] ulocale.h [edit]
[-] uformattedvalue.h [edit]
[-] uvernum.h [edit]
[-] ucal.h [edit]
[-] udateintervalformat.h [edit]
[-] bytestream.h [edit]
[-] ucharstriebuilder.h [edit]
[-] tztrans.h [edit]
[-] localebuilder.h [edit]
[-] format.h [edit]
[-] rbnf.h [edit]
[-] ucsdet.h [edit]
[-] urep.h [edit]
[-] plurfmt.h [edit]
[-] unumberrangeformatter.h [edit]
[-] symtable.h [edit]
[-] ulocbuilder.h [edit]
[-] dcfmtsym.h [edit]
[-] uconfig.h [edit]
[-] locdspnm.h [edit]
[-] utrans.h [edit]
[-] uchar.h [edit]
[-] messageformat2_formattable.h [edit]
[-] uobject.h [edit]
[-] udat.h [edit]
[-] stringoptions.h [edit]
[-] ushape.h [edit]
[-] dtitvinf.h [edit]
[-] messageformat2_arguments.h [edit]
[-] currunit.h [edit]
[-] parsepos.h [edit]
[-] decimfmt.h [edit]
[-] displayoptions.h [edit]
[-] ucnv_cb.h [edit]
[-] coleitr.h [edit]
[-] ucurr.h [edit]
[-] tmunit.h [edit]
[-] messageformat2_function_registry.h [edit]
[-] utf_old.h [edit]
[-] chariter.h [edit]
[-] ptypes.h [edit]
[-] uiter.h [edit]
[-] unorm2.h [edit]
[-] alphaindex.h [edit]
[-] uniset.h [edit]
[-] measfmt.h [edit]
[-] numberformatter.h [edit]
[-] datefmt.h [edit]
[-] simpletz.h [edit]
[-] search.h [edit]
[-] msgfmt.h [edit]
[-] schriter.h [edit]
[-] usprep.h [edit]
[-] plurrule.h [edit]
[-] brkiter.h [edit]
[-] formattedvalue.h [edit]
[-] compactdecimalformat.h [edit]
[-] ucnv_err.h [edit]
[-] enumset.h [edit]
[-] calendar.h [edit]
[-] ustring.h [edit]
[-] fmtable.h [edit]
[-] uversion.h [edit]
[-] udisplayoptions.h [edit]
[-] ucpmap.h [edit]
[-] regex.h [edit]
[-] upluralrules.h [edit]
[-] usearch.h [edit]
[-] ucoleitr.h [edit]
[-] dtintrv.h [edit]
[-] bytestrie.h [edit]
[-] ucptrie.h [edit]
[-] docmain.h [edit]
[-] caniter.h [edit]
[-] gregocal.h [edit]
[-] unumsys.h [edit]
[-] strenum.h [edit]
[-] uchriter.h [edit]
[-] dtitvfmt.h [edit]
[-] smpdtfmt.h [edit]
[-] unirepl.h [edit]
[-] icuplug.h [edit]
[-] vtzone.h [edit]
[-] ustream.h [edit]
[-] uclean.h [edit]
[-] sortkey.h [edit]
[-] errorcode.h [edit]
[-] fpositer.h [edit]
[-] char16ptr.h [edit]
[-] coll.h [edit]