chromium/third_party/icu/source/common/utf_impl.cpp

// © 2016 and later: Unicode, Inc. and others.
// License & terms of use: http://www.unicode.org/copyright.html
/*
******************************************************************************
*
*   Copyright (C) 1999-2012, International Business Machines
*   Corporation and others.  All Rights Reserved.
*
******************************************************************************
*   file name:  utf_impl.cpp
*   encoding:   UTF-8
*   tab size:   8 (not used)
*   indentation:4
*
*   created on: 1999sep13
*   created by: Markus W. Scherer
*
*   This file provides implementation functions for macros in the utfXX.h
*   that would otherwise be too long as macros.
*/

/* set import/export definitions */
#ifndef U_UTF8_IMPL
#define U_UTF8_IMPL
#endif

#include "unicode/utypes.h"
#include "unicode/utf.h"
#include "unicode/utf8.h"
#include "uassert.h"

/*
 * Table of the number of utf8 trail bytes, indexed by the lead byte.
 * Used by the deprecated macro UTF8_COUNT_TRAIL_BYTES, defined in utf_old.h
 *
 * The current macro, U8_COUNT_TRAIL_BYTES, does _not_ use this table.
 *
 * Note that this table cannot be removed, even if UTF8_COUNT_TRAIL_BYTES were
 * changed to no longer use it. References to the table from expansions of UTF8_COUNT_TRAIL_BYTES
 * may exist in old client code that must continue to run with newer icu library versions.
 *
 * This table could be replaced on many machines by
 * a few lines of assembler code using an
 * "index of first 0-bit from msb" instruction and
 * one or two more integer instructions.
 *
 * For example, on an i386, do something like
 * - MOV AL, leadByte
 * - NOT AL         (8-bit, leave b15..b8==0..0, reverse only b7..b0)
 * - MOV AH, 0
 * - BSR BX, AX     (16-bit)
 * - MOV AX, 6      (result)
 * - JZ finish      (ZF==1 if leadByte==0xff)
 * - SUB AX, BX (result)
 * -finish:
 * (BSR: Bit Scan Reverse, scans for a 1-bit, starting from the MSB)
 */
U_CAPI const uint8_t
utf8_countTrailBytes[256]= …;

static const UChar32
utf8_errorValue[6]= …;

static UChar32
errorValue(int32_t count, int8_t strict) { … }

/*
 * Handle the non-inline part of the U8_NEXT() and U8_NEXT_FFFD() macros
 * and their obsolete sibling UTF8_NEXT_CHAR_SAFE().
 *
 * U8_NEXT() supports NUL-terminated strings indicated via length<0.
 *
 * The "strict" parameter controls the error behavior:
 * <0  "Safe" behavior of U8_NEXT():
 *     -1: All illegal byte sequences yield U_SENTINEL=-1.
 *     -2: Same as -1, except for lenient treatment of surrogate code points as legal.
 *         Some implementations use this for roundtripping of
 *         Unicode 16-bit strings that are not well-formed UTF-16, that is, they
 *         contain unpaired surrogates.
 *     -3: All illegal byte sequences yield U+FFFD.
 *  0  Obsolete "safe" behavior of UTF8_NEXT_CHAR_SAFE(..., false):
 *     All illegal byte sequences yield a positive code point such that this
 *     result code point would be encoded with the same number of bytes as
 *     the illegal sequence.
 * >0  Obsolete "strict" behavior of UTF8_NEXT_CHAR_SAFE(..., true):
 *     Same as the obsolete "safe" behavior, but non-characters are also treated
 *     like illegal sequences.
 *
 * Note that a UBool is the same as an int8_t.
 */
U_CAPI UChar32 U_EXPORT2
utf8_nextCharSafeBody(const uint8_t *s, int32_t *pi, int32_t length, UChar32 c, UBool strict) { … }

U_CAPI int32_t U_EXPORT2
utf8_appendCharSafeBody(uint8_t *s, int32_t i, int32_t length, UChar32 c, UBool *pIsError) { … }

U_CAPI UChar32 U_EXPORT2
utf8_prevCharSafeBody(const uint8_t *s, int32_t start, int32_t *pi, UChar32 c, UBool strict) { … }

U_CAPI int32_t U_EXPORT2
utf8_back1SafeBody(const uint8_t *s, int32_t start, int32_t i) { … }