1.1 --- /dev/null Thu Jan 01 00:00:00 1970 +0000
1.2 +++ b/security/sandbox/chromium/base/third_party/icu/icu_utf.cc Wed Dec 31 06:09:35 2014 +0100
1.3 @@ -0,0 +1,228 @@
1.4 +/*
1.5 +******************************************************************************
1.6 +*
1.7 +* Copyright (C) 1999-2006, International Business Machines
1.8 +* Corporation and others. All Rights Reserved.
1.9 +*
1.10 +******************************************************************************
1.11 +* file name: utf_impl.c
1.12 +* encoding: US-ASCII
1.13 +* tab size: 8 (not used)
1.14 +* indentation:4
1.15 +*
1.16 +* created on: 1999sep13
1.17 +* created by: Markus W. Scherer
1.18 +*
1.19 +* This file provides implementation functions for macros in the utfXX.h
1.20 +* that would otherwise be too long as macros.
1.21 +*/
1.22 +
1.23 +#include "base/third_party/icu/icu_utf.h"
1.24 +
1.25 +namespace base_icu {
1.26 +
1.27 +/**
1.28 + * UTF8_ERROR_VALUE_1 and UTF8_ERROR_VALUE_2 are special error values for UTF-8,
1.29 + * which need 1 or 2 bytes in UTF-8:
1.30 + * \code
1.31 + * U+0015 = NAK = Negative Acknowledge, C0 control character
1.32 + * U+009f = highest C1 control character
1.33 + * \endcode
1.34 + *
1.35 + * These are used by UTF8_..._SAFE macros so that they can return an error value
1.36 + * that needs the same number of code units (bytes) as were seen by
1.37 + * a macro. They should be tested with UTF_IS_ERROR() or UTF_IS_VALID().
1.38 + *
1.39 + * @deprecated ICU 2.4. Obsolete, see utf_old.h.
1.40 + */
1.41 +#define CBUTF8_ERROR_VALUE_1 0x15
1.42 +
1.43 +/**
1.44 + * See documentation on UTF8_ERROR_VALUE_1 for details.
1.45 + *
1.46 + * @deprecated ICU 2.4. Obsolete, see utf_old.h.
1.47 + */
1.48 +#define CBUTF8_ERROR_VALUE_2 0x9f
1.49 +
1.50 +
1.51 +/**
1.52 + * Error value for all UTFs. This code point value will be set by macros with e>
1.53 + * checking if an error is detected.
1.54 + *
1.55 + * @deprecated ICU 2.4. Obsolete, see utf_old.h.
1.56 + */
1.57 +#define CBUTF_ERROR_VALUE 0xffff
1.58 +
1.59 +/*
1.60 + * This table could be replaced on many machines by
1.61 + * a few lines of assembler code using an
1.62 + * "index of first 0-bit from msb" instruction and
1.63 + * one or two more integer instructions.
1.64 + *
1.65 + * For example, on an i386, do something like
1.66 + * - MOV AL, leadByte
1.67 + * - NOT AL (8-bit, leave b15..b8==0..0, reverse only b7..b0)
1.68 + * - MOV AH, 0
1.69 + * - BSR BX, AX (16-bit)
1.70 + * - MOV AX, 6 (result)
1.71 + * - JZ finish (ZF==1 if leadByte==0xff)
1.72 + * - SUB AX, BX (result)
1.73 + * -finish:
1.74 + * (BSR: Bit Scan Reverse, scans for a 1-bit, starting from the MSB)
1.75 + *
1.76 + * In Unicode, all UTF-8 byte sequences with more than 4 bytes are illegal;
1.77 + * lead bytes above 0xf4 are illegal.
1.78 + * We keep them in this table for skipping long ISO 10646-UTF-8 sequences.
1.79 + */
1.80 +const uint8
1.81 +utf8_countTrailBytes[256]={
1.82 + 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
1.83 + 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
1.84 + 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
1.85 + 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
1.86 +
1.87 + 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
1.88 + 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
1.89 + 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
1.90 + 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
1.91 +
1.92 + 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
1.93 + 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
1.94 + 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
1.95 + 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
1.96 +
1.97 + 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,
1.98 + 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,
1.99 +
1.100 + 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2,
1.101 + 3, 3, 3, 3, 3,
1.102 + 3, 3, 3, /* illegal in Unicode */
1.103 + 4, 4, 4, 4, /* illegal in Unicode */
1.104 + 5, 5, /* illegal in Unicode */
1.105 + 0, 0 /* illegal bytes 0xfe and 0xff */
1.106 +};
1.107 +
1.108 +static const UChar32
1.109 +utf8_minLegal[4]={ 0, 0x80, 0x800, 0x10000 };
1.110 +
1.111 +static const UChar32
1.112 +utf8_errorValue[6]={
1.113 + CBUTF8_ERROR_VALUE_1, CBUTF8_ERROR_VALUE_2, CBUTF_ERROR_VALUE, 0x10ffff,
1.114 + 0x3ffffff, 0x7fffffff
1.115 +};
1.116 +
1.117 +/*
1.118 + * Handle the non-inline part of the U8_NEXT() macro and its obsolete sibling
1.119 + * UTF8_NEXT_CHAR_SAFE().
1.120 + *
1.121 + * The "strict" parameter controls the error behavior:
1.122 + * <0 "Safe" behavior of U8_NEXT(): All illegal byte sequences yield a negative
1.123 + * code point result.
1.124 + * 0 Obsolete "safe" behavior of UTF8_NEXT_CHAR_SAFE(..., FALSE):
1.125 + * All illegal byte sequences yield a positive code point such that this
1.126 + * result code point would be encoded with the same number of bytes as
1.127 + * the illegal sequence.
1.128 + * >0 Obsolete "strict" behavior of UTF8_NEXT_CHAR_SAFE(..., TRUE):
1.129 + * Same as the obsolete "safe" behavior, but non-characters are also treated
1.130 + * like illegal sequences.
1.131 + *
1.132 + * The special negative (<0) value -2 is used for lenient treatment of surrogate
1.133 + * code points as legal. Some implementations use this for roundtripping of
1.134 + * Unicode 16-bit strings that are not well-formed UTF-16, that is, they
1.135 + * contain unpaired surrogates.
1.136 + *
1.137 + * Note that a UBool is the same as an int8_t.
1.138 + */
1.139 +UChar32
1.140 +utf8_nextCharSafeBody(const uint8 *s, int32 *pi, int32 length, UChar32 c, UBool strict) {
1.141 + int32 i=*pi;
1.142 + uint8 count=CBU8_COUNT_TRAIL_BYTES(c);
1.143 + if((i)+count<=(length)) {
1.144 + uint8 trail, illegal=0;
1.145 +
1.146 + CBU8_MASK_LEAD_BYTE((c), count);
1.147 + /* count==0 for illegally leading trail bytes and the illegal bytes 0xfe and 0xff */
1.148 + switch(count) {
1.149 + /* each branch falls through to the next one */
1.150 + case 5:
1.151 + case 4:
1.152 + /* count>=4 is always illegal: no more than 3 trail bytes in Unicode's UTF-8 */
1.153 + illegal=1;
1.154 + break;
1.155 + case 3:
1.156 + trail=s[(i)++];
1.157 + (c)=((c)<<6)|(trail&0x3f);
1.158 + if(c<0x110) {
1.159 + illegal|=(trail&0xc0)^0x80;
1.160 + } else {
1.161 + /* code point>0x10ffff, outside Unicode */
1.162 + illegal=1;
1.163 + break;
1.164 + }
1.165 + case 2:
1.166 + trail=s[(i)++];
1.167 + (c)=((c)<<6)|(trail&0x3f);
1.168 + illegal|=(trail&0xc0)^0x80;
1.169 + case 1:
1.170 + trail=s[(i)++];
1.171 + (c)=((c)<<6)|(trail&0x3f);
1.172 + illegal|=(trail&0xc0)^0x80;
1.173 + break;
1.174 + case 0:
1.175 + if(strict>=0) {
1.176 + return CBUTF8_ERROR_VALUE_1;
1.177 + } else {
1.178 + return CBU_SENTINEL;
1.179 + }
1.180 + /* no default branch to optimize switch() - all values are covered */
1.181 + }
1.182 +
1.183 + /*
1.184 + * All the error handling should return a value
1.185 + * that needs count bytes so that UTF8_GET_CHAR_SAFE() works right.
1.186 + *
1.187 + * Starting with Unicode 3.0.1, non-shortest forms are illegal.
1.188 + * Starting with Unicode 3.2, surrogate code points must not be
1.189 + * encoded in UTF-8, and there are no irregular sequences any more.
1.190 + *
1.191 + * U8_ macros (new in ICU 2.4) return negative values for error conditions.
1.192 + */
1.193 +
1.194 + /* correct sequence - all trail bytes have (b7..b6)==(10)? */
1.195 + /* illegal is also set if count>=4 */
1.196 + if(illegal || (c)<utf8_minLegal[count] || (CBU_IS_SURROGATE(c) && strict!=-2)) {
1.197 + /* error handling */
1.198 + uint8 errorCount=count;
1.199 + /* don't go beyond this sequence */
1.200 + i=*pi;
1.201 + while(count>0 && CBU8_IS_TRAIL(s[i])) {
1.202 + ++(i);
1.203 + --count;
1.204 + }
1.205 + if(strict>=0) {
1.206 + c=utf8_errorValue[errorCount-count];
1.207 + } else {
1.208 + c=CBU_SENTINEL;
1.209 + }
1.210 + } else if((strict)>0 && CBU_IS_UNICODE_NONCHAR(c)) {
1.211 + /* strict: forbid non-characters like U+fffe */
1.212 + c=utf8_errorValue[count];
1.213 + }
1.214 + } else /* too few bytes left */ {
1.215 + /* error handling */
1.216 + int32 i0=i;
1.217 + /* don't just set (i)=(length) in case there is an illegal sequence */
1.218 + while((i)<(length) && CBU8_IS_TRAIL(s[i])) {
1.219 + ++(i);
1.220 + }
1.221 + if(strict>=0) {
1.222 + c=utf8_errorValue[i-i0];
1.223 + } else {
1.224 + c=CBU_SENTINEL;
1.225 + }
1.226 + }
1.227 + *pi=i;
1.228 + return c;
1.229 +}
1.230 +
1.231 +} // namespace base_icu
