1 // 2016 and later: Unicode, Inc. and others. 2 // License & terms of use: http://www.unicode.org/copyright.html#License 3 /** 4 ******************************************************************************* 5 * Copyright (C) 1996-2004, International Business Machines Corporation and * 6 * others. All Rights Reserved. * 7 ******************************************************************************* 8 */ 9 package com.ibm.icu.impl; 10 11 /** 12 * Internal character utility class for simple data type conversion and String 13 * parsing functions. Does not have an analog in the JDK. 14 * @author Syn Wee Quek 15 * @since sep2900 16 */ 17 18 public final class UCharacterUtility 19 { 20 // public methods ----------------------------------------------------- 21 22 /** 23 * Determines if codepoint is a non character 24 * @param ch codepoint 25 * @return true if codepoint is a non character false otherwise 26 */ 27 public static boolean isNonCharacter(int ch) 28 { 29 if ((ch & NON_CHARACTER_SUFFIX_MIN_3_0_) == 30 NON_CHARACTER_SUFFIX_MIN_3_0_) { 31 return true; 32 } 33 34 return ch >= NON_CHARACTER_MIN_3_1_ && ch <= NON_CHARACTER_MAX_3_1_; 35 } 36 37 // package private methods --------------------------------------------- 38 39 /** 40 * joining 2 chars to form an int 41 * @param msc most significant char 42 * @param lsc least significant char 43 * @return int form 44 */ 45 static int toInt(char msc, char lsc) 46 { 47 return ((msc << 16) | lsc); 48 } 49 50 /** 51 * Retrieves a null terminated substring from an array of bytes. 52 * Substring is a set of non-zero bytes starting from argument start to the 53 * next zero byte. If the first byte is a zero, the next byte will be taken as 54 * the first byte. 55 * @param str stringbuffer to store data in, data will be store with each 56 * byte as a char 57 * @param array byte array 58 * @param index to start substring in byte count 59 * @return the end position of the substring within the character array 60 */ 61 static int getNullTermByteSubString(StringBuffer str, byte[] array, 62 int index) 63 { 64 byte b = 1; 65 66 while (b != 0) 67 { 68 b = array[index]; 69 if (b != 0) { 70 str.append((char)(b & 0x00FF)); 71 } 72 index ++; 73 } 74 return index; 75 } 76 77 /** 78 * Compares a null terminated substring from an array of bytes. 79 * Substring is a set of non-zero bytes starting from argument start to the 80 * next zero byte. if the first byte is a zero, the next byte will be taken as 81 * the first byte. 82 * @param str string to compare 83 * @param array byte array 84 * @param strindex index within str to start comparing 85 * @param aindex array index to start in byte count 86 * @return the end position of the substring within str if matches otherwise 87 * a -1 88 */ 89 static int compareNullTermByteSubString(String str, byte[] array, 90 int strindex, int aindex) 91 { 92 byte b = 1; 93 int length = str.length(); 94 95 while (b != 0) 96 { 97 b = array[aindex]; 98 aindex ++; 99 if (b == 0) { 100 break; 101 } 102 // if we have reached the end of the string and yet the array has not 103 // reached the end of their substring yet, abort 104 if (strindex == length 105 || (str.charAt(strindex) != (char)(b & 0xFF))) { 106 return -1; 107 } 108 strindex ++; 109 } 110 return strindex; 111 } 112 113 /** 114 * Skip null terminated substrings from an array of bytes. 115 * Substring is a set of non-zero bytes starting from argument start to the 116 * next zero byte. If the first byte is a zero, the next byte will be taken as 117 * the first byte. 118 * @param array byte array 119 * @param index to start substrings in byte count 120 * @param skipcount number of null terminated substrings to skip 121 * @return the end position of the substrings within the character array 122 */ 123 static int skipNullTermByteSubString(byte[] array, int index, 124 int skipcount) 125 { 126 byte b; 127 for (int i = 0; i < skipcount; i ++) 128 { 129 b = 1; 130 while (b != 0) 131 { 132 b = array[index]; 133 index ++; 134 } 135 } 136 return index; 137 } 138 139 /** 140 * skip substrings from an array of characters, where each character is a set 141 * of 2 bytes. substring is a set of non-zero bytes starting from argument 142 * start to the byte of the argument value. skips up to a max number of 143 * characters 144 * @param array byte array to parse 145 * @param index to start substrings in byte count 146 * @param length the max number of bytes to skip 147 * @param skipend value of byte to skip to 148 * @return the number of bytes skipped 149 */ 150 static int skipByteSubString(byte[] array, int index, int length, 151 byte skipend) 152 { 153 int result; 154 byte b; 155 156 for (result = 0; result < length; result ++) 157 { 158 b = array[index + result]; 159 if (b == skipend) 160 { 161 result ++; 162 break; 163 } 164 } 165 166 return result; 167 } 168 169 // private data member -------------------------------------------------- 170 171 /** 172 * Minimum suffix value that indicates if a character is non character. 173 * Unicode 3.0 non characters 174 */ 175 private static final int NON_CHARACTER_SUFFIX_MIN_3_0_ = 0xFFFE; 176 /** 177 * New minimum non character in Unicode 3.1 178 */ 179 private static final int NON_CHARACTER_MIN_3_1_ = 0xFDD0; 180 /** 181 * New non character range in Unicode 3.1 182 */ 183 private static final int NON_CHARACTER_MAX_3_1_ = 0xFDEF; 184 185 // private constructor -------------------------------------------------- 186 187 ///CLOVER:OFF 188 /** 189 * private constructor to avoid initialisation 190 */ 191 private UCharacterUtility() 192 { 193 } 194 ///CLOVER:ON 195 } 196 197
