1/* 2********************************************************************** 3* Copyright (c) 2001-2011, International Business Machines 4* Corporation and others. All Rights Reserved. 5********************************************************************** 6* Date Name Description 7* 11/19/2001 aliu Creation. 8********************************************************************** 9*/ 10 11#include "unicode/uchar.h" 12#include "unicode/utf16.h" 13#include "patternprops.h" 14#include "util.h" 15 16U_NAMESPACE_BEGIN 17 18/** 19 * Parse an integer at pos, either of the form \d+ or of the form 20 * 0x[0-9A-Fa-f]+ or 0[0-7]+, that is, in standard decimal, hex, 21 * or octal format. 22 * @param pos INPUT-OUTPUT parameter. On input, the first 23 * character to parse. On output, the character after the last 24 * parsed character. 25 */ 26int32_t ICU_Utility::parseInteger(const UnicodeString& rule, int32_t& pos, int32_t limit) { 27 int32_t count = 0; 28 int32_t value = 0; 29 int32_t p = pos; 30 int8_t radix = 10; 31 32 if (p < limit && rule.charAt(p) == 48 /*0*/) { 33 if (p+1 < limit && (rule.charAt(p+1) == 0x78 /*x*/ || rule.charAt(p+1) == 0x58 /*X*/)) { 34 p += 2; 35 radix = 16; 36 } 37 else { 38 p++; 39 count = 1; 40 radix = 8; 41 } 42 } 43 44 while (p < limit) { 45 int32_t d = u_digit(rule.charAt(p++), radix); 46 if (d < 0) { 47 --p; 48 break; 49 } 50 ++count; 51 int32_t v = (value * radix) + d; 52 if (v <= value) { 53 // If there are too many input digits, at some point 54 // the value will go negative, e.g., if we have seen 55 // "0x8000000" already and there is another '0', when 56 // we parse the next 0 the value will go negative. 57 return 0; 58 } 59 value = v; 60 } 61 if (count > 0) { 62 pos = p; 63 } 64 return value; 65} 66 67/** 68 * Parse a pattern string starting at offset pos. Keywords are 69 * matched case-insensitively. Spaces may be skipped and may be 70 * optional or required. Integer values may be parsed, and if 71 * they are, they will be returned in the given array. If 72 * successful, the offset of the next non-space character is 73 * returned. On failure, -1 is returned. 74 * @param pattern must only contain lowercase characters, which 75 * will match their uppercase equivalents as well. A space 76 * character matches one or more required spaces. A '~' character 77 * matches zero or more optional spaces. A '#' character matches 78 * an integer and stores it in parsedInts, which the caller must 79 * ensure has enough capacity. 80 * @param parsedInts array to receive parsed integers. Caller 81 * must ensure that parsedInts.length is >= the number of '#' 82 * signs in 'pattern'. 83 * @return the position after the last character parsed, or -1 if 84 * the parse failed 85 */ 86int32_t ICU_Utility::parsePattern(const UnicodeString& rule, int32_t pos, int32_t limit, 87 const UnicodeString& pattern, int32_t* parsedInts) { 88 // TODO Update this to handle surrogates 89 int32_t p; 90 int32_t intCount = 0; // number of integers parsed 91 for (int32_t i=0; i<pattern.length(); ++i) { 92 UChar cpat = pattern.charAt(i); 93 UChar c; 94 switch (cpat) { 95 case 32 /*' '*/: 96 if (pos >= limit) { 97 return -1; 98 } 99 c = rule.charAt(pos++); 100 if (!PatternProps::isWhiteSpace(c)) { 101 return -1; 102 } 103 // FALL THROUGH to skipWhitespace 104 case 126 /*'~'*/: 105 pos = skipWhitespace(rule, pos); 106 break; 107 case 35 /*'#'*/: 108 p = pos; 109 parsedInts[intCount++] = parseInteger(rule, p, limit); 110 if (p == pos) { 111 // Syntax error; failed to parse integer 112 return -1; 113 } 114 pos = p; 115 break; 116 default: 117 if (pos >= limit) { 118 return -1; 119 } 120 c = (UChar) u_tolower(rule.charAt(pos++)); 121 if (c != cpat) { 122 return -1; 123 } 124 break; 125 } 126 } 127 return pos; 128} 129 130/** 131 * Parse a Unicode identifier from the given string at the given 132 * position. Return the identifier, or an empty string if there 133 * is no identifier. 134 * @param str the string to parse 135 * @param pos INPUT-OUPUT parameter. On INPUT, pos is the 136 * first character to examine. It must be less than str.length(), 137 * and it must not point to a whitespace character. That is, must 138 * have pos < str.length(). On 139 * OUTPUT, the position after the last parsed character. 140 * @return the Unicode identifier, or an empty string if there is 141 * no valid identifier at pos. 142 */ 143UnicodeString ICU_Utility::parseUnicodeIdentifier(const UnicodeString& str, int32_t& pos) { 144 // assert(pos < str.length()); 145 UnicodeString buf; 146 int p = pos; 147 while (p < str.length()) { 148 UChar32 ch = str.char32At(p); 149 if (buf.length() == 0) { 150 if (u_isIDStart(ch)) { 151 buf.append(ch); 152 } else { 153 buf.truncate(0); 154 return buf; 155 } 156 } else { 157 if (u_isIDPart(ch)) { 158 buf.append(ch); 159 } else { 160 break; 161 } 162 } 163 p += U16_LENGTH(ch); 164 } 165 pos = p; 166 return buf; 167} 168 169/** 170 * Parse an unsigned 31-bit integer at the given offset. Use 171 * UCharacter.digit() to parse individual characters into digits. 172 * @param text the text to be parsed 173 * @param pos INPUT-OUTPUT parameter. On entry, pos[0] is the 174 * offset within text at which to start parsing; it should point 175 * to a valid digit. On exit, pos[0] is the offset after the last 176 * parsed character. If the parse failed, it will be unchanged on 177 * exit. Must be >= 0 on entry. 178 * @param radix the radix in which to parse; must be >= 2 and <= 179 * 36. 180 * @return a non-negative parsed number, or -1 upon parse failure. 181 * Parse fails if there are no digits, that is, if pos[0] does not 182 * point to a valid digit on entry, or if the number to be parsed 183 * does not fit into a 31-bit unsigned integer. 184 */ 185int32_t ICU_Utility::parseNumber(const UnicodeString& text, 186 int32_t& pos, int8_t radix) { 187 // assert(pos[0] >= 0); 188 // assert(radix >= 2); 189 // assert(radix <= 36); 190 int32_t n = 0; 191 int32_t p = pos; 192 while (p < text.length()) { 193 UChar32 ch = text.char32At(p); 194 int32_t d = u_digit(ch, radix); 195 if (d < 0) { 196 break; 197 } 198 n = radix*n + d; 199 // ASSUME that when a 32-bit integer overflows it becomes 200 // negative. E.g., 214748364 * 10 + 8 => negative value. 201 if (n < 0) { 202 return -1; 203 } 204 ++p; 205 } 206 if (p == pos) { 207 return -1; 208 } 209 pos = p; 210 return n; 211} 212 213U_NAMESPACE_END 214 215