icuSources/common/util_props.cpp

   1 /*
   2 **********************************************************************
   3 *   Copyright (c) 2001-2016, 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
  16 U_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  */
  26 int32_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  */
  86 int32_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             U_FALLTHROUGH;
 105         case 126 /*'~'*/:
 106             pos = skipWhitespace(rule, pos);
 107             break;
 108         case 35 /*'#'*/:
 109             p = pos;
 110             parsedInts[intCount++] = parseInteger(rule, p, limit);
 111             if (p == pos) {
 112                 // Syntax error; failed to parse integer
 113                 return -1;
 114             }
 115             pos = p;
 116             break;
 117         default:
 118             if (pos >= limit) {
 119                 return -1;
 120             }
 121             c = (UChar) u_tolower(rule.charAt(pos++));
 122             if (c != cpat) {
 123                 return -1;
 124             }
 125             break;
 126         }
 127     }
 128     return pos;
 129 }
 130
 131 /**
 132  * Parse a Unicode identifier from the given string at the given
 133  * position.  Return the identifier, or an empty string if there
 134  * is no identifier.
 135  * @param str the string to parse
 136  * @param pos INPUT-OUPUT parameter.  On INPUT, pos is the
 137  * first character to examine.  It must be less than str.length(),
 138  * and it must not point to a whitespace character.  That is, must
 139  * have pos < str.length().  On
 140  * OUTPUT, the position after the last parsed character.
 141  * @return the Unicode identifier, or an empty string if there is
 142  * no valid identifier at pos.
 143  */
 144 UnicodeString ICU_Utility::parseUnicodeIdentifier(const UnicodeString& str, int32_t& pos) {
 145     // assert(pos < str.length());
 146     UnicodeString buf;
 147     int p = pos;
 148     while (p < str.length()) {
 149         UChar32 ch = str.char32At(p);
 150         if (buf.length() == 0) {
 151             if (u_isIDStart(ch)) {
 152                 buf.append(ch);
 153             } else {
 154                 buf.truncate(0);
 155                 return buf;
 156             }
 157         } else {
 158             if (u_isIDPart(ch)) {
 159                 buf.append(ch);
 160             } else {
 161                 break;
 162             }
 163         }
 164         p += U16_LENGTH(ch);
 165     }
 166     pos = p;
 167     return buf;
 168 }
 169
 170 /**
 171  * Parse an unsigned 31-bit integer at the given offset.  Use
 172  * UCharacter.digit() to parse individual characters into digits.
 173  * @param text the text to be parsed
 174  * @param pos INPUT-OUTPUT parameter.  On entry, pos[0] is the
 175  * offset within text at which to start parsing; it should point
 176  * to a valid digit.  On exit, pos[0] is the offset after the last
 177  * parsed character.  If the parse failed, it will be unchanged on
 178  * exit.  Must be >= 0 on entry.
 179  * @param radix the radix in which to parse; must be >= 2 and <=
 180  * 36.
 181  * @return a non-negative parsed number, or -1 upon parse failure.
 182  * Parse fails if there are no digits, that is, if pos[0] does not
 183  * point to a valid digit on entry, or if the number to be parsed
 184  * does not fit into a 31-bit unsigned integer.
 185  */
 186 int32_t ICU_Utility::parseNumber(const UnicodeString& text,
 187                                  int32_t& pos, int8_t radix) {
 188     // assert(pos[0] >= 0);
 189     // assert(radix >= 2);
 190     // assert(radix <= 36);
 191     int32_t n = 0;
 192     int32_t p = pos;
 193     while (p < text.length()) {
 194         UChar32 ch = text.char32At(p);
 195         int32_t d = u_digit(ch, radix);
 196         if (d < 0) {
 197             break;
 198         }
 199         n = radix*n + d;
 200         // ASSUME that when a 32-bit integer overflows it becomes
 201         // negative.  E.g., 214748364 * 10 + 8 => negative value.
 202         if (n < 0) {
 203             return -1;
 204         }
 205         ++p;
 206     }
 207     if (p == pos) {
 208         return -1;
 209     }
 210     pos = p;
 211     return n;
 212 }
 213
 214 U_NAMESPACE_END
 215