icuSources/common/util.cpp

   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/unimatch.h"
  12 #include "unicode/utf16.h"
  13 #include "patternprops.h"
  14 #include "util.h"
  15
  16 // Define UChar constants using hex for EBCDIC compatibility
  17
  18 static const UChar BACKSLASH  = 0x005C; /*\*/
  19 static const UChar UPPER_U    = 0x0055; /*U*/
  20 static const UChar LOWER_U    = 0x0075; /*u*/
  21 static const UChar APOSTROPHE = 0x0027; // '\''
  22 static const UChar SPACE      = 0x0020; // ' '
  23
  24 // "0123456789ABCDEFGHIJKLMNOPQRSTUVWXYZ"
  25 static const UChar DIGITS[] = {
  26     48,49,50,51,52,53,54,55,56,57,
  27     65,66,67,68,69,70,71,72,73,74,
  28     75,76,77,78,79,80,81,82,83,84,
  29     85,86,87,88,89,90
  30 };
  31
  32 U_NAMESPACE_BEGIN
  33
  34 UnicodeString& ICU_Utility::appendNumber(UnicodeString& result, int32_t n,
  35                                      int32_t radix, int32_t minDigits) {
  36     if (radix < 2 || radix > 36) {
  37         // Bogus radix
  38         return result.append((UChar)63/*?*/);
  39     }
  40     // Handle negatives
  41     if (n < 0) {
  42         n = -n;
  43         result.append((UChar)45/*-*/);
  44     }
  45     // First determine the number of digits
  46     int32_t nn = n;
  47     int32_t r = 1;
  48     while (nn >= radix) {
  49         nn /= radix;
  50         r *= radix;
  51         --minDigits;
  52     }
  53     // Now generate the digits
  54     while (--minDigits > 0) {
  55         result.append(DIGITS[0]);
  56     }
  57     while (r > 0) {
  58         int32_t digit = n / r;
  59         result.append(DIGITS[digit]);
  60         n -= digit * r;
  61         r /= radix;
  62     }
  63     return result;
  64 }
  65
  66 /**
  67  * Return true if the character is NOT printable ASCII.
  68  */
  69 UBool ICU_Utility::isUnprintable(UChar32 c) {
  70     return !(c >= 0x20 && c <= 0x7E);
  71 }
  72
  73 /**
  74  * Escape unprintable characters using \uxxxx notation for U+0000 to
  75  * U+FFFF and \Uxxxxxxxx for U+10000 and above.  If the character is
  76  * printable ASCII, then do nothing and return FALSE.  Otherwise,
  77  * append the escaped notation and return TRUE.
  78  */
  79 UBool ICU_Utility::escapeUnprintable(UnicodeString& result, UChar32 c) {
  80     if (isUnprintable(c)) {
  81         result.append(BACKSLASH);
  82         if (c & ~0xFFFF) {
  83             result.append(UPPER_U);
  84             result.append(DIGITS[0xF&(c>>28)]);
  85             result.append(DIGITS[0xF&(c>>24)]);
  86             result.append(DIGITS[0xF&(c>>20)]);
  87             result.append(DIGITS[0xF&(c>>16)]);
  88         } else {
  89             result.append(LOWER_U);
  90         }
  91         result.append(DIGITS[0xF&(c>>12)]);
  92         result.append(DIGITS[0xF&(c>>8)]);
  93         result.append(DIGITS[0xF&(c>>4)]);
  94         result.append(DIGITS[0xF&c]);
  95         return TRUE;
  96     }
  97     return FALSE;
  98 }
  99
 100 /**
 101  * Returns the index of a character, ignoring quoted text.
 102  * For example, in the string "abc'hide'h", the 'h' in "hide" will not be
 103  * found by a search for 'h'.
 104  */
 105 // FOR FUTURE USE.  DISABLE FOR NOW for coverage reasons.
 106 /*
 107 int32_t ICU_Utility::quotedIndexOf(const UnicodeString& text,
 108                                int32_t start, int32_t limit,
 109                                UChar charToFind) {
 110     for (int32_t i=start; i<limit; ++i) {
 111         UChar c = text.charAt(i);
 112         if (c == BACKSLASH) {
 113             ++i;
 114         } else if (c == APOSTROPHE) {
 115             while (++i < limit
 116                    && text.charAt(i) != APOSTROPHE) {}
 117         } else if (c == charToFind) {
 118             return i;
 119         }
 120     }
 121     return -1;
 122 }
 123 */
 124
 125 /**
 126  * Skip over a sequence of zero or more white space characters at pos.
 127  * @param advance if true, advance pos to the first non-white-space
 128  * character at or after pos, or str.length(), if there is none.
 129  * Otherwise leave pos unchanged.
 130  * @return the index of the first non-white-space character at or
 131  * after pos, or str.length(), if there is none.
 132  */
 133 int32_t ICU_Utility::skipWhitespace(const UnicodeString& str, int32_t& pos,
 134                                     UBool advance) {
 135     int32_t p = pos;
 136     const UChar* s = str.getBuffer();
 137     p = (int32_t)(PatternProps::skipWhiteSpace(s + p, str.length() - p) - s);
 138     if (advance) {
 139         pos = p;
 140     }
 141     return p;
 142 }
 143
 144 /**
 145  * Skip over Pattern_White_Space in a Replaceable.
 146  * Skipping may be done in the forward or
 147  * reverse direction.  In either case, the leftmost index will be
 148  * inclusive, and the rightmost index will be exclusive.  That is,
 149  * given a range defined as [start, limit), the call
 150  * skipWhitespace(text, start, limit) will advance start past leading
 151  * whitespace, whereas the call skipWhitespace(text, limit, start),
 152  * will back up limit past trailing whitespace.
 153  * @param text the text to be analyzed
 154  * @param pos either the start or limit of a range of 'text', to skip
 155  * leading or trailing whitespace, respectively
 156  * @param stop either the limit or start of a range of 'text', to skip
 157  * leading or trailing whitespace, respectively
 158  * @return the new start or limit, depending on what was passed in to
 159  * 'pos'
 160  */
 161 //?FOR FUTURE USE.  DISABLE FOR NOW for coverage reasons.
 162 //?int32_t ICU_Utility::skipWhitespace(const Replaceable& text,
 163 //?                                    int32_t pos, int32_t stop) {
 164 //?    UChar32 c;
 165 //?    UBool isForward = (stop >= pos);
 166 //?
 167 //?    if (!isForward) {
 168 //?        --pos; // pos is a limit, so back up by one
 169 //?    }
 170 //?
 171 //?    while (pos != stop &&
 172 //?           PatternProps::isWhiteSpace(c = text.char32At(pos))) {
 173 //?        if (isForward) {
 174 //?            pos += U16_LENGTH(c);
 175 //?        } else {
 176 //?            pos -= U16_LENGTH(c);
 177 //?        }
 178 //?    }
 179 //?
 180 //?    if (!isForward) {
 181 //?        ++pos; // make pos back into a limit
 182 //?    }
 183 //?
 184 //?    return pos;
 185 //?}
 186
 187 /**
 188  * Parse a single non-whitespace character 'ch', optionally
 189  * preceded by whitespace.
 190  * @param id the string to be parsed
 191  * @param pos INPUT-OUTPUT parameter.  On input, pos[0] is the
 192  * offset of the first character to be parsed.  On output, pos[0]
 193  * is the index after the last parsed character.  If the parse
 194  * fails, pos[0] will be unchanged.
 195  * @param ch the non-whitespace character to be parsed.
 196  * @return true if 'ch' is seen preceded by zero or more
 197  * whitespace characters.
 198  */
 199 UBool ICU_Utility::parseChar(const UnicodeString& id, int32_t& pos, UChar ch) {
 200     int32_t start = pos;
 201     skipWhitespace(id, pos, TRUE);
 202     if (pos == id.length() ||
 203         id.charAt(pos) != ch) {
 204         pos = start;
 205         return FALSE;
 206     }
 207     ++pos;
 208     return TRUE;
 209 }
 210
 211 /**
 212  * Parse a pattern string within the given Replaceable and a parsing
 213  * pattern.  Characters are matched literally and case-sensitively
 214  * except for the following special characters:
 215  *
 216  * ~  zero or more Pattern_White_Space chars
 217  *
 218  * If end of pattern is reached with all matches along the way,
 219  * pos is advanced to the first unparsed index and returned.
 220  * Otherwise -1 is returned.
 221  * @param pat pattern that controls parsing
 222  * @param text text to be parsed, starting at index
 223  * @param index offset to first character to parse
 224  * @param limit offset after last character to parse
 225  * @return index after last parsed character, or -1 on parse failure.
 226  */
 227 int32_t ICU_Utility::parsePattern(const UnicodeString& pat,
 228                                   const Replaceable& text,
 229                                   int32_t index,
 230                                   int32_t limit) {
 231     int32_t ipat = 0;
 232
 233     // empty pattern matches immediately
 234     if (ipat == pat.length()) {
 235         return index;
 236     }
 237
 238     UChar32 cpat = pat.char32At(ipat);
 239
 240     while (index < limit) {
 241         UChar32 c = text.char32At(index);
 242
 243         // parse \s*
 244         if (cpat == 126 /*~*/) {
 245             if (PatternProps::isWhiteSpace(c)) {
 246                 index += U16_LENGTH(c);
 247                 continue;
 248             } else {
 249                 if (++ipat == pat.length()) {
 250                     return index; // success; c unparsed
 251                 }
 252                 // fall thru; process c again with next cpat
 253             }
 254         }
 255
 256         // parse literal
 257         else if (c == cpat) {
 258             index += U16_LENGTH(c);
 259             ipat += U16_LENGTH(cpat);
 260             if (ipat == pat.length()) {
 261                 return index; // success; c parsed
 262             }
 263             // fall thru; get next cpat
 264         }
 265
 266         // match failure of literal
 267         else {
 268             return -1;
 269         }
 270
 271         cpat = pat.char32At(ipat);
 272     }
 273
 274     return -1; // text ended before end of pat
 275 }
 276
 277 /**
 278  * Append a character to a rule that is being built up.  To flush
 279  * the quoteBuf to rule, make one final call with isLiteral == TRUE.
 280  * If there is no final character, pass in (UChar32)-1 as c.
 281  * @param rule the string to append the character to
 282  * @param c the character to append, or (UChar32)-1 if none.
 283  * @param isLiteral if true, then the given character should not be
 284  * quoted or escaped.  Usually this means it is a syntactic element
 285  * such as > or $
 286  * @param escapeUnprintable if true, then unprintable characters
 287  * should be escaped using \uxxxx or \Uxxxxxxxx.  These escapes will
 288  * appear outside of quotes.
 289  * @param quoteBuf a buffer which is used to build up quoted
 290  * substrings.  The caller should initially supply an empty buffer,
 291  * and thereafter should not modify the buffer.  The buffer should be
 292  * cleared out by, at the end, calling this method with a literal
 293  * character.
 294  */
 295 void ICU_Utility::appendToRule(UnicodeString& rule,
 296                                UChar32 c,
 297                                UBool isLiteral,
 298                                UBool escapeUnprintable,
 299                                UnicodeString& quoteBuf) {
 300     // If we are escaping unprintables, then escape them outside
 301     // quotes.  \u and \U are not recognized within quotes.  The same
 302     // logic applies to literals, but literals are never escaped.
 303     if (isLiteral ||
 304         (escapeUnprintable && ICU_Utility::isUnprintable(c))) {
 305         if (quoteBuf.length() > 0) {
 306             // We prefer backslash APOSTROPHE to double APOSTROPHE
 307             // (more readable, less similar to ") so if there are
 308             // double APOSTROPHEs at the ends, we pull them outside
 309             // of the quote.
 310
 311             // If the first thing in the quoteBuf is APOSTROPHE
 312             // (doubled) then pull it out.
 313             while (quoteBuf.length() >= 2 &&
 314                    quoteBuf.charAt(0) == APOSTROPHE &&
 315                    quoteBuf.charAt(1) == APOSTROPHE) {
 316                 rule.append(BACKSLASH).append(APOSTROPHE);
 317                 quoteBuf.remove(0, 2);
 318             }
 319             // If the last thing in the quoteBuf is APOSTROPHE
 320             // (doubled) then remove and count it and add it after.
 321             int32_t trailingCount = 0;
 322             while (quoteBuf.length() >= 2 &&
 323                    quoteBuf.charAt(quoteBuf.length()-2) == APOSTROPHE &&
 324                    quoteBuf.charAt(quoteBuf.length()-1) == APOSTROPHE) {
 325                 quoteBuf.truncate(quoteBuf.length()-2);
 326                 ++trailingCount;
 327             }
 328             if (quoteBuf.length() > 0) {
 329                 rule.append(APOSTROPHE);
 330                 rule.append(quoteBuf);
 331                 rule.append(APOSTROPHE);
 332                 quoteBuf.truncate(0);
 333             }
 334             while (trailingCount-- > 0) {
 335                 rule.append(BACKSLASH).append(APOSTROPHE);
 336             }
 337         }
 338         if (c != (UChar32)-1) {
 339             /* Since spaces are ignored during parsing, they are
 340              * emitted only for readability.  We emit one here
 341              * only if there isn't already one at the end of the
 342              * rule.
 343              */
 344             if (c == SPACE) {
 345                 int32_t len = rule.length();
 346                 if (len > 0 && rule.charAt(len-1) != c) {
 347                     rule.append(c);
 348                 }
 349             } else if (!escapeUnprintable || !ICU_Utility::escapeUnprintable(rule, c)) {
 350                 rule.append(c);
 351             }
 352         }
 353     }
 354
 355     // Escape ' and '\' and don't begin a quote just for them
 356     else if (quoteBuf.length() == 0 &&
 357              (c == APOSTROPHE || c == BACKSLASH)) {
 358         rule.append(BACKSLASH);
 359         rule.append(c);
 360     }
 361
 362     // Specials (printable ascii that isn't [0-9a-zA-Z]) and
 363     // whitespace need quoting.  Also append stuff to quotes if we are
 364     // building up a quoted substring already.
 365     else if (quoteBuf.length() > 0 ||
 366              (c >= 0x0021 && c <= 0x007E &&
 367               !((c >= 0x0030/*'0'*/ && c <= 0x0039/*'9'*/) ||
 368                 (c >= 0x0041/*'A'*/ && c <= 0x005A/*'Z'*/) ||
 369                 (c >= 0x0061/*'a'*/ && c <= 0x007A/*'z'*/))) ||
 370              PatternProps::isWhiteSpace(c)) {
 371         quoteBuf.append(c);
 372         // Double ' within a quote
 373         if (c == APOSTROPHE) {
 374             quoteBuf.append(c);
 375         }
 376     }
 377
 378     // Otherwise just append
 379     else {
 380         rule.append(c);
 381     }
 382 }
 383
 384 void ICU_Utility::appendToRule(UnicodeString& rule,
 385                                const UnicodeString& text,
 386                                UBool isLiteral,
 387                                UBool escapeUnprintable,
 388                                UnicodeString& quoteBuf) {
 389     for (int32_t i=0; i<text.length(); ++i) {
 390         appendToRule(rule, text[i], isLiteral, escapeUnprintable, quoteBuf);
 391     }
 392 }
 393
 394 /**
 395  * Given a matcher reference, which may be null, append its
 396  * pattern as a literal to the given rule.
 397  */
 398 void ICU_Utility::appendToRule(UnicodeString& rule,
 399                                const UnicodeMatcher* matcher,
 400                                UBool escapeUnprintable,
 401                                UnicodeString& quoteBuf) {
 402     if (matcher != NULL) {
 403         UnicodeString pat;
 404         appendToRule(rule, matcher->toPattern(pat, escapeUnprintable),
 405                      TRUE, escapeUnprintable, quoteBuf);
 406     }
 407 }
 408
 409 U_NAMESPACE_END