1 /*
2 ******************************************************************************
3 * Copyright (C) 1997-2008, International Business Machines
4 * Corporation and others. All Rights Reserved.
5 ******************************************************************************
6 * file name: nfrs.cpp
7 * encoding: US-ASCII
8 * tab size: 8 (not used)
9 * indentation:4
10 *
11 * Modification history
12 * Date Name Comments
13 * 10/11/2001 Doug Ported from ICU4J
14 */
15
16 #include "nfrs.h"
17
18 #if U_HAVE_RBNF
19
20 #include "unicode/uchar.h"
21 #include "nfrule.h"
22 #include "nfrlist.h"
23
24 #ifdef RBNF_DEBUG
25 #include "cmemory.h"
26 #endif
27
28 #include "util.h"
29
30 U_NAMESPACE_BEGIN
31
32 #if 0
33 // euclid's algorithm works with doubles
34 // note, doubles only get us up to one quadrillion or so, which
35 // isn't as much range as we get with longs. We probably still
36 // want either 64-bit math, or BigInteger.
37
38 static int64_t
39 util_lcm(int64_t x, int64_t y)
40 {
41 x.abs();
42 y.abs();
43
44 if (x == 0 || y == 0) {
45 return 0;
46 } else {
47 do {
48 if (x < y) {
49 int64_t t = x; x = y; y = t;
50 }
51 x -= y * (x/y);
52 } while (x != 0);
53
54 return y;
55 }
56 }
57
58 #else
59 /**
60 * Calculates the least common multiple of x and y.
61 */
62 static int64_t
util_lcm(int64_t x,int64_t y)63 util_lcm(int64_t x, int64_t y)
64 {
65 // binary gcd algorithm from Knuth, "The Art of Computer Programming,"
66 // vol. 2, 1st ed., pp. 298-299
67 int64_t x1 = x;
68 int64_t y1 = y;
69
70 int p2 = 0;
71 while ((x1 & 1) == 0 && (y1 & 1) == 0) {
72 ++p2;
73 x1 >>= 1;
74 y1 >>= 1;
75 }
76
77 int64_t t;
78 if ((x1 & 1) == 1) {
79 t = -y1;
80 } else {
81 t = x1;
82 }
83
84 while (t != 0) {
85 while ((t & 1) == 0) {
86 t = t >> 1;
87 }
88 if (t > 0) {
89 x1 = t;
90 } else {
91 y1 = -t;
92 }
93 t = x1 - y1;
94 }
95
96 int64_t gcd = x1 << p2;
97
98 // x * y == gcd(x, y) * lcm(x, y)
99 return x / gcd * y;
100 }
101 #endif
102
103 static const UChar gPercent = 0x0025;
104 static const UChar gColon = 0x003a;
105 static const UChar gSemicolon = 0x003b;
106 static const UChar gLineFeed = 0x000a;
107
108 static const UChar gFourSpaces[] =
109 {
110 0x20, 0x20, 0x20, 0x20, 0
111 }; /* " " */
112 static const UChar gPercentPercent[] =
113 {
114 0x25, 0x25, 0
115 }; /* "%%" */
116
NFRuleSet(UnicodeString * descriptions,int32_t index,UErrorCode & status)117 NFRuleSet::NFRuleSet(UnicodeString* descriptions, int32_t index, UErrorCode& status)
118 : name()
119 , rules(0)
120 , negativeNumberRule(NULL)
121 , fIsFractionRuleSet(FALSE)
122 , fIsPublic(FALSE)
123 , fRecursionCount(0)
124 {
125 for (int i = 0; i < 3; ++i) {
126 fractionRules[i] = NULL;
127 }
128
129 if (U_FAILURE(status)) {
130 return;
131 }
132
133 UnicodeString& description = descriptions[index]; // !!! make sure index is valid
134
135 if (description.length() == 0) {
136 // throw new IllegalArgumentException("Empty rule set description");
137 status = U_PARSE_ERROR;
138 return;
139 }
140
141 // if the description begins with a rule set name (the rule set
142 // name can be omitted in formatter descriptions that consist
143 // of only one rule set), copy it out into our "name" member
144 // and delete it from the description
145 if (description.charAt(0) == gPercent) {
146 int32_t pos = description.indexOf(gColon);
147 if (pos == -1) {
148 // throw new IllegalArgumentException("Rule set name doesn't end in colon");
149 status = U_PARSE_ERROR;
150 } else {
151 name.setTo(description, 0, pos);
152 while (pos < description.length() && uprv_isRuleWhiteSpace(description.charAt(++pos))) {
153 }
154 description.remove(0, pos);
155 }
156 } else {
157 name.setTo(UNICODE_STRING_SIMPLE("%default"));
158 }
159
160 if (description.length() == 0) {
161 // throw new IllegalArgumentException("Empty rule set description");
162 status = U_PARSE_ERROR;
163 }
164
165 fIsPublic = name.indexOf(gPercentPercent) != 0;
166
167 // all of the other members of NFRuleSet are initialized
168 // by parseRules()
169 }
170
171 void
parseRules(UnicodeString & description,const RuleBasedNumberFormat * owner,UErrorCode & status)172 NFRuleSet::parseRules(UnicodeString& description, const RuleBasedNumberFormat* owner, UErrorCode& status)
173 {
174 // start by creating a Vector whose elements are Strings containing
175 // the descriptions of the rules (one rule per element). The rules
176 // are separated by semicolons (there's no escape facility: ALL
177 // semicolons are rule delimiters)
178
179 if (U_FAILURE(status)) {
180 return;
181 }
182
183 // dlf - the original code kept a separate description array for no reason,
184 // so I got rid of it. The loop was too complex so I simplified it.
185
186 UnicodeString currentDescription;
187 int32_t oldP = 0;
188 while (oldP < description.length()) {
189 int32_t p = description.indexOf(gSemicolon, oldP);
190 if (p == -1) {
191 p = description.length();
192 }
193 currentDescription.setTo(description, oldP, p - oldP);
194 NFRule::makeRules(currentDescription, this, rules.last(), owner, rules, status);
195 oldP = p + 1;
196 }
197
198 // for rules that didn't specify a base value, their base values
199 // were initialized to 0. Make another pass through the list and
200 // set all those rules' base values. We also remove any special
201 // rules from the list and put them into their own member variables
202 int64_t defaultBaseValue = 0;
203
204 // (this isn't a for loop because we might be deleting items from
205 // the vector-- we want to make sure we only increment i when
206 // we _didn't_ delete aything from the vector)
207 uint32_t i = 0;
208 while (i < rules.size()) {
209 NFRule* rule = rules[i];
210
211 switch (rule->getType()) {
212 // if the rule's base value is 0, fill in a default
213 // base value (this will be 1 plus the preceding
214 // rule's base value for regular rule sets, and the
215 // same as the preceding rule's base value in fraction
216 // rule sets)
217 case NFRule::kNoBase:
218 rule->setBaseValue(defaultBaseValue, status);
219 if (!isFractionRuleSet()) {
220 ++defaultBaseValue;
221 }
222 ++i;
223 break;
224
225 // if it's the negative-number rule, copy it into its own
226 // data member and delete it from the list
227 case NFRule::kNegativeNumberRule:
228 negativeNumberRule = rules.remove(i);
229 break;
230
231 // if it's the improper fraction rule, copy it into the
232 // correct element of fractionRules
233 case NFRule::kImproperFractionRule:
234 fractionRules[0] = rules.remove(i);
235 break;
236
237 // if it's the proper fraction rule, copy it into the
238 // correct element of fractionRules
239 case NFRule::kProperFractionRule:
240 fractionRules[1] = rules.remove(i);
241 break;
242
243 // if it's the master rule, copy it into the
244 // correct element of fractionRules
245 case NFRule::kMasterRule:
246 fractionRules[2] = rules.remove(i);
247 break;
248
249 // if it's a regular rule that already knows its base value,
250 // check to make sure the rules are in order, and update
251 // the default base value for the next rule
252 default:
253 if (rule->getBaseValue() < defaultBaseValue) {
254 // throw new IllegalArgumentException("Rules are not in order");
255 status = U_PARSE_ERROR;
256 return;
257 }
258 defaultBaseValue = rule->getBaseValue();
259 if (!isFractionRuleSet()) {
260 ++defaultBaseValue;
261 }
262 ++i;
263 break;
264 }
265 }
266 }
267
~NFRuleSet()268 NFRuleSet::~NFRuleSet()
269 {
270 delete negativeNumberRule;
271 delete fractionRules[0];
272 delete fractionRules[1];
273 delete fractionRules[2];
274 }
275
276 static UBool
util_equalRules(const NFRule * rule1,const NFRule * rule2)277 util_equalRules(const NFRule* rule1, const NFRule* rule2)
278 {
279 if (rule1) {
280 if (rule2) {
281 return *rule1 == *rule2;
282 }
283 } else if (!rule2) {
284 return TRUE;
285 }
286 return FALSE;
287 }
288
289 UBool
operator ==(const NFRuleSet & rhs) const290 NFRuleSet::operator==(const NFRuleSet& rhs) const
291 {
292 if (rules.size() == rhs.rules.size() &&
293 fIsFractionRuleSet == rhs.fIsFractionRuleSet &&
294 name == rhs.name &&
295 util_equalRules(negativeNumberRule, rhs.negativeNumberRule) &&
296 util_equalRules(fractionRules[0], rhs.fractionRules[0]) &&
297 util_equalRules(fractionRules[1], rhs.fractionRules[1]) &&
298 util_equalRules(fractionRules[2], rhs.fractionRules[2])) {
299
300 for (uint32_t i = 0; i < rules.size(); ++i) {
301 if (*rules[i] != *rhs.rules[i]) {
302 return FALSE;
303 }
304 }
305 return TRUE;
306 }
307 return FALSE;
308 }
309
310 #define RECURSION_LIMIT 50
311
312 void
format(int64_t number,UnicodeString & toAppendTo,int32_t pos) const313 NFRuleSet::format(int64_t number, UnicodeString& toAppendTo, int32_t pos) const
314 {
315 NFRule *rule = findNormalRule(number);
316 if (rule) { // else error, but can't report it
317 NFRuleSet* ncThis = (NFRuleSet*)this;
318 if (ncThis->fRecursionCount++ >= RECURSION_LIMIT) {
319 // stop recursion
320 ncThis->fRecursionCount = 0;
321 } else {
322 rule->doFormat(number, toAppendTo, pos);
323 ncThis->fRecursionCount--;
324 }
325 }
326 }
327
328 void
format(double number,UnicodeString & toAppendTo,int32_t pos) const329 NFRuleSet::format(double number, UnicodeString& toAppendTo, int32_t pos) const
330 {
331 NFRule *rule = findDoubleRule(number);
332 if (rule) { // else error, but can't report it
333 NFRuleSet* ncThis = (NFRuleSet*)this;
334 if (ncThis->fRecursionCount++ >= RECURSION_LIMIT) {
335 // stop recursion
336 ncThis->fRecursionCount = 0;
337 } else {
338 rule->doFormat(number, toAppendTo, pos);
339 ncThis->fRecursionCount--;
340 }
341 }
342 }
343
344 NFRule*
findDoubleRule(double number) const345 NFRuleSet::findDoubleRule(double number) const
346 {
347 // if this is a fraction rule set, use findFractionRuleSetRule()
348 if (isFractionRuleSet()) {
349 return findFractionRuleSetRule(number);
350 }
351
352 // if the number is negative, return the negative number rule
353 // (if there isn't a negative-number rule, we pretend it's a
354 // positive number)
355 if (number < 0) {
356 if (negativeNumberRule) {
357 return negativeNumberRule;
358 } else {
359 number = -number;
360 }
361 }
362
363 // if the number isn't an integer, we use one of the fraction rules...
364 if (number != uprv_floor(number)) {
365 // if the number is between 0 and 1, return the proper
366 // fraction rule
367 if (number < 1 && fractionRules[1]) {
368 return fractionRules[1];
369 }
370 // otherwise, return the improper fraction rule
371 else if (fractionRules[0]) {
372 return fractionRules[0];
373 }
374 }
375
376 // if there's a master rule, use it to format the number
377 if (fractionRules[2]) {
378 return fractionRules[2];
379 }
380
381 // and if we haven't yet returned a rule, use findNormalRule()
382 // to find the applicable rule
383 int64_t r = util64_fromDouble(number + 0.5);
384 return findNormalRule(r);
385 }
386
387 NFRule *
findNormalRule(int64_t number) const388 NFRuleSet::findNormalRule(int64_t number) const
389 {
390 // if this is a fraction rule set, use findFractionRuleSetRule()
391 // to find the rule (we should only go into this clause if the
392 // value is 0)
393 if (fIsFractionRuleSet) {
394 return findFractionRuleSetRule((double)number);
395 }
396
397 // if the number is negative, return the negative-number rule
398 // (if there isn't one, pretend the number is positive)
399 if (number < 0) {
400 if (negativeNumberRule) {
401 return negativeNumberRule;
402 } else {
403 number = -number;
404 }
405 }
406
407 // we have to repeat the preceding two checks, even though we
408 // do them in findRule(), because the version of format() that
409 // takes a long bypasses findRule() and goes straight to this
410 // function. This function does skip the fraction rules since
411 // we know the value is an integer (it also skips the master
412 // rule, since it's considered a fraction rule. Skipping the
413 // master rule in this function is also how we avoid infinite
414 // recursion)
415
416 // {dlf} unfortunately this fails if there are no rules except
417 // special rules. If there are no rules, use the master rule.
418
419 // binary-search the rule list for the applicable rule
420 // (a rule is used for all values from its base value to
421 // the next rule's base value)
422 int32_t hi = rules.size();
423 if (hi > 0) {
424 int32_t lo = 0;
425
426 while (lo < hi) {
427 int32_t mid = (lo + hi) / 2;
428 if (rules[mid]->getBaseValue() == number) {
429 return rules[mid];
430 }
431 else if (rules[mid]->getBaseValue() > number) {
432 hi = mid;
433 }
434 else {
435 lo = mid + 1;
436 }
437 }
438 if (hi == 0) { // bad rule set, minimum base > 0
439 return NULL; // want to throw exception here
440 }
441
442 NFRule *result = rules[hi - 1];
443
444 // use shouldRollBack() to see whether we need to invoke the
445 // rollback rule (see shouldRollBack()'s documentation for
446 // an explanation of the rollback rule). If we do, roll back
447 // one rule and return that one instead of the one we'd normally
448 // return
449 if (result->shouldRollBack((double)number)) {
450 if (hi == 1) { // bad rule set, no prior rule to rollback to from this base
451 return NULL;
452 }
453 result = rules[hi - 2];
454 }
455 return result;
456 }
457 // else use the master rule
458 return fractionRules[2];
459 }
460
461 /**
462 * If this rule is a fraction rule set, this function is used by
463 * findRule() to select the most appropriate rule for formatting
464 * the number. Basically, the base value of each rule in the rule
465 * set is treated as the denominator of a fraction. Whichever
466 * denominator can produce the fraction closest in value to the
467 * number passed in is the result. If there's a tie, the earlier
468 * one in the list wins. (If there are two rules in a row with the
469 * same base value, the first one is used when the numerator of the
470 * fraction would be 1, and the second rule is used the rest of the
471 * time.
472 * @param number The number being formatted (which will always be
473 * a number between 0 and 1)
474 * @return The rule to use to format this number
475 */
476 NFRule*
findFractionRuleSetRule(double number) const477 NFRuleSet::findFractionRuleSetRule(double number) const
478 {
479 // the obvious way to do this (multiply the value being formatted
480 // by each rule's base value until you get an integral result)
481 // doesn't work because of rounding error. This method is more
482 // accurate
483
484 // find the least common multiple of the rules' base values
485 // and multiply this by the number being formatted. This is
486 // all the precision we need, and we can do all of the rest
487 // of the math using integer arithmetic
488 int64_t leastCommonMultiple = rules[0]->getBaseValue();
489 int64_t numerator;
490 {
491 for (uint32_t i = 1; i < rules.size(); ++i) {
492 leastCommonMultiple = util_lcm(leastCommonMultiple, rules[i]->getBaseValue());
493 }
494 numerator = util64_fromDouble(number * (double)leastCommonMultiple + 0.5);
495 }
496 // for each rule, do the following...
497 int64_t tempDifference;
498 int64_t difference = util64_fromDouble(uprv_maxMantissa());
499 int32_t winner = 0;
500 for (uint32_t i = 0; i < rules.size(); ++i) {
501 // "numerator" is the numerator of the fraction if the
502 // denominator is the LCD. The numerator if the rule's
503 // base value is the denominator is "numerator" times the
504 // base value divided bythe LCD. Here we check to see if
505 // that's an integer, and if not, how close it is to being
506 // an integer.
507 tempDifference = numerator * rules[i]->getBaseValue() % leastCommonMultiple;
508
509
510 // normalize the result of the above calculation: we want
511 // the numerator's distance from the CLOSEST multiple
512 // of the LCD
513 if (leastCommonMultiple - tempDifference < tempDifference) {
514 tempDifference = leastCommonMultiple - tempDifference;
515 }
516
517 // if this is as close as we've come, keep track of how close
518 // that is, and the line number of the rule that did it. If
519 // we've scored a direct hit, we don't have to look at any more
520 // rules
521 if (tempDifference < difference) {
522 difference = tempDifference;
523 winner = i;
524 if (difference == 0) {
525 break;
526 }
527 }
528 }
529
530 // if we have two successive rules that both have the winning base
531 // value, then the first one (the one we found above) is used if
532 // the numerator of the fraction is 1 and the second one is used if
533 // the numerator of the fraction is anything else (this lets us
534 // do things like "one third"/"two thirds" without haveing to define
535 // a whole bunch of extra rule sets)
536 if ((unsigned)(winner + 1) < rules.size() &&
537 rules[winner + 1]->getBaseValue() == rules[winner]->getBaseValue()) {
538 double n = ((double)rules[winner]->getBaseValue()) * number;
539 if (n < 0.5 || n >= 2) {
540 ++winner;
541 }
542 }
543
544 // finally, return the winning rule
545 return rules[winner];
546 }
547
548 /**
549 * Parses a string. Matches the string to be parsed against each
550 * of its rules (with a base value less than upperBound) and returns
551 * the value produced by the rule that matched the most charcters
552 * in the source string.
553 * @param text The string to parse
554 * @param parsePosition The initial position is ignored and assumed
555 * to be 0. On exit, this object has been updated to point to the
556 * first character position this rule set didn't consume.
557 * @param upperBound Limits the rules that can be allowed to match.
558 * Only rules whose base values are strictly less than upperBound
559 * are considered.
560 * @return The numerical result of parsing this string. This will
561 * be the matching rule's base value, composed appropriately with
562 * the results of matching any of its substitutions. The object
563 * will be an instance of Long if it's an integral value; otherwise,
564 * it will be an instance of Double. This function always returns
565 * a valid object: If nothing matched the input string at all,
566 * this function returns new Long(0), and the parse position is
567 * left unchanged.
568 */
569 #ifdef RBNF_DEBUG
570 #include <stdio.h>
571
dumpUS(FILE * f,const UnicodeString & us)572 static void dumpUS(FILE* f, const UnicodeString& us) {
573 int len = us.length();
574 char* buf = (char *)uprv_malloc((len+1)*sizeof(char)); //new char[len+1];
575 if (buf != NULL) {
576 us.extract(0, len, buf);
577 buf[len] = 0;
578 fprintf(f, "%s", buf);
579 uprv_free(buf); //delete[] buf;
580 }
581 }
582 #endif
583
584 UBool
parse(const UnicodeString & text,ParsePosition & pos,double upperBound,Formattable & result) const585 NFRuleSet::parse(const UnicodeString& text, ParsePosition& pos, double upperBound, Formattable& result) const
586 {
587 // try matching each rule in the rule set against the text being
588 // parsed. Whichever one matches the most characters is the one
589 // that determines the value we return.
590
591 result.setLong(0);
592
593 // dump out if there's no text to parse
594 if (text.length() == 0) {
595 return 0;
596 }
597
598 ParsePosition highWaterMark;
599 ParsePosition workingPos = pos;
600
601 #ifdef RBNF_DEBUG
602 fprintf(stderr, "<nfrs> %x '", this);
603 dumpUS(stderr, name);
604 fprintf(stderr, "' text '");
605 dumpUS(stderr, text);
606 fprintf(stderr, "'\n");
607 fprintf(stderr, " parse negative: %d\n", this, negativeNumberRule != 0);
608 #endif
609
610 // start by trying the negative number rule (if there is one)
611 if (negativeNumberRule) {
612 Formattable tempResult;
613 #ifdef RBNF_DEBUG
614 fprintf(stderr, " <nfrs before negative> %x ub: %g\n", negativeNumberRule, upperBound);
615 #endif
616 UBool success = negativeNumberRule->doParse(text, workingPos, 0, upperBound, tempResult);
617 #ifdef RBNF_DEBUG
618 fprintf(stderr, " <nfrs after negative> success: %d wpi: %d\n", success, workingPos.getIndex());
619 #endif
620 if (success && workingPos.getIndex() > highWaterMark.getIndex()) {
621 result = tempResult;
622 highWaterMark = workingPos;
623 }
624 workingPos = pos;
625 }
626 #ifdef RBNF_DEBUG
627 fprintf(stderr, "<nfrs> continue fractional with text '");
628 dumpUS(stderr, text);
629 fprintf(stderr, "' hwm: %d\n", highWaterMark.getIndex());
630 #endif
631 // then try each of the fraction rules
632 {
633 for (int i = 0; i < 3; i++) {
634 if (fractionRules[i]) {
635 Formattable tempResult;
636 UBool success = fractionRules[i]->doParse(text, workingPos, 0, upperBound, tempResult);
637 if (success && (workingPos.getIndex() > highWaterMark.getIndex())) {
638 result = tempResult;
639 highWaterMark = workingPos;
640 }
641 workingPos = pos;
642 }
643 }
644 }
645 #ifdef RBNF_DEBUG
646 fprintf(stderr, "<nfrs> continue other with text '");
647 dumpUS(stderr, text);
648 fprintf(stderr, "' hwm: %d\n", highWaterMark.getIndex());
649 #endif
650
651 // finally, go through the regular rules one at a time. We start
652 // at the end of the list because we want to try matching the most
653 // sigificant rule first (this helps ensure that we parse
654 // "five thousand three hundred six" as
655 // "(five thousand) (three hundred) (six)" rather than
656 // "((five thousand three) hundred) (six)"). Skip rules whose
657 // base values are higher than the upper bound (again, this helps
658 // limit ambiguity by making sure the rules that match a rule's
659 // are less significant than the rule containing the substitutions)/
660 {
661 int64_t ub = util64_fromDouble(upperBound);
662 #ifdef RBNF_DEBUG
663 {
664 char ubstr[64];
665 util64_toa(ub, ubstr, 64);
666 char ubstrhex[64];
667 util64_toa(ub, ubstrhex, 64, 16);
668 fprintf(stderr, "ub: %g, i64: %s (%s)\n", upperBound, ubstr, ubstrhex);
669 }
670 #endif
671 for (int32_t i = rules.size(); --i >= 0 && highWaterMark.getIndex() < text.length();) {
672 if ((!fIsFractionRuleSet) && (rules[i]->getBaseValue() >= ub)) {
673 continue;
674 }
675 Formattable tempResult;
676 UBool success = rules[i]->doParse(text, workingPos, fIsFractionRuleSet, upperBound, tempResult);
677 if (success && workingPos.getIndex() > highWaterMark.getIndex()) {
678 result = tempResult;
679 highWaterMark = workingPos;
680 }
681 workingPos = pos;
682 }
683 }
684 #ifdef RBNF_DEBUG
685 fprintf(stderr, "<nfrs> exit\n");
686 #endif
687 // finally, update the parse postion we were passed to point to the
688 // first character we didn't use, and return the result that
689 // corresponds to that string of characters
690 pos = highWaterMark;
691
692 return 1;
693 }
694
695 void
appendRules(UnicodeString & result) const696 NFRuleSet::appendRules(UnicodeString& result) const
697 {
698 // the rule set name goes first...
699 result.append(name);
700 result.append(gColon);
701 result.append(gLineFeed);
702
703 // followed by the regular rules...
704 for (uint32_t i = 0; i < rules.size(); i++) {
705 result.append(gFourSpaces);
706 rules[i]->_appendRuleText(result);
707 result.append(gLineFeed);
708 }
709
710 // followed by the special rules (if they exist)
711 if (negativeNumberRule) {
712 result.append(gFourSpaces);
713 negativeNumberRule->_appendRuleText(result);
714 result.append(gLineFeed);
715 }
716
717 {
718 for (uint32_t i = 0; i < 3; ++i) {
719 if (fractionRules[i]) {
720 result.append(gFourSpaces);
721 fractionRules[i]->_appendRuleText(result);
722 result.append(gLineFeed);
723 }
724 }
725 }
726 }
727
728 // utility functions
729
util64_fromDouble(double d)730 int64_t util64_fromDouble(double d) {
731 int64_t result = 0;
732 if (!uprv_isNaN(d)) {
733 double mant = uprv_maxMantissa();
734 if (d < -mant) {
735 d = -mant;
736 } else if (d > mant) {
737 d = mant;
738 }
739 UBool neg = d < 0;
740 if (neg) {
741 d = -d;
742 }
743 result = (int64_t)uprv_floor(d);
744 if (neg) {
745 result = -result;
746 }
747 }
748 return result;
749 }
750
util64_pow(int32_t r,uint32_t e)751 int64_t util64_pow(int32_t r, uint32_t e) {
752 if (r == 0) {
753 return 0;
754 } else if (e == 0) {
755 return 1;
756 } else {
757 int64_t n = r;
758 while (--e > 0) {
759 n *= r;
760 }
761 return n;
762 }
763 }
764
765 static const uint8_t asciiDigits[] = {
766 0x30u, 0x31u, 0x32u, 0x33u, 0x34u, 0x35u, 0x36u, 0x37u,
767 0x38u, 0x39u, 0x61u, 0x62u, 0x63u, 0x64u, 0x65u, 0x66u,
768 0x67u, 0x68u, 0x69u, 0x6au, 0x6bu, 0x6cu, 0x6du, 0x6eu,
769 0x6fu, 0x70u, 0x71u, 0x72u, 0x73u, 0x74u, 0x75u, 0x76u,
770 0x77u, 0x78u, 0x79u, 0x7au,
771 };
772
773 static const UChar kUMinus = (UChar)0x002d;
774
775 #ifdef RBNF_DEBUG
776 static const char kMinus = '-';
777
778 static const uint8_t digitInfo[] = {
779 0, 0, 0, 0, 0, 0, 0, 0,
780 0, 0, 0, 0, 0, 0, 0, 0,
781 0, 0, 0, 0, 0, 0, 0, 0,
782 0, 0, 0, 0, 0, 0, 0, 0,
783 0, 0, 0, 0, 0, 0, 0, 0,
784 0, 0, 0, 0, 0, 0, 0, 0,
785 0x80u, 0x81u, 0x82u, 0x83u, 0x84u, 0x85u, 0x86u, 0x87u,
786 0x88u, 0x89u, 0, 0, 0, 0, 0, 0,
787 0, 0x8au, 0x8bu, 0x8cu, 0x8du, 0x8eu, 0x8fu, 0x90u,
788 0x91u, 0x92u, 0x93u, 0x94u, 0x95u, 0x96u, 0x97u, 0x98u,
789 0x99u, 0x9au, 0x9bu, 0x9cu, 0x9du, 0x9eu, 0x9fu, 0xa0u,
790 0xa1u, 0xa2u, 0xa3u, 0, 0, 0, 0, 0,
791 0, 0x8au, 0x8bu, 0x8cu, 0x8du, 0x8eu, 0x8fu, 0x90u,
792 0x91u, 0x92u, 0x93u, 0x94u, 0x95u, 0x96u, 0x97u, 0x98u,
793 0x99u, 0x9au, 0x9bu, 0x9cu, 0x9du, 0x9eu, 0x9fu, 0xa0u,
794 0xa1u, 0xa2u, 0xa3u, 0, 0, 0, 0, 0,
795 };
796
util64_atoi(const char * str,uint32_t radix)797 int64_t util64_atoi(const char* str, uint32_t radix)
798 {
799 if (radix > 36) {
800 radix = 36;
801 } else if (radix < 2) {
802 radix = 2;
803 }
804 int64_t lradix = radix;
805
806 int neg = 0;
807 if (*str == kMinus) {
808 ++str;
809 neg = 1;
810 }
811 int64_t result = 0;
812 uint8_t b;
813 while ((b = digitInfo[*str++]) && ((b &= 0x7f) < radix)) {
814 result *= lradix;
815 result += (int32_t)b;
816 }
817 if (neg) {
818 result = -result;
819 }
820 return result;
821 }
822
util64_utoi(const UChar * str,uint32_t radix)823 int64_t util64_utoi(const UChar* str, uint32_t radix)
824 {
825 if (radix > 36) {
826 radix = 36;
827 } else if (radix < 2) {
828 radix = 2;
829 }
830 int64_t lradix = radix;
831
832 int neg = 0;
833 if (*str == kUMinus) {
834 ++str;
835 neg = 1;
836 }
837 int64_t result = 0;
838 UChar c;
839 uint8_t b;
840 while (((c = *str++) < 0x0080) && (b = digitInfo[c]) && ((b &= 0x7f) < radix)) {
841 result *= lradix;
842 result += (int32_t)b;
843 }
844 if (neg) {
845 result = -result;
846 }
847 return result;
848 }
849
util64_toa(int64_t w,char * buf,uint32_t len,uint32_t radix,UBool raw)850 uint32_t util64_toa(int64_t w, char* buf, uint32_t len, uint32_t radix, UBool raw)
851 {
852 if (radix > 36) {
853 radix = 36;
854 } else if (radix < 2) {
855 radix = 2;
856 }
857 int64_t base = radix;
858
859 char* p = buf;
860 if (len && (w < 0) && (radix == 10) && !raw) {
861 w = -w;
862 *p++ = kMinus;
863 --len;
864 } else if (len && (w == 0)) {
865 *p++ = (char)raw ? 0 : asciiDigits[0];
866 --len;
867 }
868
869 while (len && w != 0) {
870 int64_t n = w / base;
871 int64_t m = n * base;
872 int32_t d = (int32_t)(w-m);
873 *p++ = raw ? (char)d : asciiDigits[d];
874 w = n;
875 --len;
876 }
877 if (len) {
878 *p = 0; // null terminate if room for caller convenience
879 }
880
881 len = p - buf;
882 if (*buf == kMinus) {
883 ++buf;
884 }
885 while (--p > buf) {
886 char c = *p;
887 *p = *buf;
888 *buf = c;
889 ++buf;
890 }
891
892 return len;
893 }
894 #endif
895
util64_tou(int64_t w,UChar * buf,uint32_t len,uint32_t radix,UBool raw)896 uint32_t util64_tou(int64_t w, UChar* buf, uint32_t len, uint32_t radix, UBool raw)
897 {
898 if (radix > 36) {
899 radix = 36;
900 } else if (radix < 2) {
901 radix = 2;
902 }
903 int64_t base = radix;
904
905 UChar* p = buf;
906 if (len && (w < 0) && (radix == 10) && !raw) {
907 w = -w;
908 *p++ = kUMinus;
909 --len;
910 } else if (len && (w == 0)) {
911 *p++ = (UChar)raw ? 0 : asciiDigits[0];
912 --len;
913 }
914
915 while (len && (w != 0)) {
916 int64_t n = w / base;
917 int64_t m = n * base;
918 int32_t d = (int32_t)(w-m);
919 *p++ = (UChar)(raw ? d : asciiDigits[d]);
920 w = n;
921 --len;
922 }
923 if (len) {
924 *p = 0; // null terminate if room for caller convenience
925 }
926
927 len = (uint32_t)(p - buf);
928 if (*buf == kUMinus) {
929 ++buf;
930 }
931 while (--p > buf) {
932 UChar c = *p;
933 *p = *buf;
934 *buf = c;
935 ++buf;
936 }
937
938 return len;
939 }
940
941
942 U_NAMESPACE_END
943
944 /* U_HAVE_RBNF */
945 #endif
946
947