1 // © 2016 and later: Unicode, Inc. and others.
2 // License & terms of use: http://www.unicode.org/copyright.html
3 /*
4 *******************************************************************************
5 * Copyright (C) 2007-2014, International Business Machines Corporation and
6 * others. All Rights Reserved.
7 ********************************************************************************
8 
9 * File PLURULTS.cpp
10 *
11 ********************************************************************************
12 */
13 
14 #include "unicode/utypes.h"
15 
16 #if !UCONFIG_NO_FORMATTING
17 
18 #include <stdlib.h>
19 #include <stdarg.h>
20 #include <string.h>
21 
22 #include "unicode/localpointer.h"
23 #include "unicode/plurrule.h"
24 #include "unicode/stringpiece.h"
25 #include "unicode/numberformatter.h"
26 #include "unicode/numberrangeformatter.h"
27 
28 #include "cmemory.h"
29 #include "cstr.h"
30 #include "plurrule_impl.h"
31 #include "plurults.h"
32 #include "uhash.h"
33 #include "number_decimalquantity.h"
34 
35 using icu::number::impl::DecimalQuantity;
36 using namespace icu::number;
37 
38 void setupResult(const int32_t testSource[], char result[], int32_t* max);
39 UBool checkEqual(const PluralRules &test, char *result, int32_t max);
40 UBool testEquality(const PluralRules &test);
41 
42 // This is an API test, not a unit test.  It doesn't test very many cases, and doesn't
43 // try to test the full functionality.  It just calls each function in the class and
44 // verifies that it works on a basic level.
45 
runIndexedTest(int32_t index,UBool exec,const char * & name,char *)46 void PluralRulesTest::runIndexedTest( int32_t index, UBool exec, const char* &name, char* /*par*/ )
47 {
48     if (exec) logln("TestSuite PluralRulesAPI");
49     TESTCASE_AUTO_BEGIN;
50     TESTCASE_AUTO(testAPI);
51     // TESTCASE_AUTO(testGetUniqueKeywordValue);
52     TESTCASE_AUTO(testGetSamples);
53     TESTCASE_AUTO(testGetFixedDecimalSamples);
54     TESTCASE_AUTO(testSamplesWithExponent);
55     TESTCASE_AUTO(testWithin);
56     TESTCASE_AUTO(testGetAllKeywordValues);
57     TESTCASE_AUTO(testCompactDecimalPluralKeyword);
58     TESTCASE_AUTO(testOrdinal);
59     TESTCASE_AUTO(testSelect);
60     TESTCASE_AUTO(testSelectRange);
61     TESTCASE_AUTO(testAvailbleLocales);
62     TESTCASE_AUTO(testParseErrors);
63     TESTCASE_AUTO(testFixedDecimal);
64     TESTCASE_AUTO(testSelectTrailingZeros);
65     TESTCASE_AUTO(testLocaleExtension);
66     TESTCASE_AUTO_END;
67 }
68 
69 
70 // Quick and dirty class for putting UnicodeStrings in char * messages.
71 //   TODO: something like this should be generally available.
72 class US {
73   private:
74     char *buf;
75   public:
US(const UnicodeString & us)76     US(const UnicodeString &us) {
77        int32_t bufLen = us.extract((int32_t)0, us.length(), (char *)NULL, (uint32_t)0) + 1;
78        buf = (char *)uprv_malloc(bufLen);
79        us.extract(0, us.length(), buf, bufLen); }
cstr()80     const char *cstr() {return buf;}
~US()81     ~US() { uprv_free(buf);}
82 };
83 
84 
85 
86 
87 
88 #define PLURAL_TEST_NUM    18
89 /**
90  * Test various generic API methods of PluralRules for API coverage.
91  */
testAPI()92 void PluralRulesTest::testAPI(/*char *par*/)
93 {
94     UnicodeString pluralTestData[PLURAL_TEST_NUM] = {
95             UNICODE_STRING_SIMPLE("a: n is 1"),
96             UNICODE_STRING_SIMPLE("a: n mod 10 is 2"),
97             UNICODE_STRING_SIMPLE("a: n is not 1"),
98             UNICODE_STRING_SIMPLE("a: n mod 3 is not 1"),
99             UNICODE_STRING_SIMPLE("a: n in 2..5"),
100             UNICODE_STRING_SIMPLE("a: n within 2..5"),
101             UNICODE_STRING_SIMPLE("a: n not in 2..5"),
102             UNICODE_STRING_SIMPLE("a: n not within 2..5"),
103             UNICODE_STRING_SIMPLE("a: n mod 10 in 2..5"),
104             UNICODE_STRING_SIMPLE("a: n mod 10 within 2..5"),
105             UNICODE_STRING_SIMPLE("a: n mod 10 is 2 and n is not 12"),
106             UNICODE_STRING_SIMPLE("a: n mod 10 in 2..3 or n mod 10 is 5"),
107             UNICODE_STRING_SIMPLE("a: n mod 10 within 2..3 or n mod 10 is 5"),
108             UNICODE_STRING_SIMPLE("a: n is 1 or n is 4 or n is 23"),
109             UNICODE_STRING_SIMPLE("a: n mod 2 is 1 and n is not 3 and n in 1..11"),
110             UNICODE_STRING_SIMPLE("a: n mod 2 is 1 and n is not 3 and n within 1..11"),
111             UNICODE_STRING_SIMPLE("a: n mod 2 is 1 or n mod 5 is 1 and n is not 6"),
112             "",
113     };
114     static const int32_t pluralTestResult[PLURAL_TEST_NUM][30] = {
115         {1, 0},
116         {2,12,22, 0},
117         {0,2,3,4,5,0},
118         {0,2,3,5,6,8,9,0},
119         {2,3,4,5,0},
120         {2,3,4,5,0},
121         {0,1,6,7,8, 0},
122         {0,1,6,7,8, 0},
123         {2,3,4,5,12,13,14,15,22,23,24,25,0},
124         {2,3,4,5,12,13,14,15,22,23,24,25,0},
125         {2,22,32,42,0},
126         {2,3,5,12,13,15,22,23,25,0},
127         {2,3,5,12,13,15,22,23,25,0},
128         {1,4,23,0},
129         {1,5,7,9,11,0},
130         {1,5,7,9,11,0},
131         {1,3,5,7,9,11,13,15,16,0},
132     };
133     UErrorCode status = U_ZERO_ERROR;
134 
135     // ======= Test constructors
136     logln("Testing PluralRules constructors");
137 
138 
139     logln("\n start default locale test case ..\n");
140 
141     PluralRules defRule(status);
142     LocalPointer<PluralRules> test(new PluralRules(status), status);
143     if(U_FAILURE(status)) {
144         dataerrln("ERROR: Could not create PluralRules (default) - exitting");
145         return;
146     }
147     LocalPointer<PluralRules> newEnPlural(test->forLocale(Locale::getEnglish(), status), status);
148     if(U_FAILURE(status)) {
149         dataerrln("ERROR: Could not create PluralRules (English) - exitting");
150         return;
151     }
152 
153     // ======= Test clone, assignment operator && == operator.
154     LocalPointer<PluralRules> dupRule(defRule.clone());
155     if (dupRule==NULL) {
156         errln("ERROR: clone plural rules test failed!");
157         return;
158     } else {
159         if ( *dupRule != defRule ) {
160             errln("ERROR:  clone plural rules test failed!");
161         }
162     }
163     *dupRule = *newEnPlural;
164     if (dupRule!=NULL) {
165         if ( *dupRule != *newEnPlural ) {
166             errln("ERROR:  clone plural rules test failed!");
167         }
168     }
169 
170     // ======= Test empty plural rules
171     logln("Testing Simple PluralRules");
172 
173     LocalPointer<PluralRules> empRule(test->createRules(UNICODE_STRING_SIMPLE("a:n"), status));
174     UnicodeString key;
175     for (int32_t i=0; i<10; ++i) {
176         key = empRule->select(i);
177         if ( key.charAt(0)!= 0x61 ) { // 'a'
178             errln("ERROR:  empty plural rules test failed! - exitting");
179         }
180     }
181 
182     // ======= Test simple plural rules
183     logln("Testing Simple PluralRules");
184 
185     char result[100];
186     int32_t max;
187 
188     for (int32_t i=0; i<PLURAL_TEST_NUM-1; ++i) {
189        LocalPointer<PluralRules> newRules(test->createRules(pluralTestData[i], status));
190        setupResult(pluralTestResult[i], result, &max);
191        if ( !checkEqual(*newRules, result, max) ) {
192             errln("ERROR:  simple plural rules failed! - exitting");
193             return;
194         }
195     }
196 
197     // ======= Test complex plural rules
198     logln("Testing Complex PluralRules");
199     // TODO: the complex test data is hard coded. It's better to implement
200     // a parser to parse the test data.
201     UnicodeString complexRule = UNICODE_STRING_SIMPLE("a: n in 2..5; b: n in 5..8; c: n mod 2 is 1");
202     UnicodeString complexRule2 = UNICODE_STRING_SIMPLE("a: n within 2..5; b: n within 5..8; c: n mod 2 is 1");
203     char cRuleResult[] =
204     {
205        0x6F, // 'o'
206        0x63, // 'c'
207        0x61, // 'a'
208        0x61, // 'a'
209        0x61, // 'a'
210        0x61, // 'a'
211        0x62, // 'b'
212        0x62, // 'b'
213        0x62, // 'b'
214        0x63, // 'c'
215        0x6F, // 'o'
216        0x63  // 'c'
217     };
218     LocalPointer<PluralRules> newRules(test->createRules(complexRule, status));
219     if ( !checkEqual(*newRules, cRuleResult, 12) ) {
220          errln("ERROR:  complex plural rules failed! - exitting");
221          return;
222     }
223     newRules.adoptInstead(test->createRules(complexRule2, status));
224     if ( !checkEqual(*newRules, cRuleResult, 12) ) {
225          errln("ERROR:  complex plural rules failed! - exitting");
226          return;
227     }
228 
229     // ======= Test decimal fractions plural rules
230     UnicodeString decimalRule= UNICODE_STRING_SIMPLE("a: n not in 0..100;");
231     UnicodeString KEYWORD_A = UNICODE_STRING_SIMPLE("a");
232     status = U_ZERO_ERROR;
233     newRules.adoptInstead(test->createRules(decimalRule, status));
234     if (U_FAILURE(status)) {
235         dataerrln("ERROR: Could not create PluralRules for testing fractions - exitting");
236         return;
237     }
238     double fData[] =     {-101, -100, -1,     -0.0,  0,     0.1,  1,     1.999,  2.0,   100,   100.001 };
239     UBool isKeywordA[] = {TRUE, FALSE, FALSE, FALSE, FALSE, TRUE, FALSE, TRUE,   FALSE, FALSE, TRUE };
240     for (int32_t i=0; i<UPRV_LENGTHOF(fData); i++) {
241         if ((newRules->select(fData[i])== KEYWORD_A) != isKeywordA[i]) {
242              errln("File %s, Line %d, ERROR: plural rules for decimal fractions test failed!\n"
243                    "  number = %g, expected %s", __FILE__, __LINE__, fData[i], isKeywordA[i]?"TRUE":"FALSE");
244         }
245     }
246 
247     // ======= Test Equality
248     logln("Testing Equality of PluralRules");
249 
250     if ( !testEquality(*test) ) {
251          errln("ERROR:  complex plural rules failed! - exitting");
252          return;
253      }
254 
255 
256     // ======= Test getStaticClassID()
257     logln("Testing getStaticClassID()");
258 
259     if(test->getDynamicClassID() != PluralRules::getStaticClassID()) {
260         errln("ERROR: getDynamicClassID() didn't return the expected value");
261     }
262     // ====== Test fallback to parent locale
263     LocalPointer<PluralRules> en_UK(test->forLocale(Locale::getUK(), status));
264     LocalPointer<PluralRules> en(test->forLocale(Locale::getEnglish(), status));
265     if (en_UK.isValid() && en.isValid()) {
266         if ( *en_UK != *en ) {
267             errln("ERROR:  test locale fallback failed!");
268         }
269     }
270 
271     LocalPointer<PluralRules> zh_Hant(test->forLocale(Locale::getTaiwan(), status));
272     LocalPointer<PluralRules> zh(test->forLocale(Locale::getChinese(), status));
273     if (zh_Hant.isValid() && zh.isValid()) {
274         if ( *zh_Hant != *zh ) {
275             errln("ERROR:  test locale fallback failed!");
276         }
277     }
278 }
279 
setupResult(const int32_t testSource[],char result[],int32_t * max)280 void setupResult(const int32_t testSource[], char result[], int32_t* max) {
281     int32_t i=0;
282     int32_t curIndex=0;
283 
284     do {
285         while (curIndex < testSource[i]) {
286             result[curIndex++]=0x6F; //'o' other
287         }
288         result[curIndex++]=0x61; // 'a'
289 
290     } while(testSource[++i]>0);
291     *max=curIndex;
292 }
293 
294 
checkEqual(const PluralRules & test,char * result,int32_t max)295 UBool checkEqual(const PluralRules &test, char *result, int32_t max) {
296     UnicodeString key;
297     UBool isEqual = TRUE;
298     for (int32_t i=0; i<max; ++i) {
299         key= test.select(i);
300         if ( key.charAt(0)!=result[i] ) {
301             isEqual = FALSE;
302         }
303     }
304     return isEqual;
305 }
306 
307 
308 
309 static const int32_t MAX_EQ_ROW = 2;
310 static const int32_t MAX_EQ_COL = 5;
testEquality(const PluralRules & test)311 UBool testEquality(const PluralRules &test) {
312     UnicodeString testEquRules[MAX_EQ_ROW][MAX_EQ_COL] = {
313         {   UNICODE_STRING_SIMPLE("a: n in 2..3"),
314             UNICODE_STRING_SIMPLE("a: n is 2 or n is 3"),
315             UNICODE_STRING_SIMPLE( "a:n is 3 and n in 2..5 or n is 2"),
316             "",
317         },
318         {   UNICODE_STRING_SIMPLE("a: n is 12; b:n mod 10 in 2..3"),
319             UNICODE_STRING_SIMPLE("b: n mod 10 in 2..3 and n is not 12; a: n in 12..12"),
320             UNICODE_STRING_SIMPLE("b: n is 13; a: n in 12..13; b: n mod 10 is 2 or n mod 10 is 3"),
321             "",
322         }
323     };
324     UErrorCode status = U_ZERO_ERROR;
325     UnicodeString key[MAX_EQ_COL];
326     UBool ret=TRUE;
327     for (int32_t i=0; i<MAX_EQ_ROW; ++i) {
328         PluralRules* rules[MAX_EQ_COL];
329 
330         for (int32_t j=0; j<MAX_EQ_COL; ++j) {
331             rules[j]=NULL;
332         }
333         int32_t totalRules=0;
334         while((totalRules<MAX_EQ_COL) && (testEquRules[i][totalRules].length()>0) ) {
335             rules[totalRules]=test.createRules(testEquRules[i][totalRules], status);
336             totalRules++;
337         }
338         for (int32_t n=0; n<300 && ret ; ++n) {
339             for(int32_t j=0; j<totalRules;++j) {
340                 key[j] = rules[j]->select(n);
341             }
342             for(int32_t j=0; j<totalRules-1;++j) {
343                 if (key[j]!=key[j+1]) {
344                     ret= FALSE;
345                     break;
346                 }
347             }
348 
349         }
350         for (int32_t j=0; j<MAX_EQ_COL; ++j) {
351             if (rules[j]!=NULL) {
352                 delete rules[j];
353             }
354         }
355     }
356 
357     return ret;
358 }
359 
360 void
assertRuleValue(const UnicodeString & rule,double expected)361 PluralRulesTest::assertRuleValue(const UnicodeString& rule, double expected) {
362     assertRuleKeyValue("a:" + rule, "a", expected);
363 }
364 
365 void
assertRuleKeyValue(const UnicodeString & rule,const UnicodeString & key,double expected)366 PluralRulesTest::assertRuleKeyValue(const UnicodeString& rule,
367                                     const UnicodeString& key, double expected) {
368     UErrorCode status = U_ZERO_ERROR;
369     PluralRules *pr = PluralRules::createRules(rule, status);
370     double result = pr->getUniqueKeywordValue(key);
371     delete pr;
372     if (expected != result) {
373         errln("expected %g but got %g", expected, result);
374     }
375 }
376 
377 // TODO: UniqueKeywordValue() is not currently supported.
378 //       If it never will be, this test code should be removed.
testGetUniqueKeywordValue()379 void PluralRulesTest::testGetUniqueKeywordValue() {
380     assertRuleValue("n is 1", 1);
381     assertRuleValue("n in 2..2", 2);
382     assertRuleValue("n within 2..2", 2);
383     assertRuleValue("n in 3..4", UPLRULES_NO_UNIQUE_VALUE);
384     assertRuleValue("n within 3..4", UPLRULES_NO_UNIQUE_VALUE);
385     assertRuleValue("n is 2 or n is 2", 2);
386     assertRuleValue("n is 2 and n is 2", 2);
387     assertRuleValue("n is 2 or n is 3", UPLRULES_NO_UNIQUE_VALUE);
388     assertRuleValue("n is 2 and n is 3", UPLRULES_NO_UNIQUE_VALUE);
389     assertRuleValue("n is 2 or n in 2..3", UPLRULES_NO_UNIQUE_VALUE);
390     assertRuleValue("n is 2 and n in 2..3", 2);
391     assertRuleKeyValue("a: n is 1", "not_defined", UPLRULES_NO_UNIQUE_VALUE); // key not defined
392     assertRuleKeyValue("a: n is 1", "other", UPLRULES_NO_UNIQUE_VALUE); // key matches default rule
393 }
394 
testGetSamples()395 void PluralRulesTest::testGetSamples() {
396     // TODO: fix samples, re-enable this test.
397 
398     // no get functional equivalent API in ICU4C, so just
399     // test every locale...
400     UErrorCode status = U_ZERO_ERROR;
401     int32_t numLocales;
402     const Locale* locales = Locale::getAvailableLocales(numLocales);
403 
404     double values[1000];
405     for (int32_t i = 0; U_SUCCESS(status) && i < numLocales; ++i) {
406         if (uprv_strcmp(locales[i].getLanguage(), "fr") == 0 &&
407                 logKnownIssue("21299", "PluralRules::getSamples cannot distinguish 1e5 from 100000")) {
408             continue;
409         }
410         LocalPointer<PluralRules> rules(PluralRules::forLocale(locales[i], status));
411         if (U_FAILURE(status)) {
412             break;
413         }
414         LocalPointer<StringEnumeration> keywords(rules->getKeywords(status));
415         if (U_FAILURE(status)) {
416             break;
417         }
418         const UnicodeString* keyword;
419         while (NULL != (keyword = keywords->snext(status))) {
420             int32_t count = rules->getSamples(*keyword, values, UPRV_LENGTHOF(values), status);
421             if (U_FAILURE(status)) {
422                 errln(UnicodeString(u"getSamples() failed for locale ") +
423                       locales[i].getName() +
424                       UnicodeString(u", keyword ") + *keyword);
425                 continue;
426             }
427             if (count == 0) {
428                 // TODO: Lots of these.
429                 //   errln(UnicodeString(u"no samples for keyword ") + *keyword + UnicodeString(u" in locale ") + locales[i].getName() );
430             }
431             if (count > UPRV_LENGTHOF(values)) {
432                 errln(UnicodeString(u"getSamples()=") + count +
433                       UnicodeString(u", too many values, for locale ") +
434                       locales[i].getName() +
435                       UnicodeString(u", keyword ") + *keyword);
436                 count = UPRV_LENGTHOF(values);
437             }
438             for (int32_t j = 0; j < count; ++j) {
439                 if (values[j] == UPLRULES_NO_UNIQUE_VALUE) {
440                     errln("got 'no unique value' among values");
441                 } else {
442                     UnicodeString resultKeyword = rules->select(values[j]);
443                     // if (strcmp(locales[i].getName(), "uk") == 0) {    // Debug only.
444                     //     std::cout << "  uk " << US(resultKeyword).cstr() << " " << values[j] << std::endl;
445                     // }
446                     if (*keyword != resultKeyword) {
447                         errln("file %s, line %d, Locale %s, sample for keyword \"%s\":  %g, select(%g) returns keyword \"%s\"",
448                               __FILE__, __LINE__, locales[i].getName(), US(*keyword).cstr(), values[j], values[j], US(resultKeyword).cstr());
449                     }
450                 }
451             }
452         }
453     }
454 }
455 
testGetFixedDecimalSamples()456 void PluralRulesTest::testGetFixedDecimalSamples() {
457     // TODO: fix samples, re-enable this test.
458 
459     // no get functional equivalent API in ICU4C, so just
460     // test every locale...
461     UErrorCode status = U_ZERO_ERROR;
462     int32_t numLocales;
463     const Locale* locales = Locale::getAvailableLocales(numLocales);
464 
465     FixedDecimal values[1000];
466     for (int32_t i = 0; U_SUCCESS(status) && i < numLocales; ++i) {
467         if (uprv_strcmp(locales[i].getLanguage(), "fr") == 0 &&
468                 logKnownIssue("21299", "PluralRules::getSamples cannot distinguish 1e5 from 100000")) {
469             continue;
470         }
471         LocalPointer<PluralRules> rules(PluralRules::forLocale(locales[i], status));
472         if (U_FAILURE(status)) {
473             break;
474         }
475         LocalPointer<StringEnumeration> keywords(rules->getKeywords(status));
476         if (U_FAILURE(status)) {
477             break;
478         }
479         const UnicodeString* keyword;
480         while (NULL != (keyword = keywords->snext(status))) {
481             int32_t count = rules->getSamples(*keyword, values, UPRV_LENGTHOF(values), status);
482             if (U_FAILURE(status)) {
483                 errln(UnicodeString(u"getSamples() failed for locale ") +
484                       locales[i].getName() +
485                       UnicodeString(u", keyword ") + *keyword);
486                 continue;
487             }
488             if (count == 0) {
489                 // TODO: Lots of these.
490                 //   errln(UnicodeString(u"no samples for keyword ") + *keyword + UnicodeString(u" in locale ") + locales[i].getName() );
491             }
492             if (count > UPRV_LENGTHOF(values)) {
493                 errln(UnicodeString(u"getSamples()=") + count +
494                       UnicodeString(u", too many values, for locale ") +
495                       locales[i].getName() +
496                       UnicodeString(u", keyword ") + *keyword);
497                 count = UPRV_LENGTHOF(values);
498             }
499             for (int32_t j = 0; j < count; ++j) {
500                 if (values[j] == UPLRULES_NO_UNIQUE_VALUE_DECIMAL) {
501                     errln("got 'no unique value' among values");
502                 } else {
503                     UnicodeString resultKeyword = rules->select(values[j]);
504                     // if (strcmp(locales[i].getName(), "uk") == 0) {    // Debug only.
505                     //     std::cout << "  uk " << US(resultKeyword).cstr() << " " << values[j] << std::endl;
506                     // }
507                     if (*keyword != resultKeyword) {
508                         errln("file %s, line %d, Locale %s, sample for keyword \"%s\":  %s, select(%s) returns keyword \"%s\"",
509                                   __FILE__, __LINE__, locales[i].getName(), US(*keyword).cstr(), values[j].toString().getBuffer(), values[j].toString().getBuffer(), US(resultKeyword).cstr());
510                     }
511                 }
512             }
513         }
514     }
515 }
516 
testSamplesWithExponent()517 void PluralRulesTest::testSamplesWithExponent() {
518     // integer samples
519     UErrorCode status = U_ZERO_ERROR;
520     UnicodeString description(
521         u"one: i = 0,1 @integer 0, 1, 1e5 @decimal 0.0~1.5, 1.1e5; "
522         u"many: e = 0 and i != 0 and i % 1000000 = 0 and v = 0 or e != 0..5"
523         u" @integer 1000000, 2e6, 3e6, 4e6, 5e6, 6e6, 7e6, … @decimal 2.1e6, 3.1e6, 4.1e6, 5.1e6, 6.1e6, 7.1e6, …; "
524         u"other:  @integer 2~17, 100, 1000, 10000, 100000, 2e5, 3e5, 4e5, 5e5, 6e5, 7e5, …"
525         u" @decimal 2.0~3.5, 10.0, 100.0, 1000.0, 10000.0, 100000.0, 1000000.0, 2.1e5, 3.1e5, 4.1e5, 5.1e5, 6.1e5, 7.1e5, …"
526     );
527     LocalPointer<PluralRules> test(PluralRules::createRules(description, status));
528     if (U_FAILURE(status)) {
529         errln("Couldn't create plural rules from a string using exponent notation, with error = %s", u_errorName(status));
530         return;
531     }
532     checkNewSamples(description, test, u"one", u"@integer 0, 1, 1e5", FixedDecimal(0));
533     checkNewSamples(description, test, u"many", u"@integer 1000000, 2e6, 3e6, 4e6, 5e6, 6e6, 7e6, …", FixedDecimal(1000000));
534     checkNewSamples(description, test, u"other", u"@integer 2~17, 100, 1000, 10000, 100000, 2e5, 3e5, 4e5, 5e5, 6e5, 7e5, …", FixedDecimal(2));
535 
536     // decimal samples
537     status = U_ZERO_ERROR;
538     UnicodeString description2(
539         u"one: i = 0,1 @decimal 0.0~1.5, 1.1e5; "
540         u"many: e = 0 and i != 0 and i % 1000000 = 0 and v = 0 or e != 0..5"
541         u" @decimal 2.1e6, 3.1e6, 4.1e6, 5.1e6, 6.1e6, 7.1e6, …; "
542         u"other:  @decimal 2.0~3.5, 10.0, 100.0, 1000.0, 10000.0, 100000.0, 1000000.0, 2.1e5, 3.1e5, 4.1e5, 5.1e5, 6.1e5, 7.1e5, …"
543     );
544     LocalPointer<PluralRules> test2(PluralRules::createRules(description2, status));
545     if (U_FAILURE(status)) {
546         errln("Couldn't create plural rules from a string using exponent notation, with error = %s", u_errorName(status));
547         return;
548     }
549     checkNewSamples(description2, test2, u"one", u"@decimal 0.0~1.5, 1.1e5", FixedDecimal(0, 1));
550     checkNewSamples(description2, test2, u"many", u"@decimal 2.1e6, 3.1e6, 4.1e6, 5.1e6, 6.1e6, 7.1e6, …", FixedDecimal::createWithExponent(2.1, 1, 6));
551     checkNewSamples(description2, test2, u"other", u"@decimal 2.0~3.5, 10.0, 100.0, 1000.0, 10000.0, 100000.0, 1000000.0, 2.1e5, 3.1e5, 4.1e5, 5.1e5, 6.1e5, 7.1e5, …", FixedDecimal(2.0, 1));
552 }
553 
checkNewSamples(UnicodeString description,const LocalPointer<PluralRules> & test,UnicodeString keyword,UnicodeString samplesString,FixedDecimal firstInRange)554 void PluralRulesTest::checkNewSamples(
555         UnicodeString description,
556         const LocalPointer<PluralRules> &test,
557         UnicodeString keyword,
558         UnicodeString samplesString,
559         FixedDecimal firstInRange) {
560 
561     UErrorCode status = U_ZERO_ERROR;
562     FixedDecimal samples[1000];
563 
564     test->getSamples(keyword, samples, UPRV_LENGTHOF(samples), status);
565     if (U_FAILURE(status)) {
566         errln("Couldn't retrieve plural samples, with error = %s", u_errorName(status));
567         return;
568     }
569     FixedDecimal actualFirstSample = samples[0];
570 
571     if (!(firstInRange == actualFirstSample)) {
572         CStr descCstr(description);
573         CStr samplesCstr(samplesString);
574         char errMsg[1000];
575         snprintf(errMsg, sizeof(errMsg), "First parsed sample FixedDecimal not equal to expected for samples: %s in rule string: %s\n", descCstr(), samplesCstr());
576         errln(errMsg);
577     }
578 }
579 
testWithin()580 void PluralRulesTest::testWithin() {
581     // goes to show you what lack of testing will do.
582     // of course, this has been broken for two years and no one has noticed...
583     UErrorCode status = U_ZERO_ERROR;
584     PluralRules *rules = PluralRules::createRules("a: n mod 10 in 5..8", status);
585     if (!rules) {
586         errln("couldn't instantiate rules");
587         return;
588     }
589 
590     UnicodeString keyword = rules->select((int32_t)26);
591     if (keyword != "a") {
592         errln("expected 'a' for 26 but didn't get it.");
593     }
594 
595     keyword = rules->select(26.5);
596     if (keyword != "other") {
597         errln("expected 'other' for 26.5 but didn't get it.");
598     }
599 
600     delete rules;
601 }
602 
603 void
testGetAllKeywordValues()604 PluralRulesTest::testGetAllKeywordValues() {
605     const char* data[] = {
606         "a: n in 2..5", "a: 2,3,4,5; other: null; b:",
607         "a: n not in 2..5", "a: null; other: null",
608         "a: n within 2..5", "a: null; other: null",
609         "a: n not within 2..5", "a: null; other: null",
610         "a: n in 2..5 or n within 6..8", "a: null", // ignore 'other' here on out, always null
611         "a: n in 2..5 and n within 6..8", "a:",
612         "a: n in 2..5 and n within 5..8", "a: 5",
613         "a: n within 2..5 and n within 6..8", "a:", // our sampling catches these
614         "a: n within 2..5 and n within 5..8", "a: 5", // ''
615         "a: n within 1..2 and n within 2..3 or n within 3..4 and n within 4..5", "a: 2,4",
616         "a: n within 1..2 and n within 2..3 or n within 3..4 and n within 4..5 "
617           "or n within 5..6 and n within 6..7", "a: null", // but not this...
618         "a: n mod 3 is 0", "a: null",
619         "a: n mod 3 is 0 and n within 1..2", "a:",
620         "a: n mod 3 is 0 and n within 0..5", "a: 0,3",
621         "a: n mod 3 is 0 and n within 0..6", "a: null", // similarly with mod, we don't catch...
622         "a: n mod 3 is 0 and n in 3..12", "a: 3,6,9,12",
623         NULL
624     };
625 
626     for (int i = 0; data[i] != NULL; i += 2) {
627         UErrorCode status = U_ZERO_ERROR;
628         UnicodeString ruleDescription(data[i], -1, US_INV);
629         const char* result = data[i+1];
630 
631         logln("[%d] %s", i >> 1, data[i]);
632 
633         PluralRules *p = PluralRules::createRules(ruleDescription, status);
634         if (p == NULL || U_FAILURE(status)) {
635             errln("file %s, line %d: could not create rules from '%s'\n"
636                   "  ErrorCode: %s\n",
637                   __FILE__, __LINE__, data[i], u_errorName(status));
638             continue;
639         }
640 
641         // TODO: fix samples implementation, re-enable test.
642         (void)result;
643         #if 0
644 
645         const char* rp = result;
646         while (*rp) {
647             while (*rp == ' ') ++rp;
648             if (!rp) {
649                 break;
650             }
651 
652             const char* ep = rp;
653             while (*ep && *ep != ':') ++ep;
654 
655             status = U_ZERO_ERROR;
656             UnicodeString keyword(rp, ep - rp, US_INV);
657             double samples[4]; // no test above should have more samples than 4
658             int32_t count = p->getAllKeywordValues(keyword, &samples[0], 4, status);
659             if (U_FAILURE(status)) {
660                 errln("error getting samples for %s", rp);
661                 break;
662             }
663 
664             if (count > 4) {
665               errln("count > 4 for keyword %s", rp);
666               count = 4;
667             }
668 
669             if (*ep) {
670                 ++ep; // skip colon
671                 while (*ep && *ep == ' ') ++ep; // and spaces
672             }
673 
674             UBool ok = TRUE;
675             if (count == -1) {
676                 if (*ep != 'n') {
677                     errln("expected values for keyword %s but got -1 (%s)", rp, ep);
678                     ok = FALSE;
679                 }
680             } else if (*ep == 'n') {
681                 errln("expected count of -1, got %d, for keyword %s (%s)", count, rp, ep);
682                 ok = FALSE;
683             }
684 
685             // We'll cheat a bit here.  The samples happend to be in order and so are our
686             // expected values, so we'll just test in order until a failure.  If the
687             // implementation changes to return samples in an arbitrary order, this test
688             // must change.  There's no actual restriction on the order of the samples.
689 
690             for (int j = 0; ok && j < count; ++j ) { // we've verified count < 4
691                 double val = samples[j];
692                 if (*ep == 0 || *ep == ';') {
693                     errln("got unexpected value[%d]: %g", j, val);
694                     ok = FALSE;
695                     break;
696                 }
697                 char* xp;
698                 double expectedVal = strtod(ep, &xp);
699                 if (xp == ep) {
700                     // internal error
701                     errln("yikes!");
702                     ok = FALSE;
703                     break;
704                 }
705                 ep = xp;
706                 if (expectedVal != val) {
707                     errln("expected %g but got %g", expectedVal, val);
708                     ok = FALSE;
709                     break;
710                 }
711                 if (*ep == ',') ++ep;
712             }
713 
714             if (ok && count != -1) {
715                 if (!(*ep == 0 || *ep == ';')) {
716                     errln("file: %s, line %d, didn't get expected value: %s", __FILE__, __LINE__, ep);
717                     ok = FALSE;
718                 }
719             }
720 
721             while (*ep && *ep != ';') ++ep;
722             if (*ep == ';') ++ep;
723             rp = ep;
724         }
725     #endif
726     delete p;
727     }
728 }
729 
730 void
testCompactDecimalPluralKeyword()731 PluralRulesTest::testCompactDecimalPluralKeyword() {
732     IcuTestErrorCode errorCode(*this, "testCompactDecimalPluralKeyword");
733 
734     LocalPointer<PluralRules> rules(PluralRules::createRules(
735         u"one: i = 0,1 @integer 0, 1 @decimal 0.0~1.5;  "
736         u"many: e = 0 and i % 1000000 = 0 and v = 0 or e != 0 .. 5;  "
737         u"other:  @integer 2~17, 100, 1000, 10000, 100000, 1000000,  "
738         u"  @decimal 2.0~3.5, 10.0, 100.0, 1000.0, 10000.0, 100000.0, 1000000.0, …", errorCode));
739 
740     if (U_FAILURE(errorCode)) {
741         errln("Couldn't instantiate plurals rules from string, with error = %s", u_errorName(errorCode));
742         return;
743     }
744 
745     const char* localeName = "fr-FR";
746     Locale locale = Locale::createFromName(localeName);
747 
748     struct TestCase {
749         const char16_t* skeleton;
750         const int input;
751         const char16_t* expectedFormattedOutput;
752         const char16_t* expectedPluralRuleKeyword;
753     } cases[] = {
754         // unlocalized formatter skeleton, input, string output, plural rule keyword
755         {u"",             0, u"0", u"one"},
756         {u"compact-long", 0, u"0", u"one"},
757 
758         {u"",             1, u"1", u"one"},
759         {u"compact-long", 1, u"1", u"one"},
760 
761         {u"",             2, u"2", u"other"},
762         {u"compact-long", 2, u"2", u"other"},
763 
764         {u"",             1000000, u"1 000 000", u"many"},
765         {u"compact-long", 1000000, u"1 million", u"many"},
766 
767         {u"",             1000001, u"1 000 001", u"other"},
768         {u"compact-long", 1000001, u"1 million", u"many"},
769 
770         {u"",             120000,  u"1 200 000",    u"other"},
771         {u"compact-long", 1200000, u"1,2 millions", u"many"},
772 
773         {u"",             1200001, u"1 200 001",    u"other"},
774         {u"compact-long", 1200001, u"1,2 millions", u"many"},
775 
776         {u"",             2000000, u"2 000 000",  u"many"},
777         {u"compact-long", 2000000, u"2 millions", u"many"},
778     };
779     for (const auto& cas : cases) {
780         const char16_t* skeleton = cas.skeleton;
781         const int input = cas.input;
782         const char16_t* expectedPluralRuleKeyword = cas.expectedPluralRuleKeyword;
783 
784         UnicodeString actualPluralRuleKeyword =
785             getPluralKeyword(rules, locale, input, skeleton);
786 
787         UnicodeString message(UnicodeString(localeName) + u" " + DoubleToUnicodeString(input));
788         assertEquals(message, expectedPluralRuleKeyword, actualPluralRuleKeyword);
789     }
790 }
791 
getPluralKeyword(const LocalPointer<PluralRules> & rules,Locale locale,double number,const char16_t * skeleton)792 UnicodeString PluralRulesTest::getPluralKeyword(const LocalPointer<PluralRules> &rules, Locale locale, double number, const char16_t* skeleton) {
793     IcuTestErrorCode errorCode(*this, "getPluralKeyword");
794     UnlocalizedNumberFormatter ulnf = NumberFormatter::forSkeleton(skeleton, errorCode);
795     if (errorCode.errIfFailureAndReset("PluralRules::getPluralKeyword(<PluralRules>, <locale>, %d, %s) failed", number, skeleton)) {
796         return nullptr;
797     }
798     LocalizedNumberFormatter formatter = ulnf.locale(locale);
799 
800     const FormattedNumber fn = formatter.formatDouble(number, errorCode);
801     if (errorCode.errIfFailureAndReset("NumberFormatter::formatDouble(%d) failed", number)) {
802         return nullptr;
803     }
804 
805     UnicodeString pluralKeyword = rules->select(fn, errorCode);
806     if (errorCode.errIfFailureAndReset("PluralRules->select(FormattedNumber of %d) failed", number)) {
807         return nullptr;
808     }
809     return pluralKeyword;
810 }
811 
testOrdinal()812 void PluralRulesTest::testOrdinal() {
813     IcuTestErrorCode errorCode(*this, "testOrdinal");
814     LocalPointer<PluralRules> pr(PluralRules::forLocale("en", UPLURAL_TYPE_ORDINAL, errorCode));
815     if (errorCode.errIfFailureAndReset("PluralRules::forLocale(en, UPLURAL_TYPE_ORDINAL) failed")) {
816         return;
817     }
818     UnicodeString keyword = pr->select(2.);
819     if (keyword != UNICODE_STRING("two", 3)) {
820         dataerrln("PluralRules(en-ordinal).select(2) failed");
821     }
822 }
823 
824 
825 static const char * END_MARK = "999.999";    // Mark end of varargs data.
826 
checkSelect(const LocalPointer<PluralRules> & rules,UErrorCode & status,int32_t line,const char * keyword,...)827 void PluralRulesTest::checkSelect(const LocalPointer<PluralRules> &rules, UErrorCode &status,
828                                   int32_t line, const char *keyword, ...) {
829     // The varargs parameters are a const char* strings, each being a decimal number.
830     //   The formatting of the numbers as strings is significant, e.g.
831     //     the difference between "2" and "2.0" can affect which rule matches (which keyword is selected).
832     // Note: rules parameter is a LocalPointer reference rather than a PluralRules * to avoid having
833     //       to write getAlias() at every (numerous) call site.
834 
835     if (U_FAILURE(status)) {
836         errln("file %s, line %d, ICU error status: %s.", __FILE__, line, u_errorName(status));
837         status = U_ZERO_ERROR;
838         return;
839     }
840 
841     if (rules == NULL) {
842         errln("file %s, line %d: rules pointer is NULL", __FILE__, line);
843         return;
844     }
845 
846     va_list ap;
847     va_start(ap, keyword);
848     for (;;) {
849         const char *num = va_arg(ap, const char *);
850         if (strcmp(num, END_MARK) == 0) {
851             break;
852         }
853 
854         // DigitList is a convenient way to parse the decimal number string and get a double.
855         DecimalQuantity  dl;
856         dl.setToDecNumber(StringPiece(num), status);
857         if (U_FAILURE(status)) {
858             errln("file %s, line %d, ICU error status: %s.", __FILE__, line, u_errorName(status));
859             status = U_ZERO_ERROR;
860             continue;
861         }
862         double numDbl = dl.toDouble();
863         const char *decimalPoint = strchr(num, '.');
864         int fractionDigitCount = decimalPoint == NULL ? 0 : static_cast<int>((num + strlen(num) - 1) - decimalPoint);
865         int fractionDigits = fractionDigitCount == 0 ? 0 : atoi(decimalPoint + 1);
866         FixedDecimal ni(numDbl, fractionDigitCount, fractionDigits);
867 
868         UnicodeString actualKeyword = rules->select(ni);
869         if (actualKeyword != UnicodeString(keyword)) {
870             errln("file %s, line %d, select(%s) returned incorrect keyword. Expected %s, got %s",
871                    __FILE__, line, num, keyword, US(actualKeyword).cstr());
872         }
873     }
874     va_end(ap);
875 }
876 
testSelect()877 void PluralRulesTest::testSelect() {
878     UErrorCode status = U_ZERO_ERROR;
879     LocalPointer<PluralRules> pr(PluralRules::createRules("s: n in 1,3,4,6", status));
880     checkSelect(pr, status, __LINE__, "s", "1.0", "3.0", "4.0", "6.0", END_MARK);
881     checkSelect(pr, status, __LINE__, "other", "0.0", "2.0", "3.1", "7.0", END_MARK);
882 
883     pr.adoptInstead(PluralRules::createRules("s: n not in 1,3,4,6", status));
884     checkSelect(pr, status, __LINE__, "other", "1.0", "3.0", "4.0", "6.0", END_MARK);
885     checkSelect(pr, status, __LINE__, "s", "0.0", "2.0", "3.1", "7.0", END_MARK);
886 
887     pr.adoptInstead(PluralRules::createRules("r: n in 1..4, 7..10, 14 .. 17;"
888                                              "s: n is 29;", status));
889     checkSelect(pr, status, __LINE__, "r", "1.0", "3.0", "7.0", "8.0", "10.0", "14.0", "17.0", END_MARK);
890     checkSelect(pr, status, __LINE__, "s", "29.0", END_MARK);
891     checkSelect(pr, status, __LINE__, "other", "28.0", "29.1", END_MARK);
892 
893     pr.adoptInstead(PluralRules::createRules("a: n mod 10 is 1;  b: n mod 100 is 0 ", status));
894     checkSelect(pr, status, __LINE__, "a", "1", "11", "41", "101", "301.00", END_MARK);
895     checkSelect(pr, status, __LINE__, "b", "0", "100", "200.0", "300.", "1000", "1100", "110000", END_MARK);
896     checkSelect(pr, status, __LINE__, "other", "0.01", "1.01", "0.99", "2", "3", "99", "102", END_MARK);
897 
898     // Rules that end with or without a ';' and with or without trailing spaces.
899     //    (There was a rule parser bug here with these.)
900     pr.adoptInstead(PluralRules::createRules("a: n is 1", status));
901     checkSelect(pr, status, __LINE__, "a", "1", END_MARK);
902     checkSelect(pr, status, __LINE__, "other", "2", END_MARK);
903 
904     pr.adoptInstead(PluralRules::createRules("a: n is 1 ", status));
905     checkSelect(pr, status, __LINE__, "a", "1", END_MARK);
906     checkSelect(pr, status, __LINE__, "other", "2", END_MARK);
907 
908     pr.adoptInstead(PluralRules::createRules("a: n is 1;", status));
909     checkSelect(pr, status, __LINE__, "a", "1", END_MARK);
910     checkSelect(pr, status, __LINE__, "other", "2", END_MARK);
911 
912     pr.adoptInstead(PluralRules::createRules("a: n is 1 ; ", status));
913     checkSelect(pr, status, __LINE__, "a", "1", END_MARK);
914     checkSelect(pr, status, __LINE__, "other", "2", END_MARK);
915 
916     // First match when rules for different keywords are not disjoint.
917     //   Also try spacing variations around ':' and '..'
918     pr.adoptInstead(PluralRules::createRules("c: n in 5..15;  b : n in 1..10 ;a:n in 10 .. 20", status));
919     checkSelect(pr, status, __LINE__, "a", "20", END_MARK);
920     checkSelect(pr, status, __LINE__, "b", "1", END_MARK);
921     checkSelect(pr, status, __LINE__, "c", "10", END_MARK);
922     checkSelect(pr, status, __LINE__, "other", "0", "21", "10.1", END_MARK);
923 
924     // in vs within
925     pr.adoptInstead(PluralRules::createRules("a: n in 2..10; b: n within 8..15", status));
926     checkSelect(pr, status, __LINE__, "a", "2", "8", "10", END_MARK);
927     checkSelect(pr, status, __LINE__, "b", "8.01", "9.5", "11", "14.99", "15", END_MARK);
928     checkSelect(pr, status, __LINE__, "other", "1", "7.7", "15.01", "16", END_MARK);
929 
930     // OR and AND chains.
931     pr.adoptInstead(PluralRules::createRules("a: n in 2..10 and n in 4..12 and n not in 5..7", status));
932     checkSelect(pr, status, __LINE__, "a", "4", "8", "9", "10", END_MARK);
933     checkSelect(pr, status, __LINE__, "other", "2", "3", "5", "7", "11", END_MARK);
934     pr.adoptInstead(PluralRules::createRules("a: n is 2 or n is 5 or n in 7..11 and n in 11..13", status));
935     checkSelect(pr, status, __LINE__, "a", "2", "5", "11", END_MARK);
936     checkSelect(pr, status, __LINE__, "other", "3", "4", "6", "8", "10", "12", "13", END_MARK);
937 
938     // Number attributes -
939     //   n: the number itself
940     //   i: integer digits
941     //   f: visible fraction digits
942     //   t: f with trailing zeros removed.
943     //   v: number of visible fraction digits
944     //   j: = n if there are no visible fraction digits
945     //      != anything if there are visible fraction digits
946 
947     pr.adoptInstead(PluralRules::createRules("a: i is 123", status));
948     checkSelect(pr, status, __LINE__, "a", "123", "123.0", "123.1", "0123.99", END_MARK);
949     checkSelect(pr, status, __LINE__, "other", "124", "122.0", END_MARK);
950 
951     pr.adoptInstead(PluralRules::createRules("a: f is 120", status));
952     checkSelect(pr, status, __LINE__, "a", "1.120", "0.120", "11123.120", "0123.120", END_MARK);
953     checkSelect(pr, status, __LINE__, "other", "1.121", "122.1200", "1.12", "120", END_MARK);
954 
955     pr.adoptInstead(PluralRules::createRules("a: t is 12", status));
956     checkSelect(pr, status, __LINE__, "a", "1.120", "0.12", "11123.12000", "0123.1200000", END_MARK);
957     checkSelect(pr, status, __LINE__, "other", "1.121", "122.1200001", "1.11", "12", END_MARK);
958 
959     pr.adoptInstead(PluralRules::createRules("a: v is 3", status));
960     checkSelect(pr, status, __LINE__, "a", "1.120", "0.000", "11123.100", "0123.124", ".666", END_MARK);
961     checkSelect(pr, status, __LINE__, "other", "1.1212", "122.12", "1.1", "122", "0.0000", END_MARK);
962 
963     pr.adoptInstead(PluralRules::createRules("a: v is 0 and i is 123", status));
964     checkSelect(pr, status, __LINE__, "a", "123", "123.", END_MARK);
965     checkSelect(pr, status, __LINE__, "other", "123.0", "123.1", "123.123", "0.123", END_MARK);
966 
967     // The reserved words from the rule syntax will also function as keywords.
968     pr.adoptInstead(PluralRules::createRules("a: n is 21; n: n is 22; i: n is 23; f: n is 24;"
969                                              "t: n is 25; v: n is 26; w: n is 27; j: n is 28"
970                                              , status));
971     checkSelect(pr, status, __LINE__, "other", "20", "29", END_MARK);
972     checkSelect(pr, status, __LINE__, "a", "21", END_MARK);
973     checkSelect(pr, status, __LINE__, "n", "22", END_MARK);
974     checkSelect(pr, status, __LINE__, "i", "23", END_MARK);
975     checkSelect(pr, status, __LINE__, "f", "24", END_MARK);
976     checkSelect(pr, status, __LINE__, "t", "25", END_MARK);
977     checkSelect(pr, status, __LINE__, "v", "26", END_MARK);
978     checkSelect(pr, status, __LINE__, "w", "27", END_MARK);
979     checkSelect(pr, status, __LINE__, "j", "28", END_MARK);
980 
981 
982     pr.adoptInstead(PluralRules::createRules("not: n=31; and: n=32; or: n=33; mod: n=34;"
983                                              "in: n=35; within: n=36;is:n=37"
984                                              , status));
985     checkSelect(pr, status, __LINE__, "other",  "30", "39", END_MARK);
986     checkSelect(pr, status, __LINE__, "not",    "31", END_MARK);
987     checkSelect(pr, status, __LINE__, "and",    "32", END_MARK);
988     checkSelect(pr, status, __LINE__, "or",     "33", END_MARK);
989     checkSelect(pr, status, __LINE__, "mod",    "34", END_MARK);
990     checkSelect(pr, status, __LINE__, "in",     "35", END_MARK);
991     checkSelect(pr, status, __LINE__, "within", "36", END_MARK);
992     checkSelect(pr, status, __LINE__, "is",     "37", END_MARK);
993 
994 // Test cases from ICU4J PluralRulesTest.parseTestData
995 
996     pr.adoptInstead(PluralRules::createRules("a: n is 1", status));
997     checkSelect(pr, status, __LINE__, "a", "1", END_MARK);
998     pr.adoptInstead(PluralRules::createRules("a: n mod 10 is 2", status));
999     checkSelect(pr, status, __LINE__, "a", "2", "12", "22", END_MARK);
1000     pr.adoptInstead(PluralRules::createRules("a: n is not 1", status));
1001     checkSelect(pr, status, __LINE__, "a", "0", "2", "3", "4", "5", END_MARK);
1002     pr.adoptInstead(PluralRules::createRules("a: n mod 3 is not 1", status));
1003     checkSelect(pr, status, __LINE__, "a", "0", "2", "3", "5", "6", "8", "9", END_MARK);
1004     pr.adoptInstead(PluralRules::createRules("a: n in 2..5", status));
1005     checkSelect(pr, status, __LINE__, "a", "2", "3", "4", "5", END_MARK);
1006     pr.adoptInstead(PluralRules::createRules("a: n within 2..5", status));
1007     checkSelect(pr, status, __LINE__, "a", "2", "3", "4", "5", END_MARK);
1008     pr.adoptInstead(PluralRules::createRules("a: n not in 2..5", status));
1009     checkSelect(pr, status, __LINE__, "a", "0", "1", "6", "7", "8", END_MARK);
1010     pr.adoptInstead(PluralRules::createRules("a: n not within 2..5", status));
1011     checkSelect(pr, status, __LINE__, "a", "0", "1", "6", "7", "8", END_MARK);
1012     pr.adoptInstead(PluralRules::createRules("a: n mod 10 in 2..5", status));
1013     checkSelect(pr, status, __LINE__, "a", "2", "3", "4", "5", "12", "13", "14", "15", "22", "23", "24", "25", END_MARK);
1014     pr.adoptInstead(PluralRules::createRules("a: n mod 10 within 2..5", status));
1015     checkSelect(pr, status, __LINE__, "a", "2", "3", "4", "5", "12", "13", "14", "15", "22", "23", "24", "25", END_MARK);
1016     pr.adoptInstead(PluralRules::createRules("a: n mod 10 is 2 and n is not 12", status));
1017     checkSelect(pr, status, __LINE__, "a", "2", "22", "32", "42", END_MARK);
1018     pr.adoptInstead(PluralRules::createRules("a: n mod 10 in 2..3 or n mod 10 is 5", status));
1019     checkSelect(pr, status, __LINE__, "a", "2", "3", "5", "12", "13", "15", "22", "23", "25", END_MARK);
1020     pr.adoptInstead(PluralRules::createRules("a: n mod 10 within 2..3 or n mod 10 is 5", status));
1021     checkSelect(pr, status, __LINE__, "a", "2", "3", "5", "12", "13", "15", "22", "23", "25", END_MARK);
1022     pr.adoptInstead(PluralRules::createRules("a: n is 1 or n is 4 or n is 23", status));
1023     checkSelect(pr, status, __LINE__, "a", "1", "4", "23", END_MARK);
1024     pr.adoptInstead(PluralRules::createRules("a: n mod 2 is 1 and n is not 3 and n in 1..11", status));
1025     checkSelect(pr, status, __LINE__, "a", "1", "5", "7", "9", "11", END_MARK);
1026     pr.adoptInstead(PluralRules::createRules("a: n mod 2 is 1 and n is not 3 and n within 1..11", status));
1027     checkSelect(pr, status, __LINE__, "a", "1", "5", "7", "9", "11", END_MARK);
1028     pr.adoptInstead(PluralRules::createRules("a: n mod 2 is 1 or n mod 5 is 1 and n is not 6", status));
1029     checkSelect(pr, status, __LINE__, "a", "1", "3", "5", "7", "9", "11", "13", "15", "16", END_MARK);
1030     pr.adoptInstead(PluralRules::createRules("a: n in 2..5; b: n in 5..8; c: n mod 2 is 1", status));
1031     checkSelect(pr, status, __LINE__, "a", "2", "3", "4", "5", END_MARK);
1032     checkSelect(pr, status, __LINE__, "b", "6", "7", "8", END_MARK);
1033     checkSelect(pr, status, __LINE__, "c", "1", "9", "11", END_MARK);
1034     pr.adoptInstead(PluralRules::createRules("a: n within 2..5; b: n within 5..8; c: n mod 2 is 1", status));
1035     checkSelect(pr, status, __LINE__, "a", "2", "3", "4", "5", END_MARK);
1036     checkSelect(pr, status, __LINE__, "b", "6", "7", "8", END_MARK);
1037     checkSelect(pr, status, __LINE__, "c", "1", "9", "11", END_MARK);
1038     pr.adoptInstead(PluralRules::createRules("a: n in 2, 4..6; b: n within 7..9,11..12,20", status));
1039     checkSelect(pr, status, __LINE__, "a", "2", "4", "5", "6", END_MARK);
1040     checkSelect(pr, status, __LINE__, "b", "7", "8", "9", "11", "12", "20", END_MARK);
1041     pr.adoptInstead(PluralRules::createRules("a: n in 2..8, 12 and n not in 4..6", status));
1042     checkSelect(pr, status, __LINE__, "a", "2", "3", "7", "8", "12", END_MARK);
1043     pr.adoptInstead(PluralRules::createRules("a: n mod 10 in 2, 3,5..7 and n is not 12", status));
1044     checkSelect(pr, status, __LINE__, "a", "2", "3", "5", "6", "7", "13", "15", "16", "17", END_MARK);
1045     pr.adoptInstead(PluralRules::createRules("a: n in 2..6, 3..7", status));
1046     checkSelect(pr, status, __LINE__, "a", "2", "3", "4", "5", "6", "7", END_MARK);
1047 
1048     // Extended Syntax, with '=', '!=' and '%' operators.
1049     pr.adoptInstead(PluralRules::createRules("a: n = 1..8 and n!= 2,3,4,5", status));
1050     checkSelect(pr, status, __LINE__, "a", "1", "6", "7", "8", END_MARK);
1051     checkSelect(pr, status, __LINE__, "other", "0", "2", "3", "4", "5", "9", END_MARK);
1052     pr.adoptInstead(PluralRules::createRules("a:n % 10 != 1", status));
1053     checkSelect(pr, status, __LINE__, "a", "2", "6", "7", "8", END_MARK);
1054     checkSelect(pr, status, __LINE__, "other", "1", "21", "211", "91", END_MARK);
1055 }
1056 
1057 
testSelectRange()1058 void PluralRulesTest::testSelectRange() {
1059     IcuTestErrorCode status(*this, "testSelectRange");
1060 
1061     int32_t d1 = 102;
1062     int32_t d2 = 201;
1063     Locale locale("sl");
1064 
1065     // Locale sl has interesting data: one + two => few
1066     auto range = NumberRangeFormatter::withLocale(locale).formatFormattableRange(d1, d2, status);
1067     auto rules = LocalPointer<PluralRules>(PluralRules::forLocale(locale, status), status);
1068     if (status.errIfFailureAndReset()) {
1069         return;
1070     }
1071 
1072     // For testing: get plural form of first and second numbers
1073     auto a = NumberFormatter::withLocale(locale).formatDouble(d1, status);
1074     auto b = NumberFormatter::withLocale(locale).formatDouble(d2, status);
1075     assertEquals("First plural", u"two", rules->select(a, status));
1076     assertEquals("Second plural", u"one", rules->select(b, status));
1077 
1078     // Check the range plural now:
1079     auto form = rules->select(range, status);
1080     assertEquals("Range plural", u"few", form);
1081 
1082     // Test after copying:
1083     PluralRules copy(*rules);
1084     form = copy.select(range, status);
1085     assertEquals("Range plural after copying", u"few", form);
1086 
1087     // Test when plural ranges data is unavailable:
1088     auto bare = LocalPointer<PluralRules>(
1089         PluralRules::createRules(u"a: i = 0,1", status), status);
1090     if (status.errIfFailureAndReset()) { return; }
1091     form = bare->select(range, status);
1092     status.expectErrorAndReset(U_UNSUPPORTED_ERROR);
1093 
1094     // However, they should not set an error when no data is available for a language.
1095     auto xyz = LocalPointer<PluralRules>(
1096         PluralRules::forLocale("xyz", status));
1097     form = xyz->select(range, status);
1098     assertEquals("Fallback form", u"other", form);
1099 }
1100 
1101 
testAvailbleLocales()1102 void PluralRulesTest::testAvailbleLocales() {
1103 
1104     // Hash set of (char *) strings.
1105     UErrorCode status = U_ZERO_ERROR;
1106     UHashtable *localeSet = uhash_open(uhash_hashUnicodeString, uhash_compareUnicodeString, uhash_compareLong, &status);
1107     uhash_setKeyDeleter(localeSet, uprv_deleteUObject);
1108     if (U_FAILURE(status)) {
1109         errln("file %s,  line %d: Error status = %s", __FILE__, __LINE__, u_errorName(status));
1110         return;
1111     }
1112 
1113     // Check that each locale returned by the iterator is unique.
1114     StringEnumeration *localesEnum = PluralRules::getAvailableLocales(status);
1115     int localeCount = 0;
1116     for (;;) {
1117         const char *locale = localesEnum->next(NULL, status);
1118         if (U_FAILURE(status)) {
1119             dataerrln("file %s,  line %d: Error status = %s", __FILE__, __LINE__, u_errorName(status));
1120             return;
1121         }
1122         if (locale == NULL) {
1123             break;
1124         }
1125         localeCount++;
1126         int32_t oldVal = uhash_puti(localeSet, new UnicodeString(locale), 1, &status);
1127         if (oldVal != 0) {
1128             errln("file %s,  line %d: locale %s was seen before.", __FILE__, __LINE__, locale);
1129         }
1130     }
1131 
1132     // Reset the iterator, verify that we get the same count.
1133     localesEnum->reset(status);
1134     int32_t localeCount2 = 0;
1135     while (localesEnum->next(NULL, status) != NULL) {
1136         if (U_FAILURE(status)) {
1137             errln("file %s,  line %d: Error status = %s", __FILE__, __LINE__, u_errorName(status));
1138             break;
1139         }
1140         localeCount2++;
1141     }
1142     if (localeCount != localeCount2) {
1143         errln("file %s,  line %d: locale counts differ. They are (%d, %d)",
1144             __FILE__, __LINE__, localeCount, localeCount2);
1145     }
1146 
1147     // Instantiate plural rules for each available locale.
1148     localesEnum->reset(status);
1149     for (;;) {
1150         status = U_ZERO_ERROR;
1151         const char *localeName = localesEnum->next(NULL, status);
1152         if (U_FAILURE(status)) {
1153             errln("file %s,  line %d: Error status = %s, locale = %s",
1154                 __FILE__, __LINE__, u_errorName(status), localeName);
1155             return;
1156         }
1157         if (localeName == NULL) {
1158             break;
1159         }
1160         Locale locale = Locale::createFromName(localeName);
1161         PluralRules *pr = PluralRules::forLocale(locale, status);
1162         if (U_FAILURE(status)) {
1163             errln("file %s,  line %d: Error %s creating plural rules for locale %s",
1164                 __FILE__, __LINE__, u_errorName(status), localeName);
1165             continue;
1166         }
1167         if (pr == NULL) {
1168             errln("file %s, line %d: Null plural rules for locale %s", __FILE__, __LINE__, localeName);
1169             continue;
1170         }
1171 
1172         // Pump some numbers through the plural rules.  Can't check for correct results,
1173         // mostly this to tickle any asserts or crashes that may be lurking.
1174         for (double n=0; n<120.0; n+=0.5) {
1175             UnicodeString keyword = pr->select(n);
1176             if (keyword.length() == 0) {
1177                 errln("file %s, line %d, empty keyword for n = %g, locale %s",
1178                     __FILE__, __LINE__, n, localeName);
1179             }
1180         }
1181         delete pr;
1182     }
1183 
1184     uhash_close(localeSet);
1185     delete localesEnum;
1186 
1187 }
1188 
1189 
testParseErrors()1190 void PluralRulesTest::testParseErrors() {
1191     // Test rules with syntax errors.
1192     // Creation of PluralRules from them should fail.
1193 
1194     static const char *testCases[] = {
1195             "a: n mod 10, is 1",
1196             "a: q is 13",
1197             "a  n is 13",
1198             "a: n is 13,",
1199             "a: n is 13, 15,   b: n is 4",
1200             "a: n is 1, 3, 4.. ",
1201             "a: n within 5..4",
1202             "A: n is 13",          // Uppercase keywords not allowed.
1203             "a: n ! = 3",          // spaces in != operator
1204             "a: n = not 3",        // '=' not exact equivalent of 'is'
1205             "a: n ! in 3..4"       // '!' not exact equivalent of 'not'
1206             "a: n % 37 ! in 3..4"
1207 
1208             };
1209     for (int i=0; i<UPRV_LENGTHOF(testCases); i++) {
1210         const char *rules = testCases[i];
1211         UErrorCode status = U_ZERO_ERROR;
1212         PluralRules *pr = PluralRules::createRules(UnicodeString(rules), status);
1213         if (U_SUCCESS(status)) {
1214             errln("file %s, line %d, expected failure with \"%s\".", __FILE__, __LINE__, rules);
1215         }
1216         if (pr != NULL) {
1217             errln("file %s, line %d, expected NULL. Rules: \"%s\"", __FILE__, __LINE__, rules);
1218         }
1219     }
1220     return;
1221 }
1222 
1223 
testFixedDecimal()1224 void PluralRulesTest::testFixedDecimal() {
1225     struct DoubleTestCase {
1226         double n;
1227         int32_t fractionDigitCount;
1228         int64_t fractionDigits;
1229     };
1230 
1231     // Check that the internal functions for extracting the decimal fraction digits from
1232     //   a double value are working.
1233     static DoubleTestCase testCases[] = {
1234         {1.0, 0, 0},
1235         {123456.0, 0, 0},
1236         {1.1, 1, 1},
1237         {1.23, 2, 23},
1238         {1.234, 3, 234},
1239         {1.2345, 4, 2345},
1240         {1.23456, 5, 23456},
1241         {.1234, 4, 1234},
1242         {.01234, 5, 1234},
1243         {.001234, 6, 1234},
1244         {.0001234, 7, 1234},
1245         {100.1234, 4, 1234},
1246         {100.01234, 5, 1234},
1247         {100.001234, 6, 1234},
1248         {100.0001234, 7, 1234}
1249     };
1250 
1251     for (int i=0; i<UPRV_LENGTHOF(testCases); ++i) {
1252         DoubleTestCase &tc = testCases[i];
1253         int32_t numFractionDigits = FixedDecimal::decimals(tc.n);
1254         if (numFractionDigits != tc.fractionDigitCount) {
1255             errln("file %s, line %d: decimals(%g) expected %d, actual %d",
1256                    __FILE__, __LINE__, tc.n, tc.fractionDigitCount, numFractionDigits);
1257             continue;
1258         }
1259         int64_t actualFractionDigits = FixedDecimal::getFractionalDigits(tc.n, numFractionDigits);
1260         if (actualFractionDigits != tc.fractionDigits) {
1261             errln("file %s, line %d: getFractionDigits(%g, %d): expected %ld, got %ld",
1262                   __FILE__, __LINE__, tc.n, numFractionDigits, tc.fractionDigits, actualFractionDigits);
1263         }
1264     }
1265 }
1266 
1267 
testSelectTrailingZeros()1268 void PluralRulesTest::testSelectTrailingZeros() {
1269     IcuTestErrorCode status(*this, "testSelectTrailingZeros");
1270     number::UnlocalizedNumberFormatter unf = number::NumberFormatter::with()
1271             .precision(number::Precision::fixedFraction(2));
1272     struct TestCase {
1273         const char* localeName;
1274         const char16_t* expectedDoubleKeyword;
1275         const char16_t* expectedFormattedKeyword;
1276         double number;
1277     } cases[] = {
1278         {"bs",  u"few",   u"other", 5.2},  // 5.2 => two, but 5.20 => other
1279         {"si",  u"one",   u"one",   0.0},
1280         {"si",  u"one",   u"one",   1.0},
1281         {"si",  u"one",   u"other", 0.1},  // 0.1 => one, but 0.10 => other
1282         {"si",  u"one",   u"one",   0.01}, // 0.01 => one
1283         {"hsb", u"few",   u"few",   1.03}, // (f % 100 == 3) => few
1284         {"hsb", u"few",   u"other", 1.3},  // 1.3 => few, but 1.30 => other
1285     };
1286     for (const auto& cas : cases) {
1287         UnicodeString message(UnicodeString(cas.localeName) + u" " + DoubleToUnicodeString(cas.number));
1288         status.setScope(message);
1289         Locale locale(cas.localeName);
1290         LocalPointer<PluralRules> rules(PluralRules::forLocale(locale, status));
1291         if (U_FAILURE(status)) {
1292             dataerrln("Failed to create PluralRules by PluralRules::forLocale(%s): %s\n",
1293                       cas.localeName, u_errorName(status));
1294             return;
1295         }
1296         assertEquals(message, cas.expectedDoubleKeyword, rules->select(cas.number));
1297         number::FormattedNumber fn = unf.locale(locale).formatDouble(cas.number, status);
1298         assertEquals(message, cas.expectedFormattedKeyword, rules->select(fn, status));
1299         status.errIfFailureAndReset();
1300     }
1301 }
1302 
compareLocaleResults(const char * loc1,const char * loc2,const char * loc3)1303 void PluralRulesTest::compareLocaleResults(const char* loc1, const char* loc2, const char* loc3) {
1304     UErrorCode status = U_ZERO_ERROR;
1305     LocalPointer<PluralRules> rules1(PluralRules::forLocale(loc1, status));
1306     LocalPointer<PluralRules> rules2(PluralRules::forLocale(loc2, status));
1307     LocalPointer<PluralRules> rules3(PluralRules::forLocale(loc3, status));
1308     if (U_FAILURE(status)) {
1309         dataerrln("Failed to create PluralRules for one of %s, %s, %s: %s\n", loc1, loc2, loc3, u_errorName(status));
1310         return;
1311     }
1312     for (int32_t value = 0; value <= 12; value++) {
1313         UnicodeString result1 = rules1->select(value);
1314         UnicodeString result2 = rules2->select(value);
1315         UnicodeString result3 = rules3->select(value);
1316         if (result1 != result2 || result1 != result3) {
1317             errln("PluralRules.select(%d) does not return the same values for %s, %s, %s\n", value, loc1, loc2, loc3);
1318         }
1319     }
1320 }
1321 
testLocaleExtension()1322 void PluralRulesTest::testLocaleExtension() {
1323     IcuTestErrorCode errorCode(*this, "testLocaleExtension");
1324     LocalPointer<PluralRules> rules(PluralRules::forLocale("pt@calendar=gregorian", errorCode));
1325     if (errorCode.errIfFailureAndReset("PluralRules::forLocale()")) { return; }
1326     UnicodeString key = rules->select(1);
1327     assertEquals("pt@calendar=gregorian select(1)", u"one", key);
1328     compareLocaleResults("ar", "ar_SA", "ar_SA@calendar=gregorian");
1329     compareLocaleResults("ru", "ru_UA", "ru-u-cu-RUB");
1330     compareLocaleResults("fr", "fr_CH", "fr@ms=uksystem");
1331 }
1332 
1333 #endif /* #if !UCONFIG_NO_FORMATTING */
1334