icuSources/test/intltest/rbbitst.h

   1 // © 2016 and later: Unicode, Inc. and others.
   2 // License & terms of use: http://www.unicode.org/copyright.html
   3 /*************************************************************************
   4  * Copyright (c) 1999-2016, International Business Machines
   5  * Corporation and others. All Rights Reserved.
   6  *************************************************************************
   7  *   Date        Name        Description
   8  *   12/15/99    Madhu        Creation.
   9  *   01/12/2000  Madhu        Updated for changed API and added new tests
  10  ************************************************************************/
  11
  12
  13 #ifndef RBBITEST_H
  14 #define RBBITEST_H
  15
  16 #include "unicode/utypes.h"
  17
  18 #if !UCONFIG_NO_BREAK_ITERATION
  19
  20 #include <memory>
  21
  22 #include "intltest.h"
  23 #include "unicode/brkiter.h"
  24 #include "unicode/rbbi.h"
  25
  26 class  Enumeration;
  27 class  BITestData;
  28 struct TestParams;
  29 class  RBBIMonkeyKind;
  30
  31 U_NAMESPACE_BEGIN
  32 class  UVector32;
  33 U_NAMESPACE_END
  34
  35 /**
  36  * Test the RuleBasedBreakIterator class giving different rules
  37  */
  38 class RBBITest: public IntlTest {
  39 public:
  40
  41     RBBITest();
  42     virtual ~RBBITest();
  43
  44     void runIndexedTest( int32_t index, UBool exec, const char* &name, char* par = NULL );
  45
  46     void TestGetAvailableLocales();
  47     void TestGetDisplayName();
  48     void TestEndBehaviour();
  49     void TestBug4153072();
  50     void TestJapaneseLineBreak();
  51     void TestThaiLineBreak();
  52     void TestMixedThaiLineBreak();
  53     void TestMaiyamok();
  54     void TestMonkey();
  55
  56     void TestExtended();
  57     UChar *ReadAndConvertFile(const char *fileName, int &ulen, const char *encoding, UErrorCode &status);
  58     void executeTest(TestParams *, UErrorCode &status);
  59
  60     void TestWordBreaks();
  61     void TestWordBoundary();
  62     void TestLineBreaks();
  63     void TestSentBreaks();
  64     void TestBug3818();
  65     void TestJapaneseWordBreak();
  66     void TestTrieDict();
  67     void TestUnicodeFiles();
  68     void TestBug5775();
  69     void TestTailoredBreaks();
  70     void TestDictRules();
  71     void TestBug5532();
  72     void TestBug9983();
  73     void TestBug7547();
  74     void TestBug12797();
  75     void TestBug12918();
  76     void TestBug12932();
  77     void TestEmoji();
  78     void TestBug12519();
  79     void TestBug12677();
  80     void TestTableRedundancies();
  81     void TestBug13447();
  82     void TestReverse();
  83     void TestReverse(std::unique_ptr<RuleBasedBreakIterator>bi);
  84     void TestBug13692();
  85
  86     void TestDebug();
  87     void TestProperties();
  88
  89 /***********************/
  90 private:
  91     /**
  92      * internal methods to prepare test data
  93      **/
  94
  95     void RunMonkey(BreakIterator *bi, RBBIMonkeyKind &mk, const char *name, uint32_t  seed,
  96         int32_t loopCount, UBool useUText);
  97
  98     // Run one of the Unicode Consortium boundary test data files.
  99     void runUnicodeTestData(const char *fileName, RuleBasedBreakIterator *bi);
 100
 101     // Run a single test case from one of the Unicode Consortium test files.
 102     void checkUnicodeTestCase(const char *testFileName, int lineNumber,
 103                          const UnicodeString &testString,
 104                          UVector32 *breakPositions,
 105                          RuleBasedBreakIterator *bi);
 106
 107     // Run the actual tests for TestTailoredBreaks()
 108     void TBTest(BreakIterator* brkitr, int type, const char *locale, const char* escapedText,
 109                 const int32_t *expectOffsets, int32_t expectOffsetsCount);
 110
 111     /** Filter for test cases from the Unicode test data files.
 112      *  Some need to be skipped because ICU is unable to fully implement the
 113      *  Unicode boundary specifications.
 114      *  @param testCase the test data string.
 115      *  @param fileName the Unicode test data file name.
 116      *  @return FALSE if the test case should be run, TRUE if it should be skipped.
 117      */
 118     UBool testCaseIsKnownIssue(const UnicodeString &testCase, const char *fileName);
 119
 120     // Test parameters, from the test framework and test invocation.
 121     const char* fTestParams;
 122 };
 123
 124 #endif /* #if !UCONFIG_NO_BREAK_ITERATION */
 125
 126 #endif