blob: a68790b27f42d4ce5e24f72d7b27acf3badccf41 [file] [log] [blame]
/********************************************************************
* COPYRIGHT:
* Copyright (c) 1999-2013, International Business Machines Corporation and
* others. All Rights Reserved.
********************************************************************/
/************************************************************************
* Date Name Description
* 12/15/99 Madhu Creation.
* 01/12/2000 Madhu Updated for changed API and added new tests
************************************************************************/
#include "utypeinfo.h" // for 'typeid' to work
#include "unicode/utypes.h"
#if !UCONFIG_NO_BREAK_ITERATION
#include "unicode/utypes.h"
#include "unicode/brkiter.h"
#include "unicode/rbbi.h"
#include "unicode/uchar.h"
#include "unicode/utf16.h"
#include "unicode/ucnv.h"
#include "unicode/schriter.h"
#include "unicode/uniset.h"
#if !UCONFIG_NO_REGULAR_EXPRESSIONS
#include "unicode/regex.h"
#endif
#include "unicode/ustring.h"
#include "unicode/utext.h"
#include "intltest.h"
#include "rbbitst.h"
#include <string.h>
#include "uvector.h"
#include "uvectr32.h"
#include <string.h>
#include <stdio.h>
#include <stdlib.h>
#include "unicode/numfmt.h"
#include "unicode/uscript.h"
#define TEST_ASSERT(x) {if (!(x)) { \
errln("Failure in file %s, line %d", __FILE__, __LINE__);}}
#define TEST_ASSERT_SUCCESS(errcode) { if (U_FAILURE(errcode)) { \
errcheckln(errcode, "Failure in file %s, line %d, status = \"%s\"", __FILE__, __LINE__, u_errorName(errcode));}}
//---------------------------------------------
// runIndexedTest
//---------------------------------------------
// Note: Before adding new tests to this file, check whether the desired test data can
// simply be added to the file testdata/rbbitest.txt. In most cases it can,
// it's much less work than writing a new test, diagnostic output in the event of failures
// is good, and the test data file will is shared with ICU4J, so eventually the test
// will run there as well, without additional effort.
void RBBITest::runIndexedTest( int32_t index, UBool exec, const char* &name, char* params )
{
if (exec) logln("TestSuite RuleBasedBreakIterator: ");
switch (index) {
#if !UCONFIG_NO_FILE_IO
case 0: name = "TestBug4153072";
if(exec) TestBug4153072(); break;
#else
case 0: name = "skip";
break;
#endif
case 1: name = "skip";
break;
case 2: name = "TestStatusReturn";
if(exec) TestStatusReturn(); break;
#if !UCONFIG_NO_FILE_IO
case 3: name = "TestUnicodeFiles";
if(exec) TestUnicodeFiles(); break;
case 4: name = "TestEmptyString";
if(exec) TestEmptyString(); break;
#else
case 3: case 4: name = "skip";
break;
#endif
case 5: name = "TestGetAvailableLocales";
if(exec) TestGetAvailableLocales(); break;
case 6: name = "TestGetDisplayName";
if(exec) TestGetDisplayName(); break;
#if !UCONFIG_NO_FILE_IO
case 7: name = "TestEndBehaviour";
if(exec) TestEndBehaviour(); break;
case 8: case 9: case 10: name = "skip";
break;
case 11: name = "TestWordBreaks";
if(exec) TestWordBreaks(); break;
case 12: name = "TestWordBoundary";
if(exec) TestWordBoundary(); break;
case 13: name = "TestLineBreaks";
if(exec) TestLineBreaks(); break;
case 14: name = "TestSentBreaks";
if(exec) TestSentBreaks(); break;
case 15: name = "TestExtended";
if(exec) TestExtended(); break;
#else
case 7: case 8: case 9: case 10: case 11: case 12: case 13: case 14: case 15: name = "skip";
break;
#endif
#if !UCONFIG_NO_REGULAR_EXPRESSIONS && !UCONFIG_NO_FILE_IO
case 16:
name = "TestMonkey"; if(exec) TestMonkey(params); break;
#else
case 16:
name = "skip"; break;
#endif
#if !UCONFIG_NO_FILE_IO
case 17: name = "TestBug3818";
if(exec) TestBug3818(); break;
#else
case 17: name = "skip";
break;
#endif
case 18: name = "skip";
break;
case 19: name = "TestDebug";
if(exec) TestDebug(); break;
case 20: name = "skip";
break;
#if !UCONFIG_NO_FILE_IO
case 21: name = "TestBug5775";
if (exec) TestBug5775(); break;
#else
case 21: name = "skip";
break;
#endif
case 22: name = "TestBug9983";
if (exec) TestBug9983(); break;
case 23: name = "TestDictRules";
if (exec) TestDictRules(); break;
case 24: name = "TestBug5532";
if (exec) TestBug5532(); break;
default: name = ""; break; //needed to end loop
}
}
//---------------------------------------------------------------------------
//
// class BITestData Holds a set of Break iterator test data and results
// Includes
// - the string data to be broken
// - a vector of the expected break positions.
// - a vector of source line numbers for the data,
// (to help see where errors occured.)
// - The expected break tag values.
// - Vectors of actual break positions and tag values.
// - Functions for comparing actual with expected and
// reporting errors.
//
//----------------------------------------------------------------------------
class BITestData {
public:
UnicodeString fDataToBreak;
UVector fExpectedBreakPositions;
UVector fExpectedTags;
UVector fLineNum;
UVector fActualBreakPositions; // Test Results.
UVector fActualTags;
BITestData(UErrorCode &status);
void addDataChunk(const char *data, int32_t tag, int32_t lineNum, UErrorCode status);
void checkResults(const char *heading, RBBITest *test);
void err(const char *heading, RBBITest *test, int32_t expectedIdx, int32_t actualIdx);
void clearResults();
};
//
// Constructor.
//
BITestData::BITestData(UErrorCode &status)
: fExpectedBreakPositions(status), fExpectedTags(status), fLineNum(status), fActualBreakPositions(status),
fActualTags(status)
{
}
//
// addDataChunk. Add a section (non-breaking) piece if data to the test data.
// The macro form collects the line number, which is helpful
// when tracking down failures.
//
// A null data item is inserted at the start of each test's data
// to put the starting zero into the data list. The position saved for
// each non-null item is its ending position.
//
#define ADD_DATACHUNK(td, data, tag, status) td.addDataChunk(data, tag, __LINE__, status);
void BITestData::addDataChunk(const char *data, int32_t tag, int32_t lineNum, UErrorCode status) {
if (U_FAILURE(status)) {return;}
if (data != NULL) {
fDataToBreak.append(CharsToUnicodeString(data));
}
fExpectedBreakPositions.addElement(fDataToBreak.length(), status);
fExpectedTags.addElement(tag, status);
fLineNum.addElement(lineNum, status);
}
//
// checkResults. Compare the actual and expected break positions, report any differences.
//
void BITestData::checkResults(const char *heading, RBBITest *test) {
int32_t expectedIndex = 0;
int32_t actualIndex = 0;
for (;;) {
// If we've run through both the expected and actual results vectors, we're done.
// break out of the loop.
if (expectedIndex >= fExpectedBreakPositions.size() &&
actualIndex >= fActualBreakPositions.size()) {
break;
}
if (expectedIndex >= fExpectedBreakPositions.size()) {
err(heading, test, expectedIndex-1, actualIndex);
actualIndex++;
continue;
}
if (actualIndex >= fActualBreakPositions.size()) {
err(heading, test, expectedIndex, actualIndex-1);
expectedIndex++;
continue;
}
if (fActualBreakPositions.elementAti(actualIndex) != fExpectedBreakPositions.elementAti(expectedIndex)) {
err(heading, test, expectedIndex, actualIndex);
// Try to resync the positions of the indices, to avoid a rash of spurious erros.
if (fActualBreakPositions.elementAti(actualIndex) < fExpectedBreakPositions.elementAti(expectedIndex)) {
actualIndex++;
} else {
expectedIndex++;
}
continue;
}
if (fActualTags.elementAti(actualIndex) != fExpectedTags.elementAti(expectedIndex)) {
test->errln("%s, tag mismatch. Test Line = %d, expected tag=%d, got %d",
heading, fLineNum.elementAt(expectedIndex),
fExpectedTags.elementAti(expectedIndex), fActualTags.elementAti(actualIndex));
}
actualIndex++;
expectedIndex++;
}
}
//
// err - An error was found. Report it, along with information about where the
// incorrectly broken test data appeared in the source file.
//
void BITestData::err(const char *heading, RBBITest *test, int32_t expectedIdx, int32_t actualIdx)
{
int32_t expected = fExpectedBreakPositions.elementAti(expectedIdx);
int32_t actual = fActualBreakPositions.elementAti(actualIdx);
int32_t o = 0;
int32_t line = fLineNum.elementAti(expectedIdx);
if (expectedIdx > 0) {
// The line numbers are off by one because a premature break occurs somewhere
// within the previous item, rather than at the start of the current (expected) item.
// We want to report the offset of the unexpected break from the start of
// this previous item.
o = actual - fExpectedBreakPositions.elementAti(expectedIdx-1);
}
if (actual < expected) {
test->errln("%s unexpected break at offset %d in test item from line %d. actual break: %d expected break: %d", heading, o, line, actual, expected);
} else {
test->errln("%s Failed to find break at end of item from line %d. actual break: %d expected break: %d", heading, line, actual, expected);
}
}
void BITestData::clearResults() {
fActualBreakPositions.removeAllElements();
fActualTags.removeAllElements();
}
//--------------------------------------------------------------------------------------
//
// RBBITest constructor and destructor
//
//--------------------------------------------------------------------------------------
RBBITest::RBBITest() {
}
RBBITest::~RBBITest() {
}
//-----------------------------------------------------------------------------------
//
// Test for status {tag} return value from break rules.
// TODO: a more thorough test.
//
//-----------------------------------------------------------------------------------
void RBBITest::TestStatusReturn() {
UnicodeString rulesString1("$Letters = [:L:];\n"
"$Numbers = [:N:];\n"
"$Letters+{1};\n"
"$Numbers+{2};\n"
"Help\\ {4}/me\\!;\n"
"[^$Letters $Numbers];\n"
"!.*;\n", -1, US_INV);
UnicodeString testString1 = "abc123..abc Help me Help me!";
// 01234567890123456789012345678
int32_t bounds1[] = {0, 3, 6, 7, 8, 11, 12, 16, 17, 19, 20, 25, 27, 28, -1};
int32_t brkStatus[] = {0, 1, 2, 0, 0, 1, 0, 1, 0, 1, 0, 4, 1, 0, -1};
UErrorCode status=U_ZERO_ERROR;
UParseError parseError;
RuleBasedBreakIterator *bi = new RuleBasedBreakIterator(rulesString1, parseError, status);
if(U_FAILURE(status)) {
dataerrln("FAIL : in construction - %s", u_errorName(status));
} else {
int32_t pos;
int32_t i = 0;
bi->setText(testString1);
for (pos=bi->first(); pos!= BreakIterator::DONE; pos=bi->next()) {
if (pos != bounds1[i]) {
errln("FAIL: expected break at %d, got %d\n", bounds1[i], pos);
break;
}
int tag = bi->getRuleStatus();
if (tag != brkStatus[i]) {
errln("FAIL: break at %d, expected tag %d, got tag %d\n", pos, brkStatus[i], tag);
break;
}
i++;
}
}
delete bi;
}
static void printStringBreaks(UnicodeString ustr, int expected[],
int expectedcount)
{
UErrorCode status = U_ZERO_ERROR;
char name[100];
printf("code alpha extend alphanum type word sent line name\n");
int j;
for (j = 0; j < ustr.length(); j ++) {
if (expectedcount > 0) {
int k;
for (k = 0; k < expectedcount; k ++) {
if (j == expected[k]) {
printf("------------------------------------------------ %d\n",
j);
}
}
}
UChar32 c = ustr.char32At(j);
if (c > 0xffff) {
j ++;
}
u_charName(c, U_UNICODE_CHAR_NAME, name, 100, &status);
printf("%7x %5d %6d %8d %4s %4s %4s %4s %s\n", (int)c,
u_isUAlphabetic(c),
u_hasBinaryProperty(c, UCHAR_GRAPHEME_EXTEND),
u_isalnum(c),
u_getPropertyValueName(UCHAR_GENERAL_CATEGORY,
u_charType(c),
U_SHORT_PROPERTY_NAME),
u_getPropertyValueName(UCHAR_WORD_BREAK,
u_getIntPropertyValue(c,
UCHAR_WORD_BREAK),
U_SHORT_PROPERTY_NAME),
u_getPropertyValueName(UCHAR_SENTENCE_BREAK,
u_getIntPropertyValue(c,
UCHAR_SENTENCE_BREAK),
U_SHORT_PROPERTY_NAME),
u_getPropertyValueName(UCHAR_LINE_BREAK,
u_getIntPropertyValue(c,
UCHAR_LINE_BREAK),
U_SHORT_PROPERTY_NAME),
name);
}
}
void RBBITest::TestBug3818() {
UErrorCode status = U_ZERO_ERROR;
// Four Thai words...
static const UChar thaiWordData[] = { 0x0E43,0x0E2B,0x0E0D,0x0E48, 0x0E43,0x0E2B,0x0E0D,0x0E48,
0x0E43,0x0E2B,0x0E0D,0x0E48, 0x0E43,0x0E2B,0x0E0D,0x0E48, 0 };
UnicodeString thaiStr(thaiWordData);
RuleBasedBreakIterator* bi =
(RuleBasedBreakIterator *)BreakIterator::createWordInstance(Locale("th"), status);
if (U_FAILURE(status) || bi == NULL) {
errcheckln(status, "Fail at file %s, line %d, status = %s", __FILE__, __LINE__, u_errorName(status));
return;
}
bi->setText(thaiStr);
int32_t startOfSecondWord = bi->following(1);
if (startOfSecondWord != 4) {
errln("Fail at file %s, line %d expected start of word at 4, got %d",
__FILE__, __LINE__, startOfSecondWord);
}
startOfSecondWord = bi->following(0);
if (startOfSecondWord != 4) {
errln("Fail at file %s, line %d expected start of word at 4, got %d",
__FILE__, __LINE__, startOfSecondWord);
}
delete bi;
}
//----------------------------------------------------------------------------
//
// generalIteratorTest Given a break iterator and a set of test data,
// Run the tests and report the results.
//
//----------------------------------------------------------------------------
void RBBITest::generalIteratorTest(RuleBasedBreakIterator& bi, BITestData &td)
{
bi.setText(td.fDataToBreak);
testFirstAndNext(bi, td);
testLastAndPrevious(bi, td);
testFollowing(bi, td);
testPreceding(bi, td);
testIsBoundary(bi, td);
doMultipleSelectionTest(bi, td);
}
//
// testFirstAndNext. Run the iterator forwards in the obvious first(), next()
// kind of loop.
//
void RBBITest::testFirstAndNext(RuleBasedBreakIterator& bi, BITestData &td)
{
UErrorCode status = U_ZERO_ERROR;
int32_t p;
int32_t lastP = -1;
int32_t tag;
logln("Test first and next");
bi.setText(td.fDataToBreak);
td.clearResults();
for (p=bi.first(); p!=RuleBasedBreakIterator::DONE; p=bi.next()) {
td.fActualBreakPositions.addElement(p, status); // Save result.
tag = bi.getRuleStatus();
td.fActualTags.addElement(tag, status);
if (p <= lastP) {
// If the iterator is not making forward progress, stop.
// No need to raise an error here, it'll be detected in the normal check of results.
break;
}
lastP = p;
}
td.checkResults("testFirstAndNext", this);
}
//
// TestLastAndPrevious. Run the iterator backwards, starting with last().
//
void RBBITest::testLastAndPrevious(RuleBasedBreakIterator& bi, BITestData &td)
{
UErrorCode status = U_ZERO_ERROR;
int32_t p;
int32_t lastP = 0x7ffffffe;
int32_t tag;
logln("Test last and previous");
bi.setText(td.fDataToBreak);
td.clearResults();
for (p=bi.last(); p!=RuleBasedBreakIterator::DONE; p=bi.previous()) {
// Save break position. Insert it at start of vector of results, shoving
// already-saved results further towards the end.
td.fActualBreakPositions.insertElementAt(p, 0, status);
// bi.previous(); // TODO: Why does this fix things up????
// bi.next();
tag = bi.getRuleStatus();
td.fActualTags.insertElementAt(tag, 0, status);
if (p >= lastP) {
// If the iterator is not making progress, stop.
// No need to raise an error here, it'll be detected in the normal check of results.
break;
}
lastP = p;
}
td.checkResults("testLastAndPrevious", this);
}
void RBBITest::testFollowing(RuleBasedBreakIterator& bi, BITestData &td)
{
UErrorCode status = U_ZERO_ERROR;
int32_t p;
int32_t tag;
int32_t lastP = -2; // A value that will never be returned as a break position.
// cannot be -1; that is returned for DONE.
int i;
logln("testFollowing():");
bi.setText(td.fDataToBreak);
td.clearResults();
// Save the starting point, since we won't get that out of following.
p = bi.first();
td.fActualBreakPositions.addElement(p, status); // Save result.
tag = bi.getRuleStatus();
td.fActualTags.addElement(tag, status);
for (i = 0; i <= td.fDataToBreak.length()+1; i++) {
p = bi.following(i);
if (p != lastP) {
if (p == RuleBasedBreakIterator::DONE) {
break;
}
// We've reached a new break position. Save it.
td.fActualBreakPositions.addElement(p, status); // Save result.
tag = bi.getRuleStatus();
td.fActualTags.addElement(tag, status);
lastP = p;
}
}
// The loop normally exits by means of the break in the middle.
// Make sure that the index was at the correct position for the break iterator to have
// returned DONE.
if (i != td.fDataToBreak.length()) {
errln("testFollowing(): iterator returned DONE prematurely.");
}
// Full check of all results.
td.checkResults("testFollowing", this);
}
void RBBITest::testPreceding(RuleBasedBreakIterator& bi, BITestData &td) {
UErrorCode status = U_ZERO_ERROR;
int32_t p;
int32_t tag;
int32_t lastP = 0x7ffffffe;
int i;
logln("testPreceding():");
bi.setText(td.fDataToBreak);
td.clearResults();
p = bi.last();
td.fActualBreakPositions.addElement(p, status);
tag = bi.getRuleStatus();
td.fActualTags.addElement(tag, status);
for (i = td.fDataToBreak.length(); i>=-1; i--) {
p = bi.preceding(i);
if (p != lastP) {
if (p == RuleBasedBreakIterator::DONE) {
break;
}
// We've reached a new break position. Save it.
td.fActualBreakPositions.insertElementAt(p, 0, status);
lastP = p;
tag = bi.getRuleStatus();
td.fActualTags.insertElementAt(tag, 0, status);
}
}
// The loop normally exits by means of the break in the middle.
// Make sure that the index was at the correct position for the break iterator to have
// returned DONE.
if (i != 0) {
errln("testPreceding(): iterator returned DONE prematurely.");
}
// Full check of all results.
td.checkResults("testPreceding", this);
}
void RBBITest::testIsBoundary(RuleBasedBreakIterator& bi, BITestData &td) {
UErrorCode status = U_ZERO_ERROR;
int i;
int32_t tag;
logln("testIsBoundary():");
bi.setText(td.fDataToBreak);
td.clearResults();
for (i = 0; i <= td.fDataToBreak.length(); i++) {
if (bi.isBoundary(i)) {
td.fActualBreakPositions.addElement(i, status); // Save result.
tag = bi.getRuleStatus();
td.fActualTags.addElement(tag, status);
}
}
td.checkResults("testIsBoundary: ", this);
}
void RBBITest::doMultipleSelectionTest(RuleBasedBreakIterator& iterator, BITestData &td)
{
iterator.setText(td.fDataToBreak);
RuleBasedBreakIterator* testIterator =(RuleBasedBreakIterator*)iterator.clone();
int32_t offset = iterator.first();
int32_t testOffset;
int32_t count = 0;
logln("doMultipleSelectionTest text of length: %d", td.fDataToBreak.length());
if (*testIterator != iterator)
errln("clone() or operator!= failed: two clones compared unequal");
do {
testOffset = testIterator->first();
testOffset = testIterator->next(count);
if (offset != testOffset)
errln(UnicodeString("next(n) and next() not returning consistent results: for step ") + count + ", next(n) returned " + testOffset + " and next() had " + offset);
if (offset != RuleBasedBreakIterator::DONE) {
count++;
offset = iterator.next();
if (offset != RuleBasedBreakIterator::DONE && *testIterator == iterator) {
errln("operator== failed: Two unequal iterators compared equal. count=%d offset=%d", count, offset);
if (count > 10000 || offset == -1) {
errln("operator== failed too many times. Stopping test.");
if (offset == -1) {
errln("Does (RuleBasedBreakIterator::DONE == -1)?");
}
return;
}
}
}
} while (offset != RuleBasedBreakIterator::DONE);
// now do it backwards...
offset = iterator.last();
count = 0;
do {
testOffset = testIterator->last();
testOffset = testIterator->next(count); // next() with a negative arg is same as previous
if (offset != testOffset)
errln(UnicodeString("next(n) and next() not returning consistent results: for step ") + count + ", next(n) returned " + testOffset + " and next() had " + offset);
if (offset != RuleBasedBreakIterator::DONE) {
count--;
offset = iterator.previous();
}
} while (offset != RuleBasedBreakIterator::DONE);
delete testIterator;
}
//---------------------------------------------
//
// other tests
//
//---------------------------------------------
void RBBITest::TestEmptyString()
{
UnicodeString text = "";
UErrorCode status = U_ZERO_ERROR;
BITestData x(status);
ADD_DATACHUNK(x, "", 0, status); // Break at start of data
RuleBasedBreakIterator* bi = (RuleBasedBreakIterator *)BreakIterator::createLineInstance(Locale::getDefault(), status);
if (U_FAILURE(status))
{
errcheckln(status, "Failed to create the BreakIterator for default locale in TestEmptyString. - %s", u_errorName(status));
return;
}
generalIteratorTest(*bi, x);
delete bi;
}
void RBBITest::TestGetAvailableLocales()
{
int32_t locCount = 0;
const Locale* locList = BreakIterator::getAvailableLocales(locCount);
if (locCount == 0)
dataerrln("getAvailableLocales() returned an empty list!");
// Just make sure that it's returning good memory.
int32_t i;
for (i = 0; i < locCount; ++i) {
logln(locList[i].getName());
}
}
//Testing the BreakIterator::getDisplayName() function
void RBBITest::TestGetDisplayName()
{
UnicodeString result;
BreakIterator::getDisplayName(Locale::getUS(), result);
if (Locale::getDefault() == Locale::getUS() && result != "English (United States)")
dataerrln("BreakIterator::getDisplayName() failed: expected \"English (United States)\", got \""
+ result);
BreakIterator::getDisplayName(Locale::getFrance(), Locale::getUS(), result);
if (result != "French (France)")
dataerrln("BreakIterator::getDisplayName() failed: expected \"French (France)\", got \""
+ result);
}
/**
* Test End Behaviour
* @bug 4068137
*/
void RBBITest::TestEndBehaviour()
{
UErrorCode status = U_ZERO_ERROR;
UnicodeString testString("boo.");
BreakIterator *wb = BreakIterator::createWordInstance(Locale::getDefault(), status);
if (U_FAILURE(status))
{
errcheckln(status, "Failed to create the BreakIterator for default locale in TestEndBehaviour. - %s", u_errorName(status));
return;
}
wb->setText(testString);
if (wb->first() != 0)
errln("Didn't get break at beginning of string.");
if (wb->next() != 3)
errln("Didn't get break before period in \"boo.\"");
if (wb->current() != 4 && wb->next() != 4)
errln("Didn't get break at end of string.");
delete wb;
}
/*
* @bug 4153072
*/
void RBBITest::TestBug4153072() {
UErrorCode status = U_ZERO_ERROR;
BreakIterator *iter = BreakIterator::createWordInstance(Locale::getDefault(), status);
if (U_FAILURE(status))
{
errcheckln(status, "Failed to create the BreakIterator for default locale in TestBug4153072 - %s", u_errorName(status));
return;
}
UnicodeString str("...Hello, World!...");
int32_t begin = 3;
int32_t end = str.length() - 3;
UBool onBoundary;
StringCharacterIterator* textIterator = new StringCharacterIterator(str, begin, end, begin);
iter->adoptText(textIterator);
int index;
// Note: with the switch to UText, there is no way to restrict the
// iteration range to begin at an index other than zero.
// String character iterators created with a non-zero bound are
// treated by RBBI as being empty.
for (index = -1; index < begin + 1; ++index) {
onBoundary = iter->isBoundary(index);
if (index == 0? !onBoundary : onBoundary) {
errln((UnicodeString)"Didn't handle isBoundary correctly with offset = " + index +
" and begin index = " + begin);
}
}
delete iter;
}
//
// Test for problem reported by Ashok Matoria on 9 July 2007
// One.<kSoftHyphen><kSpace>Two.
//
// Sentence break at start (0) and then on calling next() it breaks at
// 'T' of "Two". Now, at this point if I do next() and
// then previous(), it breaks at <kSOftHyphen> instead of 'T' of "Two".
//
void RBBITest::TestBug5775() {
UErrorCode status = U_ZERO_ERROR;
BreakIterator *bi = BreakIterator::createSentenceInstance(Locale::getEnglish(), status);
TEST_ASSERT_SUCCESS(status);
if (U_FAILURE(status)) {
return;
}
// Check for status first for better handling of no data errors.
TEST_ASSERT(bi != NULL);
if (bi == NULL) {
return;
}
UnicodeString s("One.\\u00ad Two.", -1, US_INV);
// 01234 56789
s = s.unescape();
bi->setText(s);
int pos = bi->next();
TEST_ASSERT(pos == 6);
pos = bi->next();
TEST_ASSERT(pos == 10);
pos = bi->previous();
TEST_ASSERT(pos == 6);
delete bi;
}
//------------------------------------------------------------------------------
//
// RBBITest::Extended Run RBBI Tests from an external test data file
//
//------------------------------------------------------------------------------
struct TestParams {
BreakIterator *bi;
UnicodeString dataToBreak;
UVector32 *expectedBreaks;
UVector32 *srcLine;
UVector32 *srcCol;
};
void RBBITest::executeTest(TestParams *t) {
int32_t bp;
int32_t prevBP;
int32_t i;
if (t->bi == NULL) {
return;
}
t->bi->setText(t->dataToBreak);
//
// Run the iterator forward
//
prevBP = -1;
for (bp = t->bi->first(); bp != BreakIterator::DONE; bp = t->bi->next()) {
if (prevBP == bp) {
// Fail for lack of forward progress.
errln("Forward Iteration, no forward progress. Break Pos=%4d File line,col=%4d,%4d",
bp, t->srcLine->elementAti(bp), t->srcCol->elementAti(bp));
break;
}
// Check that there were we didn't miss an expected break between the last one
// and this one.
for (i=prevBP+1; i<bp; i++) {
if (t->expectedBreaks->elementAti(i) != 0) {
int expected[] = {0, i};
printStringBreaks(t->dataToBreak, expected, 2);
errln("Forward Iteration, break expected, but not found. Pos=%4d File line,col= %4d,%4d",
i, t->srcLine->elementAti(i), t->srcCol->elementAti(i));
}
}
// Check that the break we did find was expected
if (t->expectedBreaks->elementAti(bp) == 0) {
int expected[] = {0, bp};
printStringBreaks(t->dataToBreak, expected, 2);
errln("Forward Iteration, break found, but not expected. Pos=%4d File line,col= %4d,%4d",
bp, t->srcLine->elementAti(bp), t->srcCol->elementAti(bp));
} else {
// The break was expected.
// Check that the {nnn} tag value is correct.
int32_t expectedTagVal = t->expectedBreaks->elementAti(bp);
if (expectedTagVal == -1) {
expectedTagVal = 0;
}
int32_t line = t->srcLine->elementAti(bp);
int32_t rs = ((RuleBasedBreakIterator *)t->bi)->getRuleStatus();
if (rs != expectedTagVal) {
errln("Incorrect status for forward break. Pos=%4d File line,col= %4d,%4d.\n"
" Actual, Expected status = %4d, %4d",
bp, line, t->srcCol->elementAti(bp), rs, expectedTagVal);
}
}
prevBP = bp;
}
// Verify that there were no missed expected breaks after the last one found
for (i=prevBP+1; i<t->expectedBreaks->size(); i++) {
if (t->expectedBreaks->elementAti(i) != 0) {
errln("Forward Iteration, break expected, but not found. Pos=%4d File line,col= %4d,%4d",
i, t->srcLine->elementAti(i), t->srcCol->elementAti(i));
}
}
//
// Run the iterator backwards, verify that the same breaks are found.
//
prevBP = t->dataToBreak.length()+2; // start with a phony value for the last break pos seen.
for (bp = t->bi->last(); bp != BreakIterator::DONE; bp = t->bi->previous()) {
if (prevBP == bp) {
// Fail for lack of progress.
errln("Reverse Iteration, no progress. Break Pos=%4d File line,col=%4d,%4d",
bp, t->srcLine->elementAti(bp), t->srcCol->elementAti(bp));
break;
}
// Check that there were we didn't miss an expected break between the last one
// and this one. (UVector returns zeros for index out of bounds.)
for (i=prevBP-1; i>bp; i--) {
if (t->expectedBreaks->elementAti(i) != 0) {
errln("Reverse Itertion, break expected, but not found. Pos=%4d File line,col= %4d,%4d",
i, t->srcLine->elementAti(i), t->srcCol->elementAti(i));
}
}
// Check that the break we did find was expected
if (t->expectedBreaks->elementAti(bp) == 0) {
errln("Reverse Itertion, break found, but not expected. Pos=%4d File line,col= %4d,%4d",
bp, t->srcLine->elementAti(bp), t->srcCol->elementAti(bp));
} else {
// The break was expected.
// Check that the {nnn} tag value is correct.
int32_t expectedTagVal = t->expectedBreaks->elementAti(bp);
if (expectedTagVal == -1) {
expectedTagVal = 0;
}
int line = t->srcLine->elementAti(bp);
int32_t rs = ((RuleBasedBreakIterator *)t->bi)->getRuleStatus();
if (rs != expectedTagVal) {
errln("Incorrect status for reverse break. Pos=%4d File line,col= %4d,%4d.\n"
" Actual, Expected status = %4d, %4d",
bp, line, t->srcCol->elementAti(bp), rs, expectedTagVal);
}
}
prevBP = bp;
}
// Verify that there were no missed breaks prior to the last one found
for (i=prevBP-1; i>=0; i--) {
if (t->expectedBreaks->elementAti(i) != 0) {
errln("Forward Itertion, break expected, but not found. Pos=%4d File line,col= %4d,%4d",
i, t->srcLine->elementAti(i), t->srcCol->elementAti(i));
}
}
// Check isBoundary()
for (i=0; i<t->expectedBreaks->size(); i++) {
UBool boundaryExpected = (t->expectedBreaks->elementAti(i) != 0);
UBool boundaryFound = t->bi->isBoundary(i);
if (boundaryExpected != boundaryFound) {
errln("isBoundary(%d) incorrect. File line,col= %4d,%4d\n"
" Expected, Actual= %s, %s",
i, t->srcLine->elementAti(i), t->srcCol->elementAti(i),
boundaryExpected ? "true":"false", boundaryFound? "true" : "false");
}
}
// Check following()
for (i=0; i<t->expectedBreaks->size(); i++) {
int32_t actualBreak = t->bi->following(i);
int32_t expectedBreak = BreakIterator::DONE;
for (int32_t j=i+1; j < t->expectedBreaks->size(); j++) {
if (t->expectedBreaks->elementAti(j) != 0) {
expectedBreak = j;
break;
}
}
if (expectedBreak != actualBreak) {
errln("following(%d) incorrect. File line,col= %4d,%4d\n"
" Expected, Actual= %d, %d",
i, t->srcLine->elementAti(i), t->srcCol->elementAti(i), expectedBreak, actualBreak);
}
}
// Check preceding()
for (i=t->expectedBreaks->size(); i>=0; i--) {
int32_t actualBreak = t->bi->preceding(i);
int32_t expectedBreak = BreakIterator::DONE;
for (int32_t j=i-1; j >= 0; j--) {
if (t->expectedBreaks->elementAti(j) != 0) {
expectedBreak = j;
break;
}
}
if (expectedBreak != actualBreak) {
errln("preceding(%d) incorrect. File line,col= %4d,%4d\n"
" Expected, Actual= %d, %d",
i, t->srcLine->elementAti(i), t->srcCol->elementAti(i), expectedBreak, actualBreak);
}
}
}
void RBBITest::TestExtended() {
#if !UCONFIG_NO_REGULAR_EXPRESSIONS
UErrorCode status = U_ZERO_ERROR;
Locale locale("");
UnicodeString rules;
TestParams tp;
tp.bi = NULL;
tp.expectedBreaks = new UVector32(status);
tp.srcLine = new UVector32(status);
tp.srcCol = new UVector32(status);
RegexMatcher localeMatcher(UNICODE_STRING_SIMPLE("<locale *([\\p{L}\\p{Nd}_]*) *>"), 0, status);
if (U_FAILURE(status)) {
dataerrln("Failure in file %s, line %d, status = \"%s\"", __FILE__, __LINE__, u_errorName(status));
}
//
// Open and read the test data file.
//
const char *testDataDirectory = IntlTest::getSourceTestData(status);
char testFileName[1000];
if (testDataDirectory == NULL || strlen(testDataDirectory) >= sizeof(testFileName)) {
errln("Can't open test data. Path too long.");
return;
}
strcpy(testFileName, testDataDirectory);
strcat(testFileName, "rbbitst.txt");
int len;
UChar *testFile = ReadAndConvertFile(testFileName, len, "UTF-8", status);
if (U_FAILURE(status)) {
return; /* something went wrong, error already output */
}
//
// Put the test data into a UnicodeString
//
UnicodeString testString(FALSE, testFile, len);
enum EParseState{
PARSE_COMMENT,
PARSE_TAG,
PARSE_DATA,
PARSE_NUM
}
parseState = PARSE_TAG;
EParseState savedState = PARSE_TAG;
static const UChar CH_LF = 0x0a;
static const UChar CH_CR = 0x0d;
static const UChar CH_HASH = 0x23;
/*static const UChar CH_PERIOD = 0x2e;*/
static const UChar CH_LT = 0x3c;
static const UChar CH_GT = 0x3e;
static const UChar CH_BACKSLASH = 0x5c;
static const UChar CH_BULLET = 0x2022;
int32_t lineNum = 1;
int32_t colStart = 0;
int32_t column = 0;
int32_t charIdx = 0;
int32_t tagValue = 0; // The numeric value of a <nnn> tag.
for (charIdx = 0; charIdx < len; ) {
status = U_ZERO_ERROR;
UChar c = testString.charAt(charIdx);
charIdx++;
if (c == CH_CR && charIdx<len && testString.charAt(charIdx) == CH_LF) {
// treat CRLF as a unit
c = CH_LF;
charIdx++;
}
if (c == CH_LF || c == CH_CR) {
lineNum++;
colStart = charIdx;
}
column = charIdx - colStart + 1;
switch (parseState) {
case PARSE_COMMENT:
if (c == 0x0a || c == 0x0d) {
parseState = savedState;
}
break;
case PARSE_TAG:
{
if (c == CH_HASH) {
parseState = PARSE_COMMENT;
savedState = PARSE_TAG;
break;
}
if (u_isUWhiteSpace(c)) {
break;
}
if (testString.compare(charIdx-1, 6, "<word>") == 0) {
delete tp.bi;
tp.bi = BreakIterator::createWordInstance(locale, status);
charIdx += 5;
break;
}
if (testString.compare(charIdx-1, 6, "<char>") == 0) {
delete tp.bi;
tp.bi = BreakIterator::createCharacterInstance(locale, status);
charIdx += 5;
break;
}
if (testString.compare(charIdx-1, 6, "<line>") == 0) {
delete tp.bi;
tp.bi = BreakIterator::createLineInstance(locale, status);
charIdx += 5;
break;
}
if (testString.compare(charIdx-1, 6, "<sent>") == 0) {
delete tp.bi;
tp.bi = NULL;
tp.bi = BreakIterator::createSentenceInstance(locale, status);
charIdx += 5;
break;
}
if (testString.compare(charIdx-1, 7, "<title>") == 0) {
delete tp.bi;
tp.bi = BreakIterator::createTitleInstance(locale, status);
charIdx += 6;
break;
}
// <locale loc_name>
localeMatcher.reset(testString);
if (localeMatcher.lookingAt(charIdx-1, status)) {
UnicodeString localeName = localeMatcher.group(1, status);
char localeName8[100];
localeName.extract(0, localeName.length(), localeName8, sizeof(localeName8), 0);
locale = Locale::createFromName(localeName8);
charIdx += localeMatcher.group(0, status).length() - 1;
TEST_ASSERT_SUCCESS(status);
break;
}
if (testString.compare(charIdx-1, 6, "<data>") == 0) {
parseState = PARSE_DATA;
charIdx += 5;
tp.dataToBreak = "";
tp.expectedBreaks->removeAllElements();
tp.srcCol ->removeAllElements();
tp.srcLine->removeAllElements();
break;
}
errln("line %d: Tag expected in test file.", lineNum);
parseState = PARSE_COMMENT;
savedState = PARSE_DATA;
goto end_test; // Stop the test.
}
break;
case PARSE_DATA:
if (c == CH_BULLET) {
int32_t breakIdx = tp.dataToBreak.length();
tp.expectedBreaks->setSize(breakIdx+1);
tp.expectedBreaks->setElementAt(-1, breakIdx);
tp.srcLine->setSize(breakIdx+1);
tp.srcLine->setElementAt(lineNum, breakIdx);
tp.srcCol ->setSize(breakIdx+1);
tp.srcCol ->setElementAt(column, breakIdx);
break;
}
if (testString.compare(charIdx-1, 7, "</data>") == 0) {
// Add final entry to mappings from break location to source file position.
// Need one extra because last break position returned is after the
// last char in the data, not at the last char.
tp.srcLine->addElement(lineNum, status);
tp.srcCol ->addElement(column, status);
parseState = PARSE_TAG;
charIdx += 6;
// RUN THE TEST!
executeTest(&tp);
break;
}
if (testString.compare(charIdx-1, 3, UNICODE_STRING_SIMPLE("\\N{")) == 0) {
// Named character, e.g. \N{COMBINING GRAVE ACCENT}
// Get the code point from the name and insert it into the test data.
// (Damn, no API takes names in Unicode !!!
// we've got to take it back to char *)
int32_t nameEndIdx = testString.indexOf((UChar)0x7d/*'}'*/, charIdx);
int32_t nameLength = nameEndIdx - (charIdx+2);
char charNameBuf[200];
UChar32 theChar = -1;
if (nameEndIdx != -1) {
UErrorCode status = U_ZERO_ERROR;
testString.extract(charIdx+2, nameLength, charNameBuf, sizeof(charNameBuf));
charNameBuf[sizeof(charNameBuf)-1] = 0;
theChar = u_charFromName(U_UNICODE_CHAR_NAME, charNameBuf, &status);
if (U_FAILURE(status)) {
theChar = -1;
}
}
if (theChar == -1) {
errln("Error in named character in test file at line %d, col %d",
lineNum, column);
} else {
// Named code point was recognized. Insert it
// into the test data.
tp.dataToBreak.append(theChar);
while (tp.dataToBreak.length() > tp.srcLine->size()) {
tp.srcLine->addElement(lineNum, status);
tp.srcCol ->addElement(column, status);
}
}
if (nameEndIdx > charIdx) {
charIdx = nameEndIdx+1;
}
break;
}
if (testString.compare(charIdx-1, 2, "<>") == 0) {
charIdx++;
int32_t breakIdx = tp.dataToBreak.length();
tp.expectedBreaks->setSize(breakIdx+1);
tp.expectedBreaks->setElementAt(-1, breakIdx);
tp.srcLine->setSize(breakIdx+1);
tp.srcLine->setElementAt(lineNum, breakIdx);
tp.srcCol ->setSize(breakIdx+1);
tp.srcCol ->setElementAt(column, breakIdx);
break;
}
if (c == CH_LT) {
tagValue = 0;
parseState = PARSE_NUM;
break;
}
if (c == CH_HASH && column==3) { // TODO: why is column off so far?
parseState = PARSE_COMMENT;
savedState = PARSE_DATA;
break;
}
if (c == CH_BACKSLASH) {
// Check for \ at end of line, a line continuation.
// Advance over (discard) the newline
UChar32 cp = testString.char32At(charIdx);
if (cp == CH_CR && charIdx<len && testString.charAt(charIdx+1) == CH_LF) {
// We have a CR LF
// Need an extra increment of the input ptr to move over both of them
charIdx++;
}
if (cp == CH_LF || cp == CH_CR) {
lineNum++;
colStart = charIdx;
charIdx++;
break;
}
// Let unescape handle the back slash.
cp = testString.unescapeAt(charIdx);
if (cp != -1) {
// Escape sequence was recognized. Insert the char
// into the test data.
tp.dataToBreak.append(cp);
while (tp.dataToBreak.length() > tp.srcLine->size()) {
tp.srcLine->addElement(lineNum, status);
tp.srcCol ->addElement(column, status);
}
break;
}
// Not a recognized backslash escape sequence.
// Take the next char as a literal.
// TODO: Should this be an error?
c = testString.charAt(charIdx);
charIdx = testString.moveIndex32(charIdx, 1);
}
// Normal, non-escaped data char.
tp.dataToBreak.append(c);
// Save the mapping from offset in the data to line/column numbers in
// the original input file. Will be used for better error messages only.
// If there's an expected break before this char, the slot in the mapping
// vector will already be set for this char; don't overwrite it.
if (tp.dataToBreak.length() > tp.srcLine->size()) {
tp.srcLine->addElement(lineNum, status);
tp.srcCol ->addElement(column, status);
}
break;
case PARSE_NUM:
// We are parsing an expected numeric tag value, like <1234>,
// within a chunk of data.
if (u_isUWhiteSpace(c)) {
break;
}
if (c == CH_GT) {
// Finished the number. Add the info to the expected break data,
// and switch parse state back to doing plain data.
parseState = PARSE_DATA;
if (tagValue == 0) {
tagValue = -1;
}
int32_t breakIdx = tp.dataToBreak.length();
tp.expectedBreaks->setSize(breakIdx+1);
tp.expectedBreaks->setElementAt(tagValue, breakIdx);
tp.srcLine->setSize(breakIdx+1);
tp.srcLine->setElementAt(lineNum, breakIdx);
tp.srcCol ->setSize(breakIdx+1);
tp.srcCol ->setElementAt(column, breakIdx);
break;
}
if (u_isdigit(c)) {
tagValue = tagValue*10 + u_charDigitValue(c);
break;
}
errln("Syntax Error in test file at line %d, col %d",
lineNum, column);
parseState = PARSE_COMMENT;
goto end_test; // Stop the test
break;
}
if (U_FAILURE(status)) {
dataerrln("ICU Error %s while parsing test file at line %d.",
u_errorName(status), lineNum);
status = U_ZERO_ERROR;
goto end_test; // Stop the test
}
}
end_test:
delete tp.bi;
delete tp.expectedBreaks;
delete tp.srcLine;
delete tp.srcCol;
delete [] testFile;
#endif
}
//-------------------------------------------------------------------------------
//
// TestDictRules create a break iterator from source rules that includes a
// dictionary range. Regression for bug #7130. Source rules
// do not declare a break iterator type (word, line, sentence, etc.
// but the dictionary code, without a type, would loop.
//
//-------------------------------------------------------------------------------
void RBBITest::TestDictRules() {
const char *rules = "$dictionary = [a-z]; \n"
"!!forward; \n"
"$dictionary $dictionary; \n"
"!!reverse; \n"
"$dictionary $dictionary; \n";
const char *text = "aa";
UErrorCode status = U_ZERO_ERROR;
UParseError parseError;
RuleBasedBreakIterator bi(rules, parseError, status);
if (U_SUCCESS(status)) {
UnicodeString utext = text;
bi.setText(utext);
int32_t position;
int32_t loops;
for (loops = 0; loops<10; loops++) {
position = bi.next();
if (position == RuleBasedBreakIterator::DONE) {
break;
}
}
TEST_ASSERT(loops == 1);
} else {
dataerrln("Error creating RuleBasedBreakIterator: %s", u_errorName(status));
}
}
//-------------------------------------------------------------------------------
//
// ReadAndConvertFile Read a text data file, convert it to UChars, and
// return the datain one big UChar * buffer, which the caller must delete.
//
// parameters:
// fileName: the name of the file, with no directory part. The test data directory
// is assumed.
// ulen an out parameter, receives the actual length (in UChars) of the file data.
// encoding The file encoding. If the file contains a BOM, that will override the encoding
// specified here. The BOM, if it exists, will be stripped from the returned data.
// Pass NULL for the system default encoding.
// status
// returns:
// The file data, converted to UChar.
// The caller must delete this when done with
// delete [] theBuffer;
//
// TODO: This is a clone of RegexTest::ReadAndConvertFile.
// Move this function to some common place.
//
//--------------------------------------------------------------------------------
UChar *RBBITest::ReadAndConvertFile(const char *fileName, int &ulen, const char *encoding, UErrorCode &status) {
UChar *retPtr = NULL;
char *fileBuf = NULL;
UConverter* conv = NULL;
FILE *f = NULL;
ulen = 0;
if (U_FAILURE(status)) {
return retPtr;
}
//
// Open the file.
//
f = fopen(fileName, "rb");
if (f == 0) {
dataerrln("Error opening test data file %s\n", fileName);
status = U_FILE_ACCESS_ERROR;
return NULL;
}
//
// Read it in
//
int fileSize;
int amt_read;
fseek( f, 0, SEEK_END);
fileSize = ftell(f);
fileBuf = new char[fileSize];
fseek(f, 0, SEEK_SET);
amt_read = fread(fileBuf, 1, fileSize, f);
if (amt_read != fileSize || fileSize <= 0) {
errln("Error reading test data file.");
goto cleanUpAndReturn;
}
//
// Look for a Unicode Signature (BOM) on the data just read
//
int32_t signatureLength;
const char * fileBufC;
const char* bomEncoding;
fileBufC = fileBuf;
bomEncoding = ucnv_detectUnicodeSignature(
fileBuf, fileSize, &signatureLength, &status);
if(bomEncoding!=NULL ){
fileBufC += signatureLength;
fileSize -= signatureLength;
encoding = bomEncoding;
}
//
// Open a converter to take the rule file to UTF-16
//
conv = ucnv_open(encoding, &status);
if (U_FAILURE(status)) {
goto cleanUpAndReturn;
}
//
// Convert the rules to UChar.
// Preflight first to determine required buffer size.
//
ulen = ucnv_toUChars(conv,
NULL, // dest,
0, // destCapacity,
fileBufC,
fileSize,
&status);
if (status == U_BUFFER_OVERFLOW_ERROR) {
// Buffer Overflow is expected from the preflight operation.
status = U_ZERO_ERROR;
retPtr = new UChar[ulen+1];
ucnv_toUChars(conv,
retPtr, // dest,
ulen+1,
fileBufC,
fileSize,
&status);
}
cleanUpAndReturn:
fclose(f);
delete []fileBuf;
ucnv_close(conv);
if (U_FAILURE(status)) {
errln("ucnv_toUChars: ICU Error \"%s\"\n", u_errorName(status));
delete []retPtr;
retPtr = 0;
ulen = 0;
};
return retPtr;
}
//--------------------------------------------------------------------------------------------
//
// Run tests from each of the boundary test data files distributed by the Unicode Consortium
//
//-------------------------------------------------------------------------------------------
void RBBITest::TestUnicodeFiles() {
RuleBasedBreakIterator *bi;
UErrorCode status = U_ZERO_ERROR;
bi = (RuleBasedBreakIterator *)BreakIterator::createCharacterInstance(Locale::getEnglish(), status);
TEST_ASSERT_SUCCESS(status);
if (U_SUCCESS(status)) {
runUnicodeTestData("GraphemeBreakTest.txt", bi);
}
delete bi;
bi = (RuleBasedBreakIterator *)BreakIterator::createWordInstance(Locale::getEnglish(), status);
TEST_ASSERT_SUCCESS(status);
if (U_SUCCESS(status)) {
runUnicodeTestData("WordBreakTest.txt", bi);
}
delete bi;
bi = (RuleBasedBreakIterator *)BreakIterator::createSentenceInstance(Locale::getEnglish(), status);
TEST_ASSERT_SUCCESS(status);
if (U_SUCCESS(status)) {
runUnicodeTestData("SentenceBreakTest.txt", bi);
}
delete bi;
bi = (RuleBasedBreakIterator *)BreakIterator::createLineInstance(Locale::getEnglish(), status);
TEST_ASSERT_SUCCESS(status);
if (U_SUCCESS(status)) {
runUnicodeTestData("LineBreakTest.txt", bi);
}
delete bi;
}
//--------------------------------------------------------------------------------------------
//
// Run tests from one of the boundary test data files distributed by the Unicode Consortium
//
//-------------------------------------------------------------------------------------------
void RBBITest::runUnicodeTestData(const char *fileName, RuleBasedBreakIterator *bi) {
#if !UCONFIG_NO_REGULAR_EXPRESSIONS
// TODO(andy): Match line break behavior to Unicode 6.0 and remove this time bomb. Ticket #7270
UBool isTicket7270Fixed = isICUVersionAtLeast(52, 1);
UBool isLineBreak = 0 == strcmp(fileName, "LineBreakTest.txt");
UErrorCode status = U_ZERO_ERROR;
//
// Open and read the test data file, put it into a UnicodeString.
//
const char *testDataDirectory = IntlTest::getSourceTestData(status);
char testFileName[1000];
if (testDataDirectory == NULL || strlen(testDataDirectory) >= sizeof(testFileName)) {
dataerrln("Can't open test data. Path too long.");
return;
}
strcpy(testFileName, testDataDirectory);
strcat(testFileName, fileName);
logln("Opening data file %s\n", fileName);
int len;
UChar *testFile = ReadAndConvertFile(testFileName, len, "UTF-8", status);
if (status != U_FILE_ACCESS_ERROR) {
TEST_ASSERT_SUCCESS(status);
TEST_ASSERT(testFile != NULL);
}
if (U_FAILURE(status) || testFile == NULL) {
return; /* something went wrong, error already output */
}
UnicodeString testFileAsString(TRUE, testFile, len);
//
// Parse the test data file using a regular expression.
// Each kind of token is recognized in its own capture group; what type of item was scanned
// is identified by which group had a match.
//
// Caputure Group # 1 2 3 4 5
// Parses this item: divide x hex digits comment \n unrecognized \n
//
UnicodeString tokenExpr("[ \t]*(?:(\\u00F7)|(\\u00D7)|([0-9a-fA-F]+)|((?:#.*?)?$.)|(.*?$.))", -1, US_INV);
RegexMatcher tokenMatcher(tokenExpr, testFileAsString, UREGEX_MULTILINE | UREGEX_DOTALL, status);
UnicodeString testString;
UVector32 breakPositions(status);
int lineNumber = 1;
TEST_ASSERT_SUCCESS(status);
if (U_FAILURE(status)) {
return;
}
//
// Scan through each test case, building up the string to be broken in testString,
// and the positions that should be boundaries in the breakPositions vector.
//
int spin = 0;
while (tokenMatcher.find()) {
if(tokenMatcher.hitEnd()) {
/* Shouldnt Happen(TM). This means we didn't find the symbols we were looking for.
This occurred when the text file was corrupt (wasn't marked as UTF-8)
and caused an infinite loop here on EBCDIC systems!
*/
fprintf(stderr,"FAIL: hit end of file %s for the %8dth time- corrupt data file?\r", fileName, ++spin);
// return;
}
if (tokenMatcher.start(1, status) >= 0) {
// Scanned a divide sign, indicating a break position in the test data.
if (testString.length()>0) {
breakPositions.addElement(testString.length(), status);
}
}
else if (tokenMatcher.start(2, status) >= 0) {
// Scanned an 'x', meaning no break at this position in the test data
// Nothing to be done here.
}
else if (tokenMatcher.start(3, status) >= 0) {
// Scanned Hex digits. Convert them to binary, append to the character data string.
const UnicodeString &hexNumber = tokenMatcher.group(3, status);
int length = hexNumber.length();
if (length<=8) {
char buf[10];
hexNumber.extract (0, length, buf, sizeof(buf), US_INV);
UChar32 c = (UChar32)strtol(buf, NULL, 16);
if (c<=0x10ffff) {
testString.append(c);
} else {
errln("Error: Unicode Character value out of range. \'%s\', line %d.\n",
fileName, lineNumber);
}
} else {
errln("Syntax Error: Hex Unicode Character value must have no more than 8 digits at \'%s\', line %d.\n",
fileName, lineNumber);
}
}
else if (tokenMatcher.start(4, status) >= 0) {
// Scanned to end of a line, possibly skipping over a comment in the process.
// If the line from the file contained test data, run the test now.
//
if (testString.length() > 0) {
// TODO(andy): Remove this time bomb code. Note: Failing line numbers may change when updating to new Unicode data.
// Rule 8
// ZW SP* <break>
// is not yet implemented.
if (!(isLineBreak && !isTicket7270Fixed && (5198 == lineNumber ||
5202 == lineNumber ||
5214 == lineNumber ||
5246 == lineNumber ||
5298 == lineNumber ||
5302 == lineNumber ))) {
checkUnicodeTestCase(fileName, lineNumber, testString, &breakPositions, bi);
}
}
// Clear out this test case.
// The string and breakPositions vector will be refilled as the next
// test case is parsed.
testString.remove();
breakPositions.removeAllElements();
lineNumber++;
} else {
// Scanner catchall. Something unrecognized appeared on the line.
char token[16];
UnicodeString uToken = tokenMatcher.group(0, status);
uToken.extract(0, uToken.length(), token, (uint32_t)sizeof(token));
token[sizeof(token)-1] = 0;
errln("Syntax error in test data file \'%s\', line %d. Scanning \"%s\"\n", fileName, lineNumber, token);
// Clean up, in preparation for continuing with the next line.
testString.remove();
breakPositions.removeAllElements();
lineNumber++;
}
TEST_ASSERT_SUCCESS(status);
if (U_FAILURE(status)) {
break;
}
}
delete [] testFile;
#endif // !UCONFIG_NO_REGULAR_EXPRESSIONS
}
//--------------------------------------------------------------------------------------------
//
// checkUnicodeTestCase() Run one test case from one of the Unicode Consortium
// test data files. Do only a simple, forward-only check -
// this test is mostly to check that ICU and the Unicode
// data agree with each other.
//
//--------------------------------------------------------------------------------------------
void RBBITest::checkUnicodeTestCase(const char *testFileName, int lineNumber,
const UnicodeString &testString, // Text data to be broken
UVector32 *breakPositions, // Positions where breaks should be found.
RuleBasedBreakIterator *bi) {
int32_t pos; // Break Position in the test string
int32_t expectedI = 0; // Index of expected break position in the vector of expected results.
int32_t expectedPos; // Expected break position (index into test string)
bi->setText(testString);
pos = bi->first();
pos = bi->next();
while (pos != BreakIterator::DONE) {
if (expectedI >= breakPositions->size()) {
errln("Test file \"%s\", line %d, unexpected break found at position %d",
testFileName, lineNumber, pos);
break;
}
expectedPos = breakPositions->elementAti(expectedI);
if (pos < expectedPos) {
errln("Test file \"%s\", line %d, unexpected break found at position %d",
testFileName, lineNumber, pos);
break;
}
if (pos > expectedPos) {
errln("Test file \"%s\", line %d, failed to find expected break at position %d",
testFileName, lineNumber, expectedPos);
break;
}
pos = bi->next();
expectedI++;
}
if (pos==BreakIterator::DONE && expectedI<breakPositions->size()) {
errln("Test file \"%s\", line %d, failed to find expected break at position %d",
testFileName, lineNumber, breakPositions->elementAti(expectedI));
}
}
#if !UCONFIG_NO_REGULAR_EXPRESSIONS
//---------------------------------------------------------------------------------------
//
// classs RBBIMonkeyKind
//
// Monkey Test for Break Iteration
// Abstract interface class. Concrete derived classes independently
// implement the break rules for different iterator types.
//
// The Monkey Test itself uses doesn't know which type of break iterator it is
// testing, but works purely in terms of the interface defined here.
//
//---------------------------------------------------------------------------------------
class RBBIMonkeyKind {
public:
// Return a UVector of UnicodeSets, representing the character classes used
// for this type of iterator.
virtual UVector *charClasses() = 0;
// Set the test text on which subsequent calls to next() will operate
virtual void setText(const UnicodeString &s) = 0;
// Find the next break postion, starting from the prev break position, or from zero.
// Return -1 after reaching end of string.
virtual int32_t next(int32_t i) = 0;
virtual ~RBBIMonkeyKind();
UErrorCode deferredStatus;
protected:
RBBIMonkeyKind();
private:
};
RBBIMonkeyKind::RBBIMonkeyKind() {
deferredStatus = U_ZERO_ERROR;
}
RBBIMonkeyKind::~RBBIMonkeyKind() {
}
//----------------------------------------------------------------------------------------
//
// Random Numbers. Similar to standard lib rand() and srand()
// Not using library to
// 1. Get same results on all platforms.
// 2. Get access to current seed, to more easily reproduce failures.
//
//---------------------------------------------------------------------------------------
static uint32_t m_seed = 1;
static uint32_t m_rand()
{
m_seed = m_seed * 1103515245 + 12345;
return (uint32_t)(m_seed/65536) % 32768;
}
//------------------------------------------------------------------------------------------
//
// class RBBICharMonkey Character (Grapheme Cluster) specific implementation
// of RBBIMonkeyKind.
//
//------------------------------------------------------------------------------------------
class RBBICharMonkey: public RBBIMonkeyKind {
public:
RBBICharMonkey();
virtual ~RBBICharMonkey();
virtual UVector *charClasses();
virtual void setText(const UnicodeString &s);
virtual int32_t next(int32_t i);
private:
UVector *fSets;
UnicodeSet *fCRLFSet;
UnicodeSet *fControlSet;
UnicodeSet *fExtendSet;
UnicodeSet *fRegionalIndicatorSet;
UnicodeSet *fPrependSet;
UnicodeSet *fSpacingSet;
UnicodeSet *fLSet;
UnicodeSet *fVSet;
UnicodeSet *fTSet;
UnicodeSet *fLVSet;
UnicodeSet *fLVTSet;
UnicodeSet *fHangulSet;
UnicodeSet *fAnySet;
const UnicodeString *fText;
};
RBBICharMonkey::RBBICharMonkey() {
UErrorCode status = U_ZERO_ERROR;
fText = NULL;
fCRLFSet = new UnicodeSet(UNICODE_STRING_SIMPLE("[\\r\\n]"), status);
fControlSet = new UnicodeSet(UNICODE_STRING_SIMPLE("[\\p{Grapheme_Cluster_Break = Control}]"), status);
fExtendSet = new UnicodeSet(UNICODE_STRING_SIMPLE("[\\p{Grapheme_Cluster_Break = Extend}]"), status);
fRegionalIndicatorSet = new UnicodeSet(UNICODE_STRING_SIMPLE("[\\p{Grapheme_Cluster_Break = Regional_Indicator}]"), status);
fPrependSet = new UnicodeSet(UNICODE_STRING_SIMPLE("[\\p{Grapheme_Cluster_Break = Prepend}]"), status);
fSpacingSet = new UnicodeSet(UNICODE_STRING_SIMPLE("[\\p{Grapheme_Cluster_Break = SpacingMark}]"), status);
fLSet = new UnicodeSet(UNICODE_STRING_SIMPLE("[\\p{Grapheme_Cluster_Break = L}]"), status);
fVSet = new UnicodeSet(UNICODE_STRING_SIMPLE("[\\p{Grapheme_Cluster_Break = V}]"), status);
fTSet = new UnicodeSet(UNICODE_STRING_SIMPLE("[\\p{Grapheme_Cluster_Break = T}]"), status);
fLVSet = new UnicodeSet(UNICODE_STRING_SIMPLE("[\\p{Grapheme_Cluster_Break = LV}]"), status);
fLVTSet = new UnicodeSet(UNICODE_STRING_SIMPLE("[\\p{Grapheme_Cluster_Break = LVT}]"), status);
fHangulSet = new UnicodeSet();
fHangulSet->addAll(*fLSet);
fHangulSet->addAll(*fVSet);
fHangulSet->addAll(*fTSet);
fHangulSet->addAll(*fLVSet);
fHangulSet->addAll(*fLVTSet);
fAnySet = new UnicodeSet(0, 0x10ffff);
fSets = new UVector(status);
fSets->addElement(fCRLFSet, status);
fSets->addElement(fControlSet, status);
fSets->addElement(fExtendSet, status);
fSets->addElement(fRegionalIndicatorSet, status);
if (!fPrependSet->isEmpty()) {
fSets->addElement(fPrependSet, status);
}
fSets->addElement(fSpacingSet, status);
fSets->addElement(fHangulSet, status);
fSets->addElement(fAnySet, status);
if (U_FAILURE(status)) {
deferredStatus = status;
}
}
void RBBICharMonkey::setText(const UnicodeString &s) {
fText = &s;
}
int32_t RBBICharMonkey::next(int32_t prevPos) {
int p0, p1, p2, p3; // Indices of the significant code points around the
// break position being tested. The candidate break
// location is before p2.
int breakPos = -1;
UChar32 c0, c1, c2, c3; // The code points at p0, p1, p2 & p3.
if (U_FAILURE(deferredStatus)) {
return -1;
}
// Previous break at end of string. return DONE.
if (prevPos >= fText->length()) {
return -1;
}
p0 = p1 = p2 = p3 = prevPos;
c3 = fText->char32At(prevPos);
c0 = c1 = c2 = 0;
// Loop runs once per "significant" character position in the input text.
for (;;) {
// Move all of the positions forward in the input string.
p0 = p1; c0 = c1;
p1 = p2; c1 = c2;
p2 = p3; c2 = c3;
// Advancd p3 by one codepoint
p3 = fText->moveIndex32(p3, 1);
c3 = fText->char32At(p3);
if (p1 == p2) {
// Still warming up the loop. (won't work with zero length strings, but we don't care)
continue;
}
if (p2 == fText->length()) {
// Reached end of string. Always a break position.
break;
}
// Rule GB3 CR x LF
// No Extend or Format characters may appear between the CR and LF,
// which requires the additional check for p2 immediately following p1.
//
if (c1==0x0D && c2==0x0A && p1==(p2-1)) {
continue;
}
// Rule (GB4). ( Control | CR | LF ) <break>
if (fControlSet->contains(c1) ||
c1 == 0x0D ||
c1 == 0x0A) {
break;
}
// Rule (GB5) <break> ( Control | CR | LF )
//
if (fControlSet->contains(c2) ||
c2 == 0x0D ||
c2 == 0x0A) {
break;
}
// Rule (GB6) L x ( L | V | LV | LVT )
if (fLSet->contains(c1) &&
(fLSet->contains(c2) ||
fVSet->contains(c2) ||
fLVSet->contains(c2) ||
fLVTSet->contains(c2))) {
continue;
}
// Rule (GB7) ( LV | V ) x ( V | T )
if ((fLVSet->contains(c1) || fVSet->contains(c1)) &&
(fVSet->contains(c2) || fTSet->contains(c2))) {
continue;
}
// Rule (GB8) ( LVT | T) x T
if ((fLVTSet->contains(c1) || fTSet->contains(c1)) &&
fTSet->contains(c2)) {
continue;
}
// Rule (GB8a) Regional_Indicator x Regional_Indicator
if (fRegionalIndicatorSet->contains(c1) && fRegionalIndicatorSet->contains(c2)) {
continue;
}
// Rule (GB9) Numeric x ALetter
if (fExtendSet->contains(c2)) {
continue;
}
// Rule (GB9a) x SpacingMark
if (fSpacingSet->contains(c2)) {
continue;
}
// Rule (GB9b) Prepend x
if (fPrependSet->contains(c1)) {
continue;
}
// Rule (GB10) Any <break> Any
break;
}
breakPos = p2;
return breakPos;
}
UVector *RBBICharMonkey::charClasses() {
return fSets;
}
RBBICharMonkey::~RBBICharMonkey() {
delete fSets;
delete fCRLFSet;
delete fControlSet;
delete fExtendSet;
delete fRegionalIndicatorSet;
delete fPrependSet;
delete fSpacingSet;
delete fLSet;
delete fVSet;
delete fTSet;
delete fLVSet;
delete fLVTSet;
delete fHangulSet;
delete fAnySet;
}
//------------------------------------------------------------------------------------------
//
// class RBBIWordMonkey Word Break specific implementation
// of RBBIMonkeyKind.
//
//------------------------------------------------------------------------------------------
class RBBIWordMonkey: public RBBIMonkeyKind {
public:
RBBIWordMonkey();
virtual ~RBBIWordMonkey();
virtual UVector *charClasses();
virtual void setText(const UnicodeString &s);
virtual int32_t next(int32_t i);
private:
UVector *fSets;
UnicodeSet *fCRSet;
UnicodeSet *fLFSet;
UnicodeSet *fNewlineSet;
UnicodeSet *fKatakanaSet;
UnicodeSet *fALetterSet;
// TODO(jungshik): Do we still need this change?
// UnicodeSet *fALetterSet; // matches ALetterPlus in word.txt
UnicodeSet *fMidNumLetSet;
UnicodeSet *fMidLetterSet;
UnicodeSet *fMidNumSet;
UnicodeSet *fNumericSet;
UnicodeSet *fFormatSet;
UnicodeSet *fOtherSet;
UnicodeSet *fExtendSet;
UnicodeSet *fExtendNumLetSet;
UnicodeSet *fRegionalIndicatorSet;
UnicodeSet *fDictionaryCjkSet;
RegexMatcher *fMatcher;
const UnicodeString *fText;
};
RBBIWordMonkey::RBBIWordMonkey()
{
UErrorCode status = U_ZERO_ERROR;
fSets = new UVector(status);
fCRSet = new UnicodeSet(UNICODE_STRING_SIMPLE("[\\p{Word_Break = CR}]"), status);
fLFSet = new UnicodeSet(UNICODE_STRING_SIMPLE("[\\p{Word_Break = LF}]"), status);
fNewlineSet = new UnicodeSet(UNICODE_STRING_SIMPLE("[\\p{Word_Break = Newline}]"), status);
fDictionaryCjkSet= new UnicodeSet("[[\\uac00-\\ud7a3][:Han:][:Hiragana:][:Katakana:]]", status);
// Exclude Hangul syllables from ALetterSet during testing.
// Leave CJK dictionary characters out from the monkey tests!
#if 0
fALetterSet = new UnicodeSet("[\\p{Word_Break = ALetter}"
"[\\p{Line_Break = Complex_Context}"
"-\\p{Grapheme_Cluster_Break = Extend}"
"-\\p{Grapheme_Cluster_Break = Control}"
"]]",
status);
#endif
fALetterSet = new UnicodeSet(UNICODE_STRING_SIMPLE("[\\p{Word_Break = ALetter}]"), status);
fALetterSet->removeAll(*fDictionaryCjkSet);
fKatakanaSet = new UnicodeSet(UNICODE_STRING_SIMPLE("[\\p{Word_Break = Katakana}]"), status);
fMidNumLetSet = new UnicodeSet(UNICODE_STRING_SIMPLE("[\\p{Word_Break = MidNumLet}]"), status);
fMidLetterSet = new UnicodeSet(UNICODE_STRING_SIMPLE("[\\p{Word_Break = MidLetter}]"), status);
fMidNumSet = new UnicodeSet(UNICODE_STRING_SIMPLE("[\\p{Word_Break = MidNum}]"), status);
// TODO: this set used to contain [\\uff10-\\uff19] (fullwidth digits), but this breaks the test
// we should figure out why
fNumericSet = new UnicodeSet(UNICODE_STRING_SIMPLE("[\\p{Word_Break = Numeric}]"), status);
fFormatSet = new UnicodeSet(UNICODE_STRING_SIMPLE("[\\p{Word_Break = Format}]"), status);
fExtendNumLetSet = new UnicodeSet(UNICODE_STRING_SIMPLE("[\\p{Word_Break = ExtendNumLet}]"), status);
fExtendSet = new UnicodeSet(UNICODE_STRING_SIMPLE("[\\p{Word_Break = Extend}]"), status);
fRegionalIndicatorSet = new UnicodeSet(UNICODE_STRING_SIMPLE("[\\p{Word_Break = Regional_Indicator}]"), status);
fOtherSet = new UnicodeSet();
if(U_FAILURE(status)) {
deferredStatus = status;
return;
}
fOtherSet->complement();
fOtherSet->removeAll(*fCRSet);
fOtherSet->removeAll(*fLFSet);
fOtherSet->removeAll(*fNewlineSet);
fOtherSet->removeAll(*fKatakanaSet);
fOtherSet->removeAll(*fALetterSet);
fOtherSet->removeAll(*fMidLetterSet);
fOtherSet->removeAll(*fMidNumSet);
fOtherSet->removeAll(*fNumericSet);
fOtherSet->removeAll(*fExtendNumLetSet);
fOtherSet->removeAll(*fFormatSet);
fOtherSet->removeAll(*fExtendSet);
fOtherSet->removeAll(*fRegionalIndicatorSet);
// Inhibit dictionary characters from being tested at all.
fOtherSet->removeAll(*fDictionaryCjkSet);
fOtherSet->removeAll(UnicodeSet(UNICODE_STRING_SIMPLE("[\\p{LineBreak = Complex_Context}]"), status));
fSets->addElement(fCRSet, status);
fSets->addElement(fLFSet, status);
fSets->addElement(fNewlineSet, status);
fSets->addElement(fALetterSet, status);
//fSets->addElement(fKatakanaSet, status); //TODO: work out how to test katakana
fSets->addElement(fMidLetterSet, status);
fSets->addElement(fMidNumLetSet, status);
fSets->addElement(fMidNumSet, status);
fSets->addElement(fNumericSet, status);
fSets->addElement(fFormatSet, status);
fSets->addElement(fExtendSet, status);
fSets->addElement(fOtherSet, status);
fSets->addElement(fExtendNumLetSet, status);
fSets->addElement(fRegionalIndicatorSet, status);
if (U_FAILURE(status)) {
deferredStatus = status;
}
}
void RBBIWordMonkey::setText(const UnicodeString &s) {
fText = &s;
}
int32_t RBBIWordMonkey::next(int32_t prevPos) {
int p0, p1, p2, p3; // Indices of the significant code points around the
// break position being tested. The candidate break
// location is before p2.
int breakPos = -1;
UChar32 c0, c1, c2, c3; // The code points at p0, p1, p2 & p3.
if (U_FAILURE(deferredStatus)) {
return -1;
}
// Prev break at end of string. return DONE.
if (prevPos >= fText->length()) {
return -1;
}
p0 = p1 = p2 = p3 = prevPos;
c3 = fText->char32At(prevPos);
c0 = c1 = c2 = 0;
// Loop runs once per "significant" character position in the input text.
for (;;) {
// Move all of the positions forward in the input string.
p0 = p1; c0 = c1;
p1 = p2; c1 = c2;
p2 = p3; c2 = c3;
// Advancd p3 by X(Extend | Format)* Rule 4
// But do not advance over Extend & Format following a new line. (Unicode 5.1 change)
do {
p3 = fText->moveIndex32(p3, 1);
c3 = fText->char32At(p3);
if (fCRSet->contains(c2) || fLFSet->contains(c2) || fNewlineSet->contains(c2)) {
break;
};
}
while (fFormatSet->contains(c3) || fExtendSet->contains(c3));
if (p1 == p2) {
// Still warming up the loop. (won't work with zero length strings, but we don't care)
continue;
}
if (p2 == fText->length()) {
// Reached end of string. Always a break position.
break;
}
// Rule (3) CR x LF
// No Extend or Format characters may appear between the CR and LF,
// which requires the additional check for p2 immediately following p1.
//
if (c1==0x0D && c2==0x0A) {
continue;
}
// Rule (3a) Break before and after newlines (including CR and LF)
//
if (fCRSet->contains(c1) || fLFSet->contains(c1) || fNewlineSet->contains(c1)) {
break;
};
if (fCRSet->contains(c2) || fLFSet->contains(c2) || fNewlineSet->contains(c2)) {
break;
};
// Rule (5). ALetter x ALetter
if (fALetterSet->contains(c1) &&
fALetterSet->contains(c2)) {
continue;
}
// Rule (6) ALetter x (MidLetter | MidNumLet) ALetter
//
if ( fALetterSet->contains(c1) &&
(fMidLetterSet->contains(c2) || fMidNumLetSet->contains(c2)) &&
fALetterSet->contains(c3)) {
continue;
}
// Rule (7) ALetter (MidLetter | MidNumLet) x ALetter
if (fALetterSet->contains(c0) &&
(fMidLetterSet->contains(c1) || fMidNumLetSet->contains(c1)) &&
fALetterSet->contains(c2)) {
continue;
}
// Rule (8) Numeric x Numeric
if (fNumericSet->contains(c1) &&
fNumericSet->contains(c2)) {
continue;
}
// Rule (9) ALetter x Numeric
if (fALetterSet->contains(c1) &&
fNumericSet->contains(c2)) {
continue;
}
// Rule (10) Numeric x ALetter
if (fNumericSet->contains(c1) &&
fALetterSet->contains(c2)) {
continue;
}
// Rule (11) Numeric (MidNum | MidNumLet) x Numeric
if (fNumericSet->contains(c0) &&
(fMidNumSet->contains(c1) || fMidNumLetSet->contains(c1)) &&
fNumericSet->contains(c2)) {
continue;
}
// Rule (12) Numeric x (MidNum | MidNumLet) Numeric
if (fNumericSet->contains(c1) &&
(fMidNumSet->contains(c2) || fMidNumLetSet->contains(c2)) &&
fNumericSet->contains(c3)) {
continue;
}
// Rule (13) Katakana x Katakana
if (fKatakanaSet->contains(c1) &&
fKatakanaSet->contains(c2)) {
continue;
}
// Rule 13a
if ((fALetterSet->contains(c1) || fNumericSet->contains(c1) ||
fKatakanaSet->contains(c1) || fExtendNumLetSet->contains(c1)) &&
fExtendNumLetSet->contains(c2)) {
continue;
}
// Rule 13b
if (fExtendNumLetSet->contains(c1) &&
(fALetterSet->contains(c2) || fNumericSet->contains(c2) ||
fKatakanaSet->contains(c2))) {
continue;
}
// Rule 13c
if (fRegionalIndicatorSet->contains(c1) && fRegionalIndicatorSet->contains(c2)) {
continue;
}
// Rule 14. Break found here.
break;
}
breakPos = p2;
return breakPos;
}
UVector *RBBIWordMonkey::charClasses() {
return fSets;
}
RBBIWordMonkey::~RBBIWordMonkey() {
delete fSets;
delete fCRSet;
delete fLFSet;
delete fNewlineSet;
delete fKatakanaSet;
delete fALetterSet;
delete fMidNumLetSet;
delete fMidLetterSet;
delete fMidNumSet;
delete fNumericSet;
delete fFormatSet;
delete fExtendSet;
delete fExtendNumLetSet;
delete fRegionalIndicatorSet;
delete fDictionaryCjkSet;
delete fOtherSet;
}
//------------------------------------------------------------------------------------------
//
// class RBBISentMonkey Sentence Break specific implementation
// of RBBIMonkeyKind.
//
//------------------------------------------------------------------------------------------
class RBBISentMonkey: public RBBIMonkeyKind {
public:
RBBISentMonkey();
virtual ~RBBISentMonkey();
virtual UVector *charClasses();
virtual void setText(const UnicodeString &s);
virtual int32_t next(int32_t i);
private:
int moveBack(int posFrom);
int moveForward(int posFrom);
UChar32 cAt(int pos);
UVector *fSets;
UnicodeSet *fSepSet;
UnicodeSet *fFormatSet;
UnicodeSet *fSpSet;
UnicodeSet *fLowerSet;
UnicodeSet *fUpperSet;
UnicodeSet *fOLetterSet;
UnicodeSet *fNumericSet;
UnicodeSet *fATermSet;
UnicodeSet *fSContinueSet;
UnicodeSet *fSTermSet;
UnicodeSet *fCloseSet;
UnicodeSet *fOtherSet;
UnicodeSet *fExtendSet;
const UnicodeString *fText;
};
RBBISentMonkey::RBBISentMonkey()
{
UErrorCode status = U_ZERO_ERROR;
fSets = new UVector(status);
// Separator Set Note: Beginning with Unicode 5.1, CR and LF were removed from the separator
// set and made into character classes of their own. For the monkey impl,
// they remain in SEP, since Sep always appears with CR and LF in the rules.
fSepSet = new UnicodeSet(UNICODE_STRING_SIMPLE("[\\p{Sentence_Break = Sep} \\u000a \\u000d]"), status);
fFormatSet = new UnicodeSet(UNICODE_STRING_SIMPLE("[\\p{Sentence_Break = Format}]"), status);
fSpSet = new UnicodeSet(UNICODE_STRING_SIMPLE("[\\p{Sentence_Break = Sp}]"), status);
fLowerSet = new UnicodeSet(UNICODE_STRING_SIMPLE("[\\p{Sentence_Break = Lower}]"), status);
fUpperSet = new UnicodeSet(UNICODE_STRING_SIMPLE("[\\p{Sentence_Break = Upper}]"), status);
fOLetterSet = new UnicodeSet(UNICODE_STRING_SIMPLE("[\\p{Sentence_Break = OLetter}]"), status);
fNumericSet = new UnicodeSet(UNICODE_STRING_SIMPLE("[\\p{Sentence_Break = Numeric}]"), status);
fATermSet = new UnicodeSet(UNICODE_STRING_SIMPLE("[\\p{Sentence_Break = ATerm}]"), status);
fSContinueSet = new UnicodeSet(UNICODE_STRING_SIMPLE("[\\p{Sentence_Break = SContinue}]"), status);
fSTermSet = new UnicodeSet(UNICODE_STRING_SIMPLE("[\\p{Sentence_Break = STerm}]"), status);
fCloseSet = new UnicodeSet(UNICODE_STRING_SIMPLE("[\\p{Sentence_Break = Close}]"), status);
fExtendSet = new UnicodeSet(UNICODE_STRING_SIMPLE("[\\p{Sentence_Break = Extend}]"), status);
fOtherSet = new UnicodeSet();
if(U_FAILURE(status)) {
deferredStatus = status;
return;
}
fOtherSet->complement();
fOtherSet->removeAll(*fSepSet);
fOtherSet->removeAll(*fFormatSet);
fOtherSet->removeAll(*fSpSet);
fOtherSet->removeAll(*fLowerSet);
fOtherSet->removeAll(*fUpperSet);
fOtherSet->removeAll(*fOLetterSet);
fOtherSet->removeAll(*fNumericSet);
fOtherSet->removeAll(*fATermSet);
fOtherSet->removeAll(*fSContinueSet);
fOtherSet->removeAll(*fSTermSet);
fOtherSet->removeAll(*fCloseSet);
fOtherSet->removeAll(*fExtendSet);
fSets->addElement(fSepSet, status);
fSets->addElement(fFormatSet, status);
fSets->addElement(fSpSet, status);
fSets->addElement(fLowerSet, status);
fSets->addElement(fUpperSet, status);
fSets->addElement(fOLetterSet, status);
fSets->addElement(fNumericSet, status);
fSets->addElement(fATermSet, status);
fSets->addElement(fSContinueSet, status);
fSets->addElement(fSTermSet, status);
fSets->addElement(fCloseSet, status);
fSets->addElement(fOtherSet, status);
fSets->addElement(fExtendSet, status);
if (U_FAILURE(status)) {
deferredStatus = status;
}
}
void RBBISentMonkey::setText(const UnicodeString &s) {
fText = &s;
}
UVector *RBBISentMonkey::charClasses() {
return fSets;
}
// moveBack() Find the "significant" code point preceding the index i.
// Skips over ($Extend | $Format)* .
//
int RBBISentMonkey::moveBack(int i) {
if (i <= 0) {
return -1;
}
UChar32 c;
int32_t j = i;
do {
j = fText->moveIndex32(j, -1);
c = fText->char32At(j);
}
while (j>0 &&(fFormatSet->contains(c) || fExtendSet->contains(c)));
return j;
}
int RBBISentMonkey::moveForward(int i) {
if (i>=fText->length()) {
return fText->length();
}
UChar32 c;
int32_t j = i;
do {
j = fText->moveIndex32(j, 1);
c = cAt(j);
}
while (fFormatSet->contains(c) || fExtendSet->contains(c));
return j;
}
UChar32 RBBISentMonkey::cAt(int pos) {
if (pos<0 || pos>=fText->length()) {
return -1;
} else {
return fText->char32At(pos);
}
}
int32_t RBBISentMonkey::next(int32_t prevPos) {
int p0, p1, p2, p3; // Indices of the significant code points around the
// break position being tested. The candidate break
// location is before p2.
int breakPos = -1;
UChar32 c0, c1, c2, c3; // The code points at p0, p1, p2 & p3.
UChar32 c;
if (U_FAILURE(deferredStatus)) {
return -1;
}
// Prev break at end of string. return DONE.
if (prevPos >= fText->length()) {
return -1;
}
p0 = p1 = p2 = p3 = prevPos;
c3 = fText->char32At(prevPos);
c0 = c1 = c2 = 0;
// Loop runs once per "significant" character position in the input text.
for (;;) {
// Move all of the positions forward in the input string.
p0 = p1; c0 = c1;
p1 = p2; c1 = c2;
p2 = p3; c2 = c3;
// Advancd p3 by X(Extend | Format)* Rule 4
p3 = moveForward(p3);
c3 = cAt(p3);
// Rule (3) CR x LF
if (c1==0x0d && c2==0x0a && p2==(p1+1)) {
continue;
}
// Rule (4). Sep <break>
if (fSepSet->contains(c1)) {
p2 = p1+1; // Separators don't combine with Extend or Format.
break;
}
if (p2 >= fText->length()) {
// Reached end of string. Always a break position.
break;
}
if (p2 == prevPos) {
// Still warming up the loop. (won't work with zero length strings, but we don't care)
continue;
}
// Rule (6). ATerm x Numeric
if (fATermSet->contains(c1) && fNumericSet->contains(c2)) {
continue;
}
// Rule (7). Upper ATerm x Uppper
if (fUpperSet->contains(c0) && fATermSet->contains(c1) && fUpperSet->contains(c2)) {
continue;
}
// Rule (8) ATerm Close* Sp* x (not (OLettter | Upper | Lower | Sep | STerm | ATerm))* Lower
// Note: STerm | ATerm are added to the negated part of the expression by a
// note to the Unicode 5.0 documents.
int p8 = p1;
while (fSpSet->contains(cAt(p8))) {
p8 = moveBack(p8);
}
while (fCloseSet->contains(cAt(p8))) {
p8 = moveBack(p8);
}
if (fATermSet->contains(cAt(p8))) {
p8=p2;
for (;;) {
c = cAt(p8);
if (c==-1 || fOLetterSet->contains(c) || fUpperSet->contains(c) ||
fLowerSet->contains(c) || fSepSet->contains(c) ||
fATermSet->contains(c) || fSTermSet->contains(c)) {
break;
}
p8 = moveForward(p8);
}
if (fLowerSet->contains(cAt(p8))) {
continue;
}
}
// Rule 8a (STerm | ATerm) Close* Sp* x (SContinue | STerm | ATerm);
if (fSContinueSet->contains(c2) || fSTermSet->contains(c2) || fATermSet->contains(c2)) {
p8 = p1;
while (fSpSet->contains(cAt(p8))) {
p8 = moveBack(p8);
}
while (fCloseSet->contains(cAt(p8))) {
p8 = moveBack(p8);
}
c = cAt(p8);
if (fSTermSet->contains(c) || fATermSet->contains(c)) {
continue;
}
}
// Rule (9) (STerm | ATerm) Close* x (Close | Sp | Sep | CR | LF)
int p9 = p1;
while (fCloseSet->contains(cAt(p9))) {
p9 = moveBack(p9);
}
c = cAt(p9);
if ((fSTermSet->contains(c) || fATermSet->contains(c))) {
if (fCloseSet->contains(c2) || fSpSet->contains(c2) || fSepSet->contains(c2)) {
continue;
}
}
// Rule (10) (Sterm | ATerm) Close* Sp* x (Sp | Sep | CR | LF)
int p10 = p1;
while (fSpSet->contains(cAt(p10))) {
p10 = moveBack(p10);
}
while (fCloseSet->contains(cAt(p10))) {
p10 = moveBack(p10);
}
if (fSTermSet->contains(cAt(p10)) || fATermSet->contains(cAt(p10))) {
if (fSpSet->contains(c2) || fSepSet->contains(c2)) {
continue;
}
}
// Rule (11) (STerm | ATerm) Close* Sp* (Sep | CR | LF)? <break>
int p11 = p1;
if (fSepSet->contains(cAt(p11))) {
p11 = moveBack(p11);
}
while (fSpSet->contains(cAt(p11))) {
p11 = moveBack(p11);
}
while (fCloseSet->