16f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org/******************************************************************** 26f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org * COPYRIGHT: 36f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org * Copyright (c) 1997-2013, International Business Machines Corporation and 46f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org * others. All Rights Reserved. 56f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org ********************************************************************/ 66f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org/******************************************************************************** 76f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org* 86f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org* File CBIAPTS.C 96f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org* 106f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org* Modification History: 116f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org* Name Description 126f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org* Madhu Katragadda Creation 136f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org*********************************************************************************/ 146f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org/*C API TEST FOR BREAKITERATOR */ 156f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org/** 166f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org* This is an API test. It doesn't test very many cases, and doesn't 176f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org* try to test the full functionality. It just calls each function in the class and 186f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org* verifies that it works on a basic level. 196f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org**/ 206f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org 216f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org#include "unicode/utypes.h" 226f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org 236f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org#if !UCONFIG_NO_BREAK_ITERATION 246f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org 256f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org#include <stdlib.h> 266f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org#include <string.h> 276f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org#include "unicode/uloc.h" 286f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org#include "unicode/ubrk.h" 296f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org#include "unicode/ustring.h" 306f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org#include "unicode/ucnv.h" 316f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org#include "unicode/utext.h" 326f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org#include "cintltst.h" 336f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org#include "cbiapts.h" 346f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org 356f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org#define TEST_ASSERT_SUCCESS(status) {if (U_FAILURE(status)) { \ 366f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.orglog_data_err("Failure at file %s, line %d, error = %s (Are you missing data?)\n", __FILE__, __LINE__, u_errorName(status));}} 376f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org 386f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org#define TEST_ASSERT(expr) {if ((expr)==FALSE) { \ 396f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.orglog_data_err("Test Failure at file %s, line %d (Are you missing data?)\n", __FILE__, __LINE__);}} 406f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org 416f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org#if !UCONFIG_NO_FILE_IO 426f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.orgstatic void TestBreakIteratorSafeClone(void); 436f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org#endif 446f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.orgstatic void TestBreakIteratorRules(void); 456f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.orgstatic void TestBreakIteratorRuleError(void); 466f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.orgstatic void TestBreakIteratorStatusVec(void); 476f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.orgstatic void TestBreakIteratorUText(void); 486f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.orgstatic void TestBreakIteratorTailoring(void); 496f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.orgstatic void TestBreakIteratorRefresh(void); 506f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org 516f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.orgvoid addBrkIterAPITest(TestNode** root); 526f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org 536f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.orgvoid addBrkIterAPITest(TestNode** root) 546f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org{ 556f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org#if !UCONFIG_NO_FILE_IO 566f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org addTest(root, &TestBreakIteratorCAPI, "tstxtbd/cbiapts/TestBreakIteratorCAPI"); 576f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org addTest(root, &TestBreakIteratorSafeClone, "tstxtbd/cbiapts/TestBreakIteratorSafeClone"); 586f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org addTest(root, &TestBreakIteratorUText, "tstxtbd/cbiapts/TestBreakIteratorUText"); 596f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org#endif 606f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org addTest(root, &TestBreakIteratorRules, "tstxtbd/cbiapts/TestBreakIteratorRules"); 616f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org addTest(root, &TestBreakIteratorRuleError, "tstxtbd/cbiapts/TestBreakIteratorRuleError"); 626f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org addTest(root, &TestBreakIteratorStatusVec, "tstxtbd/cbiapts/TestBreakIteratorStatusVec"); 636f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org addTest(root, &TestBreakIteratorTailoring, "tstxtbd/cbiapts/TestBreakIteratorTailoring"); 646f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org addTest(root, &TestBreakIteratorRefresh, "tstxtbd/cbiapts/TestBreakIteratorRefresh"); 656f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org} 666f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org 676f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org#define CLONETEST_ITERATOR_COUNT 2 686f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org 696f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org/* 706f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org * Utility function for converting char * to UChar * strings, to 716f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org * simplify the test code. Converted strings are put in heap allocated 726f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org * storage. A hook (probably a local in the caller's code) allows all 736f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org * strings converted with that hook to be freed with a single call. 746f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org */ 756f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.orgtypedef struct StringStruct { 766f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org struct StringStruct *link; 776f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org UChar str[1]; 786f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org } StringStruct; 796f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org 806f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org 816f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.orgstatic UChar* toUChar(const char *src, void **freeHook) { 826f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org /* Structure of the memory that we allocate on the heap */ 836f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org 846f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org int32_t numUChars; 856f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org int32_t destSize; 866f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org UChar stackBuf[2000 + sizeof(void *)/sizeof(UChar)]; 876f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org StringStruct *dest; 886f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org UConverter *cnv; 896f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org 906f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org UErrorCode status = U_ZERO_ERROR; 916f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org if (src == NULL) { 926f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org return NULL; 936f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org }; 946f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org 956f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org cnv = ucnv_open(NULL, &status); 966f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org if(U_FAILURE(status) || cnv == NULL) { 976f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org return NULL; 986f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org } 996f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org ucnv_reset(cnv); 1006f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org numUChars = ucnv_toUChars(cnv, 1016f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org stackBuf, 1026f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org 2000, 1036f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org src, -1, 1046f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org &status); 1056f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org 1066f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org destSize = (numUChars+1) * sizeof(UChar) + sizeof(struct StringStruct); 1076f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org dest = (StringStruct *)malloc(destSize); 1086f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org if (dest != NULL) { 1096f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org if (status == U_BUFFER_OVERFLOW_ERROR || status == U_STRING_NOT_TERMINATED_WARNING) { 1106f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org ucnv_toUChars(cnv, dest->str, numUChars+1, src, -1, &status); 1116f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org } else if (status == U_ZERO_ERROR) { 1126f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org u_strcpy(dest->str, stackBuf); 1136f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org } else { 1146f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org free(dest); 1156f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org dest = NULL; 1166f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org } 1176f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org } 1186f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org 1196f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org ucnv_reset(cnv); /* be good citizens */ 1206f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org ucnv_close(cnv); 1216f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org if (dest == NULL) { 1226f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org return NULL; 1236f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org } 1246f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org 1256f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org dest->link = (StringStruct*)(*freeHook); 1266f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org *freeHook = dest; 1276f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org return dest->str; 1286f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org} 1296f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org 1306f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.orgstatic void freeToUCharStrings(void **hook) { 1316f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org StringStruct *s = *(StringStruct **)hook; 1326f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org while (s != NULL) { 1336f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org StringStruct *next = s->link; 1346f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org free(s); 1356f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org s = next; 1366f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org } 1376f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org} 1386f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org 1396f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org 1406f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org#if !UCONFIG_NO_FILE_IO 1416f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.orgstatic void TestBreakIteratorCAPI() 1426f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org{ 1436f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org UErrorCode status = U_ZERO_ERROR; 1446f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org UBreakIterator *word, *sentence, *line, *character, *b, *bogus; 1456f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org int32_t start,pos,end,to; 1466f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org int32_t i; 1476f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org int32_t count = 0; 1486f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org 1496f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org UChar text[50]; 1506f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org 1516f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org /* Note: the adjacent "" are concatenating strings, not adding a \" to the 1526f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org string, which is probably what whoever wrote this intended. Don't fix, 1536f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org because it would throw off the hard coded break positions in the following 1546f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org tests. */ 1556f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org u_uastrcpy(text, "He's from Africa. ""Mr. Livingston, I presume?"" Yeah"); 1566f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org 1576f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org 1586f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org/*test ubrk_open()*/ 1596f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org log_verbose("\nTesting BreakIterator open functions\n"); 1606f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org 1616f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org /* Use french for fun */ 1626f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org word = ubrk_open(UBRK_WORD, "en_US", text, u_strlen(text), &status); 1636f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org if(status == U_FILE_ACCESS_ERROR) { 1646f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org log_data_err("Check your data - it doesn't seem to be around\n"); 1656f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org return; 1666f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org } else if(U_FAILURE(status)){ 1676f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org log_err_status(status, "FAIL: Error in ubrk_open() for word breakiterator: %s\n", myErrorName(status)); 1686f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org } 1696f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org else{ 1706f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org log_verbose("PASS: Successfully opened word breakiterator\n"); 1716f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org } 1726f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org 1736f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org sentence = ubrk_open(UBRK_SENTENCE, "en_US", text, u_strlen(text), &status); 1746f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org if(U_FAILURE(status)){ 1756f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org log_err_status(status, "FAIL: Error in ubrk_open() for sentence breakiterator: %s\n", myErrorName(status)); 1766f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org return; 1776f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org } 1786f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org else{ 1796f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org log_verbose("PASS: Successfully opened sentence breakiterator\n"); 1806f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org } 1816f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org 1826f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org line = ubrk_open(UBRK_LINE, "en_US", text, u_strlen(text), &status); 1836f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org if(U_FAILURE(status)){ 1846f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org log_err("FAIL: Error in ubrk_open() for line breakiterator: %s\n", myErrorName(status)); 1856f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org return; 1866f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org } 1876f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org else{ 1886f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org log_verbose("PASS: Successfully opened line breakiterator\n"); 1896f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org } 1906f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org 1916f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org character = ubrk_open(UBRK_CHARACTER, "en_US", text, u_strlen(text), &status); 1926f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org if(U_FAILURE(status)){ 1936f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org log_err("FAIL: Error in ubrk_open() for character breakiterator: %s\n", myErrorName(status)); 1946f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org return; 1956f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org } 1966f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org else{ 1976f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org log_verbose("PASS: Successfully opened character breakiterator\n"); 1986f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org } 1996f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org /*trying to open an illegal iterator*/ 2006f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org bogus = ubrk_open((UBreakIteratorType)5, "en_US", text, u_strlen(text), &status); 2016f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org if(bogus != NULL) { 2026f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org log_err("FAIL: expected NULL from opening an invalid break iterator.\n"); 2036f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org } 2046f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org if(U_SUCCESS(status)){ 2056f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org log_err("FAIL: Error in ubrk_open() for BOGUS breakiterator. Expected U_ILLEGAL_ARGUMENT_ERROR\n"); 2066f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org } 2076f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org if(U_FAILURE(status)){ 2086f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org if(status != U_ILLEGAL_ARGUMENT_ERROR){ 2096f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org log_err("FAIL: Error in ubrk_open() for BOGUS breakiterator. Expected U_ILLEGAL_ARGUMENT_ERROR\n Got %s\n", myErrorName(status)); 2106f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org } 2116f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org } 2126f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org status=U_ZERO_ERROR; 2136f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org 2146f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org 2156f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org/* ======= Test ubrk_countAvialable() and ubrk_getAvialable() */ 2166f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org 2176f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org log_verbose("\nTesting ubrk_countAvailable() and ubrk_getAvailable()\n"); 2186f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org count=ubrk_countAvailable(); 2196f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org /* use something sensible w/o hardcoding the count */ 2206f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org if(count < 0){ 2216f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org log_err("FAIL: Error in ubrk_countAvialable() returned %d\n", count); 2226f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org } 2236f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org else{ 2246f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org log_verbose("PASS: ubrk_countAvialable() successful returned %d\n", count); 2256f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org } 2266f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org for(i=0;i<count;i++) 2276f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org { 2286f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org log_verbose("%s\n", ubrk_getAvailable(i)); 2296f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org if (ubrk_getAvailable(i) == 0) 2306f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org log_err("No locale for which breakiterator is applicable\n"); 2316f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org else 2326f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org log_verbose("A locale %s for which breakiterator is applicable\n",ubrk_getAvailable(i)); 2336f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org } 2346f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org 2356f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org/*========Test ubrk_first(), ubrk_last()...... and other functions*/ 2366f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org 2376f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org log_verbose("\nTesting the functions for word\n"); 2386f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org start = ubrk_first(word); 2396f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org if(start!=0) 2406f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org log_err("error ubrk_start(word) did not return 0\n"); 2416f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org log_verbose("first (word = %d\n", (int32_t)start); 2426f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org pos=ubrk_next(word); 2436f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org if(pos!=4) 2446f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org log_err("error ubrk_next(word) did not return 4\n"); 2456f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org log_verbose("next (word = %d\n", (int32_t)pos); 2466f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org pos=ubrk_following(word, 4); 2476f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org if(pos!=5) 2486f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org log_err("error ubrl_following(word,4) did not return 6\n"); 2496f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org log_verbose("next (word = %d\n", (int32_t)pos); 2506f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org end=ubrk_last(word); 2516f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org if(end!=49) 2526f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org log_err("error ubrk_last(word) did not return 49\n"); 2536f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org log_verbose("last (word = %d\n", (int32_t)end); 2546f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org 2556f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org pos=ubrk_previous(word); 2566f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org log_verbose("%d %d\n", end, pos); 2576f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org 2586f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org pos=ubrk_previous(word); 2596f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org log_verbose("%d \n", pos); 2606f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org 2616f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org if (ubrk_isBoundary(word, 2) != FALSE) { 2626f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org log_err("error ubrk_isBoundary(word, 2) did not return FALSE\n"); 2636f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org } 2646f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org pos=ubrk_current(word); 2656f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org if (pos != 4) { 2666f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org log_err("error ubrk_current() != 4 after ubrk_isBoundary(word, 2)\n"); 2676f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org } 2686f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org if (ubrk_isBoundary(word, 4) != TRUE) { 2696f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org log_err("error ubrk_isBoundary(word, 4) did not return TRUE\n"); 2706f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org } 2716f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org 2726f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org 2736f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org 2746f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org log_verbose("\nTesting the functions for character\n"); 2756f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org ubrk_first(character); 2766f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org pos = ubrk_following(character, 5); 2776f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org if(pos!=6) 2786f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org log_err("error ubrk_following(character,5) did not return 6\n"); 2796f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org log_verbose("Following (character,5) = %d\n", (int32_t)pos); 2806f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org pos=ubrk_following(character, 18); 2816f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org if(pos!=19) 2826f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org log_err("error ubrk_following(character,18) did not return 19\n"); 2836f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org log_verbose("Followingcharacter,18) = %d\n", (int32_t)pos); 2846f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org pos=ubrk_preceding(character, 22); 2856f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org if(pos!=21) 2866f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org log_err("error ubrk_preceding(character,22) did not return 21\n"); 2876f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org log_verbose("preceding(character,22) = %d\n", (int32_t)pos); 2886f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org 2896f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org 2906f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org log_verbose("\nTesting the functions for line\n"); 2916f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org pos=ubrk_first(line); 2926f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org if(pos != 0) 2936f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org log_err("error ubrk_first(line) returned %d, expected 0\n", (int32_t)pos); 2946f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org pos = ubrk_next(line); 2956f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org pos=ubrk_following(line, 18); 2966f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org if(pos!=22) 2976f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org log_err("error ubrk_following(line) did not return 22\n"); 2986f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org log_verbose("following (line) = %d\n", (int32_t)pos); 2996f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org 3006f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org 3016f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org log_verbose("\nTesting the functions for sentence\n"); 3026f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org ubrk_first(sentence); 3036f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org pos = ubrk_current(sentence); 3046f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org log_verbose("Current(sentence) = %d\n", (int32_t)pos); 3056f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org pos = ubrk_last(sentence); 3066f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org if(pos!=49) 3076f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org log_err("error ubrk_last for sentence did not return 49\n"); 3086f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org log_verbose("Last (sentence) = %d\n", (int32_t)pos); 3096f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org ubrk_first(sentence); 3106f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org to = ubrk_following( sentence, 0 ); 3116f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org if (to == 0) log_err("ubrk_following returned 0\n"); 3126f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org to = ubrk_preceding( sentence, to ); 3136f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org if (to != 0) log_err("ubrk_preceding didn't return 0\n"); 3146f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org if (ubrk_first(sentence)!=ubrk_current(sentence)) { 3156f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org log_err("error in ubrk_first() or ubrk_current()\n"); 3166f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org } 3176f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org 3186f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org 3196f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org /*---- */ 3206f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org /*Testing ubrk_open and ubrk_close()*/ 3216f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org log_verbose("\nTesting open and close for us locale\n"); 3226f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org b = ubrk_open(UBRK_WORD, "fr_FR", text, u_strlen(text), &status); 3236f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org if (U_FAILURE(status)) { 3246f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org log_err("ubrk_open for word returned NULL: %s\n", myErrorName(status)); 3256f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org } 3266f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org ubrk_close(b); 3276f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org 3286f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org /* Test setText and setUText */ 3296f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org { 3306f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org UChar s1[] = {0x41, 0x42, 0x20, 0}; 3316f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org UChar s2[] = {0x41, 0x42, 0x43, 0x44, 0x45, 0}; 3326f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org UText *ut = NULL; 3336f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org UBreakIterator *bb; 3346f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org int j; 3356f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org 3366f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org log_verbose("\nTesting ubrk_setText() and ubrk_setUText()\n"); 3376f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org status = U_ZERO_ERROR; 3386f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org bb = ubrk_open(UBRK_WORD, "en_US", NULL, 0, &status); 3396f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org TEST_ASSERT_SUCCESS(status); 3406f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org ubrk_setText(bb, s1, -1, &status); 3416f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org TEST_ASSERT_SUCCESS(status); 3426f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org ubrk_first(bb); 3436f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org j = ubrk_next(bb); 3446f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org TEST_ASSERT(j == 2); 3456f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org ut = utext_openUChars(ut, s2, -1, &status); 3466f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org ubrk_setUText(bb, ut, &status); 3476f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org TEST_ASSERT_SUCCESS(status); 3486f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org j = ubrk_next(bb); 3496f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org TEST_ASSERT(j == 5); 3506f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org 3516f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org ubrk_close(bb); 3526f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org utext_close(ut); 3536f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org } 3546f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org 3556f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org ubrk_close(word); 3566f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org ubrk_close(sentence); 3576f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org ubrk_close(line); 3586f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org ubrk_close(character); 3596f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org} 3606f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org 3616f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.orgstatic void TestBreakIteratorSafeClone(void) 3626f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org{ 3636f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org UChar text[51]; /* Keep this odd to test for 64-bit memory alignment */ 3646f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org /* NOTE: This doesn't reliably force mis-alignment of following items. */ 3656f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org uint8_t buffer [CLONETEST_ITERATOR_COUNT] [U_BRK_SAFECLONE_BUFFERSIZE]; 3666f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org int32_t bufferSize = U_BRK_SAFECLONE_BUFFERSIZE; 3676f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org 3686f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org UBreakIterator * someIterators [CLONETEST_ITERATOR_COUNT]; 3696f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org UBreakIterator * someClonedIterators [CLONETEST_ITERATOR_COUNT]; 3706f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org 3716f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org UBreakIterator * brk; 3726f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org UErrorCode status = U_ZERO_ERROR; 3736f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org int32_t start,pos; 3746f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org int32_t i; 3756f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org 3766f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org /*Testing ubrk_safeClone */ 3776f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org 3786f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org /* Note: the adjacent "" are concatenating strings, not adding a \" to the 3796f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org string, which is probably what whoever wrote this intended. Don't fix, 3806f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org because it would throw off the hard coded break positions in the following 3816f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org tests. */ 3826f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org u_uastrcpy(text, "He's from Africa. ""Mr. Livingston, I presume?"" Yeah"); 3836f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org 3846f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org /* US & Thai - rule-based & dictionary based */ 3856f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org someIterators[0] = ubrk_open(UBRK_WORD, "en_US", text, u_strlen(text), &status); 3866f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org if(!someIterators[0] || U_FAILURE(status)) { 3876f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org log_data_err("Couldn't open en_US word break iterator - %s\n", u_errorName(status)); 3886f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org return; 3896f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org } 3906f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org 3916f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org someIterators[1] = ubrk_open(UBRK_WORD, "th_TH", text, u_strlen(text), &status); 3926f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org if(!someIterators[1] || U_FAILURE(status)) { 3936f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org log_data_err("Couldn't open th_TH word break iterator - %s\n", u_errorName(status)); 3946f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org return; 3956f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org } 3966f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org 3976f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org /* test each type of iterator */ 3986f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org for (i = 0; i < CLONETEST_ITERATOR_COUNT; i++) 3996f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org { 4006f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org 4016f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org /* Check the various error & informational states */ 4026f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org 4036f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org /* Null status - just returns NULL */ 4046f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org if (NULL != ubrk_safeClone(someIterators[i], buffer[i], &bufferSize, NULL)) 4056f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org { 4066f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org log_err("FAIL: Cloned Iterator failed to deal correctly with null status\n"); 4076f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org } 4086f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org /* error status - should return 0 & keep error the same */ 4096f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org status = U_MEMORY_ALLOCATION_ERROR; 4106f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org if (NULL != ubrk_safeClone(someIterators[i], buffer[i], &bufferSize, &status) || status != U_MEMORY_ALLOCATION_ERROR) 4116f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org { 4126f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org log_err("FAIL: Cloned Iterator failed to deal correctly with incoming error status\n"); 4136f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org } 4146f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org status = U_ZERO_ERROR; 4156f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org 4166f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org /* Null buffer size pointer is ok */ 4176f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org if (NULL == (brk = ubrk_safeClone(someIterators[i], buffer[i], NULL, &status)) || U_FAILURE(status)) 4186f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org { 4196f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org log_err("FAIL: Cloned Iterator failed to deal correctly with null bufferSize pointer\n"); 4206f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org } 4216f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org ubrk_close(brk); 4226f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org status = U_ZERO_ERROR; 4236f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org 4246f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org /* buffer size pointer is 0 - fill in pbufferSize with a size */ 4256f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org bufferSize = 0; 4266f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org if (NULL != ubrk_safeClone(someIterators[i], buffer[i], &bufferSize, &status) || 4276f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org U_FAILURE(status) || bufferSize <= 0) 4286f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org { 4296f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org log_err("FAIL: Cloned Iterator failed a sizing request ('preflighting')\n"); 4306f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org } 4316f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org /* Verify our define is large enough */ 4326f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org if (U_BRK_SAFECLONE_BUFFERSIZE < bufferSize) 4336f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org { 4346f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org log_err("FAIL: Pre-calculated buffer size is too small - %d but needed %d\n", U_BRK_SAFECLONE_BUFFERSIZE, bufferSize); 4356f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org } 4366f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org /* Verify we can use this run-time calculated size */ 4376f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org if (NULL == (brk = ubrk_safeClone(someIterators[i], buffer[i], &bufferSize, &status)) || U_FAILURE(status)) 4386f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org { 4396f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org log_err("FAIL: Iterator can't be cloned with run-time size\n"); 4406f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org } 4416f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org if (brk) 4426f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org ubrk_close(brk); 4436f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org /* size one byte too small - should allocate & let us know */ 4446f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org if (bufferSize > 1) { 4456f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org --bufferSize; 4466f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org } 4476f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org if (NULL == (brk = ubrk_safeClone(someIterators[i], NULL, &bufferSize, &status)) || status != U_SAFECLONE_ALLOCATED_WARNING) 4486f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org { 4496f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org log_err("FAIL: Cloned Iterator failed to deal correctly with too-small buffer size\n"); 4506f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org } 4516f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org if (brk) 4526f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org ubrk_close(brk); 4536f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org status = U_ZERO_ERROR; 4546f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org bufferSize = U_BRK_SAFECLONE_BUFFERSIZE; 4556f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org 4566f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org /* Null buffer pointer - return Iterator & set error to U_SAFECLONE_ALLOCATED_ERROR */ 4576f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org if (NULL == (brk = ubrk_safeClone(someIterators[i], NULL, &bufferSize, &status)) || status != U_SAFECLONE_ALLOCATED_WARNING) 4586f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org { 4596f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org log_err("FAIL: Cloned Iterator failed to deal correctly with null buffer pointer\n"); 4606f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org } 4616f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org if (brk) 4626f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org ubrk_close(brk); 4636f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org status = U_ZERO_ERROR; 4646f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org 4656f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org /* Mis-aligned buffer pointer. */ 4666f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org { 4676f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org char stackBuf[U_BRK_SAFECLONE_BUFFERSIZE+sizeof(void *)]; 4686f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org 4696f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org brk = ubrk_safeClone(someIterators[i], &stackBuf[1], &bufferSize, &status); 4706f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org if (U_FAILURE(status) || brk == NULL) { 4716f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org log_err("FAIL: Cloned Iterator failed with misaligned buffer pointer\n"); 4726f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org } 4736f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org if (status == U_SAFECLONE_ALLOCATED_WARNING) { 4746f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org log_verbose("Cloned Iterator allocated when using a mis-aligned buffer.\n"); 4756f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org } 4766f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org if (brk) 4776f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org ubrk_close(brk); 4786f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org } 4796f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org 4806f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org 4816f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org /* Null Iterator - return NULL & set U_ILLEGAL_ARGUMENT_ERROR */ 4826f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org if (NULL != ubrk_safeClone(NULL, buffer[i], &bufferSize, &status) || status != U_ILLEGAL_ARGUMENT_ERROR) 4836f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org { 4846f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org log_err("FAIL: Cloned Iterator failed to deal correctly with null Iterator pointer\n"); 4856f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org } 4866f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org status = U_ZERO_ERROR; 4876f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org 4886f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org /* Do these cloned Iterators work at all - make a first & next call */ 4896f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org bufferSize = U_BRK_SAFECLONE_BUFFERSIZE; 4906f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org someClonedIterators[i] = ubrk_safeClone(someIterators[i], buffer[i], &bufferSize, &status); 4916f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org 4926f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org start = ubrk_first(someClonedIterators[i]); 4936f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org if(start!=0) 4946f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org log_err("error ubrk_start(clone) did not return 0\n"); 4956f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org pos=ubrk_next(someClonedIterators[i]); 4966f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org if(pos!=4) 4976f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org log_err("error ubrk_next(clone) did not return 4\n"); 4986f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org 4996f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org ubrk_close(someClonedIterators[i]); 5006f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org ubrk_close(someIterators[i]); 5016f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org } 5026f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org} 5036f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org#endif 5046f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org 5056f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org 5066f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org/* 5076f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org// Open a break iterator from char * rules. Take care of conversion 5086f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org// of the rules and error checking. 5096f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org*/ 5106f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.orgstatic UBreakIterator * testOpenRules(char *rules) { 5116f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org UErrorCode status = U_ZERO_ERROR; 5126f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org UChar *ruleSourceU = NULL; 5136f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org void *strCleanUp = NULL; 5146f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org UParseError parseErr; 5156f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org UBreakIterator *bi; 5166f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org 5176f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org ruleSourceU = toUChar(rules, &strCleanUp); 5186f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org 5196f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org bi = ubrk_openRules(ruleSourceU, -1, /* The rules */ 5206f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org NULL, -1, /* The text to be iterated over. */ 5216f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org &parseErr, &status); 5226f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org 5236f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org if (U_FAILURE(status)) { 5246f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org log_data_err("FAIL: ubrk_openRules: ICU Error \"%s\" (Are you missing data?)\n", u_errorName(status)); 5256f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org bi = 0; 5266f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org }; 5276f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org freeToUCharStrings(&strCleanUp); 5286f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org return bi; 5296f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org 5306f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org} 5316f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org 5326f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org/* 5336f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org * TestBreakIteratorRules - Verify that a break iterator can be created from 5346f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org * a set of source rules. 5356f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org */ 5366f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.orgstatic void TestBreakIteratorRules() { 5376f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org /* Rules will keep together any run of letters not including 'a', OR 5386f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org * keep together 'abc', but only when followed by 'def', OTHERWISE 5396f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org * just return one char at a time. 5406f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org */ 5416f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org char rules[] = "abc{666}/def;\n [\\p{L} - [a]]* {2}; . {1};"; 5426f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org /* 0123456789012345678 */ 5436f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org char data[] = "abcdex abcdefgh-def"; /* the test data string */ 5446f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org char breaks[] = "** ** * ** *"; /* * the expected break positions */ 5456f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org char tags[] = "01 21 6 21 2"; /* expected tag values at break positions */ 5466f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org int32_t tagMap[] = {0, 1, 2, 3, 4, 5, 666}; 5476f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org 5486f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org UChar *uData; 5496f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org void *freeHook = NULL; 5506f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org UErrorCode status = U_ZERO_ERROR; 5516f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org int32_t pos; 5526f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org int i; 5536f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org 5546f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org UBreakIterator *bi = testOpenRules(rules); 5556f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org if (bi == NULL) {return;} 5566f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org uData = toUChar(data, &freeHook); 5576f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org ubrk_setText(bi, uData, -1, &status); 5586f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org 5596f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org pos = ubrk_first(bi); 5606f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org for (i=0; i<sizeof(breaks); i++) { 5616f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org if (pos == i && breaks[i] != '*') { 5626f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org log_err("FAIL: unexpected break at position %d found\n", pos); 5636f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org break; 5646f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org } 5656f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org if (pos != i && breaks[i] == '*') { 5666f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org log_err("FAIL: expected break at position %d not found.\n", i); 5676f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org break; 5686f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org } 5696f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org if (pos == i) { 5706f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org int32_t tag, expectedTag; 5716f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org tag = ubrk_getRuleStatus(bi); 5726f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org expectedTag = tagMap[tags[i]&0xf]; 5736f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org if (tag != expectedTag) { 5746f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org log_err("FAIL: incorrect tag value. Position = %d; expected tag %d, got %d", 5756f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org pos, expectedTag, tag); 5766f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org break; 5776f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org } 5786f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org pos = ubrk_next(bi); 5796f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org } 5806f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org } 5816f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org 5826f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org freeToUCharStrings(&freeHook); 5836f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org ubrk_close(bi); 5846f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org} 5856f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org 5866f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.orgstatic void TestBreakIteratorRuleError() { 5876f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org/* 5886f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org * TestBreakIteratorRuleError - Try to create a BI from rules with syntax errors, 5896f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org * check that the error is reported correctly. 5906f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org */ 5916f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org char rules[] = " # This is a rule comment on line 1\n" 5926f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org "[:L:]; # this rule is OK.\n" 5936f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org "abcdefg); # Error, mismatched parens\n"; 5946f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org UChar *uRules; 5956f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org void *freeHook = NULL; 5966f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org UErrorCode status = U_ZERO_ERROR; 5976f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org UParseError parseErr; 5986f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org UBreakIterator *bi; 5996f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org 6006f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org uRules = toUChar(rules, &freeHook); 6016f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org bi = ubrk_openRules(uRules, -1, /* The rules */ 6026f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org NULL, -1, /* The text to be iterated over. */ 6036f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org &parseErr, &status); 6046f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org if (U_SUCCESS(status)) { 6056f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org log_err("FAIL: construction of break iterator succeeded when it should have failed.\n"); 6066f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org ubrk_close(bi); 6076f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org } else { 6086f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org if (parseErr.line != 3 || parseErr.offset != 8) { 6096f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org log_data_err("FAIL: incorrect error position reported. Got line %d, char %d, expected line 3, char 7 (Are you missing data?)\n", 6106f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org parseErr.line, parseErr.offset); 6116f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org } 6126f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org } 6136f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org freeToUCharStrings(&freeHook); 6146f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org} 6156f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org 6166f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org 6176f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org/* 6186f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org* TestsBreakIteratorStatusVals() Test the ubrk_getRuleStatusVec() funciton 6196f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org*/ 6206f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.orgstatic void TestBreakIteratorStatusVec() { 6216f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org #define RULE_STRING_LENGTH 200 6226f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org UChar rules[RULE_STRING_LENGTH]; 6236f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org 6246f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org #define TEST_STRING_LENGTH 25 6256f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org UChar testString[TEST_STRING_LENGTH]; 6266f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org UBreakIterator *bi = NULL; 6276f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org int32_t pos = 0; 6286f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org int32_t vals[10]; 6296f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org int32_t numVals; 6306f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org UErrorCode status = U_ZERO_ERROR; 6316f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org 6326f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org u_uastrncpy(rules, "[A-N]{100}; \n" 6336f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org "[a-w]{200}; \n" 6346f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org "[\\p{L}]{300}; \n" 6356f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org "[\\p{N}]{400}; \n" 6366f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org "[0-5]{500}; \n" 6376f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org "!.*;\n", RULE_STRING_LENGTH); 6386f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org u_uastrncpy(testString, "ABC", TEST_STRING_LENGTH); 6396f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org 6406f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org 6416f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org bi = ubrk_openRules(rules, -1, testString, -1, NULL, &status); 6426f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org TEST_ASSERT_SUCCESS(status); 6436f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org TEST_ASSERT(bi != NULL); 6446f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org 6456f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org /* The TEST_ASSERT above should change too... */ 6466f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org if (bi != NULL) { 6476f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org pos = ubrk_next(bi); 6486f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org TEST_ASSERT(pos == 1); 6496f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org 6506f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org memset(vals, -1, sizeof(vals)); 6516f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org numVals = ubrk_getRuleStatusVec(bi, vals, 10, &status); 6526f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org TEST_ASSERT_SUCCESS(status); 6536f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org TEST_ASSERT(numVals == 2); 6546f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org TEST_ASSERT(vals[0] == 100); 6556f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org TEST_ASSERT(vals[1] == 300); 6566f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org TEST_ASSERT(vals[2] == -1); 6576f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org 6586f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org numVals = ubrk_getRuleStatusVec(bi, vals, 0, &status); 6596f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org TEST_ASSERT(status == U_BUFFER_OVERFLOW_ERROR); 6606f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org TEST_ASSERT(numVals == 2); 6616f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org } 6626f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org 6636f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org ubrk_close(bi); 6646f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org} 6656f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org 6666f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org 6676f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org/* 6686f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org * static void TestBreakIteratorUText(void); 6696f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org * 6706f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org * Test that ubrk_setUText() is present and works for a simple case. 6716f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org */ 6726f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.orgstatic void TestBreakIteratorUText(void) { 6736f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org const char *UTF8Str = "\x41\xc3\x85\x5A\x20\x41\x52\x69\x6E\x67"; /* c3 85 is utf-8 for A with a ring on top */ 6746f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org /* 0 1 2 34567890 */ 6756f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org 6766f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org UErrorCode status = U_ZERO_ERROR; 6776f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org UBreakIterator *bi = NULL; 6786f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org int32_t pos = 0; 6796f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org 6806f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org 6816f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org UText *ut = utext_openUTF8(NULL, UTF8Str, -1, &status); 6826f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org TEST_ASSERT_SUCCESS(status); 6836f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org 6846f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org bi = ubrk_open(UBRK_WORD, "en_US", NULL, 0, &status); 6856f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org if (U_FAILURE(status)) { 6866f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org log_err_status(status, "Failure at file %s, line %d, error = %s\n", __FILE__, __LINE__, u_errorName(status)); 6876f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org return; 6886f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org } 6896f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org 6906f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org ubrk_setUText(bi, ut, &status); 6916f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org if (U_FAILURE(status)) { 6926f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org log_err("Failure at file %s, line %d, error = %s\n", __FILE__, __LINE__, u_errorName(status)); 6936f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org return; 6946f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org } 6956f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org 6966f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org pos = ubrk_first(bi); 6976f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org TEST_ASSERT(pos == 0); 6986f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org 6996f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org pos = ubrk_next(bi); 7006f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org TEST_ASSERT(pos == 4); 7016f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org 7026f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org pos = ubrk_next(bi); 7036f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org TEST_ASSERT(pos == 5); 7046f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org 7056f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org pos = ubrk_next(bi); 7066f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org TEST_ASSERT(pos == 10); 7076f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org 7086f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org pos = ubrk_next(bi); 7096f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org TEST_ASSERT(pos == UBRK_DONE); 7106f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org ubrk_close(bi); 7116f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org utext_close(ut); 7126f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org} 7136f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org 7146f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org/* 7156f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org * static void TestBreakIteratorTailoring(void); 7166f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org * 7176f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org * Test break iterator tailorings from CLDR data. 7186f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org */ 7196f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org 7206f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org/* Thai/Lao grapheme break tailoring */ 7216f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.orgstatic const UChar thTest[] = { 0x0020, 0x0E40, 0x0E01, 0x0020, 7226f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org 0x0E01, 0x0E30, 0x0020, 0x0E01, 0x0E33, 0x0020, 0 }; 7236f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org/*in Unicode 6.1 en should behave just like th for this*/ 7246f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org/*static const int32_t thTestOffs_enFwd[] = { 1, 3, 4, 6, 7, 9, 10 };*/ 7256f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.orgstatic const int32_t thTestOffs_thFwd[] = { 1, 2, 3, 4, 5, 6, 7, 9, 10 }; 7266f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org/*static const int32_t thTestOffs_enRev[] = { 9, 7, 6, 4, 3, 1, 0 };*/ 7276f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.orgstatic const int32_t thTestOffs_thRev[] = { 9, 7, 6, 5, 4, 3, 2, 1, 0 }; 7286f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org 7296f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org/* Hebrew line break tailoring, for cldrbug 3028 */ 7306f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.orgstatic const UChar heTest[] = { 0x0020, 0x002D, 0x0031, 0x0032, 0x0020, 7316f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org 0x0061, 0x002D, 0x006B, 0x0020, 7326f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org 0x0061, 0x0300, 0x2010, 0x006B, 0x0020, 7336f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org 0x05DE, 0x05D4, 0x002D, 0x0069, 0x0020, 7346f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org 0x05D1, 0x05BC, 0x2010, 0x0047, 0x0020, 0 }; 7356f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org/*in Unicode 6.1 en should behave just like he for this*/ 7366f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org/*static const int32_t heTestOffs_enFwd[] = { 1, 5, 7, 9, 12, 14, 17, 19, 22, 24 };*/ 7376f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.orgstatic const int32_t heTestOffs_heFwd[] = { 1, 5, 7, 9, 12, 14, 19, 24 }; 7386f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org/*static const int32_t heTestOffs_enRev[] = { 22, 19, 17, 14, 12, 9, 7, 5, 1, 0 };*/ 7396f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.orgstatic const int32_t heTestOffs_heRev[] = { 19, 14, 12, 9, 7, 5, 1, 0 }; 7406f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org 7416f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org/* Finnish line break tailoring, for cldrbug 3029 */ 7426f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.orgstatic const UChar fiTest[] = { /* 00 */ 0x0020, 0x002D, 0x0031, 0x0032, 0x0020, 7436f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org /* 05 */ 0x0061, 0x002D, 0x006B, 0x0020, 7446f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org /* 09 */ 0x0061, 0x0300, 0x2010, 0x006B, 0x0020, 7456f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org /* 14 */ 0x0061, 0x0020, 0x002D, 0x006B, 0x0020, 7466f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org /* 19 */ 0x0061, 0x0300, 0x0020, 0x2010, 0x006B, 0x0020, 0 }; 7476f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.orgstatic const int32_t fiTestOffs_enFwd[] = { 1, 5, 7, 9, 12, 14, 16, 17, 19, 22, 23, 25 }; 7486f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.orgstatic const int32_t fiTestOffs_fiFwd[] = { 1, 5, 7, 9, 12, 14, 16, 19, 22, 25 }; 7496f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.orgstatic const int32_t fiTestOffs_enRev[] = { 23, 22, 19, 17, 16, 14, 12, 9, 7, 5, 1, 0 }; 7506f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.orgstatic const int32_t fiTestOffs_fiRev[] = { 22, 19, 16, 14, 12, 9, 7, 5, 1, 0 }; 7516f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org 7526f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org/* Khmer dictionary-based work break, for ICU ticket #8329 */ 7536f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.orgstatic const UChar kmTest[] = { /* 00 */ 0x179F, 0x17BC, 0x1798, 0x1785, 0x17C6, 0x178E, 0x17B6, 0x1799, 0x1796, 0x17C1, 7546f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org /* 10 */ 0x179B, 0x1794, 0x1793, 0x17D2, 0x178F, 0x17B7, 0x1785, 0x178A, 0x17BE, 0x1798, 7556f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org /* 20 */ 0x17D2, 0x1794, 0x17B8, 0x17A2, 0x1792, 0x17B7, 0x179F, 0x17D2, 0x178B, 0x17B6, 7566f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org /* 30 */ 0x1793, 0x17A2, 0x179A, 0x1796, 0x17D2, 0x179A, 0x17C7, 0x1782, 0x17BB, 0x178E, 7576f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org /* 40 */ 0x178A, 0x179B, 0x17CB, 0x1796, 0x17D2, 0x179A, 0x17C7, 0x17A2, 0x1784, 0x17D2, 7586f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org /* 50 */ 0x1782, 0 }; 7596f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.orgstatic const int32_t kmTestOffs_kmFwd[] = { 3, /*8,*/ 11, 17, 23, 31, /*33,*/ 40, 43, 51 }; /* TODO: Investigate failure to break at offset 8 */ 7606f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.orgstatic const int32_t kmTestOffs_kmRev[] = { 43, 40, /*33,*/ 31, 23, 17, 11, /*8,*/ 3, 0 }; 7616f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org 7626f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.orgtypedef struct { 7636f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org const char * locale; 7646f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org UBreakIteratorType type; 7656f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org const UChar * test; 7666f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org const int32_t * offsFwd; 7676f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org const int32_t * offsRev; 7686f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org int32_t numOffsets; 7696f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org} RBBITailoringTest; 7706f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org 7716f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.orgstatic const RBBITailoringTest tailoringTests[] = { 7726f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org { "en", UBRK_CHARACTER, thTest, thTestOffs_thFwd, thTestOffs_thRev, sizeof(thTestOffs_thFwd)/sizeof(thTestOffs_thFwd[0]) }, 7736f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org { "en_US_POSIX", UBRK_CHARACTER, thTest, thTestOffs_thFwd, thTestOffs_thRev, sizeof(thTestOffs_thFwd)/sizeof(thTestOffs_thFwd[0]) }, 7746f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org { "en", UBRK_LINE, heTest, heTestOffs_heFwd, heTestOffs_heRev, sizeof(heTestOffs_heFwd)/sizeof(heTestOffs_heFwd[0]) }, 7756f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org { "he", UBRK_LINE, heTest, heTestOffs_heFwd, heTestOffs_heRev, sizeof(heTestOffs_heFwd)/sizeof(heTestOffs_heFwd[0]) }, 7766f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org { "en", UBRK_LINE, fiTest, fiTestOffs_enFwd, fiTestOffs_enRev, sizeof(fiTestOffs_enFwd)/sizeof(fiTestOffs_enFwd[0]) }, 7776f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org { "fi", UBRK_LINE, fiTest, fiTestOffs_fiFwd, fiTestOffs_fiRev, sizeof(fiTestOffs_fiFwd)/sizeof(fiTestOffs_fiFwd[0]) }, 7786f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org { "km", UBRK_WORD, kmTest, kmTestOffs_kmFwd, kmTestOffs_kmRev, sizeof(kmTestOffs_kmFwd)/sizeof(kmTestOffs_kmFwd[0]) }, 7796f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org { NULL, 0, NULL, NULL, NULL, 0 }, 7806f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org}; 7816f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org 7826f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.orgstatic void TestBreakIteratorTailoring(void) { 7836f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org const RBBITailoringTest * testPtr; 7846f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org for (testPtr = tailoringTests; testPtr->locale != NULL; ++testPtr) { 7856f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org UErrorCode status = U_ZERO_ERROR; 7866f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org UBreakIterator* ubrkiter = ubrk_open(testPtr->type, testPtr->locale, testPtr->test, -1, &status); 7876f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org if ( U_SUCCESS(status) ) { 7886f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org int32_t offset, offsindx; 7896f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org UBool foundError; 7906f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org 7916f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org foundError = FALSE; 7926f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org for (offsindx = 0; (offset = ubrk_next(ubrkiter)) != UBRK_DONE; ++offsindx) { 7936f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org if (!foundError && offsindx >= testPtr->numOffsets) { 7946f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org log_err("FAIL: locale %s, break type %d, ubrk_next expected UBRK_DONE, got %d\n", 7956f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org testPtr->locale, testPtr->type, offset); 7966f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org foundError = TRUE; 7976f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org } else if (!foundError && offset != testPtr->offsFwd[offsindx]) { 7986f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org log_err("FAIL: locale %s, break type %d, ubrk_next expected %d, got %d\n", 7996f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org testPtr->locale, testPtr->type, testPtr->offsFwd[offsindx], offset); 8006f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org foundError = TRUE; 8016f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org } 8026f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org } 8036f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org if (!foundError && offsindx < testPtr->numOffsets) { 8046f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org log_err("FAIL: locale %s, break type %d, ubrk_next expected %d, got UBRK_DONE\n", 8056f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org testPtr->locale, testPtr->type, testPtr->offsFwd[offsindx]); 8066f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org } 8076f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org 8086f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org foundError = FALSE; 8096f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org for (offsindx = 0; (offset = ubrk_previous(ubrkiter)) != UBRK_DONE; ++offsindx) { 8106f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org if (!foundError && offsindx >= testPtr->numOffsets) { 8116f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org log_err("FAIL: locale %s, break type %d, ubrk_previous expected UBRK_DONE, got %d\n", 8126f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org testPtr->locale, testPtr->type, offset); 8136f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org foundError = TRUE; 8146f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org } else if (!foundError && offset != testPtr->offsRev[offsindx]) { 8156f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org log_err("FAIL: locale %s, break type %d, ubrk_previous expected %d, got %d\n", 8166f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org testPtr->locale, testPtr->type, testPtr->offsRev[offsindx], offset); 8176f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org foundError = TRUE; 8186f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org } 8196f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org } 8206f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org if (!foundError && offsindx < testPtr->numOffsets) { 8216f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org log_err("FAIL: locale %s, break type %d, ubrk_previous expected %d, got UBRK_DONE\n", 8226f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org testPtr->locale, testPtr->type, testPtr->offsRev[offsindx]); 8236f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org } 8246f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org 8256f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org ubrk_close(ubrkiter); 8266f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org } else { 8276f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org log_err_status(status, "FAIL: locale %s, break type %d, ubrk_open status: %s\n", testPtr->locale, testPtr->type, u_errorName(status)); 8286f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org } 8296f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org } 8306f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org} 8316f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org 8326f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org 8336f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.orgstatic void TestBreakIteratorRefresh(void) { 8346f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org /* 8356f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org * RefreshInput changes out the input of a Break Iterator without 8366f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org * changing anything else in the iterator's state. Used with Java JNI, 8376f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org * when Java moves the underlying string storage. This test 8386f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org * runs a ubrk_next() repeatedly, moving the text in the middle of the sequence. 8396f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org * The right set of boundaries should still be found. 8406f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org */ 8416f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org UChar testStr[] = {0x20, 0x41, 0x20, 0x42, 0x20, 0x43, 0x20, 0x44, 0x0}; /* = " A B C D" */ 8426f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org UChar movedStr[] = {0x20, 0x20, 0x20, 0x20, 0x20, 0x20, 0x20, 0x20, 0}; 8436f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org UErrorCode status = U_ZERO_ERROR; 8446f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org UBreakIterator *bi; 8456f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org UText ut1 = UTEXT_INITIALIZER; 8466f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org UText ut2 = UTEXT_INITIALIZER; 8476f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org 8486f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org bi = ubrk_open(UBRK_LINE, "en_US", NULL, 0, &status); 8496f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org TEST_ASSERT_SUCCESS(status); 8506f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org 8516f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org utext_openUChars(&ut1, testStr, -1, &status); 8526f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org TEST_ASSERT_SUCCESS(status); 8536f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org ubrk_setUText(bi, &ut1, &status); 8546f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org TEST_ASSERT_SUCCESS(status); 8556f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org 8566f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org if (U_SUCCESS(status)) { 8576f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org /* Line boundaries will occur before each letter in the original string */ 8586f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org TEST_ASSERT(1 == ubrk_next(bi)); 8596f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org TEST_ASSERT(3 == ubrk_next(bi)); 8606f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org 8616f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org /* Move the string, kill the original string. */ 8626f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org u_strcpy(movedStr, testStr); 8636f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org u_memset(testStr, 0x20, u_strlen(testStr)); 8646f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org utext_openUChars(&ut2, movedStr, -1, &status); 8656f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org TEST_ASSERT_SUCCESS(status); 8666f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org ubrk_refreshUText(bi, &ut2, &status); 8676f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org TEST_ASSERT_SUCCESS(status); 8686f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org 8696f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org /* Find the following matches, now working in the moved string. */ 8706f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org TEST_ASSERT(5 == ubrk_next(bi)); 8716f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org TEST_ASSERT(7 == ubrk_next(bi)); 8726f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org TEST_ASSERT(8 == ubrk_next(bi)); 8736f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org TEST_ASSERT(UBRK_DONE == ubrk_next(bi)); 8746f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org TEST_ASSERT_SUCCESS(status); 8756f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org 8766f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org utext_close(&ut1); 8776f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org utext_close(&ut2); 8786f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org } 8796f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org ubrk_close(bi); 8806f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org} 8816f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org 8826f31ac30b9092fd02a8c97e5216cf53f3e4fae4jshin@chromium.org#endif /* #if !UCONFIG_NO_BREAK_ITERATION */ 883