icuSources/samples/break/break.cpp

   1 /*
   2 *******************************************************************************
   3 *
   4 *   © 2016 and later: Unicode, Inc. and others.
   5 *   License & terms of use: http://www.unicode.org/copyright.html#License
   6 *
   7 *******************************************************************************
   8 *******************************************************************************
   9 *
  10 *   Copyright (C) 2002-2003, International Business Machines
  11 *   Corporation and others.  All Rights Reserved.
  12 *
  13 *******************************************************************************
  14 */
  15
  16 #include <stdio.h>
  17 #include <unicode/brkiter.h>
  18 #include <stdlib.h>
  19
  20 using namespace icu;
  21
  22 U_CFUNC int c_main(void);
  23
  24 void printUnicodeString(const UnicodeString &s) {
  25     char charBuf[1000];
  26     s.extract(0, s.length(), charBuf, sizeof(charBuf)-1, 0);
  27     charBuf[sizeof(charBuf)-1] = 0;
  28     printf("%s", charBuf);
  29 }
  30
  31
  32 void printTextRange( BreakIterator& iterator,
  33                     int32_t start, int32_t end )
  34 {
  35     CharacterIterator *strIter = iterator.getText().clone();
  36     UnicodeString  s;
  37     strIter->getText(s);
  38
  39     printf(" %ld %ld\t", (long)start, (long)end);
  40     printUnicodeString(UnicodeString(s, 0, start));
  41     printf("|");
  42     printUnicodeString(UnicodeString(s, start, end-start));
  43     printf("|");
  44     printUnicodeString(UnicodeString(s, end));
  45     puts("");
  46     delete strIter;
  47 }
  48
  49
  50 /* Print each element in order: */
  51 void printEachForward( BreakIterator& boundary)
  52 {
  53     int32_t start = boundary.first();
  54     for (int32_t end = boundary.next();
  55          end != BreakIterator::DONE;
  56          start = end, end = boundary.next())
  57     {
  58         printTextRange( boundary, start, end );
  59     }
  60 }
  61
  62 /* Print each element in reverse order: */
  63 void printEachBackward( BreakIterator& boundary)
  64 {
  65     int32_t end = boundary.last();
  66     for (int32_t start = boundary.previous();
  67          start != BreakIterator::DONE;
  68          end = start, start = boundary.previous())
  69     {
  70         printTextRange( boundary, start, end );
  71     }
  72 }
  73
  74 /* Print the first element */
  75 void printFirst(BreakIterator& boundary)
  76 {
  77     int32_t start = boundary.first();
  78     int32_t end = boundary.next();
  79     printTextRange( boundary, start, end );
  80 }
  81
  82 /* Print the last element */
  83 void printLast(BreakIterator& boundary)
  84 {
  85     int32_t end = boundary.last();
  86     int32_t start = boundary.previous();
  87     printTextRange( boundary, start, end );
  88 }
  89
  90 /* Print the element at a specified position */
  91 void printAt(BreakIterator &boundary, int32_t pos )
  92 {
  93     int32_t end = boundary.following(pos);
  94     int32_t start = boundary.previous();
  95     printTextRange( boundary, start, end );
  96 }
  97
  98 /* Creating and using text boundaries */
  99 int main( void )
 100 {
 101     puts("ICU Break Iterator Sample Program\n");
 102     puts("C++ Break Iteration\n");
 103     BreakIterator* boundary;
 104     UnicodeString stringToExamine("Aaa bbb ccc. Ddd eee fff.");
 105     printf("Examining: ");
 106     printUnicodeString(stringToExamine);
 107     puts("");
 108
 109     //print each sentence in forward and reverse order
 110     UErrorCode status = U_ZERO_ERROR;
 111     boundary = BreakIterator::createSentenceInstance(
 112         Locale::getUS(), status );
 113     if (U_FAILURE(status)) {
 114         printf("failed to create sentence break iterator.  status = %s",
 115             u_errorName(status));
 116         exit(1);
 117     }
 118
 119     boundary->setText(stringToExamine);
 120     puts("\n Sentence Boundaries... ");
 121     puts("----- forward: -----------");
 122     printEachForward(*boundary);
 123     puts("----- backward: ----------");
 124     printEachBackward(*boundary);
 125     delete boundary;
 126
 127     //print each word in order
 128     printf("\n Word Boundaries... \n");
 129     boundary = BreakIterator::createWordInstance(
 130         Locale::getUS(), status);
 131     boundary->setText(stringToExamine);
 132     puts("----- forward: -----------");
 133     printEachForward(*boundary);
 134     //print first element
 135     puts("----- first: -------------");
 136     printFirst(*boundary);
 137     //print last element
 138     puts("----- last: --------------");
 139     printLast(*boundary);
 140     //print word at charpos 10
 141     puts("----- at pos 10: ---------");
 142     printAt(*boundary, 10 );
 143
 144     delete boundary;
 145
 146     puts("\nEnd C++ Break Iteration");
 147
 148     // Call the C version
 149     return c_main();
 150 }