icuSources/samples/break/ubreak.c

   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, International Business Machines
  11 *   Corporation and others.  All Rights Reserved.
  12 *
  13 *******************************************************************************
  14 */
  15
  16 #include <stdio.h>
  17 #include <stdlib.h>
  18 #include <unicode/ustring.h>
  19 #include <unicode/ubrk.h>
  20
  21 U_CFUNC int c_main(void);
  22
  23 void printTextRange(UChar* str, int32_t start, int32_t end)
  24 {
  25   char    charBuf[1000];
  26   UChar   savedEndChar;
  27
  28   savedEndChar = str[end];
  29   str[end] = 0;
  30   u_austrncpy(charBuf, str+start, sizeof(charBuf)-1);
  31   charBuf[sizeof(charBuf)-1]=0;
  32   printf("string[%2d..%2d] \"%s\"\n", start, end-1, charBuf);
  33   str[end] = savedEndChar;
  34 }
  35
  36
  37
  38 /* Print each element in order: */
  39 void printEachForward( UBreakIterator* boundary, UChar* str) {
  40   int32_t end;
  41   int32_t start = ubrk_first(boundary);
  42   for (end = ubrk_next(boundary); end != UBRK_DONE; start = end, end =
  43          ubrk_next(boundary)) {
  44     printTextRange(str, start, end );
  45   }
  46 }
  47
  48
  49 /* Print each element in reverse order: */
  50 void printEachBackward( UBreakIterator* boundary, UChar* str) {
  51   int32_t start;
  52   int32_t end = ubrk_last(boundary);
  53   for (start = ubrk_previous(boundary); start != UBRK_DONE;  end = start,
  54          start =ubrk_previous(boundary)) {
  55     printTextRange( str, start, end );
  56   }
  57 }
  58
  59 /* Print first element */
  60 void printFirst(UBreakIterator* boundary, UChar* str) {
  61   int32_t end;
  62   int32_t start = ubrk_first(boundary);
  63   end = ubrk_next(boundary);
  64   printTextRange( str, start, end );
  65 }
  66
  67 /* Print last element */
  68 void printLast(UBreakIterator* boundary, UChar* str) {
  69   int32_t start;
  70   int32_t end = ubrk_last(boundary);
  71   start = ubrk_previous(boundary);
  72   printTextRange(str, start, end );
  73 }
  74
  75 /* Print the element at a specified position */
  76
  77 void printAt(UBreakIterator* boundary, int32_t pos , UChar* str) {
  78   int32_t start;
  79   int32_t end = ubrk_following(boundary, pos);
  80   start = ubrk_previous(boundary);
  81   printTextRange(str, start, end );
  82 }
  83
  84 /* Creating and using text boundaries*/
  85
  86 int c_main( void ) {
  87   UBreakIterator *boundary;
  88   char           cStringToExamine[] = "Aaa bbb ccc. Ddd eee fff.";
  89   UChar          stringToExamine[sizeof(cStringToExamine)+1];
  90   UErrorCode     status = U_ZERO_ERROR;
  91
  92   printf("\n\n"
  93          "C Boundary Analysis\n"
  94          "-------------------\n\n");
  95
  96   printf("Examining: %s\n", cStringToExamine);
  97   u_uastrcpy(stringToExamine, cStringToExamine);
  98
  99   /*print each sentence in forward and reverse order*/
 100   boundary = ubrk_open(UBRK_SENTENCE, "en_us", stringToExamine,
 101                        -1, &status);
 102   if (U_FAILURE(status)) {
 103     printf("ubrk_open error: %s\n", u_errorName(status));
 104     exit(1);
 105   }
 106
 107   printf("\n----- Sentence Boundaries, forward: -----------\n");
 108   printEachForward(boundary, stringToExamine);
 109   printf("\n----- Sentence Boundaries, backward: ----------\n");
 110   printEachBackward(boundary, stringToExamine);
 111   ubrk_close(boundary);
 112
 113   /*print each word in order*/
 114   boundary = ubrk_open(UBRK_WORD, "en_us", stringToExamine,
 115                        u_strlen(stringToExamine), &status);
 116   printf("\n----- Word Boundaries, forward: -----------\n");
 117   printEachForward(boundary, stringToExamine);
 118   printf("\n----- Word Boundaries, backward: ----------\n");
 119   printEachBackward(boundary, stringToExamine);
 120   /*print first element*/
 121   printf("\n----- first: -------------\n");
 122   printFirst(boundary, stringToExamine);
 123   /*print last element*/
 124   printf("\n----- last: --------------\n");
 125   printLast(boundary, stringToExamine);
 126   /*print word at charpos 10 */
 127   printf("\n----- at pos 10: ---------\n");
 128   printAt(boundary, 10 , stringToExamine);
 129
 130   ubrk_close(boundary);
 131
 132   printf("\nEnd of C boundary analysis\n");
 133   return 0;
 134 }