icuSources/samples/break/ubreak.c

   1 /*
   2 *******************************************************************************
   3 *
   4 *   Copyright (C) 2002, International Business Machines
   5 *   Corporation and others.  All Rights Reserved.
   6 *
   7 *******************************************************************************
   8 */
   9
  10 #include <stdio.h>
  11 #include <stdlib.h>
  12 #include <unicode/ustring.h>
  13 #include <unicode/ubrk.h>
  14
  15 U_CFUNC int c_main(void);
  16
  17 void printTextRange(UChar* str, int32_t start, int32_t end)
  18 {
  19   char    charBuf[1000];
  20   UChar   savedEndChar;
  21
  22   savedEndChar = str[end];
  23   str[end] = 0;
  24   u_austrncpy(charBuf, str+start, sizeof(charBuf)-1);
  25   charBuf[sizeof(charBuf)-1]=0;
  26   printf("string[%2d..%2d] \"%s\"\n", start, end-1, charBuf);
  27   str[end] = savedEndChar;
  28 }
  29
  30
  31
  32 /* Print each element in order: */
  33 void printEachForward( UBreakIterator* boundary, UChar* str) {
  34   int32_t end;
  35   int32_t start = ubrk_first(boundary);
  36   for (end = ubrk_next(boundary); end != UBRK_DONE; start = end, end =
  37          ubrk_next(boundary)) {
  38     printTextRange(str, start, end );
  39   }
  40 }
  41
  42
  43 /* Print each element in reverse order: */
  44 void printEachBackward( UBreakIterator* boundary, UChar* str) {
  45   int32_t start;
  46   int32_t end = ubrk_last(boundary);
  47   for (start = ubrk_previous(boundary); start != UBRK_DONE;  end = start,
  48          start =ubrk_previous(boundary)) {
  49     printTextRange( str, start, end );
  50   }
  51 }
  52
  53 /* Print first element */
  54 void printFirst(UBreakIterator* boundary, UChar* str) {
  55   int32_t end;
  56   int32_t start = ubrk_first(boundary);
  57   end = ubrk_next(boundary);
  58   printTextRange( str, start, end );
  59 }
  60
  61 /* Print last element */
  62 void printLast(UBreakIterator* boundary, UChar* str) {
  63   int32_t start;
  64   int32_t end = ubrk_last(boundary);
  65   start = ubrk_previous(boundary);
  66   printTextRange(str, start, end );
  67 }
  68
  69 /* Print the element at a specified position */
  70
  71 void printAt(UBreakIterator* boundary, int32_t pos , UChar* str) {
  72   int32_t start;
  73   int32_t end = ubrk_following(boundary, pos);
  74   start = ubrk_previous(boundary);
  75   printTextRange(str, start, end );
  76 }
  77
  78 /* Creating and using text boundaries*/
  79
  80 int c_main( void ) {
  81   UBreakIterator *boundary;
  82   char           cStringToExamine[] = "Aaa bbb ccc. Ddd eee fff.";
  83   UChar          stringToExamine[sizeof(cStringToExamine)+1];
  84   UErrorCode     status = U_ZERO_ERROR;
  85
  86   printf("\n\n"
  87          "C Boundary Analysis\n"
  88          "-------------------\n\n");
  89
  90   printf("Examining: %s\n", cStringToExamine);
  91   u_uastrcpy(stringToExamine, cStringToExamine);
  92
  93   /*print each sentence in forward and reverse order*/
  94   boundary = ubrk_open(UBRK_SENTENCE, "en_us", stringToExamine,
  95                        -1, &status);
  96   if (U_FAILURE(status)) {
  97     printf("ubrk_open error: %s\n", u_errorName(status));
  98     exit(1);
  99   }
 100
 101   printf("\n----- Sentence Boundaries, forward: -----------\n");
 102   printEachForward(boundary, stringToExamine);
 103   printf("\n----- Sentence Boundaries, backward: ----------\n");
 104   printEachBackward(boundary, stringToExamine);
 105   ubrk_close(boundary);
 106
 107   /*print each word in order*/
 108   boundary = ubrk_open(UBRK_WORD, "en_us", stringToExamine,
 109                        u_strlen(stringToExamine), &status);
 110   printf("\n----- Word Boundaries, forward: -----------\n");
 111   printEachForward(boundary, stringToExamine);
 112   printf("\n----- Word Boundaries, backward: ----------\n");
 113   printEachBackward(boundary, stringToExamine);
 114   /*print first element*/
 115   printf("\n----- first: -------------\n");
 116   printFirst(boundary, stringToExamine);
 117   /*print last element*/
 118   printf("\n----- last: --------------\n");
 119   printLast(boundary, stringToExamine);
 120   /*print word at charpos 10 */
 121   printf("\n----- at pos 10: ---------\n");
 122   printAt(boundary, 10 , stringToExamine);
 123
 124   ubrk_close(boundary);
 125
 126   printf("\nEnd of C boundary analysis\n");
 127   return 0;
 128 }