icuSources/samples/csdet/csdet.c

   1 /*
   2  ********************************************************************************
   3  *   Copyright (C) 2005-2006, International Business Machines
   4  *   Corporation and others.  All Rights Reserved.
   5  ********************************************************************************
   6  */
   7
   8 #include "unicode/utypes.h"
   9 #include "unicode/ucsdet.h"
  10
  11 #include <string.h>
  12 #include <stdio.h>
  13
  14 #define BUFFER_SIZE 8192
  15
  16 int main(int argc, char *argv[])
  17 {
  18     static char buffer[BUFFER_SIZE];
  19     int32_t arg;
  20
  21     if( argc <= 1 ) {
  22         printf("Usage: %s [filename]...\n", argv[0]);
  23         return -1;
  24     }
  25
  26     for(arg = 1; arg < argc; arg += 1) {
  27         FILE *file;
  28         char *filename = argv[arg];
  29         int32_t inputLength, match, matchCount = 0;
  30         UCharsetDetector* csd;
  31         const UCharsetMatch **csm;
  32         UErrorCode status = U_ZERO_ERROR;
  33
  34         if (arg > 1) {
  35             printf("\n");
  36         }
  37
  38         file = fopen(filename, "rb");
  39
  40         if (file == NULL) {
  41             printf("Cannot open file \"%s\"\n\n", filename);
  42             continue;
  43         }
  44
  45         printf("%s:\n", filename);
  46
  47         inputLength = (int32_t) fread(buffer, 1, BUFFER_SIZE, file);
  48
  49         fclose(file);
  50
  51         csd = ucsdet_open(&status);
  52         ucsdet_setText(csd, buffer, inputLength, &status);
  53
  54         csm = ucsdet_detectAll(csd, &matchCount, &status);
  55
  56         for(match = 0; match < matchCount; match += 1) {
  57             const char *name = ucsdet_getName(csm[match], &status);
  58             const char *lang = ucsdet_getLanguage(csm[match], &status);
  59             int32_t confidence = ucsdet_getConfidence(csm[match], &status);
  60
  61             if (lang == NULL || strlen(lang) == 0) {
  62                 lang = "**";
  63             }
  64
  65             printf("%s (%s) %d\n", name, lang, confidence);
  66         }
  67
  68         ucsdet_close(csd);
  69     }
  70
  71     return 0;
  72 }
  73