icuSources/samples/csdet/csdet.c

   1 /*
   2 ********************************************************************************
   3 *   © 2016 and later: Unicode, Inc. and others.
   4 *   License & terms of use: http://www.unicode.org/copyright.html#License
   5 ********************************************************************************
   6 ********************************************************************************
   7  *   Copyright (C) 2005-2006, International Business Machines
   8  *   Corporation and others.  All Rights Reserved.
   9  *******************************************************************************
  10  */
  11
  12 #include "unicode/utypes.h"
  13 #include "unicode/ucsdet.h"
  14
  15 #include <string.h>
  16 #include <stdio.h>
  17
  18 #define BUFFER_SIZE 8192
  19
  20 int main(int argc, char *argv[])
  21 {
  22     static char buffer[BUFFER_SIZE];
  23     int32_t arg;
  24
  25     if( argc <= 1 ) {
  26         printf("Usage: %s [filename]...\n", argv[0]);
  27         return -1;
  28     }
  29
  30     for(arg = 1; arg < argc; arg += 1) {
  31         FILE *file;
  32         char *filename = argv[arg];
  33         int32_t inputLength, match, matchCount = 0;
  34         UCharsetDetector* csd;
  35         const UCharsetMatch **csm;
  36         UErrorCode status = U_ZERO_ERROR;
  37
  38         if (arg > 1) {
  39             printf("\n");
  40         }
  41
  42         file = fopen(filename, "rb");
  43
  44         if (file == NULL) {
  45             printf("Cannot open file \"%s\"\n\n", filename);
  46             continue;
  47         }
  48
  49         printf("%s:\n", filename);
  50
  51         inputLength = (int32_t) fread(buffer, 1, BUFFER_SIZE, file);
  52
  53         fclose(file);
  54
  55         csd = ucsdet_open(&status);
  56         ucsdet_setText(csd, buffer, inputLength, &status);
  57
  58         csm = ucsdet_detectAll(csd, &matchCount, &status);
  59
  60         for(match = 0; match < matchCount; match += 1) {
  61             const char *name = ucsdet_getName(csm[match], &status);
  62             const char *lang = ucsdet_getLanguage(csm[match], &status);
  63             int32_t confidence = ucsdet_getConfidence(csm[match], &status);
  64
  65             if (lang == NULL || strlen(lang) == 0) {
  66                 lang = "**";
  67             }
  68
  69             printf("%s (%s) %d\n", name, lang, confidence);
  70         }
  71
  72         ucsdet_close(csd);
  73     }
  74
  75     return 0;
  76 }
  77