icuSources/samples/layout/UnicodeReader.cpp

   1 /*
   2  ******************************************************************************
   3  * Copyright (C) 1998-2005, International Business Machines Corporation and   *
   4  * others. All Rights Reserved.                                               *
   5  ******************************************************************************
   6  */
   7
   8 #include <errno.h>
   9 #include <stdio.h>
  10 #include <string.h>
  11
  12 #include "unicode/utypes.h"
  13 #include "unicode/unistr.h"
  14
  15 #include "layout/LETypes.h"
  16
  17 #include "GUISupport.h"
  18 #include "UnicodeReader.h"
  19
  20 #define BYTE(b) (((int) b) & 0xFF)
  21
  22 /*
  23  * Read the text from a file. The text must start with a Unicode Byte
  24  * Order Mark (BOM) so that we know what order to read the bytes in.
  25  */
  26 const UChar *UnicodeReader::readFile(const char *fileName, GUISupport *guiSupport, int32_t &charCount)
  27 {
  28     FILE *f;
  29     int32_t fileSize;
  30
  31     UChar *charBuffer;
  32     char *byteBuffer;
  33     char startBytes[4] = {'\xA5', '\xA5', '\xA5', '\xA5'};
  34     char errorMessage[128];
  35     const char *cp = "";
  36     int32_t signatureLength = 0;
  37
  38     f = fopen(fileName, "rb");
  39
  40     if( f == NULL ) {
  41         sprintf(errorMessage,"Couldn't open %s: %s \n", fileName, strerror(errno));
  42         guiSupport->postErrorMessage(errorMessage, "Text File Error");
  43         return 0;
  44     }
  45
  46     fseek(f, 0, SEEK_END);
  47     fileSize = ftell(f);
  48
  49     fseek(f, 0, SEEK_SET);
  50     fread(startBytes, sizeof(char), 4, f);
  51
  52     if (startBytes[0] == '\xFE' && startBytes[1] == '\xFF') {
  53         cp = "UTF-16BE";
  54         signatureLength = 2;
  55     } else if (startBytes[0] == '\xFF' && startBytes[1] == '\xFE') {
  56         if (startBytes[2] == '\x00' && startBytes[3] == '\x00') {
  57             cp = "UTF-32LE";
  58             signatureLength = 4;
  59         } else {
  60             cp = "UTF-16LE";
  61             signatureLength = 2;
  62         }
  63     } else if (startBytes[0] == '\xEF' && startBytes[1] == '\xBB' && startBytes[2] == '\xBF') {
  64         cp = "UTF-8";
  65         signatureLength = 3;
  66     } else if (startBytes[0] == '\x0E' && startBytes[1] == '\xFE' && startBytes[2] == '\xFF') {
  67         cp = "SCSU";
  68         signatureLength = 3;
  69     } else if (startBytes[0] == '\x00' && startBytes[1] == '\x00' &&
  70         startBytes[2] == '\xFE' && startBytes[3] == '\xFF') {
  71         cp = "UTF-32BE";
  72         signatureLength = 4;
  73     } else {
  74         sprintf(errorMessage, "Couldn't detect the encoding of %s: (%2.2X, %2.2X, %2.2X, %2.2X)\n", fileName,
  75                     BYTE(startBytes[0]), BYTE(startBytes[1]), BYTE(startBytes[2]), BYTE(startBytes[3]));
  76         guiSupport->postErrorMessage(errorMessage, "Text File Error");
  77         fclose(f);
  78         return 0;
  79     }
  80
  81     fileSize -= signatureLength;
  82     fseek(f, signatureLength, SEEK_SET);
  83     byteBuffer = new char[fileSize];
  84
  85     if(byteBuffer == 0) {
  86         sprintf(errorMessage,"Couldn't get memory for reading %s: %s \n", fileName, strerror(errno));
  87         guiSupport->postErrorMessage(errorMessage, "Text File Error");
  88         fclose(f);
  89         return 0;
  90     }
  91
  92     fread(byteBuffer, sizeof(char), fileSize, f);
  93     if( ferror(f) ) {
  94         sprintf(errorMessage,"Couldn't read %s: %s \n", fileName, strerror(errno));
  95         guiSupport->postErrorMessage(errorMessage, "Text File Error");
  96         fclose(f);
  97         delete[] byteBuffer;
  98         return 0;
  99     }
 100     fclose(f);
 101
 102     UnicodeString myText(byteBuffer, fileSize, cp);
 103
 104     delete[] byteBuffer;
 105
 106     charCount = myText.length();
 107     charBuffer = LE_NEW_ARRAY(UChar, charCount + 1);
 108     if(charBuffer == 0) {
 109         sprintf(errorMessage,"Couldn't get memory for reading %s: %s \n", fileName, strerror(errno));
 110         guiSupport->postErrorMessage(errorMessage, "Text File Error");
 111         return 0;
 112     }
 113
 114     myText.extract(0, myText.length(), charBuffer);
 115     charBuffer[charCount] = 0;    // NULL terminate for easier reading in the debugger
 116
 117     return charBuffer;
 118 }
 119