icuSources/io/ustream.cpp

   1 /*
   2 **********************************************************************
   3 *   Copyright (C) 2001-2011, International Business Machines
   4 *   Corporation and others.  All Rights Reserved.
   5 **********************************************************************
   6 *  FILE NAME : ustream.cpp
   7 *
   8 *   Modification History:
   9 *
  10 *   Date        Name        Description
  11 *   06/25/2001  grhoten     Move iostream from unistr.h to here
  12 ******************************************************************************
  13 */
  14
  15 #include "unicode/utypes.h"
  16 #include "unicode/uobject.h"
  17 #include "unicode/ustream.h"
  18 #include "unicode/ucnv.h"
  19 #include "unicode/uchar.h"
  20 #include "unicode/utf16.h"
  21 #include "ustr_cnv.h"
  22 #include "cmemory.h"
  23 #include <string.h>
  24
  25 // console IO
  26
  27 #if U_IOSTREAM_SOURCE >= 199711
  28
  29 #define STD_NAMESPACE std::
  30
  31 #define STD_OSTREAM STD_NAMESPACE ostream
  32 #define STD_ISTREAM STD_NAMESPACE istream
  33
  34 U_NAMESPACE_BEGIN
  35
  36 U_IO_API STD_OSTREAM & U_EXPORT2
  37 operator<<(STD_OSTREAM& stream, const UnicodeString& str)
  38 {
  39     if(str.length() > 0) {
  40         char buffer[200];
  41         UConverter *converter;
  42         UErrorCode errorCode = U_ZERO_ERROR;
  43
  44         // use the default converter to convert chunks of text
  45         converter = u_getDefaultConverter(&errorCode);
  46         if(U_SUCCESS(errorCode)) {
  47             const UChar *us = str.getBuffer();
  48             const UChar *uLimit = us + str.length();
  49             char *s, *sLimit = buffer + (sizeof(buffer) - 1);
  50             do {
  51                 errorCode = U_ZERO_ERROR;
  52                 s = buffer;
  53                 ucnv_fromUnicode(converter, &s, sLimit, &us, uLimit, 0, FALSE, &errorCode);
  54                 *s = 0;
  55
  56                 // write this chunk
  57                 if(s > buffer) {
  58                     stream << buffer;
  59                 }
  60             } while(errorCode == U_BUFFER_OVERFLOW_ERROR);
  61             u_releaseDefaultConverter(converter);
  62         }
  63     }
  64
  65 /*    stream.flush();*/
  66     return stream;
  67 }
  68
  69 U_IO_API STD_ISTREAM & U_EXPORT2
  70 operator>>(STD_ISTREAM& stream, UnicodeString& str)
  71 {
  72     // This is like ICU status checking.
  73     if (stream.fail()) {
  74         return stream;
  75     }
  76
  77     /* ipfx should eat whitespace when ios::skipws is set */
  78     UChar uBuffer[16];
  79     char buffer[16];
  80     int32_t idx = 0;
  81     UConverter *converter;
  82     UErrorCode errorCode = U_ZERO_ERROR;
  83
  84     // use the default converter to convert chunks of text
  85     converter = u_getDefaultConverter(&errorCode);
  86     if(U_SUCCESS(errorCode)) {
  87         UChar *us = uBuffer;
  88         const UChar *uLimit = uBuffer + sizeof(uBuffer)/sizeof(*uBuffer);
  89         const char *s, *sLimit;
  90         char ch;
  91         UChar ch32;
  92         UBool initialWhitespace = TRUE;
  93         UBool continueReading = TRUE;
  94
  95         /* We need to consume one byte at a time to see what is considered whitespace. */
  96         while (continueReading) {
  97             ch = stream.get();
  98             if (stream.eof()) {
  99                 // The EOF is only set after the get() of an unavailable byte.
 100                 if (!initialWhitespace) {
 101                     stream.clear(stream.eofbit);
 102                 }
 103                 continueReading = FALSE;
 104             }
 105             sLimit = &ch + (int)continueReading;
 106             us = uBuffer;
 107             s = &ch;
 108             errorCode = U_ZERO_ERROR;
 109             /*
 110             Since we aren't guaranteed to see the state before this call,
 111             this code won't work on stateful encodings like ISO-2022 or an EBCDIC stateful encoding.
 112             We flush on the last byte to ensure that we output truncated multibyte characters.
 113             */
 114             ucnv_toUnicode(converter, &us, uLimit, &s, sLimit, 0, !continueReading, &errorCode);
 115             if(U_FAILURE(errorCode)) {
 116                 /* Something really bad happened. setstate() isn't always an available API */
 117                 stream.clear(stream.failbit);
 118                 goto STOP_READING;
 119             }
 120             /* Was the character consumed? */
 121             if (us != uBuffer) {
 122                 /* Reminder: ibm-1390 & JISX0213 can output 2 Unicode code points */
 123                 int32_t uBuffSize = us-uBuffer;
 124                 int32_t uBuffIdx = 0;
 125                 while (uBuffIdx < uBuffSize) {
 126                     U16_NEXT(uBuffer, uBuffIdx, uBuffSize, ch32);
 127                     if (u_isWhitespace(ch32)) {
 128                         if (!initialWhitespace) {
 129                             buffer[idx++] = ch;
 130                             while (idx > 0) {
 131                                 stream.putback(buffer[--idx]);
 132                             }
 133                             goto STOP_READING;
 134                         }
 135                         /* else skip intialWhitespace */
 136                     }
 137                     else {
 138                         if (initialWhitespace) {
 139                             /*
 140                             When initialWhitespace is TRUE, we haven't appended any
 141                             character yet.  This is where we truncate the string,
 142                             to avoid modifying the string before we know if we can
 143                             actually read from the stream.
 144                             */
 145                             str.truncate(0);
 146                             initialWhitespace = FALSE;
 147                         }
 148                         str.append(ch32);
 149                     }
 150                 }
 151                 idx = 0;
 152             }
 153             else {
 154                 buffer[idx++] = ch;
 155             }
 156         }
 157 STOP_READING:
 158         u_releaseDefaultConverter(converter);
 159     }
 160
 161 /*    stream.flush();*/
 162     return stream;
 163 }
 164
 165 U_NAMESPACE_END
 166
 167 #endif