icuSources/io/ustream.cpp

   1 /*
   2 **********************************************************************
   3 *   Copyright (C) 2001-2007, 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
  16 #include "unicode/utypes.h"
  17 #include "unicode/uobject.h"
  18 #include "unicode/ustream.h"
  19 #include "unicode/ucnv.h"
  20 #include "unicode/uchar.h"
  21 #include "ustr_cnv.h"
  22 #include <string.h>
  23
  24 // console IO
  25
  26 #if U_IOSTREAM_SOURCE >= 198506
  27
  28 #if U_IOSTREAM_SOURCE >= 199711
  29 #define STD_NAMESPACE std::
  30 #else
  31 #define STD_NAMESPACE
  32 #endif
  33
  34 #define STD_OSTREAM STD_NAMESPACE ostream
  35 #define STD_ISTREAM STD_NAMESPACE istream
  36
  37 U_NAMESPACE_BEGIN
  38
  39 U_IO_API STD_OSTREAM & U_EXPORT2
  40 operator<<(STD_OSTREAM& stream, const UnicodeString& str)
  41 {
  42     if(str.length() > 0) {
  43         char buffer[200];
  44         UConverter *converter;
  45         UErrorCode errorCode = U_ZERO_ERROR;
  46
  47         // use the default converter to convert chunks of text
  48         converter = u_getDefaultConverter(&errorCode);
  49         if(U_SUCCESS(errorCode)) {
  50             const UChar *us = str.getBuffer();
  51             const UChar *uLimit = us + str.length();
  52             char *s, *sLimit = buffer + sizeof(buffer);
  53             do {
  54                 errorCode = U_ZERO_ERROR;
  55                 s = buffer;
  56                 ucnv_fromUnicode(converter, &s, sLimit, &us, uLimit, 0, FALSE, &errorCode);
  57
  58                 // write this chunk
  59                 if(s > buffer) {
  60                     stream.write(buffer, (int32_t)(s - buffer));
  61                 }
  62             } while(errorCode == U_BUFFER_OVERFLOW_ERROR);
  63             u_releaseDefaultConverter(converter);
  64         }
  65     }
  66
  67 /*    stream.flush();*/
  68     return stream;
  69 }
  70
  71 U_IO_API STD_ISTREAM & U_EXPORT2
  72 operator>>(STD_ISTREAM& stream, UnicodeString& str)
  73 {
  74     // This is like ICU status checking.
  75     if (stream.fail()) {
  76         return stream;
  77     }
  78
  79     /* ipfx should eat whitespace when ios::skipws is set */
  80     UChar uBuffer[16];
  81     char buffer[16];
  82     int32_t idx = 0;
  83     UConverter *converter;
  84     UErrorCode errorCode = U_ZERO_ERROR;
  85
  86     // use the default converter to convert chunks of text
  87     converter = u_getDefaultConverter(&errorCode);
  88     if(U_SUCCESS(errorCode)) {
  89         UChar *us = uBuffer;
  90         const UChar *uLimit = uBuffer + sizeof(uBuffer)/sizeof(*uBuffer);
  91         const char *s, *sLimit;
  92         char ch;
  93         UChar ch32;
  94         UBool initialWhitespace = TRUE;
  95         UBool continueReading = TRUE;
  96
  97         /* We need to consume one byte at a time to see what is considered whitespace. */
  98         while (continueReading) {
  99             ch = stream.get();
 100             if (stream.eof()) {
 101                 // The EOF is only set after the get() of an unavailable byte.
 102                 if (!initialWhitespace) {
 103                     stream.clear(stream.eofbit);
 104                 }
 105                 continueReading = FALSE;
 106             }
 107             sLimit = &ch + (int)continueReading;
 108             us = uBuffer;
 109             s = &ch;
 110             errorCode = U_ZERO_ERROR;
 111             /*
 112             Since we aren't guaranteed to see the state before this call,
 113             this code won't work on stateful encodings like ISO-2022 or an EBCDIC stateful encoding.
 114             We flush on the last byte to ensure that we output truncated multibyte characters.
 115             */
 116             ucnv_toUnicode(converter, &us, uLimit, &s, sLimit, 0, !continueReading, &errorCode);
 117             if(U_FAILURE(errorCode)) {
 118                 /* Something really bad happened. setstate() isn't always an available API */
 119                 stream.clear(stream.failbit);
 120                 goto STOP_READING;
 121             }
 122             /* Was the character consumed? */
 123             if (us != uBuffer) {
 124                 /* Reminder: ibm-1390 & JISX0213 can output 2 Unicode code points */
 125                 int32_t uBuffSize = us-uBuffer;
 126                 int32_t uBuffIdx = 0;
 127                 while (uBuffIdx < uBuffSize) {
 128                     U16_NEXT(uBuffer, uBuffIdx, uBuffSize, ch32);
 129                     if (u_isWhitespace(ch32)) {
 130                         if (!initialWhitespace) {
 131                             buffer[idx++] = ch;
 132                             while (idx > 0) {
 133                                 stream.putback(buffer[--idx]);
 134                             }
 135                             goto STOP_READING;
 136                         }
 137                         /* else skip intialWhitespace */
 138                     }
 139                     else {
 140                         if (initialWhitespace) {
 141                             /*
 142                             When initialWhitespace is TRUE, we haven't appended any
 143                             character yet.  This is where we truncate the string,
 144                             to avoid modifying the string before we know if we can
 145                             actually read from the stream.
 146                             */
 147                             str.truncate(0);
 148                             initialWhitespace = FALSE;
 149                         }
 150                         str.append(ch32);
 151                     }
 152                 }
 153                 idx = 0;
 154             }
 155             else {
 156                 buffer[idx++] = ch;
 157             }
 158         }
 159 STOP_READING:
 160         u_releaseDefaultConverter(converter);
 161     }
 162
 163 /*    stream.flush();*/
 164     return stream;
 165 }
 166
 167 U_NAMESPACE_END
 168
 169 #endif
 170