intl/icu/source/io/ustream.cpp@b8a032363ba2 (annotated)
intl/icu/source/io/ustream.cpp
Thu, 22 Jan 2015 13:21:57 +0100
- author
- Michael Schloh von Bennewitz <michael@schloh.com>
- date
- Thu, 22 Jan 2015 13:21:57 +0100
- branch
- TOR_BUG_9701
- changeset 15
- b8a032363ba2
- permissions
- -rw-r--r--
Incorporate requested changes from Mozilla in review:
https://bugzilla.mozilla.org/show_bug.cgi?id=1123480#c6
| michael@0 | 1 | /* |
| michael@0 | 2 | ********************************************************************** |
| michael@0 | 3 | * Copyright (C) 2001-2011, International Business Machines |
| michael@0 | 4 | * Corporation and others. All Rights Reserved. |
| michael@0 | 5 | ********************************************************************** |
| michael@0 | 6 | * FILE NAME : ustream.cpp |
| michael@0 | 7 | * |
| michael@0 | 8 | * Modification History: |
| michael@0 | 9 | * |
| michael@0 | 10 | * Date Name Description |
| michael@0 | 11 | * 06/25/2001 grhoten Move iostream from unistr.h to here |
| michael@0 | 12 | ****************************************************************************** |
| michael@0 | 13 | */ |
| michael@0 | 14 | |
| michael@0 | 15 | #include "unicode/utypes.h" |
| michael@0 | 16 | #include "unicode/uobject.h" |
| michael@0 | 17 | #include "unicode/ustream.h" |
| michael@0 | 18 | #include "unicode/ucnv.h" |
| michael@0 | 19 | #include "unicode/uchar.h" |
| michael@0 | 20 | #include "unicode/utf16.h" |
| michael@0 | 21 | #include "ustr_cnv.h" |
| michael@0 | 22 | #include "cmemory.h" |
| michael@0 | 23 | #include <string.h> |
| michael@0 | 24 | |
| michael@0 | 25 | // console IO |
| michael@0 | 26 | |
| michael@0 | 27 | #if U_IOSTREAM_SOURCE >= 199711 |
| michael@0 | 28 | |
| michael@0 | 29 | #define STD_NAMESPACE std:: |
| michael@0 | 30 | |
| michael@0 | 31 | #define STD_OSTREAM STD_NAMESPACE ostream |
| michael@0 | 32 | #define STD_ISTREAM STD_NAMESPACE istream |
| michael@0 | 33 | |
| michael@0 | 34 | U_NAMESPACE_BEGIN |
| michael@0 | 35 | |
| michael@0 | 36 | U_IO_API STD_OSTREAM & U_EXPORT2 |
| michael@0 | 37 | operator<<(STD_OSTREAM& stream, const UnicodeString& str) |
| michael@0 | 38 | { |
| michael@0 | 39 | if(str.length() > 0) { |
| michael@0 | 40 | char buffer[200]; |
| michael@0 | 41 | UConverter *converter; |
| michael@0 | 42 | UErrorCode errorCode = U_ZERO_ERROR; |
| michael@0 | 43 | |
| michael@0 | 44 | // use the default converter to convert chunks of text |
| michael@0 | 45 | converter = u_getDefaultConverter(&errorCode); |
| michael@0 | 46 | if(U_SUCCESS(errorCode)) { |
| michael@0 | 47 | const UChar *us = str.getBuffer(); |
| michael@0 | 48 | const UChar *uLimit = us + str.length(); |
| michael@0 | 49 | char *s, *sLimit = buffer + (sizeof(buffer) - 1); |
| michael@0 | 50 | do { |
| michael@0 | 51 | errorCode = U_ZERO_ERROR; |
| michael@0 | 52 | s = buffer; |
| michael@0 | 53 | ucnv_fromUnicode(converter, &s, sLimit, &us, uLimit, 0, FALSE, &errorCode); |
| michael@0 | 54 | *s = 0; |
| michael@0 | 55 | |
| michael@0 | 56 | // write this chunk |
| michael@0 | 57 | if(s > buffer) { |
| michael@0 | 58 | stream << buffer; |
| michael@0 | 59 | } |
| michael@0 | 60 | } while(errorCode == U_BUFFER_OVERFLOW_ERROR); |
| michael@0 | 61 | u_releaseDefaultConverter(converter); |
| michael@0 | 62 | } |
| michael@0 | 63 | } |
| michael@0 | 64 | |
| michael@0 | 65 | /* stream.flush();*/ |
| michael@0 | 66 | return stream; |
| michael@0 | 67 | } |
| michael@0 | 68 | |
| michael@0 | 69 | U_IO_API STD_ISTREAM & U_EXPORT2 |
| michael@0 | 70 | operator>>(STD_ISTREAM& stream, UnicodeString& str) |
| michael@0 | 71 | { |
| michael@0 | 72 | // This is like ICU status checking. |
| michael@0 | 73 | if (stream.fail()) { |
| michael@0 | 74 | return stream; |
| michael@0 | 75 | } |
| michael@0 | 76 | |
| michael@0 | 77 | /* ipfx should eat whitespace when ios::skipws is set */ |
| michael@0 | 78 | UChar uBuffer[16]; |
| michael@0 | 79 | char buffer[16]; |
| michael@0 | 80 | int32_t idx = 0; |
| michael@0 | 81 | UConverter *converter; |
| michael@0 | 82 | UErrorCode errorCode = U_ZERO_ERROR; |
| michael@0 | 83 | |
| michael@0 | 84 | // use the default converter to convert chunks of text |
| michael@0 | 85 | converter = u_getDefaultConverter(&errorCode); |
| michael@0 | 86 | if(U_SUCCESS(errorCode)) { |
| michael@0 | 87 | UChar *us = uBuffer; |
| michael@0 | 88 | const UChar *uLimit = uBuffer + sizeof(uBuffer)/sizeof(*uBuffer); |
| michael@0 | 89 | const char *s, *sLimit; |
| michael@0 | 90 | char ch; |
| michael@0 | 91 | UChar ch32; |
| michael@0 | 92 | UBool initialWhitespace = TRUE; |
| michael@0 | 93 | UBool continueReading = TRUE; |
| michael@0 | 94 | |
| michael@0 | 95 | /* We need to consume one byte at a time to see what is considered whitespace. */ |
| michael@0 | 96 | while (continueReading) { |
| michael@0 | 97 | ch = stream.get(); |
| michael@0 | 98 | if (stream.eof()) { |
| michael@0 | 99 | // The EOF is only set after the get() of an unavailable byte. |
| michael@0 | 100 | if (!initialWhitespace) { |
| michael@0 | 101 | stream.clear(stream.eofbit); |
| michael@0 | 102 | } |
| michael@0 | 103 | continueReading = FALSE; |
| michael@0 | 104 | } |
| michael@0 | 105 | sLimit = &ch + (int)continueReading; |
| michael@0 | 106 | us = uBuffer; |
| michael@0 | 107 | s = &ch; |
| michael@0 | 108 | errorCode = U_ZERO_ERROR; |
| michael@0 | 109 | /* |
| michael@0 | 110 | Since we aren't guaranteed to see the state before this call, |
| michael@0 | 111 | this code won't work on stateful encodings like ISO-2022 or an EBCDIC stateful encoding. |
| michael@0 | 112 | We flush on the last byte to ensure that we output truncated multibyte characters. |
| michael@0 | 113 | */ |
| michael@0 | 114 | ucnv_toUnicode(converter, &us, uLimit, &s, sLimit, 0, !continueReading, &errorCode); |
| michael@0 | 115 | if(U_FAILURE(errorCode)) { |
| michael@0 | 116 | /* Something really bad happened. setstate() isn't always an available API */ |
| michael@0 | 117 | stream.clear(stream.failbit); |
| michael@0 | 118 | goto STOP_READING; |
| michael@0 | 119 | } |
| michael@0 | 120 | /* Was the character consumed? */ |
| michael@0 | 121 | if (us != uBuffer) { |
| michael@0 | 122 | /* Reminder: ibm-1390 & JISX0213 can output 2 Unicode code points */ |
| michael@0 | 123 | int32_t uBuffSize = us-uBuffer; |
| michael@0 | 124 | int32_t uBuffIdx = 0; |
| michael@0 | 125 | while (uBuffIdx < uBuffSize) { |
| michael@0 | 126 | U16_NEXT(uBuffer, uBuffIdx, uBuffSize, ch32); |
| michael@0 | 127 | if (u_isWhitespace(ch32)) { |
| michael@0 | 128 | if (!initialWhitespace) { |
| michael@0 | 129 | buffer[idx++] = ch; |
| michael@0 | 130 | while (idx > 0) { |
| michael@0 | 131 | stream.putback(buffer[--idx]); |
| michael@0 | 132 | } |
| michael@0 | 133 | goto STOP_READING; |
| michael@0 | 134 | } |
| michael@0 | 135 | /* else skip intialWhitespace */ |
| michael@0 | 136 | } |
| michael@0 | 137 | else { |
| michael@0 | 138 | if (initialWhitespace) { |
| michael@0 | 139 | /* |
| michael@0 | 140 | When initialWhitespace is TRUE, we haven't appended any |
| michael@0 | 141 | character yet. This is where we truncate the string, |
| michael@0 | 142 | to avoid modifying the string before we know if we can |
| michael@0 | 143 | actually read from the stream. |
| michael@0 | 144 | */ |
| michael@0 | 145 | str.truncate(0); |
| michael@0 | 146 | initialWhitespace = FALSE; |
| michael@0 | 147 | } |
| michael@0 | 148 | str.append(ch32); |
| michael@0 | 149 | } |
| michael@0 | 150 | } |
| michael@0 | 151 | idx = 0; |
| michael@0 | 152 | } |
| michael@0 | 153 | else { |
| michael@0 | 154 | buffer[idx++] = ch; |
| michael@0 | 155 | } |
| michael@0 | 156 | } |
| michael@0 | 157 | STOP_READING: |
| michael@0 | 158 | u_releaseDefaultConverter(converter); |
| michael@0 | 159 | } |
| michael@0 | 160 | |
| michael@0 | 161 | /* stream.flush();*/ |
| michael@0 | 162 | return stream; |
| michael@0 | 163 | } |
| michael@0 | 164 | |
| michael@0 | 165 | U_NAMESPACE_END |
| michael@0 | 166 | |
| michael@0 | 167 | #endif |
