summaryrefslogtreecommitdiffstats
path: root/qtools/qtextstream.cpp
diff options
context:
space:
mode:
Diffstat (limited to 'qtools/qtextstream.cpp')
-rw-r--r--qtools/qtextstream.cpp206
1 files changed, 149 insertions, 57 deletions
diff --git a/qtools/qtextstream.cpp b/qtools/qtextstream.cpp
index 774b730..6686f81 100644
--- a/qtools/qtextstream.cpp
+++ b/qtools/qtextstream.cpp
@@ -193,10 +193,14 @@ const int QTextStream::floatfield = ( QTextStream::scientific |
class QTextStreamPrivate {
public:
- QTextStreamPrivate(): decoder( 0 ), sourceType( NotSet ) {}
+#ifndef QT_NO_TEXTCODEC
+ QTextStreamPrivate() : decoder( 0 ), sourceType( NotSet ) {}
~QTextStreamPrivate() { delete decoder; }
-
QTextDecoder *decoder; //???
+#else
+ QTextStreamPrivate() : sourceType( NotSet ) {}
+ ~QTextStreamPrivate() { }
+#endif
QString ungetcBuf;
enum SourceType { NotSet, IODevice, String, ByteArray, File };
@@ -664,14 +668,41 @@ uint QTextStream::ts_getbuf( QChar* buf, uint len )
}
}
+#ifndef QT_NO_TEXTCODEC
if ( mapper ) {
+ bool shortRead = FALSE;
if ( !d->decoder )
d->decoder = mapper->makeDecoder();
while( rnum < len ) {
QString s;
- while ( s.isEmpty() ) {
- // TODO: can this getch() call be optimized to read
- // more than one character after another? YES!
+ bool readBlock = !( len == 1+rnum );
+ while ( TRUE ) {
+ // for efficiency: normally read a whole block
+ if ( readBlock ) {
+ // guess buffersize; this may be wrong (too small or too
+ // big). But we can handle this (either iterate reading
+ // or use ungetcBuf).
+ // Note that this might cause problems for codecs where
+ // one byte can result in >1 Unicode Characters if bytes
+ // are written to the stream in the meantime (loss of
+ // synchronicity).
+ uint rlen = len - rnum;
+ char *cbuf = new char[ rlen ];
+ if ( ungetHack != EOF ) {
+ rlen = 1+dev->readBlock( cbuf+1, rlen-1 );
+ cbuf[0] = (char)ungetHack;
+ ungetHack = EOF;
+ } else {
+ rlen = dev->readBlock( cbuf, rlen );
+ }
+ s += d->decoder->toUnicode( cbuf, rlen );
+ delete[] cbuf;
+ // use buffered reading only for the first time, because we
+ // have to get the stream synchronous again (this is easier
+ // with single character reading)
+ readBlock = FALSE;
+ }
+ // get stream (and codec) in sync
int c;
if ( ungetHack == EOF ) {
c = dev->getch();
@@ -679,10 +710,15 @@ uint QTextStream::ts_getbuf( QChar* buf, uint len )
c = ungetHack;
ungetHack = EOF;
}
- if ( c == EOF )
- return rnum;
+ if ( c == EOF ) {
+ shortRead = TRUE;
+ break;
+ }
char b = c;
- s = d->decoder->toUnicode( &b, 1 );
+ uint lengthBefore = s.length();
+ s += d->decoder->toUnicode( &b, 1 );
+ if ( s.length() > lengthBefore )
+ break; // it seems we are in sync now
}
uint i = 0;
while( rnum < len && i < s.length() )
@@ -690,8 +726,12 @@ uint QTextStream::ts_getbuf( QChar* buf, uint len )
if ( s.length() > i )
// could be = but append is clearer
d->ungetcBuf.append( s.mid( i ) );
+ if ( shortRead )
+ return rnum;
}
- } else if ( latin1 ) {
+ } else
+#endif
+ if ( latin1 ) {
if ( len == 1+rnum ) {
// use this method for one character because it is more efficient
// (arnt doubts whether it makes a difference, but lets it stand)
@@ -699,14 +739,18 @@ uint QTextStream::ts_getbuf( QChar* buf, uint len )
if ( c != EOF )
buf[rnum++] = (char)c;
} else {
- if ( (QChar)ungetHack != QEOF )
+ if ( ungetHack != EOF ) {
buf[rnum++] = (char)ungetHack;
- uint rlen = len - rnum;
- char *cbuf = new char[rlen];
- rlen = dev->readBlock( cbuf, rlen );
- uint i = 0;
- while( i < rlen )
- buf[rnum++] = cbuf[i++];
+ ungetHack = EOF;
+ }
+ char *cbuf = new char[len - rnum];
+ while ( !dev->atEnd() && rnum < len ) {
+ uint rlen = len - rnum;
+ rlen = dev->readBlock( cbuf, rlen );
+ uint i = 0;
+ while( i < rlen )
+ buf[rnum++] = cbuf[i++];
+ }
delete[] cbuf;
}
} else { // UCS-2 or UTF-16
@@ -722,29 +766,34 @@ uint QTextStream::ts_getbuf( QChar* buf, uint len )
else
buf[rnum++] = QChar( c1, c2 );
} else {
- uint rlen = 2 * ( len-rnum );
- char *cbuf = new char[rlen]; // for paranoids: overflow possible
- if ( (QChar)ungetHack != QEOF ) {
- rlen = 1+dev->readBlock( cbuf+1, rlen-1 );
- cbuf[0] = (char)ungetHack;
- } else {
- rlen = dev->readBlock( cbuf, rlen );
- }
- // is this right? we can't use an odd number of bytes, but
- // if there -is- an odd number, with this code we'll never
- // get to EOF.
- if ( (rlen & 1) == 1 )
- dev->ungetch( cbuf[--rlen] );
- uint i = 0;
- if ( isNetworkOrder() ) {
- while( i < rlen ) {
- buf[rnum++] = QChar( cbuf[i+1], cbuf[i] );
- i+=2;
+ char *cbuf = new char[ 2*( len - rnum ) ]; // for paranoids: overflow possible
+ while ( !dev->atEnd() && rnum < len ) {
+ uint rlen = 2 * ( len-rnum );
+ if ( ungetHack != EOF ) {
+ rlen = 1+dev->readBlock( cbuf+1, rlen-1 );
+ cbuf[0] = (char)ungetHack;
+ ungetHack = EOF;
+ } else {
+ rlen = dev->readBlock( cbuf, rlen );
}
- } else {
- while( i < rlen ) {
- buf[rnum++] = QChar( cbuf[i], cbuf[i+1] );
- i+=2;
+ // We can't use an odd number of bytes, so put it back. But
+ // do it only if we are capable of reading more -- normally
+ // there should not be an odd number, but the file might be
+ // truncated or not in UTF-16...
+ if ( (rlen & 1) == 1 )
+ if ( !dev->atEnd() )
+ dev->ungetch( cbuf[--rlen] );
+ uint i = 0;
+ if ( isNetworkOrder() ) {
+ while( i < rlen ) {
+ buf[rnum++] = QChar( cbuf[i+1], cbuf[i] );
+ i+=2;
+ }
+ } else {
+ while( i < rlen ) {
+ buf[rnum++] = QChar( cbuf[i], cbuf[i+1] );
+ i+=2;
+ }
}
}
delete[] cbuf;
@@ -759,12 +808,15 @@ uint QTextStream::ts_getbuf( QChar* buf, uint len )
*/
void QTextStream::ts_putc( QChar c )
{
+#ifndef QT_NO_TEXTCODEC
if ( mapper ) {
int len = 1;
QString s = c;
QCString block = mapper->fromUnicode( s, len );
dev->writeBlock( block, len );
- } else if ( latin1 ) {
+ } else
+#endif
+ if ( latin1 ) {
if( c.row() )
dev->putch( '?' ); //######unknown character???
else
@@ -820,7 +872,7 @@ void QTextStream::ts_ungetc( QChar c )
The buffer \e s must be preallocated.
- \note No Encoding is done by this function.
+ Note that no encoding is done by this function.
\warning The behaviour of this function is undefined unless the
stream's encoding is set to Unicode or Latin1.
@@ -838,7 +890,7 @@ QTextStream &QTextStream::readRawBytes( char *s, uint len )
Writes the \e len bytes from \e s to the stream and returns a reference to
the stream.
- \note No Encoding is done by this function.
+ Note that no encoding is done by this function.
\sa QIODevice::writeBlock()
*/
@@ -1428,19 +1480,28 @@ QString QTextStream::readLine()
return QString::null;
}
#endif
- QChar c = ts_getc();
- if ( c == QEOF )
+ QString result( "" );
+ const int buf_size = 256;
+ QChar c[buf_size];
+ int pos = 0;
+
+ c[pos] = ts_getc();
+ if ( c[pos] == QEOF )
return QString::null;
- QString result( "" );
- while ( c != QEOF && c != '\n' ) {
- result += c;
- c = ts_getc();
+ while ( c[pos] != QEOF && c[pos] != '\n' ) {
+ pos++;
+ if ( pos >= buf_size ) {
+ result += QString( c, pos );
+ pos = 0;
+ }
+ c[pos] = ts_getc();
}
+ result += QString( c, pos );
int len = (int)result.length();
if ( len && result[len-1] == '\r' )
- result.truncate(len-1); // (if there are two \r, let one stay)
+ result.truncate(len-1); // (if there are two \r, let one stay)
return result;
}
@@ -1456,7 +1517,7 @@ QString QTextStream::read()
{
#if defined(CHECK_STATE)
if ( !dev ) {
- qWarning( "QTextStream::readLine: No device" );
+ qWarning( "QTextStream::read: No device" );
return QString::null;
}
#endif
@@ -1507,8 +1568,20 @@ QString QTextStream::read()
/*!
Writes a \c char to the stream and returns a reference to the stream.
+
+ The character \a c is assumed to be Latin1 encoded independent of the Encoding set
+ for the QTextStream.
*/
+QTextStream &QTextStream::operator<<( QChar c )
+{
+ CHECK_STREAM_PRECOND
+ ts_putc( c );
+ return *this;
+}
+/*!
+ Writes a \c char to the stream and returns a reference to the stream.
+*/
QTextStream &QTextStream::operator<<( char c )
{
CHECK_STREAM_PRECOND
@@ -1737,6 +1810,9 @@ QTextStream &QTextStream::operator<<( double f )
/*!
Writes a string to the stream and returns a reference to the stream.
+
+ The string \a s is assumed to be Latin1 encoded independent of the Encoding set
+ for the QTextStream.
*/
QTextStream &QTextStream::operator<<( const char* s )
@@ -1774,6 +1850,9 @@ QTextStream &QTextStream::operator<<( const char* s )
/*!
Writes \a s to the stream and returns a reference to the stream.
+
+ The string \a s is assumed to be Latin1 encoded independent of the Encoding set
+ for the QTextStream.
*/
QTextStream &QTextStream::operator<<( const QCString & s )
@@ -2064,11 +2143,12 @@ QTextStream &reset( QTextStream &s )
writing to non-persistent storage used by a single process.
</ul>
- \c Locale and all Unicode encodings, except \c RawUnicode, will look
- at the first two bytes in a input stream to determine the byte
- order. The initial byte order marker will be stripped off before data is read.
+ \c Locale and all Unicode encodings, except \c RawUnicode, will look at
+ the first two bytes in a input stream to determine the byte order. The
+ initial byte order marker will be stripped off before data is read.
- \note This function should be called before any data is read to/written from the stream.
+ Note that this function should be called before any data is read
+ to/written from the stream.
\sa setCodec()
*/
@@ -2084,10 +2164,16 @@ void QTextStream::setEncoding( Encoding e )
internalOrder = TRUE;
break;
case UnicodeUTF8:
+#ifndef QT_NO_CODECS
mapper = QTextCodec::codecForMib( 106 );
latin1 = FALSE;
doUnicodeHeader = TRUE;
internalOrder = TRUE;
+#else
+ mapper = 0;
+ latin1 = TRUE;
+ doUnicodeHeader = TRUE;
+#endif
break;
case UnicodeNetworkOrder:
mapper = 0;
@@ -2109,12 +2195,14 @@ void QTextStream::setEncoding( Encoding e )
break;
case Locale:
latin1 = TRUE; // fallback to Latin 1
+#ifndef QT_NO_TEXTCODEC
mapper = QTextCodec::codecForLocale();
#if defined(_OS_WIN32_)
if ( GetACP() == 1252 )
mapper = 0; // Optimized latin1 processing
#endif
if ( mapper && mapper->mibEnum() == 4 )
+#endif
mapper = 0; // Optimized latin1 processing
doUnicodeHeader = TRUE; // If it reads as Unicode, accept it
break;
@@ -2127,11 +2215,14 @@ void QTextStream::setEncoding( Encoding e )
}
-/*!
- Sets the codec for this stream to \a codec. Will not try to autodetect Unicode.
+#ifndef QT_NO_TEXTCODEC
+/*! Sets the codec for this stream to \a codec. Will not try to
+ autodetect Unicode.
- \note This function should be called before any data is read to/written from the stream.
- \sa setEncoding()
+ Note that this function should be called before any data is read
+ to/written from the stream.
+
+ \sa setEncoding()
*/
void QTextStream::setCodec( QTextCodec *codec )
@@ -2141,5 +2232,6 @@ void QTextStream::setCodec( QTextCodec *codec )
mapper = codec;
doUnicodeHeader = FALSE;
}
+#endif
#endif // QT_NO_TEXTSTREAM
generated by cgit v0.12 at 2024-11-18 22:06:32 (GMT)