kviewshell

UnicodeByteStream.cpp

00001 //C-  -*- C++ -*-
00002 //C- -------------------------------------------------------------------
00003 //C- DjVuLibre-3.5
00004 //C- Copyright (c) 2002  Leon Bottou and Yann Le Cun.
00005 //C- Copyright (c) 2001  AT&T
00006 //C-
00007 //C- This software is subject to, and may be distributed under, the
00008 //C- GNU General Public License, Version 2. The license should have
00009 //C- accompanied the software or you may obtain a copy of the license
00010 //C- from the Free Software Foundation at http://www.fsf.org .
00011 //C-
00012 //C- This program is distributed in the hope that it will be useful,
00013 //C- but WITHOUT ANY WARRANTY; without even the implied warranty of
00014 //C- MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
00015 //C- GNU General Public License for more details.
00016 //C- 
00017 //C- DjVuLibre-3.5 is derived from the DjVu(r) Reference Library
00018 //C- distributed by Lizardtech Software.  On July 19th 2002, Lizardtech 
00019 //C- Software authorized us to replace the original DjVu(r) Reference 
00020 //C- Library notice by the following text (see doc/lizard2002.djvu):
00021 //C-
00022 //C-  ------------------------------------------------------------------
00023 //C- | DjVu (r) Reference Library (v. 3.5)
00024 //C- | Copyright (c) 1999-2001 LizardTech, Inc. All Rights Reserved.
00025 //C- | The DjVu Reference Library is protected by U.S. Pat. No.
00026 //C- | 6,058,214 and patents pending.
00027 //C- |
00028 //C- | This software is subject to, and may be distributed under, the
00029 //C- | GNU General Public License, Version 2. The license should have
00030 //C- | accompanied the software or you may obtain a copy of the license
00031 //C- | from the Free Software Foundation at http://www.fsf.org .
00032 //C- |
00033 //C- | The computer code originally released by LizardTech under this
00034 //C- | license and unmodified by other parties is deemed "the LIZARDTECH
00035 //C- | ORIGINAL CODE."  Subject to any third party intellectual property
00036 //C- | claims, LizardTech grants recipient a worldwide, royalty-free, 
00037 //C- | non-exclusive license to make, use, sell, or otherwise dispose of 
00038 //C- | the LIZARDTECH ORIGINAL CODE or of programs derived from the 
00039 //C- | LIZARDTECH ORIGINAL CODE in compliance with the terms of the GNU 
00040 //C- | General Public License.   This grant only confers the right to 
00041 //C- | infringe patent claims underlying the LIZARDTECH ORIGINAL CODE to 
00042 //C- | the extent such infringement is reasonably necessary to enable 
00043 //C- | recipient to make, have made, practice, sell, or otherwise dispose 
00044 //C- | of the LIZARDTECH ORIGINAL CODE (or portions thereof) and not to 
00045 //C- | any greater extent that may be necessary to utilize further 
00046 //C- | modifications or combinations.
00047 //C- |
00048 //C- | The LIZARDTECH ORIGINAL CODE is provided "AS IS" WITHOUT WARRANTY
00049 //C- | OF ANY KIND, EITHER EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED
00050 //C- | TO ANY WARRANTY OF NON-INFRINGEMENT, OR ANY IMPLIED WARRANTY OF
00051 //C- | MERCHANTABILITY OR FITNESS FOR A PARTICULAR PURPOSE.
00052 //C- +------------------------------------------------------------------
00053 //
00054 // $Id: UnicodeByteStream.cpp,v 1.8 2003/11/07 22:08:22 leonb Exp $
00055 // $Name: release_3_5_15 $
00056 
00057 #ifdef HAVE_CONFIG_H
00058 # include "config.h"
00059 #endif
00060 #if NEED_GNUG_PRAGMAS
00061 # pragma implementation
00062 #endif
00063 
00064 #include "UnicodeByteStream.h"
00065 #include "ByteStream.h"
00066 
00067 
00068 #ifdef HAVE_NAMESPACES
00069 namespace DJVU {
00070 # ifdef NOT_DEFINED // Just to fool emacs c++ mode
00071 }
00072 #endif
00073 #endif
00074 
00075 UnicodeByteStream::UnicodeByteStream(const UnicodeByteStream &uni)
00076 : bs(uni.bs), buffer(uni.buffer), bufferpos(uni.bufferpos), linesread(0)
00077 {
00078   startpos=bs->tell();
00079 }
00080 
00081 UnicodeByteStream::UnicodeByteStream(
00082   GP<ByteStream> ibs,const GStringRep::EncodeType et)
00083 : bs(ibs), bufferpos(0), linesread(0)
00084 {
00085   buffer=GUTF8String::create(0,0,et);
00086   startpos=bs->tell();
00087 }
00088 
00089 UnicodeByteStream::~UnicodeByteStream()
00090 {}
00091 
00092 static int
00093 CountLines(const GUTF8String &str)
00094 {
00095   int retval=0;
00096   static const unsigned long lf='\n';
00097   for(int pos=0;(pos=str.search(lf,pos)+1)>0;++retval)
00098     EMPTY_LOOP;
00099   return retval;
00100 }
00101 
00102 void
00103 UnicodeByteStream::set_encodetype(const GStringRep::EncodeType et)
00104 {
00105   seek(startpos,SEEK_SET);
00106   bufferpos=0;
00107   buffer=GUTF8String::create(0,0,et);
00108 }
00109 
00110 void
00111 UnicodeByteStream::set_encoding(const GUTF8String &xencoding)
00112 {
00113   seek(startpos,SEEK_SET);
00114   bufferpos=0;
00115   buffer=GUTF8String::create(0,0,xencoding);
00116 }
00117 
00118 size_t
00119 UnicodeByteStream::read(void *buf, size_t size)
00120 {
00121   bufferpos=0;
00122   const int retval=bs->read(buf,size);
00123   if(retval)
00124   {
00125     buffer=GUTF8String::create(
00126       (unsigned char const *)buf,retval,buffer.get_remainder());
00127   }else
00128   {
00129     buffer=GUTF8String::create(0,0,buffer.get_remainder());
00130   }
00131   return retval;
00132 }
00133 
00134 size_t
00135 UnicodeByteStream::write(const void *buf, size_t size)
00136 {
00137   bufferpos=0;
00138   buffer=GUTF8String::create(0,0,buffer.get_remainder());
00139   return bs->write(buf,size);
00140 }
00141 
00142 long 
00143 UnicodeByteStream::tell(void) const
00144 {
00145   return bs->tell();
00146 }
00147 
00148 UnicodeByteStream & 
00149 UnicodeByteStream::operator=(UnicodeByteStream &uni)
00150 {
00151   bs=uni.bs;
00152   bufferpos=uni.bufferpos;
00153   buffer=uni.buffer;
00154   return *this;
00155 }
00156 
00157 int 
00158 UnicodeByteStream::seek
00159 (long offset, int whence, bool nothrow)
00160 {
00161   int retval=bs->seek(offset,whence,nothrow);
00162   bufferpos=0;
00163   buffer=GUTF8String::create(0,0,buffer.get_remainder());
00164   return retval;
00165 }
00166 
00167 void 
00168 UnicodeByteStream::flush(void)
00169 {
00170   bs->flush();
00171   bufferpos=0;
00172   buffer=GUTF8String::create(0,0,buffer.get_remainder());
00173 }
00174 
00175 
00176 
00177 GUTF8String
00178 UnicodeByteStream::gets(
00179   size_t const t,unsigned long const stopat,bool const inclusive)
00180 {
00181   GUTF8String retval;
00182   unsigned int len=buffer.length()-bufferpos;
00183   if(!len)
00184   {
00185     int i;
00186     char *buf;
00187     static const size_t bufsize=327680;
00188     GPBuffer<char> gbuf(buf,bufsize);
00189     while((i=read(buf,bufsize)>0))
00190     {
00191       if((len=buffer.length()-bufferpos))
00192         break;
00193     }
00194   }
00195   if(len)
00196   {
00197     int i=buffer.search((char)stopat,bufferpos);
00198     if(i>=0)
00199     {
00200       if(inclusive)
00201       {
00202         ++i;
00203       }
00204       if(t&&(i>(int)t+bufferpos))
00205       {
00206         i=t+bufferpos;
00207       }
00208       if(i>bufferpos)
00209       {
00210         retval=buffer.substr(bufferpos,i-bufferpos);
00211       }
00212       bufferpos=i;
00213       linesread+=CountLines(retval);
00214     }else
00215     {
00216       retval=buffer.substr(bufferpos,len);
00217       bufferpos=buffer.length();
00218       linesread+=CountLines(retval);
00219       retval+=gets(t?(t-(i-bufferpos)):0,stopat,inclusive);
00220     }
00221   }
00222   return retval;
00223 }
00224 
00225 XMLByteStream::XMLByteStream(UnicodeByteStream &uni)
00226 : UnicodeByteStream(uni) {}
00227 
00228 XMLByteStream::XMLByteStream(GP<ByteStream> &ibs) 
00229 : UnicodeByteStream(ibs,GStringRep::XOTHER)
00230 {}
00231 
00232 GP<XMLByteStream>
00233 XMLByteStream::create(GP<ByteStream> ibs) 
00234 {
00235   XMLByteStream *xml=new XMLByteStream(ibs);
00236   GP<XMLByteStream> retval=xml;
00237   xml->init();
00238   return retval;
00239 }
00240 
00241 void
00242 XMLByteStream::init(void)
00243 {
00244   unsigned char buf[4];
00245   GP<ByteStream> ibs=bs;
00246   bufferpos=0;
00247   bs->readall(buf,sizeof(buf));
00248   const unsigned int i=(buf[0]<<8)+buf[1];
00249   switch(i)
00250   {
00251     case 0x0000:
00252     {
00253       const unsigned int j=(buf[2]<<8)+buf[3];
00254       switch(j)
00255       {
00256         case 0x003C:
00257         {
00258           buffer=GUTF8String::create(buf,sizeof(buf),GStringRep::XUCS4BE);
00259           break;
00260         }
00261         case 0x3C00:
00262         {
00263           buffer=GUTF8String::create(buf,sizeof(buf),GStringRep::XUCS4_2143);
00264           break;
00265         }
00266         case 0xFEFF:
00267         {
00268           buffer=GUTF8String::create(0,0,GStringRep::XUCS4BE);
00269           startpos+=sizeof(buf);
00270           break;
00271         }
00272         case 0xFFFE:
00273         {
00274           buffer=GUTF8String::create(0,0,GStringRep::XUCS4_2143);
00275           startpos+=sizeof(buf);
00276           break;
00277         }
00278         default:
00279         {
00280           buffer=GUTF8String::create(buf,sizeof(buf),GStringRep::XUTF8);
00281           break;
00282         }
00283       }
00284     }
00285     case 0x003C:
00286     {
00287       const unsigned int j=(buf[2]<<8)+buf[3];
00288       switch(j)
00289       {
00290         case 0x0000:
00291           buffer=GUTF8String::create(buf,sizeof(buf),GStringRep::XUCS4_3412);
00292           break;
00293         case 0x003F:
00294           buffer=GUTF8String::create(buf,sizeof(buf),GStringRep::XUTF16BE);
00295           break;
00296         default:
00297           buffer=GUTF8String::create(buf,sizeof(buf),GStringRep::XUTF8);
00298           break;
00299       }
00300       break;
00301     }
00302     case 0x3C00:
00303     {
00304       const unsigned int j=(buf[2]<<8)+buf[3];
00305       switch(j)
00306       {
00307         case 0x0000:
00308           buffer=GUTF8String::create(buf,sizeof(buf),GStringRep::XUCS4LE);
00309           break;
00310         case 0x3F00:
00311           buffer=GUTF8String::create(buf,sizeof(buf),GStringRep::XUTF16LE);
00312           break;
00313         default:
00314           buffer=GUTF8String::create(buf,sizeof(buf),GStringRep::XUTF8);
00315           break;
00316       }
00317       break;
00318     }
00319     case 0x4C6F:
00320     {
00321       const unsigned int j=(buf[2]<<8)+buf[3];
00322       buffer=GUTF8String::create(buf,sizeof(buf),
00323          (j == 0xA794)?(GStringRep::XEBCDIC):(GStringRep::XUTF8));
00324       break;
00325     }
00326     case 0xFFFE:
00327     {
00328       buffer=GUTF8String::create(buf+2,sizeof(buf)-2,GStringRep::XUTF16LE);
00329       startpos+=2;
00330       break;
00331     }
00332     case 0xFEFF:
00333     {
00334       buffer=GUTF8String::create(buf+2,sizeof(buf)-2,GStringRep::XUTF16BE);
00335       startpos+=2;
00336       break;
00337     }
00338     case 0xEFBB:
00339     {
00340       if(buf[2] == 0xBF)
00341       {
00342         buffer=GUTF8String::create(buf+3,sizeof(buf)-3,GStringRep::XUTF8);
00343         startpos+=3;
00344       }else
00345       {
00346         buffer=GUTF8String::create(buf,sizeof(buf),GStringRep::XUTF8);
00347       }
00348       break;
00349     }
00350     case 0x3C3F:
00351     default:
00352     {
00353       buffer=GUTF8String::create(buf,sizeof(buf),GStringRep::XUTF8);
00354     }
00355   }
00356   bs=ibs;
00357 }
00358 
00359 XMLByteStream::~XMLByteStream()
00360 {}
00361 
00362 
00363 #ifdef HAVE_NAMESPACES
00364 }
00365 # ifndef NOT_USING_DJVU_NAMESPACE
00366 using namespace DJVU;
00367 # endif
00368 #endif
kviewshell

UnicodeByteStream.cpp

kviewshell

API Reference
