libchmfileimpl.cpp

chm阅读器 用于在linux系统下chm文档的阅读

QCString LCHMFileImpl::convertSearchWord( const QString & src )
{
	static const char * searchwordtable[128] =
	{
		0, 0, 0, 0, 0, 0, 0, 0, 0, 0, "s", 0, "oe", 0, 0, 0,
  0, 0, 0, 0, 0, 0, 0, 0, 0, 0, "s", 0, "oe", 0, 0, "y",
  0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
  0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
  "a", "a", "a", "a", "a", "a", "ae", "c", "e", "e", "e", "e", "i", "i", "i", "i",
  "d", "n", "o", "o", "o", "o", "o", 0, "o", "u", "u", "u", "u", "y", "\xDE", "ss",
  "a", "a", "a", "a", "a", "a", "ae", "c", "e", "e", "e", "e", "i", "i", "i", "i",
  "o", "n", "o", "o", "o", "o", "o", 0, "o", "u", "u", "u", "u", "y", "\xFE", "y"
	};

	if ( !m_textCodec )
		return (QCString) src.lower();

	QCString dest = m_textCodec->fromUnicode (src);

	for ( unsigned int i = 0; i < dest.size(); i++ )
	{
		if ( dest[i] & 0x80 )
		{
			int index = dest[i] & 0x7F;
			if ( searchwordtable[index] )
				dest.replace (i, 1, searchwordtable[index]);
			else
				dest.remove (i, 1);
		}
	}

	return dest.lower();
}



void LCHMFileImpl::getSearchResults( const LCHMSearchProgressResults& tempres, 
									 QStringList * results, 
		  							 unsigned int limit_results )
{
	unsigned char combuf [COMMON_BUF_LEN];
	QMap<u_int32_t, u_int32_t> urlsmap;  // used to prevent duplicated urls
	
	for ( unsigned int i = 0; i < tempres.size(); i++ )
	{
		if ( urlsmap.find (tempres[i].urloff) != urlsmap.end() )
			continue;
		
		urlsmap[tempres[i].urloff] = 1;
		
		if ( RetrieveObject (&m_chmURLSTR, combuf, tempres[i].urloff + 8, COMMON_BUF_LEN - 1) == 0 )
			continue;

		combuf[COMMON_BUF_LEN - 1] = 0;
		results->push_back( LCHMUrlFactory::makeURLabsoluteIfNeeded( (const char*) combuf ) );
		
		if ( --limit_results == 0 )
			break;
	}
}


QString LCHMFileImpl::normalizeUrl( const QString & path ) const
{
	int pos = path.find ('#');
	QString fixedpath = pos == -1 ? path : path.left (pos);
	
	return LCHMUrlFactory::makeURLabsoluteIfNeeded( fixedpath );
}


/*
 * FIXME: <OBJECT type="text/sitemap"><param name="Merge" value="hhaxref.chm::/HHOCX_c.hhc"></OBJECT>
 *  (from htmlhelp.chm)
*/
bool LCHMFileImpl::parseFileAndFillArray( const QString & file, QT34VECTOR< LCHMParsedEntry > * data, bool asIndex )
{
	QString src;
	const int MAX_NEST_DEPTH = 256;

	if ( !getFileContentAsString( &src, file ) || src.isEmpty() )
		return false;

	KCHMShowWaitCursor wc;
		
/*
	// Save the index for debugging purposes
	QFile outfile( "parsed.htm" );
	
	if ( outfile.open( IO_WriteOnly ) )
	{
		QTextStream textstream( &outfile );
		textstream << src;
		outfile.close();
	}
*/
	
	unsigned int defaultimagenum = asIndex ? LCHMBookIcons::IMAGE_INDEX : LCHMBookIcons::IMAGE_AUTO;
	int pos = 0, indent = 0, root_indent_offset = 0;
	bool in_object = false, root_indent_offset_set = false;
	
	LCHMParsedEntry entry;
	entry.imageid = defaultimagenum;
	
	// Split the HHC file by HTML tags
	int stringlen = src.length();
	
	while ( pos < stringlen && (pos = src.find ('<', pos)) != -1 )
	{
		int i, word_end = 0;
		
		for ( i = ++pos; i < stringlen; i++ )
		{
			// If a " or ' is found, skip to the next one.
			if ( (src[i] == '"' || src[i] == '\'') )
			{
				// find where quote ends, either by another quote, or by '>' symbol (some people don't know HTML)
				int nextpos = src.find (src[i], i+1);
				if ( nextpos == -1 	&& (nextpos = src.find ('>', i+1)) == -1 )
				{
					qWarning ("LCHMFileImpl::ParseHhcAndFillTree: corrupted TOC: %s", src.mid(i).ascii());
					return false;
				}

				i =  nextpos;
			}
			else if ( src[i] == '>'  )
				break;
			else if ( !src[i].isLetterOrNumber() && src[i] != '/' && !word_end )
				word_end = i;
		}
		
		QString tagword, tag = src.mid (pos, i - pos);
		 
		if ( word_end )
			tagword = src.mid (pos, word_end - pos).lower();
		else
			tagword = tag.lower();

		//qDebug ("tag: '%s', tagword: '%s'\n", tag.ascii(), tagword.ascii());
						
		// <OBJECT type="text/sitemap"> - a topic entry
		if ( tagword == "object" && tag.find ("text/sitemap", 0, false) != -1 )
			in_object = true;
		else if ( tagword == "/object" && in_object ) 
		{
			// a topic entry closed. Add a tree item
			if ( entry.name )
			{
				if ( !root_indent_offset_set )
				{
					root_indent_offset_set = true;
					root_indent_offset = indent;
					
					if ( root_indent_offset > 1 )
						qWarning("CHM has improper index; root indent offset is %d", root_indent_offset);
				}

				int real_indent = indent - root_indent_offset;
				
				entry.indent = real_indent;
				data->push_back( entry );
			}
			else
			{
				if ( !entry.urls.isEmpty() )
					qDebug ("LCHMFileImpl::ParseAndFillTopicsTree: <object> tag with url \"%s\" is parsed, but name is empty.", entry.urls[0].ascii());
				else
					qDebug ("LCHMFileImpl::ParseAndFillTopicsTree: <object> tag is parsed, but both name and url are empty.");	
			}

			entry.name = QString::null;
			entry.urls.clear();
			entry.imageid = defaultimagenum;
			in_object = false;
		}
		else if ( tagword == "param" && in_object )
		{
			// <param name="Name" value="First Page">
			int offset; // strlen("param ")
			QString name_pattern = "name=", value_pattern = "value=";
			QString pname, pvalue;

			if ( (offset = tag.find (name_pattern, 0, FALSE)) == -1 )
				qFatal ("LCHMFileImpl::ParseAndFillTopicsTree: bad <param> tag '%s': no name=\n", tag.ascii());

			// offset+5 skips 'name='
			offset = findStringInQuotes (tag, offset + name_pattern.length(), pname, TRUE, FALSE);
			pname = pname.lower();

			if ( (offset = tag.find (value_pattern, offset, FALSE)) == -1 )
				qFatal ("LCHMFileImpl::ParseAndFillTopicsTree: bad <param> tag '%s': no value=\n", tag.ascii());

			// offset+6 skips 'value='
			findStringInQuotes (tag, offset + value_pattern.length(), pvalue, FALSE, TRUE);

			//qDebug ("<param>: name '%s', value '%s'", pname.ascii(), pvalue.ascii());

			if ( pname == "name" )
			{
				// Some help files contain duplicate names, where the second name is empty. Work it around by keeping the first one
				if ( !pvalue.isEmpty() )
					entry.name = pvalue;
			}
			else if ( pname == "local" )
			{
				// Check for URL duplication
				QString url = LCHMUrlFactory::makeURLabsoluteIfNeeded( pvalue );
				
				if ( entry.urls.find( url ) == entry.urls.end() )
					entry.urls.push_back( url );
			}
			else if ( pname == "see also" && asIndex && entry.name != pvalue )
				entry.urls.push_back (":" + pvalue);
			else if ( pname == "imagenumber" )
			{
				bool bok;
				int imgnum = pvalue.toInt (&bok);
	
				if ( bok && imgnum >= 0 && imgnum < LCHMBookIcons::MAX_BUILTIN_ICONS )
					entry.imageid = imgnum;
			}
		}
		else if ( tagword == "ul" ) // increase indent level
		{
			// Fix for buggy help files		
			if ( ++indent >= MAX_NEST_DEPTH )
				qFatal("LCHMFileImpl::ParseAndFillTopicsTree: max nest depth (%d) is reached, error in help file", MAX_NEST_DEPTH);

			// This intended to fix <ul><ul>, which was seen in some buggy chm files,
			// and brokes rootentry[indent-1] check
		}
		else if ( tagword == "/ul" ) // decrease indent level
		{
			if ( --indent < root_indent_offset )
				indent = root_indent_offset;
			
			DEBUGPARSER(("</ul>: new intent is %d\n", indent - root_indent_offset));
		}

		pos = i;	
	}
	
	return true;
}


bool LCHMFileImpl::getFileContentAsBinary( QByteArray * data, const QString & url ) const
{
	chmUnitInfo ui;

	if( !ResolveObject( url, &ui ) )
		return false;

	data->resize( ui.length );
			
	if ( RetrieveObject( &ui, (unsigned char*) data->data(), 0, ui.length ) )
		return true;
	else
		return false;
}

	
bool LCHMFileImpl::getFileContentAsString( QString * str, const QString & url, bool internal_encoding )
{
	QByteArray buf;
	
	if ( getFileContentAsBinary( &buf, url ) )
	{
		unsigned int length = buf.size();
		
		if ( length > 0 )
		{
			buf.resize( length + 1 );
			buf [length] = '\0';
			
			*str = internal_encoding ? (QString)((const char*) buf) :  encodeWithCurrentCodec((const char*) buf);
			return true;
		}
	}
	
	return false;
}


QString LCHMFileImpl::getTopicByUrl( const QString & url ) const
{
	QMap< QString, QString >::const_iterator it = m_url2topics.find( url );
	
	if ( it == m_url2topics.end() )
		return QString::null;
	
	return it.data();
}


static int chm_enumerator_callback( struct chmFile*, struct chmUnitInfo *ui, void *context )
{
	((QStringList*) context)->push_back( ui->path );
	return CHM_ENUMERATOR_CONTINUE;
}

bool LCHMFileImpl::enumerateFiles( QStringList * files )
{
	files->clear();
	return chm_enumerate( m_chmFile, CHM_ENUMERATE_ALL, chm_enumerator_callback, files );
}

const QPixmap * LCHMFileImpl::getBookIconPixmap( unsigned int imagenum )
{
	return m_imagesKeeper.getImage( imagenum );
}

bool LCHMFileImpl::setCurrentEncoding( const LCHMTextEncoding * encoding )
{
	m_currentEncoding = encoding;
	return changeFileEncoding( encoding->qtcodec );
}


bool LCHMFileImpl::guessTextEncoding( )
{
	const LCHMTextEncoding * enc = 0;

	if ( !m_detectedLCID || (enc = lookupByLCID (m_detectedLCID)) == 0 )
		qFatal ("Could not detect text encoding by LCID");
	
	if ( changeFileEncoding (enc->qtcodec) )
	{
		m_currentEncoding = enc;
		return true;
	}
	
	return false;
}

bool LCHMFileImpl::changeFileEncoding( const char *qtencoding  )
{
	// Encoding could be either simple Qt codepage, or set like CP1251/KOI8, which allows to
	// set up encodings separately for text (first) and internal files (second)
	const char * p = strchr( qtencoding, '/' );
	if ( p )
	{
		char buf[128]; // much bigger that any encoding possible. No DoS; all encodings are hardcoded.
		strcpy( buf, qtencoding );
		buf[p - qtencoding] = '\0';
		
		m_textCodec = QTextCodec::codecForName( buf );
	
		if ( !m_textCodec )
		{
			qWarning( "Could not set up Text Codec for encoding '%s'", buf );
			return false;
		}
		
		m_textCodecForSpecialFiles = QTextCodec::codecForName( p + 1 );
	
		if ( !m_textCodecForSpecialFiles )
		{
			qWarning( "Could not set up Text Codec for encoding '%s'", p + 1 );
			return false;
		}
	}
	else
	{
		m_textCodecForSpecialFiles = m_textCodec = QTextCodec::codecForName (qtencoding);
	
		if ( !m_textCodec )
		{
			qWarning( "Could not set up Text Codec for encoding '%s'", qtencoding );
			return false;
		}
	}
	
	m_entityDecodeMap.clear();
	return true;
}


void LCHMFileImpl::fillTopicsUrlMap()
{
	if ( !m_lookupTablesValid )
		return;

	// Read those tables
	QByteArray topics( m_chmTOPICS.length ), urltbl( m_chmURLTBL.length ), urlstr( m_chmURLSTR.length ), strings( m_chmSTRINGS.length );

	if ( !RetrieveObject( &m_chmTOPICS, (unsigned char*) topics.data(), 0, m_chmTOPICS.length )
	|| !RetrieveObject( &m_chmURLTBL, (unsigned char*) urltbl.data(), 0, m_chmURLTBL.length )
	|| !RetrieveObject( &m_chmURLSTR, (unsigned char*) urlstr.data(), 0, m_chmURLSTR.length )
	|| !RetrieveObject( &m_chmSTRINGS, (unsigned char*) strings.data(), 0, m_chmSTRINGS.length ) )
		return;
	
	for ( unsigned int i = 0; i < m_chmTOPICS.length; i += TOPICS_ENTRY_LEN )
	{
		u_int32_t off_title = get_int32_le( (u_int32_t *)(topics.data() + i + 4) );
		u_int32_t off_url = get_int32_le( (u_int32_t *)(topics.data() + i + 8) );
		off_url = get_int32_le( (u_int32_t *)( urltbl.data() + off_url + 8) ) + 8;

		QString url = LCHMUrlFactory::makeURLabsoluteIfNeeded( (const char*) urlstr.data() + off_url );

		if ( off_title < strings.size() )
			m_url2topics[url] = encodeWithCurrentCodec ( (const char*) strings.data() + off_title );
		else
			m_url2topics[url] = "Untitled";
	}
}


bool LCHMFileImpl::getFileSize(unsigned int * size, const QString & url)
{
	chmUnitInfo ui;

	if( !ResolveObject( url, &ui ) )
		return false;

	*size = ui.length;
	return true;
}