xchmfile.cpp

okular

		else
			tagword = tag.toLower();

// qDebug ("tag: '%s', tagword: '%s'\n", tag.toAscii().constData(), tagword.toAscii().constData());
	 					
		// <OBJECT type="text/sitemap"> - a topic entry
		if ( tagword == "object" && tag.indexOf ("text/sitemap", 0, Qt::CaseInsensitive) != -1 )
			in_object = true;
		else if ( tagword == "/object" && in_object ) 
		{
			// a topic entry closed. Add a tree item
			if ( !name.isNull() )
			{
				QDomElement * item = new QDomElement();

				if ( !root_created )
					indent = 0;

				QString url = urls.join ("|");

				// Add item into the tree
				if ( !indent )
				{
					*item = tree->createElement(name);
					item->setAttribute("ViewportName",url);
					item->setAttribute("Icon",imagenum);
                    tree->appendChild(*item);
				}
				else
				{
					if ( !rootentry[indent-1] )
						qFatal("CHMFile::ParseAndFillTopicsTree: child entry %d with no root entry!", indent-1);

					*item = tree->createElement(name);
					item->setAttribute("ViewportName",url);
					item->setAttribute("Icon",imagenum);
                    rootentry[indent-1]->appendChild(*item);
				}

				lastchild[indent] = item;

				if ( indent == 0 || !rootentry[indent] )
				{
					rootentry[indent] = item;
					root_created = true;
				}

				// There are no 'titles' in index file
				if ( add2treemap  )
				{
					insertIntoUrlMaps(m_UrlPage,m_PageUrl,url,pnum);
				}
			}
			else
			{
				if ( !urls.isEmpty() )
					qDebug ("CHMFile::ParseAndFillTopicsTree: <object> tag with url \"%s\" is parsed, but name is empty.", urls[0].toAscii().constData());
				else
					qDebug ("CHMFile::ParseAndFillTopicsTree: <object> tag is parsed, but both name and url are empty.");	
			}

			name = QString::null;
			urls.clear();
			in_object = false;
			imagenum = defaultimagenum;
		}
		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.indexOf (name_pattern, 0, Qt::CaseInsensitive)) == -1 )
				qFatal ("CHMFile::ParseAndFillTopicsTree: bad <param> tag '%s': no name=\n", tag.toAscii().constData());

			// offset+5 skips 'name='
			offset = findStringInQuotes (tag, offset + name_pattern.length(), pname, true, false);
			pname = pname.toLower();

			if ( (offset = tag.indexOf (value_pattern, offset, Qt::CaseInsensitive)) == -1 )
				qFatal ("CHMFile::ParseAndFillTopicsTree: bad <param> tag '%s': no value=\n", tag.toAscii().constData());

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

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

			if ( pname == "name" )
				name = pvalue;
			else if ( pname == "local" )
				urls.push_back (KCHMUrl::makeURLabsoluteIfNeeded(pvalue));
			else if ( pname == "see also" && asIndex && name != pvalue )
				urls.push_back (KCHMUrl::makeURLabsoluteIfNeeded(":" + pvalue));
			else if ( pname == "imagenumber" )
			{
				bool bok;
				int imgnum = pvalue.toInt (&bok);
	
				if ( bok && imgnum >= 0 && (unsigned) imgnum < MAX_BUILTIN_ICONS )
					imagenum = imgnum;
			}
		}
		else if ( tagword == "ul" ) // increase indent level
		{
			// Fix for buggy help files		
			if ( ++indent >= MAX_NEST_DEPTH )
				qFatal("CHMFile::ParseAndFillTopicsTree: max nest depth (%d) is reached, error in help file", MAX_NEST_DEPTH);
				
			lastchild[indent] = 0;
			rootentry[indent] = 0;
		}
		else if ( tagword == "/ul" ) // decrease indent level
		{
			if ( --indent < 0 )
				indent = 0;
				
			rootentry[indent] = 0;
		}

		pos = i;	
	}
	
	return true;
}

bool CHMFile::ParseHhcAndFillTree (const QString& file, K3ListView *tree, bool asIndex)
{
	chmUnitInfo ui;
	const int MAX_NEST_DEPTH = 256;

	if(file.isEmpty() || !ResolveObject(file, &ui))
		return false;

	QString src;
	GetFileContentAsString(src, &ui);

	if(src.isEmpty())
		return false;

	unsigned int defaultimagenum = asIndex ? KCHMImageType::IMAGE_INDEX : KCHMImageType::IMAGE_AUTO;
	unsigned int imagenum = defaultimagenum;
	int pos = 0, indent = 0;
	bool in_object = false, root_created = false;
	QString name;
	QStringList urls;
	KCHMMainTreeViewItem * rootentry[MAX_NEST_DEPTH];
	KCHMMainTreeViewItem * lastchild[MAX_NEST_DEPTH];
	
	memset (lastchild, 0, sizeof(*lastchild));
	memset (rootentry, 0, sizeof(*rootentry));
	
	// Split the HHC file by HTML tags
	int stringlen = src.length();
	
	while ( pos < stringlen 
	&& (pos = src.indexOf ('<', 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.indexOf (src[i], i+1);
				if ( nextpos == -1 	&& (nextpos = src.indexOf ('>', i+1)) == -1 )
				{
					qWarning ("CHMFile::ParseHhcAndFillTree: corrupted TOC: %s", src.mid(i).toAscii().constData());
					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).toLower();
		else
			tagword = tag.toLower();

//qDebug ("tag: '%s', tagword: '%s'\n", tag.toAscii().constData(), tagword.toAscii().constData());
						
		// <OBJECT type="text/sitemap"> - a topic entry
		if ( tagword == "object" && tag.indexOf ("text/sitemap", 0, Qt::CaseInsensitive) != -1 )
			in_object = true;
		else if ( tagword == "/object" && in_object ) 
		{
			// a topic entry closed. Add a tree item
			if ( !name.isNull() )
			{
				KCHMMainTreeViewItem * item;

				if ( !root_created )
					indent = 0;

				QString url = urls.join ("|");

				// Add item into the tree
				if ( !indent )
				{
					item = new KCHMMainTreeViewItem (tree, lastchild[indent], name, url, imagenum);
				}
				else
				{
					if ( !rootentry[indent-1] )
						qFatal("CHMFile::ParseAndFillTopicsTree: child entry %d with no root entry!", indent-1);

					item = new KCHMMainTreeViewItem (rootentry[indent-1], lastchild[indent], name, url,  imagenum);
				}

				lastchild[indent] = item;

				if ( indent == 0 || !rootentry[indent] )
				{
					rootentry[indent] = item;
					root_created = true;

					if ( asIndex  )
						rootentry[indent]->setOpen(true);
				}
			}
			else
			{
				if ( !urls.isEmpty() )
					qDebug ("CHMFile::ParseAndFillTopicsTree: <object> tag with url \"%s\" is parsed, but name is empty.", urls[0].toAscii().constData());
				else
					qDebug ("CHMFile::ParseAndFillTopicsTree: <object> tag is parsed, but both name and url are empty.");	
			}

			name = QString::null;
			urls.clear();
			in_object = false;
			imagenum = defaultimagenum;
		}
		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.indexOf (name_pattern, 0, Qt::CaseInsensitive)) == -1 )
				qFatal ("CHMFile::ParseAndFillTopicsTree: bad <param> tag '%s': no name=\n", tag.toAscii().constData());

			// offset+5 skips 'name='
			offset = findStringInQuotes (tag, offset + name_pattern.length(), pname, true, false);
			pname = pname.toLower();

			if ( (offset = tag.indexOf (value_pattern, offset, Qt::CaseInsensitive)) == -1 )
				qFatal ("CHMFile::ParseAndFillTopicsTree: bad <param> tag '%s': no value=\n", tag.toAscii().constData());

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

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

			if ( pname == "name" )
				name = pvalue;
			else if ( pname == "local" )
				urls.push_back (KCHMUrl::makeURLabsoluteIfNeeded (pvalue));
			else if ( pname == "see also" && asIndex && name != pvalue )
				urls.push_back (":" + pvalue);
			else if ( pname == "imagenumber" )
			{
				bool bok;
				int imgnum = pvalue.toInt (&bok);
	
				if ( bok && imgnum >= 0 && (unsigned) imgnum < MAX_BUILTIN_ICONS )
					imagenum = imgnum;
			}
		}
		else if ( tagword == "ul" ) // increase indent level
		{
			// Fix for buggy help files		
			if ( ++indent >= MAX_NEST_DEPTH )
				qFatal("CHMFile::ParseAndFillTopicsTree: max nest depth (%d) is reached, error in help file", MAX_NEST_DEPTH);
				
			lastchild[indent] = 0;
			rootentry[indent] = 0;
		}
		else if ( tagword == "/ul" ) // decrease indent level
		{
			if ( --indent < 0 )
				indent = 0;
				
			rootentry[indent] = 0;
		}

		pos = i;	
	}
	
	return true;
}


bool CHMFile::ParseAndFillTopicsTree(QDomDocument *tree)
{
	return ParseHhcAndFillTree (m_topicsFile, tree, false);
}

bool CHMFile::ParseAndFillIndex(K3ListView *indexlist)
{
	return ParseHhcAndFillTree (m_indexFile, indexlist, true);
}

bool CHMFile::SearchWord (const QString& text, bool wholeWords, bool titlesOnly, KCHMSearchProgressResults_t& results,  bool phrase_search)
{
	bool partial = false;

	if ( text.isEmpty() || !m_searchAvailable )
		return false;

	QString searchword = (QString) convertSearchWord (text);

#define FTS_HEADER_LEN 0x32
	unsigned char header[FTS_HEADER_LEN];

	if ( RetrieveObject (&m_chmFIftiMain, header, 0, FTS_HEADER_LEN) == 0 )
		return false;
	
	unsigned char doc_index_s = header[0x1E], doc_index_r = header[0x1F];
	unsigned char code_count_s = header[0x20], code_count_r = header[0x21];
	unsigned char loc_codes_s = header[0x22], loc_codes_r = header[0x23];

	if(doc_index_s != 2 || code_count_s != 2 || loc_codes_s != 2)
	{
		// Don't know how to use values other than 2 yet. Maybe next chmspec.
		return false;
	}

	unsigned char* cursor32 = header + 0x14;
	u_int32_t node_offset = UINT32ARRAY(cursor32);

	cursor32 = header + 0x2e;
	u_int32_t node_len = UINT32ARRAY(cursor32);

	unsigned char* cursor16 = header + 0x18;
	u_int16_t tree_depth = UINT16ARRAY(cursor16);

	unsigned char word_len, pos;
	QString word;
	u_int32_t i = sizeof(u_int16_t);
	u_int16_t free_space;

	QVector<unsigned char> buffer(node_len);

	node_offset = GetLeafNodeOffset (searchword, node_offset, node_len, tree_depth);

	if ( !node_offset )
		return false;

	do
	{
		// got a leaf node here.
		if ( RetrieveObject (&m_chmFIftiMain, buffer.data(), node_offset, node_len) == 0 )
			return false;

		cursor16 = buffer.data() + 6;
		free_space = UINT16ARRAY(cursor16);

		i = sizeof(u_int32_t) + sizeof(u_int16_t) + sizeof(u_int16_t);
		u_int64_t wlc_count, wlc_size;
		u_int32_t wlc_offset;

		while (i < node_len - free_space)
		{
			word_len = *(buffer.data() + i);
			pos = *(buffer.data() + i + 1);

			char *wrd_buf = new char[word_len];
			memcpy (wrd_buf, buffer.data() + i + 2, word_len - 1);
			wrd_buf[word_len - 1] = 0;

			if ( pos == 0 )
				word = wrd_buf;
			else
				word = word.mid (0, pos) + wrd_buf;

			delete[] wrd_buf;

			i += 2 + word_len;
			unsigned char title = *(buffer.data() + i - 1);

			size_t encsz;
			wlc_count = be_encint (buffer.data() + i, encsz);
			i += encsz;
		
			cursor32 = buffer.data() + i;
			wlc_offset = UINT32ARRAY(cursor32);

			i += sizeof(u_int32_t) + sizeof(u_int16_t);
			wlc_size =  be_encint (buffer.data() + i, encsz);
			i += encsz;

			cursor32 = buffer.data();
			node_offset = UINT32ARRAY(cursor32);
		
			if ( !title && titlesOnly )