C++ Matches::addMatches方法代码示例

//.........这里部分代码省略.........
	const char *rootTitleBuf    = NULL;
	const char *rootTitleBufEnd = NULL;

	// get the root title if we are not root!
	if (filteredRootTitleBuf) {
#ifdef _VALGRIND_
		VALGRIND_CHECK_MEM_IS_DEFINED(filteredRootTitleBuf,filteredRootTitleBufSize);
#endif
		// point to list of \0 separated titles
		rootTitleBuf    = filteredRootTitleBuf;
		rootTitleBufEnd =  filteredRootTitleBuf + filteredRootTitleBufSize;
	}

	{
		Matches m;
		if ( rootTitleBuf && query ) {
			m.setQuery ( query );
		}

		// convert into an array
		int32_t nr = 0;
		const char *pr = rootTitleBuf;
		const char *rootTitles[20];
		int32_t  rootTitleLens[20];

		// loop over each root title segment
		for ( ; pr && pr < rootTitleBufEnd ; pr += strnlen(pr,rootTitleBufEnd-pr) + 1 ) {
			// if we had a query...
			if ( query ) {
				// reset it
				m.reset();

				// see if root title segment has query terms in it
				m.addMatches ( const_cast<char*>(pr), strnlen(pr,rootTitleBufEnd-pr), MF_TITLEGEN, m_niceness );

				// if matches query, do NOT add it, we only add it for
				// removing from the title of the page...
				if ( m.getNumMatches() ) {
					continue;
				}
			}
			// point to it. it should start with an alnum already
			// since it is the "filtered" list of root titles...
			// if not, fix it in xmldoc then.
			rootTitles   [nr] = pr;
			rootTitleLens[nr] = gbstrlen(pr);
			// advance
			nr++;
			// no breaching
			if ( nr >= 20 ) break;
		}

		// now split up candidates in children candidates by tokenizing
		// using :, | and - as delimters.
		// the hyphen must have a space on at least one side, so "cd-rom" does
		// not create a pair of tokens...
		// FIX: for the title:
		// Best Careers 2009: Librarian - US News and World Report
		// we need to recognize "Best Careers 2009: Librarian" as a subtitle
		// otherwise we don't get it as the title. so my question is are we
		// going to have to do all the permutations at some point? for now
		// let's just add in pairs...
		for ( int32_t i = 0 ; i < oldn && n + 3 < MAX_TIT_CANDIDATES ; i++ ) {
			// stop if no root title segments
			if ( nr <= 0 ) break;
			// get the word info

//.........这里部分代码省略.........
	//sprintf ( x, "&seq=%li&rtq=%lid=%lli",
	//	  (long)st->m_seq,(long)st->m_rtq,st->m_msg22.getDocId());
	sprintf ( x, "&d=%lli",st->m_docId );
	x += gbstrlen(x);		
	// set our query for highlighting
	Query qq;
	qq.set2 ( q, st->m_langId , true );

	// print the query terms into our highlight buffer
	Highlight hi;
	// make words so we can set the scores to ignore fielded terms
	Words qw;
	qw.set ( q            ,  // content being highlighted, utf8
		 qlen         ,  // content being highlighted, utf8
		 TITLEREC_CURRENT_VERSION,
		 true         ,  // computeIds
		 false        ); // hasHtmlEntities?
	// . assign scores of 0 to query words that should be ignored
	// . TRICKY: loop over words in qq.m_qwords, but they should be 1-1
	//   with words in qw.
	// . sanity check
	//if ( qw.getNumWords() != qq.m_numWords ) { char *xx = NULL; *xx = 0;}
	// declare up here
	Matches m;
	// do the loop
	//Scores ss;
	//ss.set ( &qw , NULL );
	//for ( long i = 0 ; i < qq.m_numWords ; i++ )
	//	if ( ! m.matchWord ( &qq.m_qwords[i],i ) ) ss.m_scores[i] = 0;
	// now set m.m_matches[] to those words in qw that match a query word
	// or phrase in qq.
	m.setQuery ( &qq );
	//m.addMatches ( &qw , &ss , true );
	m.addMatches ( &qw );
	long hilen = 0;

	// CNS: if ( ! st->m_clickNScroll ) {
	// and highlight the matches
	if ( printDisclaimer ) {
		hilen = hi.set ( //p       ,
				 //avail   ,
				sb ,
				 &qw     , // words to highlight
				 &m      , // matches relative to qw
				 false   , // doSteming
				 false   , // st->m_clickAndScroll , 
				 (char *)thisUrl );// base url for ClcknScrll
		//p += hilen;
		// now an hr
		//memcpy ( p , "</span></table></table>\n" , 24 );   p += 24;
		sb->safeStrcpy("</span></table></table>\n");
	}


	bool includeHeader = st->m_includeHeader;

	// do not show header for json object display
	if ( xd->m_contentType == CT_JSON )
		includeHeader = false;

	if ( format == FORMAT_XML ) includeHeader = false;
	if ( format == FORMAT_JSON ) includeHeader = false;

	//mfree(uq, uqCapacity, "PageGet");
	// undo the header writes if we should
	if ( ! includeHeader ) {

// . return length stored into "buf"
// . content must be NULL terminated
// . if "useAnchors" is true we do click and scroll
// . if "isQueryTerms" is true, we do typical anchors in a special way
int32_t Highlight::set ( SafeBuf *sb,
		      //char        *buf          ,
		      //int32_t         bufLen       ,
		      char        *content      ,
		      int32_t         contentLen   ,
		      // primary language of the document (for synonyms)
		      char         docLangId    ,
		      Query       *q            ,
		      bool         doStemming   ,
		      bool         useAnchors   ,
		      const char  *baseUrl      ,
		      const char  *frontTag     ,
		      const char  *backTag      ,
		      int32_t         fieldCode    ,
		      int32_t         niceness      ) {

	Words words;
	if ( ! words.set ( content      , 
			   contentLen   , 
			   TITLEREC_CURRENT_VERSION,
			   true         , // computeId
			   true         ) ) // has html entites?
		return -1;

	int32_t version = TITLEREC_CURRENT_VERSION;

	Bits bits;
	if ( ! bits.set (&words,version,niceness) ) return -1;

	Phrases phrases;
	if ( !phrases.set(&words,&bits,true,false,version,niceness))return -1;

	//SafeBuf langBuf;
	//if ( !setLangVec ( &words , &langBuf , niceness )) return 0;
	//uint8_t *langVec = (uint8_t *)langBuf.getBufStart();

	// make synonyms
	//Synonyms syns;
	//if(!syns.set(&words,NULL,docLangId,&phrases,niceness,NULL)) return 0;

	Matches matches;
	matches.setQuery ( q );

	if ( ! matches.addMatches ( &words , &phrases ) ) return -1;

	// store
	m_numMatches = matches.getNumMatches();

	return set ( sb , 
		     //buf         ,
		     //bufLen      , 
		     &words      ,
		     &matches    ,
		     doStemming  ,
		     useAnchors  ,
		     baseUrl     ,
		     frontTag    ,
		     backTag     ,
		     fieldCode   ,
		     q		 );
}