C++ Words::getTagIds方法代码示例

//.........这里部分代码省略.........

	// * remove all imgs. just src them to dev null.

	// * allow for entering a custom title for an event or all events
	//   that are or will ever appear on the page. 

	// * when displaying the text of the events, use hyphens to
	//   delineate the section topology. strike out text as a section
	//   fence is activated.

	// * when a section is activated is it easier to just redownload
	//   the whole text of the page? maybe just the text frame?

	// * clicking on an individual sentence section should just remove
	//   that sentence. that is kinda a special content hash removal
	//   tag. like "Click here for video."

	// * when an event id is selected i guess activate its bgcolor to
	//   be light blue for all sentences currently in the event that
	//   are not in activated sections. (make exception for designated 
	//   title sections). so we need multiple tags for each events
	//   sentence div section. if sentence is split use multiple div tags
	//   then to keep the order. so each event sentence would have 
	//   <div ev1=1 ev2=1 ev10=1>...</div> if it is in event ids 1,2 and
	//   10. that way we can activate it when one of those event ids is
	//   activated.


	SafeBuf sb;

	// int16_tcuts
	if ( ! xd->m_wordsValid ) { char *xx=NULL;*xx=0; }
	Words     *words = &xd->m_words;
	int32_t       nw    = words->getNumWords();
	char     **wptrs = words->getWords();
	int32_t      *wlens = words->getWordLens();
	nodeid_t  *tids  = words->getTagIds();

	// a special array for printing </div> tags
	char *endCounts = (char *)mcalloc ( nw ,"endcounts");
	if ( ! endCounts ) return sendErrorReply ( st , g_errno );


	// 
	// now loop over all the words. if word starts a section that has
	// SEC_CONTROL bit set, and print out the section hash and a color
	// tag to be activated if the turkey activates us.
	// CAUTION: word may start multiple sections.
	//
	for ( int32_t i = 0 ; i < nw ; i++ ) { 
		// get section ptr
		Section *sj = ss->m_sectionPtrs[i];
		// sanity check. sj must be first section ptr that starts @ a
		if ( sj && sj->m_a==i && sj->m_prev && sj->m_prev->m_a==i ) {
			char *xx=NULL;*xx=0; }
		// . does word #i start a section?
		// . if section is control, print out the control
		while ( sj && sj->m_a == i ) {
			// print this section's hash
			if ( sj->m_flags & SEC_CONTROL) {
				// after the turkeys have made all the edits
				// they need to submit the changes they made.
				// how can we get that data sent back to the
				// back end? we need to send back the colors
				// of the sections that have been activated
				// i guess. just do a loop over them.
				sb.safePrintf("<div nobreak gbsecid=%"UINT32" "
					      "bgcolor=#%"XINT32" "
					      "onclick=gbtogglecolor()>",
					      (uint32_t)sj->m_tagHash,
					      (uint32_t)sj->m_tagHash);
				// sanity check
				if ( sj->m_b < 0  ) { char *xx=NULL;*xx=0; }
				if ( sj->m_b > nw ) { char *xx=NULL;*xx=0; }
				// and inc the /div count for that word
				endCounts[sj->m_b-1]++;
			}
			// try next section too
			sj = sj->m_next;
		}
		// if this is a tag, remove any coloring
		if ( tids[i] ) {
		}
		// print the word, be it a tag, alnum, punct
		sb.safeMemcpy ( wptrs[i] , wlens[i] );
		// end a div tag?
		if ( ! endCounts[i] ) continue;
		// might be many so loop it
		for ( int32_t j = 0 ; j < endCounts[i] ; j++ )
			sb.safePrintf("</div>");
	}			







	return false;
}

// . return the score of the highest-scoring window containing match #m
// . window is defined by the half-open interval [a,b) where a and b are 
//   word #'s in the Words array indicated by match #m
// . return -1 and set g_errno on error
int64_t Summary::getBestWindow ( Matches *matches, int32_t mm, int32_t *lasta,
                                 int32_t *besta, int32_t *bestb, char *gotIt,
                                 char *retired, int32_t maxExcerptLen ) {
	// get the window around match #mm
	Match *m = &matches->m_matches[mm];

	// what is the word # of match #mm?
	int32_t matchWordNum = m->m_wordNum;

	// what Words/Pos/Bits classes is this match in?
	Words *words = m->m_words;
	Section **sp = NULL;
	int32_t *pos = m->m_pos->m_pos;

	// use "m_swbits" not "m_bits", that is what Bits::setForSummary() uses
	const swbit_t *bb = m->m_bits->m_swbits;

	// shortcut
	if ( m->m_sections ) {
		sp = m->m_sections->m_sectionPtrs;
	}

	int32_t nw = words->getNumWords();
	int64_t *wids = words->getWordIds();
	nodeid_t *tids = words->getTagIds();

	// . sanity check
	// . this prevents a core i've seen
	if ( matchWordNum >= nw ) {
		log("summary: got overflow condition for q=%s",m_q->m_orig);

		// assume no best window
		*besta = -1;
		*bestb = -1;
		*lasta = matchWordNum;
		return 0;
	}

	// . we NULLify the section ptrs if we already used the word in another summary.
	int32_t badFlags = SEC_SCRIPT|SEC_STYLE|SEC_SELECT|SEC_IN_TITLE;
	if ( (bb[matchWordNum] & D_USED) || ( sp && (sp[matchWordNum]->m_flags & badFlags) ) ) {
		// assume no best window
		*besta = -1;
		*bestb = -1;
		*lasta = matchWordNum;
		return 0;
	}

	// . "a" is the left fence post of the window (it is a word # in Words)
	// . go to the left as far as we can 
	// . thus we decrement "a"
	int32_t a = matchWordNum;

	// "posa" is the character position of the END of word #a
	int32_t posa = pos[a+1];
	int32_t firstFrag = -1;
	bool startOnQuote = false;
	bool goodStart = false;
	int32_t wordCount = 0;

	// . decrease "a" as int32_t as we stay within maxNumCharsPerLine
	// . avoid duplicating windows by using "lasta", the last "a" of the
	//   previous call to getBestWindow(). This can happen if our last
	//   central query term was close to this one.
	for ( ; a > 0 && posa - pos[a-1] < maxExcerptLen && a > *lasta; a-- ) {
		// . don't include any "dead zone", 
		// . dead zones have already been used for the summary, and
		//   we are getting a second/third/... excerpt here now then
		// stop if its the start of a sentence, too
		// stop before title word
		if ( (bb[a-1] & D_USED) || (bb[a] & D_STARTS_SENTENCE) || ( bb[a-1] & D_IN_TITLE )) {
			goodStart = true;
			break;
		}

		// don't go beyond an LI, TR, P tag
		if ( tids && ( tids[a-1] == TAG_LI ||
		               tids[a-1] == TAG_TR ||
		               tids[a-1] == TAG_P  ||
		               tids[a-1] == TAG_DIV ) ) {
			goodStart = true;
			break;
		}

		// stop if its the start of a quoted sentence
		if ( a+1<nw && (bb[a+1] & D_IN_QUOTES) && 
		     words->getWord(a)[0] == '\"' ){
			startOnQuote = true;
			goodStart    = true;
			break;
		}

		// find out the first instance of a fragment (comma, etc)
		// watch out! because frag also means 's' in there's
		if ( ( bb[a] & D_STARTS_FRAG ) && !(bb[a-1] & D_IS_STRONG_CONNECTOR) && firstFrag == -1 ) {
			firstFrag = a;
//.........这里部分代码省略.........