Function makeTeaser

docs/searcher.js:159–253

(body, searchterms)

    }
    
    function makeTeaser(body, searchterms) {
        // The strategy is as follows:
        // First, assign a value to each word in the document:
        //  Words that correspond to search terms (stemmer aware): 40
        //  Normal words: 2
        //  First word in a sentence: 8
        // Then use a sliding window with a constant number of words and count the
        // sum of the values of the words within the window. Then use the window that got the
        // maximum sum. If there are multiple maximas, then get the last one.
        // Enclose the terms in <em>.
        var stemmed_searchterms = searchterms.map(function(w) {
            return elasticlunr.stemmer(w.toLowerCase());
        });
        var searchterm_weight = 40;
        var weighted = []; // contains elements of ["word", weight, index_in_document]
        // split in sentences, then words
        var sentences = body.toLowerCase().split('. ');
        var index = 0;
        var value = 0;
        var searchterm_found = false;
        for (var sentenceindex in sentences) {
            var words = sentences[sentenceindex].split(' ');
            value = 8;
            for (var wordindex in words) {
                var word = words[wordindex];
                if (word.length > 0) {
                    for (var searchtermindex in stemmed_searchterms) {
                        if (elasticlunr.stemmer(word).startsWith(stemmed_searchterms[searchtermindex])) {
                            value = searchterm_weight;
                            searchterm_found = true;
                        }
                    };
                    weighted.push([word, value, index]);
                    value = 2;
                }
                index += word.length;
                index += 1; // ' ' or '.' if last word in sentence
            };
            index += 1; // because we split at a two-char boundary '. '
        };

        if (weighted.length == 0) {
            return body;
        }

        var window_weight = [];
        var window_size = Math.min(weighted.length, results_options.teaser_word_count);

        var cur_sum = 0;
        for (var wordindex = 0; wordindex < window_size; wordindex++) {
            cur_sum += weighted[wordindex][1];
        };
        window_weight.push(cur_sum);
        for (var wordindex = 0; wordindex < weighted.length - window_size; wordindex++) {
            cur_sum -= weighted[wordindex][1];
            cur_sum += weighted[wordindex + window_size][1];
            window_weight.push(cur_sum);
        };