Search in sources :

Example 1 with SimpleSpanFragmenter

use of org.apache.lucene.search.highlight.SimpleSpanFragmenter in project elasticsearch by elastic.

the class PlainHighlighter method highlight.

@Override
public HighlightField highlight(HighlighterContext highlighterContext) {
    SearchContextHighlight.Field field = highlighterContext.field;
    SearchContext context = highlighterContext.context;
    FetchSubPhase.HitContext hitContext = highlighterContext.hitContext;
    FieldMapper mapper = highlighterContext.mapper;
    Encoder encoder = field.fieldOptions().encoder().equals("html") ? HighlightUtils.Encoders.HTML : HighlightUtils.Encoders.DEFAULT;
    if (!hitContext.cache().containsKey(CACHE_KEY)) {
        Map<FieldMapper, org.apache.lucene.search.highlight.Highlighter> mappers = new HashMap<>();
        hitContext.cache().put(CACHE_KEY, mappers);
    }
    @SuppressWarnings("unchecked") Map<FieldMapper, org.apache.lucene.search.highlight.Highlighter> cache = (Map<FieldMapper, org.apache.lucene.search.highlight.Highlighter>) hitContext.cache().get(CACHE_KEY);
    org.apache.lucene.search.highlight.Highlighter entry = cache.get(mapper);
    if (entry == null) {
        QueryScorer queryScorer = new CustomQueryScorer(highlighterContext.query, field.fieldOptions().requireFieldMatch() ? mapper.fieldType().name() : null);
        queryScorer.setExpandMultiTermQuery(true);
        Fragmenter fragmenter;
        if (field.fieldOptions().numberOfFragments() == 0) {
            fragmenter = new NullFragmenter();
        } else if (field.fieldOptions().fragmenter() == null) {
            fragmenter = new SimpleSpanFragmenter(queryScorer, field.fieldOptions().fragmentCharSize());
        } else if ("simple".equals(field.fieldOptions().fragmenter())) {
            fragmenter = new SimpleFragmenter(field.fieldOptions().fragmentCharSize());
        } else if ("span".equals(field.fieldOptions().fragmenter())) {
            fragmenter = new SimpleSpanFragmenter(queryScorer, field.fieldOptions().fragmentCharSize());
        } else {
            throw new IllegalArgumentException("unknown fragmenter option [" + field.fieldOptions().fragmenter() + "] for the field [" + highlighterContext.fieldName + "]");
        }
        Formatter formatter = new SimpleHTMLFormatter(field.fieldOptions().preTags()[0], field.fieldOptions().postTags()[0]);
        entry = new org.apache.lucene.search.highlight.Highlighter(formatter, encoder, queryScorer);
        entry.setTextFragmenter(fragmenter);
        // always highlight across all data
        entry.setMaxDocCharsToAnalyze(Integer.MAX_VALUE);
        cache.put(mapper, entry);
    }
    // a HACK to make highlighter do highlighting, even though its using the single frag list builder
    int numberOfFragments = field.fieldOptions().numberOfFragments() == 0 ? 1 : field.fieldOptions().numberOfFragments();
    ArrayList<TextFragment> fragsList = new ArrayList<>();
    List<Object> textsToHighlight;
    Analyzer analyzer = context.mapperService().documentMapper(hitContext.hit().getType()).mappers().indexAnalyzer();
    try {
        textsToHighlight = HighlightUtils.loadFieldValues(field, mapper, context, hitContext);
        for (Object textToHighlight : textsToHighlight) {
            String text;
            if (textToHighlight instanceof BytesRef) {
                text = mapper.fieldType().valueForDisplay(textToHighlight).toString();
            } else {
                text = textToHighlight.toString();
            }
            try (TokenStream tokenStream = analyzer.tokenStream(mapper.fieldType().name(), text)) {
                if (!tokenStream.hasAttribute(CharTermAttribute.class) || !tokenStream.hasAttribute(OffsetAttribute.class)) {
                    // can't perform highlighting if the stream has no terms (binary token stream) or no offsets
                    continue;
                }
                TextFragment[] bestTextFragments = entry.getBestTextFragments(tokenStream, text, false, numberOfFragments);
                for (TextFragment bestTextFragment : bestTextFragments) {
                    if (bestTextFragment != null && bestTextFragment.getScore() > 0) {
                        fragsList.add(bestTextFragment);
                    }
                }
            }
        }
    } catch (Exception e) {
        if (ExceptionsHelper.unwrap(e, BytesRefHash.MaxBytesLengthExceededException.class) != null) {
            // the plain highlighter will parse the source and try to analyze it.
            return null;
        } else {
            throw new FetchPhaseExecutionException(context, "Failed to highlight field [" + highlighterContext.fieldName + "]", e);
        }
    }
    if (field.fieldOptions().scoreOrdered()) {
        CollectionUtil.introSort(fragsList, new Comparator<TextFragment>() {

            @Override
            public int compare(TextFragment o1, TextFragment o2) {
                return Math.round(o2.getScore() - o1.getScore());
            }
        });
    }
    String[] fragments;
    // number_of_fragments is set to 0 but we have a multivalued field
    if (field.fieldOptions().numberOfFragments() == 0 && textsToHighlight.size() > 1 && fragsList.size() > 0) {
        fragments = new String[fragsList.size()];
        for (int i = 0; i < fragsList.size(); i++) {
            fragments[i] = fragsList.get(i).toString();
        }
    } else {
        // refine numberOfFragments if needed
        numberOfFragments = fragsList.size() < numberOfFragments ? fragsList.size() : numberOfFragments;
        fragments = new String[numberOfFragments];
        for (int i = 0; i < fragments.length; i++) {
            fragments[i] = fragsList.get(i).toString();
        }
    }
    if (fragments.length > 0) {
        return new HighlightField(highlighterContext.fieldName, Text.convertFromStringArray(fragments));
    }
    int noMatchSize = highlighterContext.field.fieldOptions().noMatchSize();
    if (noMatchSize > 0 && textsToHighlight.size() > 0) {
        // Pull an excerpt from the beginning of the string but make sure to split the string on a term boundary.
        String fieldContents = textsToHighlight.get(0).toString();
        int end;
        try {
            end = findGoodEndForNoHighlightExcerpt(noMatchSize, analyzer, mapper.fieldType().name(), fieldContents);
        } catch (Exception e) {
            throw new FetchPhaseExecutionException(context, "Failed to highlight field [" + highlighterContext.fieldName + "]", e);
        }
        if (end > 0) {
            return new HighlightField(highlighterContext.fieldName, new Text[] { new Text(fieldContents.substring(0, end)) });
        }
    }
    return null;
}
Also used : TokenStream(org.apache.lucene.analysis.TokenStream) HashMap(java.util.HashMap) Formatter(org.apache.lucene.search.highlight.Formatter) SimpleHTMLFormatter(org.apache.lucene.search.highlight.SimpleHTMLFormatter) ArrayList(java.util.ArrayList) SearchContext(org.elasticsearch.search.internal.SearchContext) TextFragment(org.apache.lucene.search.highlight.TextFragment) Analyzer(org.apache.lucene.analysis.Analyzer) SimpleFragmenter(org.apache.lucene.search.highlight.SimpleFragmenter) Encoder(org.apache.lucene.search.highlight.Encoder) SimpleFragmenter(org.apache.lucene.search.highlight.SimpleFragmenter) Fragmenter(org.apache.lucene.search.highlight.Fragmenter) SimpleSpanFragmenter(org.apache.lucene.search.highlight.SimpleSpanFragmenter) NullFragmenter(org.apache.lucene.search.highlight.NullFragmenter) FetchSubPhase(org.elasticsearch.search.fetch.FetchSubPhase) BytesRefHash(org.apache.lucene.util.BytesRefHash) BytesRef(org.apache.lucene.util.BytesRef) SimpleSpanFragmenter(org.apache.lucene.search.highlight.SimpleSpanFragmenter) QueryScorer(org.apache.lucene.search.highlight.QueryScorer) Text(org.elasticsearch.common.text.Text) NullFragmenter(org.apache.lucene.search.highlight.NullFragmenter) FetchPhaseExecutionException(org.elasticsearch.search.fetch.FetchPhaseExecutionException) IOException(java.io.IOException) FetchPhaseExecutionException(org.elasticsearch.search.fetch.FetchPhaseExecutionException) SimpleHTMLFormatter(org.apache.lucene.search.highlight.SimpleHTMLFormatter) FieldMapper(org.elasticsearch.index.mapper.FieldMapper) HashMap(java.util.HashMap) Map(java.util.Map)

Example 2 with SimpleSpanFragmenter

use of org.apache.lucene.search.highlight.SimpleSpanFragmenter in project gitblit by gitblit.

the class LuceneService method getHighlightedFragment.

/**
	 *
	 * @param analyzer
	 * @param query
	 * @param content
	 * @param result
	 * @return
	 * @throws IOException
	 * @throws InvalidTokenOffsetsException
	 */
private String getHighlightedFragment(Analyzer analyzer, Query query, String content, SearchResult result) throws IOException, InvalidTokenOffsetsException {
    if (content == null) {
        content = "";
    }
    int tabLength = storedSettings.getInteger(Keys.web.tabLength, 4);
    int fragmentLength = SearchObjectType.commit == result.type ? 512 : 150;
    QueryScorer scorer = new QueryScorer(query, "content");
    Fragmenter fragmenter = new SimpleSpanFragmenter(scorer, fragmentLength);
    // use an artificial delimiter for the token
    String termTag = "!!--[";
    String termTagEnd = "]--!!";
    SimpleHTMLFormatter formatter = new SimpleHTMLFormatter(termTag, termTagEnd);
    Highlighter highlighter = new Highlighter(formatter, scorer);
    highlighter.setTextFragmenter(fragmenter);
    String[] fragments = highlighter.getBestFragments(analyzer, "content", content, 3);
    if (ArrayUtils.isEmpty(fragments)) {
        if (SearchObjectType.blob == result.type) {
            return "";
        }
        // clip commit message
        String fragment = content;
        if (fragment.length() > fragmentLength) {
            fragment = fragment.substring(0, fragmentLength) + "...";
        }
        return "<pre class=\"text\">" + StringUtils.escapeForHtml(fragment, true, tabLength) + "</pre>";
    }
    // make sure we have unique fragments
    Set<String> uniqueFragments = new LinkedHashSet<String>();
    for (String fragment : fragments) {
        uniqueFragments.add(fragment);
    }
    fragments = uniqueFragments.toArray(new String[uniqueFragments.size()]);
    StringBuilder sb = new StringBuilder();
    for (int i = 0, len = fragments.length; i < len; i++) {
        String fragment = fragments[i];
        String tag = "<pre class=\"text\">";
        // resurrect the raw fragment from removing the artificial delimiters
        String raw = fragment.replace(termTag, "").replace(termTagEnd, "");
        // determine position of the raw fragment in the content
        int pos = content.indexOf(raw);
        // restore complete first line of fragment
        int c = pos;
        while (c > 0) {
            c--;
            if (content.charAt(c) == '\n') {
                break;
            }
        }
        if (c > 0) {
            // inject leading chunk of first fragment line
            fragment = content.substring(c + 1, pos) + fragment;
        }
        if (SearchObjectType.blob == result.type) {
            // count lines as offset into the content for this fragment
            int line = Math.max(1, StringUtils.countLines(content.substring(0, pos)));
            // create fragment tag with line number and language
            String lang = "";
            String ext = StringUtils.getFileExtension(result.path).toLowerCase();
            if (!StringUtils.isEmpty(ext)) {
                // maintain leading space!
                lang = " lang-" + ext;
            }
            tag = MessageFormat.format("<pre class=\"prettyprint linenums:{0,number,0}{1}\">", line, lang);
        }
        sb.append(tag);
        // replace the artificial delimiter with html tags
        String html = StringUtils.escapeForHtml(fragment, false);
        html = html.replace(termTag, "<span class=\"highlight\">").replace(termTagEnd, "</span>");
        sb.append(html);
        sb.append("</pre>");
        if (i < len - 1) {
            sb.append("<span class=\"ellipses\">...</span><br/>");
        }
    }
    return sb.toString();
}
Also used : LinkedHashSet(java.util.LinkedHashSet) SimpleSpanFragmenter(org.apache.lucene.search.highlight.SimpleSpanFragmenter) QueryScorer(org.apache.lucene.search.highlight.QueryScorer) Fragmenter(org.apache.lucene.search.highlight.Fragmenter) SimpleSpanFragmenter(org.apache.lucene.search.highlight.SimpleSpanFragmenter) SimpleHTMLFormatter(org.apache.lucene.search.highlight.SimpleHTMLFormatter) Highlighter(org.apache.lucene.search.highlight.Highlighter)

Aggregations

Fragmenter (org.apache.lucene.search.highlight.Fragmenter)2 QueryScorer (org.apache.lucene.search.highlight.QueryScorer)2 SimpleHTMLFormatter (org.apache.lucene.search.highlight.SimpleHTMLFormatter)2 SimpleSpanFragmenter (org.apache.lucene.search.highlight.SimpleSpanFragmenter)2 IOException (java.io.IOException)1 ArrayList (java.util.ArrayList)1 HashMap (java.util.HashMap)1 LinkedHashSet (java.util.LinkedHashSet)1 Map (java.util.Map)1 Analyzer (org.apache.lucene.analysis.Analyzer)1 TokenStream (org.apache.lucene.analysis.TokenStream)1 Encoder (org.apache.lucene.search.highlight.Encoder)1 Formatter (org.apache.lucene.search.highlight.Formatter)1 Highlighter (org.apache.lucene.search.highlight.Highlighter)1 NullFragmenter (org.apache.lucene.search.highlight.NullFragmenter)1 SimpleFragmenter (org.apache.lucene.search.highlight.SimpleFragmenter)1 TextFragment (org.apache.lucene.search.highlight.TextFragment)1 BytesRef (org.apache.lucene.util.BytesRef)1 BytesRefHash (org.apache.lucene.util.BytesRefHash)1 Text (org.elasticsearch.common.text.Text)1