Examples with Snippet org.apache.lucene.search.highlight.Snippet used on opensource projects

Example 1 with Snippet

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

the class PostingsHighlighter method highlight.

@Override
public HighlightField highlight(HighlighterContext highlighterContext) {
    FieldMapper fieldMapper = highlighterContext.mapper;
    SearchContextHighlight.Field field = highlighterContext.field;
    if (canHighlight(fieldMapper) == false) {
        throw new IllegalArgumentException("the field [" + highlighterContext.fieldName + "] should be indexed with positions and offsets in the postings list to be used with postings highlighter");
    }
    SearchContext context = highlighterContext.context;
    FetchSubPhase.HitContext hitContext = highlighterContext.hitContext;
    if (!hitContext.cache().containsKey(CACHE_KEY)) {
        hitContext.cache().put(CACHE_KEY, new HighlighterEntry());
    }
    HighlighterEntry highlighterEntry = (HighlighterEntry) hitContext.cache().get(CACHE_KEY);
    MapperHighlighterEntry mapperHighlighterEntry = highlighterEntry.mappers.get(fieldMapper);
    if (mapperHighlighterEntry == null) {
        Encoder encoder = field.fieldOptions().encoder().equals("html") ? Encoders.HTML : Encoders.DEFAULT;
        CustomPassageFormatter passageFormatter = new CustomPassageFormatter(field.fieldOptions().preTags()[0], field.fieldOptions().postTags()[0], encoder);
        mapperHighlighterEntry = new MapperHighlighterEntry(passageFormatter);
    }
    List<Snippet> snippets = new ArrayList<>();
    int numberOfFragments;
    try {
        Analyzer analyzer = context.mapperService().documentMapper(hitContext.hit().getType()).mappers().indexAnalyzer();
        List<Object> fieldValues = HighlightUtils.loadFieldValues(field, fieldMapper, context, hitContext);
        CustomPostingsHighlighter highlighter;
        if (field.fieldOptions().numberOfFragments() == 0) {
            //we use a control char to separate values, which is the only char that the custom break iterator breaks the text on,
            //so we don't lose the distinction between the different values of a field and we get back a snippet per value
            String fieldValue = mergeFieldValues(fieldValues, HighlightUtils.NULL_SEPARATOR);
            CustomSeparatorBreakIterator breakIterator = new CustomSeparatorBreakIterator(HighlightUtils.NULL_SEPARATOR);
            highlighter = new CustomPostingsHighlighter(analyzer, mapperHighlighterEntry.passageFormatter, breakIterator, fieldValue, field.fieldOptions().noMatchSize() > 0);
            //we are highlighting the whole content, one snippet per value
            numberOfFragments = fieldValues.size();
        } else {
            //using paragraph separator we make sure that each field value holds a discrete passage for highlighting
            String fieldValue = mergeFieldValues(fieldValues, HighlightUtils.PARAGRAPH_SEPARATOR);
            highlighter = new CustomPostingsHighlighter(analyzer, mapperHighlighterEntry.passageFormatter, fieldValue, field.fieldOptions().noMatchSize() > 0);
            numberOfFragments = field.fieldOptions().numberOfFragments();
        }
        IndexSearcher searcher = new IndexSearcher(hitContext.reader());
        Snippet[] fieldSnippets = highlighter.highlightField(fieldMapper.fieldType().name(), highlighterContext.query, searcher, hitContext.docId(), numberOfFragments);
        for (Snippet fieldSnippet : fieldSnippets) {
            if (Strings.hasText(fieldSnippet.getText())) {
                snippets.add(fieldSnippet);
            }
        }
    } catch (IOException e) {
        throw new FetchPhaseExecutionException(context, "Failed to highlight field [" + highlighterContext.fieldName + "]", e);
    }
    snippets = filterSnippets(snippets, field.fieldOptions().numberOfFragments());
    if (field.fieldOptions().scoreOrdered()) {
        //let's sort the snippets by score if needed
        CollectionUtil.introSort(snippets, new Comparator<Snippet>() {

            @Override
            public int compare(Snippet o1, Snippet o2) {
                return (int) Math.signum(o2.getScore() - o1.getScore());
            }
        });
    }
    String[] fragments = new String[snippets.size()];
    for (int i = 0; i < fragments.length; i++) {
        fragments[i] = snippets.get(i).getText();
    }
    if (fragments.length > 0) {
        return new HighlightField(highlighterContext.fieldName, Text.convertFromStringArray(fragments));
    }
    return null;
}

Example 2 with Snippet

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

the class PostingsHighlighter method filterSnippets.

static List<Snippet> filterSnippets(List<Snippet> snippets, int numberOfFragments) {
    //We need to filter the snippets as due to no_match_size we could have
    //either highlighted snippets or non highlighted ones and we don't want to mix those up
    List<Snippet> filteredSnippets = new ArrayList<>(snippets.size());
    for (Snippet snippet : snippets) {
        if (snippet.isHighlighted()) {
            filteredSnippets.add(snippet);
        }
    }
    //otherwise we return the first non highlighted one if available
    if (filteredSnippets.size() == 0) {
        if (snippets.size() > 0) {
            Snippet snippet = snippets.get(0);
            //we need to return the first sentence of the content rather than the whole content
            if (numberOfFragments == 0) {
                BreakIterator bi = BreakIterator.getSentenceInstance(Locale.ROOT);
                String text = snippet.getText();
                bi.setText(text);
                int next = bi.next();
                if (next != BreakIterator.DONE) {
                    String newText = text.substring(0, next).trim();
                    snippet = new Snippet(newText, snippet.getScore(), snippet.isHighlighted());
                }
            }
            filteredSnippets.add(snippet);
        }
    }
    return filteredSnippets;
}

Example 3 with Snippet

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

the class UnifiedHighlighter method highlight.

@Override
public HighlightField highlight(HighlighterContext highlighterContext) {
    FieldMapper fieldMapper = highlighterContext.mapper;
    SearchContextHighlight.Field field = highlighterContext.field;
    SearchContext context = highlighterContext.context;
    FetchSubPhase.HitContext hitContext = highlighterContext.hitContext;
    if (!hitContext.cache().containsKey(CACHE_KEY)) {
        hitContext.cache().put(CACHE_KEY, new HighlighterEntry());
    }
    HighlighterEntry highlighterEntry = (HighlighterEntry) hitContext.cache().get(CACHE_KEY);
    MapperHighlighterEntry mapperHighlighterEntry = highlighterEntry.mappers.get(fieldMapper);
    if (mapperHighlighterEntry == null) {
        Encoder encoder = field.fieldOptions().encoder().equals("html") ? HighlightUtils.Encoders.HTML : HighlightUtils.Encoders.DEFAULT;
        CustomPassageFormatter passageFormatter = new CustomPassageFormatter(field.fieldOptions().preTags()[0], field.fieldOptions().postTags()[0], encoder);
        mapperHighlighterEntry = new MapperHighlighterEntry(passageFormatter);
    }
    List<Snippet> snippets = new ArrayList<>();
    int numberOfFragments;
    try {
        Analyzer analyzer = context.mapperService().documentMapper(hitContext.hit().getType()).mappers().indexAnalyzer();
        List<Object> fieldValues = HighlightUtils.loadFieldValues(field, fieldMapper, context, hitContext);
        fieldValues = fieldValues.stream().map(obj -> {
            if (obj instanceof BytesRef) {
                return fieldMapper.fieldType().valueForDisplay(obj).toString();
            } else {
                return obj;
            }
        }).collect(Collectors.toList());
        IndexSearcher searcher = new IndexSearcher(hitContext.reader());
        CustomUnifiedHighlighter highlighter;
        if (field.fieldOptions().numberOfFragments() == 0) {
            // we use a control char to separate values, which is the only char that the custom break iterator
            // breaks the text on, so we don't lose the distinction between the different values of a field and we
            // get back a snippet per value
            String fieldValue = mergeFieldValues(fieldValues, MULTIVAL_SEP_CHAR);
            org.apache.lucene.search.postingshighlight.CustomSeparatorBreakIterator breakIterator = new org.apache.lucene.search.postingshighlight.CustomSeparatorBreakIterator(MULTIVAL_SEP_CHAR);
            highlighter = new CustomUnifiedHighlighter(searcher, analyzer, mapperHighlighterEntry.passageFormatter, field.fieldOptions().boundaryScannerLocale(), breakIterator, fieldValue, field.fieldOptions().noMatchSize());
            // we are highlighting the whole content, one snippet per value
            numberOfFragments = fieldValues.size();
        } else {
            //using paragraph separator we make sure that each field value holds a discrete passage for highlighting
            String fieldValue = mergeFieldValues(fieldValues, MULTIVAL_SEP_CHAR);
            BreakIterator bi = getBreakIterator(field);
            highlighter = new CustomUnifiedHighlighter(searcher, analyzer, mapperHighlighterEntry.passageFormatter, field.fieldOptions().boundaryScannerLocale(), bi, fieldValue, field.fieldOptions().noMatchSize());
            numberOfFragments = field.fieldOptions().numberOfFragments();
        }
        if (field.fieldOptions().requireFieldMatch()) {
            final String fieldName = highlighterContext.fieldName;
            highlighter.setFieldMatcher((name) -> fieldName.equals(name));
        } else {
            highlighter.setFieldMatcher((name) -> true);
        }
        Snippet[] fieldSnippets = highlighter.highlightField(highlighterContext.fieldName, highlighterContext.query, hitContext.docId(), numberOfFragments);
        for (Snippet fieldSnippet : fieldSnippets) {
            if (Strings.hasText(fieldSnippet.getText())) {
                snippets.add(fieldSnippet);
            }
        }
    } catch (IOException e) {
        throw new FetchPhaseExecutionException(context, "Failed to highlight field [" + highlighterContext.fieldName + "]", e);
    }
    snippets = filterSnippets(snippets, field.fieldOptions().numberOfFragments());
    if (field.fieldOptions().scoreOrdered()) {
        //let's sort the snippets by score if needed
        CollectionUtil.introSort(snippets, (o1, o2) -> Double.compare(o2.getScore(), o1.getScore()));
    }
    String[] fragments = new String[snippets.size()];
    for (int i = 0; i < fragments.length; i++) {
        fragments[i] = snippets.get(i).getText();
    }
    if (fragments.length > 0) {
        return new HighlightField(highlighterContext.fieldName, Text.convertFromStringArray(fragments));
    }
    return null;
}

Example 4 with Snippet

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

the class CustomPassageFormatterTests method testHtmlEncodeFormat.

public void testHtmlEncodeFormat() {
    String content = "<b>This is a really cool highlighter.</b> Unified highlighter gives nice snippets back.";
    CustomPassageFormatter passageFormatter = new CustomPassageFormatter("<em>", "</em>", new SimpleHTMLEncoder());
    Passage[] passages = new Passage[2];
    String match = "highlighter";
    BytesRef matchBytesRef = new BytesRef(match);
    Passage passage1 = new Passage();
    int start = content.indexOf(match);
    int end = start + match.length();
    passage1.setStartOffset(0);
    //lets include the whitespace at the end to make sure we trim it
    passage1.setEndOffset(end + 6);
    passage1.addMatch(start, end, matchBytesRef);
    passages[0] = passage1;
    Passage passage2 = new Passage();
    start = content.lastIndexOf(match);
    end = start + match.length();
    passage2.setStartOffset(passage1.getEndOffset());
    passage2.setEndOffset(content.length());
    passage2.addMatch(start, end, matchBytesRef);
    passages[1] = passage2;
    Snippet[] fragments = passageFormatter.format(passages, content);
    assertThat(fragments, notNullValue());
    assertThat(fragments.length, equalTo(2));
    assertThat(fragments[0].getText(), equalTo("&lt;b&gt;This is a really cool <em>highlighter</em>.&lt;&#x2F;b&gt;"));
    assertThat(fragments[1].getText(), equalTo("Unified <em>highlighter</em> gives nice snippets back."));
}

Example 5 with Snippet

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

the class CustomPassageFormatterTests method testSimpleFormat.

public void testSimpleFormat() {
    String content = "This is a really cool highlighter. Unified highlighter gives nice snippets back. No matches here.";
    CustomPassageFormatter passageFormatter = new CustomPassageFormatter("<em>", "</em>", new DefaultEncoder());
    Passage[] passages = new Passage[3];
    String match = "highlighter";
    BytesRef matchBytesRef = new BytesRef(match);
    Passage passage1 = new Passage();
    int start = content.indexOf(match);
    int end = start + match.length();
    passage1.setStartOffset(0);
    //lets include the whitespace at the end to make sure we trim it
    passage1.setEndOffset(end + 2);
    passage1.addMatch(start, end, matchBytesRef);
    passages[0] = passage1;
    Passage passage2 = new Passage();
    start = content.lastIndexOf(match);
    end = start + match.length();
    passage2.setStartOffset(passage1.getEndOffset());
    passage2.setEndOffset(end + 26);
    passage2.addMatch(start, end, matchBytesRef);
    passages[1] = passage2;
    Passage passage3 = new Passage();
    passage3.setStartOffset(passage2.getEndOffset());
    passage3.setEndOffset(content.length());
    passages[2] = passage3;
    Snippet[] fragments = passageFormatter.format(passages, content);
    assertThat(fragments, notNullValue());
    assertThat(fragments.length, equalTo(3));
    assertThat(fragments[0].getText(), equalTo("This is a really cool <em>highlighter</em>."));
    assertThat(fragments[0].isHighlighted(), equalTo(true));
    assertThat(fragments[1].getText(), equalTo("Unified <em>highlighter</em> gives nice snippets back."));
    assertThat(fragments[1].isHighlighted(), equalTo(true));
    assertThat(fragments[2].getText(), equalTo("No matches here."));
    assertThat(fragments[2].isHighlighted(), equalTo(false));
}