/**
* Retourne la liste des champs "snippet".
*
* @param node
* @param source
* @param target
* @throws IllegalAccessException
* @throws InstantiationException
*/
public static void copySnippetFields(Node node, SchemaFieldList source, SnippetFieldList target)
throws InstantiationException, IllegalAccessException {
String fieldName = XPathParser.getAttributeString(node, "name");
String tag = XPathParser.getAttributeString(node, "tag");
if (tag == null) tag = "em";
int maxSnippetNumber = XPathParser.getAttributeValue(node, "maxSnippetNumber");
if (maxSnippetNumber == 0) maxSnippetNumber = 1;
int maxSnippetSize = XPathParser.getAttributeValue(node, "maxSnippetSize");
if (maxSnippetSize == 0) maxSnippetSize = 200;
int timeLimit = DomUtils.getAttributeInteger(node, "timeLimit", 0);
FragmenterAbstract fragmenter =
FragmenterAbstract.newInstance(XPathParser.getAttributeString(node, "fragmenterClass"));
fragmenter.setAttributes(node.getAttributes());
String separator = XPathParser.getAttributeString(node, "separator");
if (separator == null) separator = "...";
SchemaField schemaField = source.get(fieldName);
if (schemaField == null) return;
SnippetField field =
new SnippetField(
schemaField.getName(),
tag,
separator,
maxSnippetSize,
maxSnippetNumber,
fragmenter,
timeLimit);
target.put(field);
}
public void setFragmenter(String fragmenterName) throws SearchLibException {
try {
fragmenterTemplate = FragmenterAbstract.newInstance(fragmenterName);
} catch (InstantiationException e) {
throw new SearchLibException(e);
} catch (IllegalAccessException e) {
throw new SearchLibException(e);
}
}
@Override
public int compareTo(SnippetField f) {
int c = super.compareTo(f);
if (c != 0) return c;
if ((c =
fragmenterTemplate
.getClass()
.getName()
.compareTo(f.fragmenterTemplate.getClass().getName()))
!= 0) return c;
if ((c = tag.compareTo(f.tag)) != 0) return c;
if ((c = separator.compareTo(f.separator)) != 0) return c;
if ((c = maxSnippetSize - f.maxSnippetSize) != 0) return c;
if ((c = maxSnippetNumber - f.maxSnippetNumber) != 0) return c;
return 0;
}
@Override
public void writeXmlConfig(XmlWriter xmlWriter) throws SAXException {
xmlWriter.startElement(
"field",
"name",
name,
"tag",
tag,
"separator",
separator,
"maxSnippetSize",
Integer.toString(maxSnippetSize),
"maxSnippetNumber",
Integer.toString(maxSnippetNumber),
"fragmenterClass",
fragmenterTemplate != null ? fragmenterTemplate.getClass().getSimpleName() : null,
"timeLimit",
Long.toString(timeLimit));
xmlWriter.endElement();
}
public String getFragmenter() {
return fragmenterTemplate.getClass().getSimpleName();
}
public boolean getSnippets(
final int docId,
final ReaderInterface reader,
final List<FieldValueItem> values,
final List<FieldValueItem> snippets,
final Timer parentTimer)
throws IOException, ParseException, SyntaxError, SearchLibException {
if (values == null) return false;
final Timer timer = new Timer(parentTimer, "SnippetField " + this.name);
final long halfTimeExpiration =
this.timeLimit == 0 ? 0 : timer.getStartOffset(this.timeLimit / 2);
final long expiration = this.timeLimit == 0 ? 0 : timer.getStartOffset(this.timeLimit);
FragmenterAbstract fragmenter = fragmenterTemplate.newInstance();
SnippetVector currentVector = null;
Timer t = new Timer(timer, "extractTermVectorIterator");
Iterator<SnippetVector> vectorIterator =
SnippetVectors.extractTermVectorIterator(
docId, reader, snippetQueries, name, values, indexAnalyzer, t, halfTimeExpiration);
if (vectorIterator != null)
currentVector = vectorIterator.hasNext() ? vectorIterator.next() : null;
t.end(null);
t = new Timer(timer, "getFraments");
int startOffset = 0;
FragmentList fragments = new FragmentList();
int vectorOffset = 0;
for (FieldValueItem valueItem : values) {
String value = valueItem.getValue();
if (value != null) {
// VectorOffset++ depends of EndOffset bug #patch Lucene 579 and
// 1458
fragmenter.getFragments(value, fragments, vectorOffset++);
}
}
t.end(null);
if (fragments.size() == 0) {
timer.end(null);
return false;
}
t = new Timer(timer, "checkValue");
Fragment fragment = fragments.first();
while (fragment != null) {
currentVector = checkValue(currentVector, vectorIterator, startOffset, fragment);
startOffset += fragment.getOriginalText().length();
fragment = fragment.next();
}
t.end(null);
Timer sbTimer = new Timer(timer, "snippetBuilder");
boolean result = false;
int snippetCounter = maxSnippetNumber;
int scoredFragment = 0;
while (snippetCounter-- != 0) {
Fragment bestScoreFragment = null;
fragment = Fragment.findNextHighlightedFragment(fragments.first());
List<Fragment> scoreFragments = new ArrayList<Fragment>(0);
double maxSearchScore = 0;
t = new Timer(sbTimer, "fragmentScore");
boolean expired = false;
while (fragment != null) {
double sc = fragment.searchScore(name, queryAnalyzer, query);
if (sc > maxSearchScore) maxSearchScore = sc;
scoreFragments.add(fragment);
fragment = Fragment.findNextHighlightedFragment(fragment.next());
scoredFragment++;
if (expiration != 0) {
if (System.currentTimeMillis() > expiration) {
expired = true;
break;
}
}
}
t.end("fragmentScore " + scoredFragment + " " + expired);
for (Fragment frag : scoreFragments)
bestScoreFragment =
Fragment.bestScore(bestScoreFragment, frag, maxSearchScore, maxSnippetSize);
if (bestScoreFragment != null) {
SnippetBuilder snippetBuilder =
new SnippetBuilder(maxSnippetSize, unescapedSeparator, tags, bestScoreFragment);
if (snippetBuilder.length() > 0)
snippets.add(new FieldValueItem(FieldValueOriginEnum.SNIPPET, snippetBuilder.toString()));
fragments.remove(snippetBuilder.getFragments());
result = true;
continue;
}
if (fragments.first() == null) break;
SnippetBuilder snippetBuilder =
new SnippetBuilder(maxSnippetSize, unescapedSeparator, tags, fragments.first());
if (snippetBuilder.length() > 0) {
snippets.add(new FieldValueItem(FieldValueOriginEnum.SNIPPET, snippetBuilder.toString()));
fragments.remove(snippetBuilder.getFragments());
}
}
sbTimer.end(null);
timer.end(null);
return result;
}
