@Override
public void modifyIndex(final IndexWriter writer, final IndexSearcher searcher)
throws ModifyKnowledgeBaseException {
for (final Map.Entry<String, HashMap<String, String>> entry : this.attributes.entrySet()) {
final String key = entry.getKey();
final HashMap<String, String> hash = entry.getValue();
final QueryParser qp = new QueryParser(this.docPrimaryKey, new DoserIDAnalyzer());
try {
final TopDocs top = searcher.search(qp.parse(QueryParserBase.escape(key)), 1);
final ScoreDoc[] scores = top.scoreDocs;
if (scores.length > 0) {
final Document doc = new Document();
final Document currentDoc = searcher.getIndexReader().document(scores[0].doc);
// BugFix create new Document und copy Fields.
final List<IndexableField> fields = currentDoc.getFields();
for (final IndexableField field : fields) {
if (field.stringValue() != null) {
if (field.name().equalsIgnoreCase(docPrimaryKey)) {
doc.add(new StringField(field.name(), field.stringValue(), Field.Store.YES));
} else {
doc.add(new TextField(field.name(), field.stringValue(), Field.Store.YES));
}
}
}
final List<Document> docListToAdd = new LinkedList<Document>();
docListToAdd.add(doc);
for (final Map.Entry<String, String> subentry : hash.entrySet()) {
final IndexableField field = doc.getField(subentry.getKey());
if (field == null) {
throw new ModifyKnowledgeBaseException("UpdateField no found", null);
}
if (this.action.equals(KBModifications.OVERRIDEFIELD)) {
doc.removeFields(subentry.getKey());
String[] newentries = generateSeperatedFieldStrings(subentry.getValue());
for (int i = 0; i < newentries.length; i++) {
doc.add(new TextField(subentry.getKey(), newentries[i], Field.Store.YES));
}
} else if (this.action.equals(KBModifications.UPDATERELATEDLABELS)) {
doc.removeFields(subentry.getKey());
doc.add(updateOccurrences(subentry.getValue(), field, "surroundinglabels"));
} else if (this.action.equals(KBModifications.UPDATEOCCURRENCES)) {
doc.removeFields(subentry.getKey());
IndexableField f = updateOccurrences(subentry.getValue(), field, "occurrences");
doc.add(f);
}
}
writer.updateDocuments(new Term(this.docPrimaryKey, key), docListToAdd);
} else {
throw new ModifyKnowledgeBaseException("Document not found", null);
}
} catch (final IOException e) {
throw new ModifyKnowledgeBaseException("IOException in IndexSearcher", e);
} catch (ParseException e) {
throw new ModifyKnowledgeBaseException("Queryparser Exception", e);
}
}
}
public static void recoverSrcData(
SenseiResult res, SenseiHit[] hits, boolean isFetchStoredFields) {
if (hits != null) {
for (SenseiHit hit : hits) {
try {
byte[] dataBytes = hit.getStoredValue();
if (dataBytes == null || dataBytes.length == 0) {
Document doc = hit.getStoredFields();
if (doc != null) {
dataBytes = doc.getBinaryValue(AbstractZoieIndexable.DOCUMENT_STORE_FIELD);
if (dataBytes == null || dataBytes.length == 0) {
dataBytes = doc.getBinaryValue(SenseiSchema.SRC_DATA_COMPRESSED_FIELD_NAME);
if (dataBytes == null || dataBytes.length == 0) {
dataBytes = doc.getBinaryValue(SenseiSchema.SRC_DATA_FIELD_NAME);
if (dataBytes != null && dataBytes.length > 0) {
hit.setSrcData(new String(dataBytes, "UTF-8"));
dataBytes = null; // set to null to avoid gunzip.
}
}
doc.removeFields(SenseiSchema.SRC_DATA_COMPRESSED_FIELD_NAME);
doc.removeFields(SenseiSchema.SRC_DATA_FIELD_NAME);
}
}
}
if (dataBytes != null && dataBytes.length > 0) {
byte[] data;
try {
data = DefaultJsonSchemaInterpreter.decompress(dataBytes);
} catch (Exception ex) {
data = dataBytes;
}
hit.setSrcData(new String(data, "UTF-8"));
}
} catch (Exception e) {
logger.error(e.getMessage(), e);
res.getErrors().add(new SenseiError(e.getMessage(), ErrorType.BrokerGatherError));
}
recoverSrcData(res, hit.getSenseiGroupHits(), isFetchStoredFields);
// Remove stored fields since the user is not requesting:
if (!isFetchStoredFields) hit.setStoredFields(null);
}
}
}
public void deleteFieldFromIndex(String fieldName, int docId, Analyzer analyzer)
throws IOException, ConfigurationException {
Document doc = reader.document(docId);
doc.removeFields(fieldName);
Field uri = doc.getField("URI");
Term term = new Term("URI", uri.stringValue());
writer.updateDocument(term, doc, analyzer);
}
protected void shiftContextInTime(IndexingContext ctx, int shiftDays) throws IOException {
if (shiftDays != 0) {
IndexWriter iw = ctx.getIndexWriter();
for (int docNum = 0; docNum < ctx.getIndexReader().maxDoc(); docNum++) {
if (!ctx.getIndexReader().isDeleted(docNum)) {
Document doc = ctx.getIndexReader().document(docNum);
String lastModified = doc.get(ArtifactInfo.LAST_MODIFIED);
if (lastModified != null) {
long lm = Long.parseLong(lastModified);
lm = lm + (shiftDays * A_DAY_MILLIS);
doc.removeFields(ArtifactInfo.LAST_MODIFIED);
doc.add(
new Field(
ArtifactInfo.LAST_MODIFIED,
Long.toString(lm),
Field.Store.YES,
Field.Index.NO));
iw.updateDocument(new Term(ArtifactInfo.UINFO, doc.get(ArtifactInfo.UINFO)), doc);
}
}
}
iw.optimize();
iw.close();
// shift timestamp too
if (ctx.getTimestamp() != null) {
ctx.updateTimestamp(
true, new Date(ctx.getTimestamp().getTime() + (shiftDays * A_DAY_MILLIS)));
} else {
ctx.updateTimestamp(
true, new Date(System.currentTimeMillis() + (shiftDays * A_DAY_MILLIS)));
}
}
}
@Override
DocWriter processDocument() throws IOException {
// this is where we process the geo-search components of the document
Document doc = docState.doc;
int docID = docState.docID;
List<Fieldable> fields = doc.getFields();
List<GeoCoordinateField> geoFields = new Vector<GeoCoordinateField>();
for (Fieldable field : fields) {
if (field instanceof GeoCoordinateField) {
geoFields.add((GeoCoordinateField) field);
}
}
for (GeoCoordinateField geoField : geoFields) {
// process field into GeoIndex here
geoIndexer.index(docID, geoField);
doc.removeFields(geoField.name());
}
return defaultDocConsumerPerThread.processDocument();
}
@Override
public Document handleDocument(IndexWriter indexWriter, IndexState state, File file)
throws Exception {
Document doc = super.handleDocument(indexWriter, state, file);
/* create/update document */
if (doc != null && file.length() > 0) {
/* ultra fast header analysis, with fileinputstream - too many filedescripor
* exeption on samba network */
try {
ImageInfo imageInfo = Sanselan.getImageInfo(file);
/* default image meta informations */
int w = imageInfo.getWidth();
int h = imageInfo.getHeight();
/* image width */
get(ImageWidthField.NAME).add(doc, w);
/* image height */
get(ImageHeightField.NAME).add(doc, h);
/* bits per pixel */
get(BitsPerPixelField.NAME).add(doc, imageInfo.getBitsPerPixel());
/* complete size */
get(PixelSizeField.NAME).add(doc, w * h);
/* image aspect ratio */
get(AspectRatioField.NAME).add(doc, w / (float) h);
/* try to extract metadata from jpeg files */
String mime = imageInfo.getMimeType();
/* override because they are different content types than extensions */
doc.removeFields(MimeTypeField.NAME);
get(MimeTypeField.NAME).add(doc, mime);
/* special jpeg handling */
if (mime != null && mime.equals("image/jpeg")) {
IImageMetadata metadata = Sanselan.getMetadata(file);
if (metadata instanceof JpegImageMetadata) {
JpegImageMetadata jpegMetadata = (JpegImageMetadata) metadata;
/* create date of the image */
TiffField field = jpegMetadata.findEXIFValue(ExifTagConstants.EXIF_TAG_CREATE_DATE);
if (field != null) {
Date date = DateUtil.parseEXIFFormat(field.getStringValue());
/* else unknown format */
if (date != null) {
/* override file last modified date */
doc.removeFields(LastModifiedField.NAME);
get(LastModifiedField.NAME).add(doc, date.getTime());
}
}
/* user comment tag */
field = jpegMetadata.findEXIFValue(ExifTagConstants.EXIF_TAG_USER_COMMENT);
if (field != null) {
get(CommentField.NAME).add(doc, field.getStringValue().trim());
}
/* make tag */
field = jpegMetadata.findEXIFValue(ExifTagConstants.EXIF_TAG_MAKE);
if (field != null) {
get(ExifMakeField.NAME).add(doc, field.getStringValue().trim());
}
/* model tag */
field = jpegMetadata.findEXIFValue(ExifTagConstants.EXIF_TAG_MODEL);
if (field != null) {
get(ExifModelField.NAME).add(doc, field.getStringValue().trim());
}
/* date time tag */
field = jpegMetadata.findEXIFValue(ExifTagConstants.EXIF_TAG_CREATE_DATE);
if (field != null) {
Date date = DateUtil.parseEXIFFormat(field.getStringValue());
/* else unknown format */
if (date != null) {
get(ExifDateField.NAME).add(doc, date.getTime());
}
}
/* try to find gps informations */
TiffImageMetadata exifMetadata = jpegMetadata.getExif();
if (exifMetadata != null) {
try {
TiffImageMetadata.GPSInfo gpsInfo = exifMetadata.getGPS();
if (null != gpsInfo) {
get(LatField.NAME).add(doc, gpsInfo.getLatitudeAsDegreesNorth());
get(LonField.NAME).add(doc, gpsInfo.getLongitudeAsDegreesEast());
// doc.add(new
// NumericField(FIELD_EXIF_GPS_IMG_DIRECTION, Field.Store.YES,
// true).setIntValue(gps.getRational(
//
// GpsDirectory.TAG_GPS_IMG_DIRECTION).intValue()));
//
// System.out.println(exifMetadata.findField(GPSTagConstants.GPS_TAG_GPS_IMG_DIRECTION));
}
} catch (ImageReadException e) {
}
}
}
}
/* run some classification */
if (classifyImage) {
int[] v = ImageUtil.analyzeRGB(file);
for (int i = 0; i < v.length; i++) {
doc.add(new IntPoint(COLORMEAN + i, v[i]));
}
}
} catch (Throwable e) {
/* only log */
log.log(Level.INFO, "can not extract image informations!", e);
}
}
return doc;
}
