@Test()
public void testSonyDSCHXV5()
throws IOException, CatalogTransformerException, UnsupportedQueryException,
SourceUnavailableException, FederationException, ParseException {
File file = new File(TEST_DATA_PATH + "Sony DSC-HX5V.jpg");
FileInputStream fis = FileUtils.openInputStream(file);
Metacard metacard = createTransformer().transform(fis);
assertNotNull(metacard);
assertNotNull(metacard.getCreatedDate());
assertThat(metacard.getCreatedDate().getYear() + 1900, is(2010));
assertThat(metacard.getCreatedDate().getMonth() + 1, is(7));
assertThat(metacard.getCreatedDate().getDate(), is(14));
assertThat(metacard.getCreatedDate().getHours(), is(11));
assertThat(metacard.getCreatedDate().getMinutes(), is(00));
assertThat(metacard.getCreatedDate().getSeconds(), is(23));
assertNotNull(metacard.getModifiedDate());
assertThat(metacard.getModifiedDate().getYear() + 1900, is(2010));
assertThat(metacard.getModifiedDate().getMonth() + 1, is(7));
assertThat(metacard.getModifiedDate().getDate(), is(14));
assertThat(metacard.getModifiedDate().getHours(), is(11));
assertThat(metacard.getModifiedDate().getMinutes(), is(00));
assertThat(metacard.getModifiedDate().getSeconds(), is(23));
WKTReader reader = new WKTReader();
Geometry geometry = reader.read(metacard.getLocation());
assertEquals(-104.303846389, geometry.getCoordinate().x, 0.00001);
assertEquals(39.5698783333, geometry.getCoordinate().y, 0.00001);
byte[] thumbnail = metacard.getThumbnail();
assertNotNull(thumbnail);
assertThat(thumbnail.length, is(11490));
}
protected void applyUpdateOperationRules(
UpdateRequest updateRequest,
Entry<Serializable, Metacard> anUpdate,
Metacard aMetacard,
Date now) {
if (UpdateRequest.UPDATE_BY_ID.equals(updateRequest.getAttributeName())
&& !anUpdate.getKey().toString().equals(aMetacard.getId())) {
LOGGER.info(
"{} in metacard must match the Update {}, overwriting metacard {} [{}] with the update identifier [{}]",
Metacard.ID,
Metacard.ID,
Metacard.ID,
aMetacard.getId(),
anUpdate.getKey());
aMetacard.setAttribute(new AttributeImpl(Metacard.ID, anUpdate.getKey()));
}
if (aMetacard.getCreatedDate() == null) {
LOGGER.info(
"{} date should match the original metacard. Changing date to current timestamp so it is at least not null.",
Metacard.CREATED);
aMetacard.setAttribute(new AttributeImpl(Metacard.CREATED, now));
}
if (aMetacard.getModifiedDate() == null) {
aMetacard.setAttribute(new AttributeImpl(Metacard.MODIFIED, now));
}
if (aMetacard.getEffectiveDate() == null) {
aMetacard.setAttribute(new AttributeImpl(Metacard.EFFECTIVE, now));
}
}
@Test()
public void testIPhone()
throws IOException, CatalogTransformerException, UnsupportedQueryException,
SourceUnavailableException, FederationException, ParseException {
File file = new File(TEST_DATA_PATH + "Apple iPhone 4.jpg");
FileInputStream fis = FileUtils.openInputStream(file);
Metacard metacard = createTransformer().transform(fis);
assertNotNull(metacard);
assertNotNull(metacard.getCreatedDate());
assertThat(metacard.getCreatedDate().getYear() + 1900, is(2011));
assertThat(metacard.getCreatedDate().getMonth() + 1, is(1));
assertThat(metacard.getCreatedDate().getDate(), is(13));
assertThat(metacard.getCreatedDate().getHours(), is(14));
assertThat(metacard.getCreatedDate().getMinutes(), is(33));
assertThat(metacard.getCreatedDate().getSeconds(), is(39));
assertEquals(metacard.getCreatedDate(), metacard.getModifiedDate());
WKTReader reader = new WKTReader();
Geometry geometry = reader.read(metacard.getLocation());
assertEquals(12.488833, geometry.getCoordinate().x, 0.00001);
assertEquals(41.853, geometry.getCoordinate().y, 0.00001);
}
@Test
public void testGeoTaggedJpeg() throws Exception {
InputStream stream =
Thread.currentThread().getContextClassLoader().getResourceAsStream("testJPEG_GEO.jpg");
/*
* The dates in testJPED_GEO.jpg do not contain timezones. If no timezone is specified,
* the Tika input transformer assumes the local time zone. Set the system timezone to UTC
* so we can do assertions.
*/
TimeZone defaultTimeZone = TimeZone.getDefault();
TimeZone.setDefault(TimeZone.getTimeZone("UTC"));
Metacard metacard = transform(stream);
assertNotNull(metacard);
assertNotNull(metacard.getMetadata());
assertThat(
metacard.getMetadata(), containsString("<meta name=\"Model\" content=\"Canon EOS 40D\"/>"));
assertThat(metacard.getContentTypeName(), is("image/jpeg"));
assertThat(convertDate(metacard.getCreatedDate()), is("2009-08-11 09:09:45 UTC"));
assertThat(convertDate(metacard.getModifiedDate()), is("2009-10-02 23:02:49 UTC"));
assertThat(
(String) metacard.getAttribute(Metacard.GEOGRAPHY).getValue(),
is("POINT(-54.1234 12.54321)"));
// Reset timezone back to local time zone.
TimeZone.setDefault(defaultTimeZone);
}
@Test
public void testOpenOffice() throws Exception {
InputStream stream =
Thread.currentThread().getContextClassLoader().getResourceAsStream("testOpenOffice2.odt");
/*
* The dates in testOpenOffice2.odt do not contain timezones. If no timezone is specified,
* the Tika input transformer assumes the local time zone. Set the system timezone to UTC
* so we can do assertions.
*/
TimeZone defaultTimeZone = TimeZone.getDefault();
TimeZone.setDefault(TimeZone.getTimeZone("UTC"));
Metacard metacard = transform(stream);
assertNotNull(metacard);
assertThat(metacard.getTitle(), is("Test OpenOffice2 Document"));
assertThat(convertDate(metacard.getCreatedDate()), is("2007-09-14 11:06:08 UTC"));
assertThat(convertDate(metacard.getModifiedDate()), is("2013-02-13 06:52:10 UTC"));
assertNotNull(metacard.getMetadata());
assertThat(
metacard.getMetadata(),
containsString("This is a sample Open Office document, written in NeoOffice 2.2.1"));
assertThat(metacard.getContentTypeName(), is("application/vnd.oasis.opendocument.text"));
// Reset timezone back to local time zone.
TimeZone.setDefault(defaultTimeZone);
}
@Test
public void testPpt() throws Exception {
InputStream stream =
Thread.currentThread().getContextClassLoader().getResourceAsStream("testPPT.ppt");
Metacard metacard = transform(stream);
assertNotNull(metacard);
assertThat(metacard.getTitle(), is("Sample Powerpoint Slide"));
assertThat(convertDate(metacard.getCreatedDate()), is("2007-09-14 17:33:12 UTC"));
assertThat(convertDate(metacard.getModifiedDate()), is("2007-09-14 19:16:39 UTC"));
assertNotNull(metacard.getMetadata());
assertThat(metacard.getMetadata(), containsString("Created with Microsoft"));
assertThat(metacard.getContentTypeName(), is("application/vnd.ms-powerpoint"));
}
@Test
public void testPDF() throws Exception {
InputStream stream =
Thread.currentThread().getContextClassLoader().getResourceAsStream("testPDF.pdf");
Metacard metacard = transform(stream);
assertNotNull(metacard);
assertThat(metacard.getTitle(), is("Apache Tika - Apache Tika"));
assertThat(convertDate(metacard.getCreatedDate()), is("2007-09-15 09:02:31 UTC"));
assertThat(convertDate(metacard.getModifiedDate()), is("2007-09-15 09:02:31 UTC"));
assertNotNull(metacard.getMetadata());
assertThat(
metacard.getMetadata(), containsString("<meta name=\"xmpTPg:NPages\" content=\"1\"/>"));
assertThat(metacard.getContentTypeName(), is("application/pdf"));
}
@Test
public void testXls() throws Exception {
InputStream stream =
Thread.currentThread().getContextClassLoader().getResourceAsStream("testEXCEL.xls");
Metacard metacard = transform(stream);
assertNotNull(metacard);
assertThat(metacard.getTitle(), is("Simple Excel document"));
assertThat(convertDate(metacard.getCreatedDate()), is("2007-10-01 16:13:56 UTC"));
assertThat(convertDate(metacard.getModifiedDate()), is("2007-10-01 16:31:43 UTC"));
assertNotNull(metacard.getMetadata());
assertThat(
metacard.getMetadata(),
containsString("Written and saved in Microsoft Excel X for Mac Service Release 1."));
assertThat(metacard.getContentTypeName(), is("application/vnd.ms-excel"));
}
@Test
public void testMp4() throws Exception {
InputStream stream =
Thread.currentThread().getContextClassLoader().getResourceAsStream("testMP4.m4a");
Metacard metacard = transform(stream);
assertNotNull(metacard);
assertThat(metacard.getTitle(), is("Test Title"));
assertThat(convertDate(metacard.getCreatedDate()), is("2012-01-28 18:39:18 UTC"));
assertThat(convertDate(metacard.getModifiedDate()), is("2012-01-28 18:40:25 UTC"));
assertNotNull(metacard.getMetadata());
assertThat(
metacard.getMetadata(),
containsString("<meta name=\"xmpDM:artist\" content=\"Test Artist\"/>"));
assertThat(metacard.getContentTypeName(), is("audio/mp4"));
}
@Test
public void testWordDoc() throws Exception {
InputStream stream =
Thread.currentThread().getContextClassLoader().getResourceAsStream("testWORD.docx");
Metacard metacard = transform(stream);
assertNotNull(metacard);
assertThat(metacard.getTitle(), is("Sample Word Document"));
assertThat(convertDate(metacard.getCreatedDate()), is("2008-12-11 16:04:00 UTC"));
assertThat(convertDate(metacard.getModifiedDate()), is("2010-11-12 16:21:00 UTC"));
assertNotNull(metacard.getMetadata());
assertThat(
metacard.getMetadata(), containsString("<p>This is a sample Microsoft Word Document.</p>"));
assertThat(
metacard.getContentTypeName(),
is("application/vnd.openxmlformats-officedocument.wordprocessingml.document"));
}
@Test
public void testXlsx() throws Exception {
InputStream stream =
Thread.currentThread().getContextClassLoader().getResourceAsStream("testEXCEL.xlsx");
Metacard metacard = transform(stream);
assertNotNull(metacard);
assertThat(metacard.getTitle(), is("Simple Excel document"));
assertThat(convertDate(metacard.getCreatedDate()), is("2007-10-01 16:13:56 UTC"));
assertThat(convertDate(metacard.getModifiedDate()), is("2008-12-11 16:02:17 UTC"));
assertNotNull(metacard.getMetadata());
assertThat(
metacard.getMetadata(),
containsString("Sample Excel Worksheet - Numbers and their Squares"));
assertThat(
metacard.getContentTypeName(),
is("application/vnd.openxmlformats-officedocument.spreadsheetml.sheet"));
}
@Test
public void testPptx() throws Exception {
InputStream stream =
Thread.currentThread().getContextClassLoader().getResourceAsStream("testPPT.pptx");
Metacard metacard = transform(stream);
assertNotNull(metacard);
assertThat(metacard.getTitle(), is("Attachment Test"));
assertThat(convertDate(metacard.getCreatedDate()), is("2010-05-04 06:43:54 UTC"));
assertThat(convertDate(metacard.getModifiedDate()), is("2010-06-29 06:34:35 UTC"));
assertNotNull(metacard.getMetadata());
assertThat(
metacard.getMetadata(),
containsString("content as every other file being tested for tika content parsing"));
assertThat(
metacard.getContentTypeName(),
is("application/vnd.openxmlformats-officedocument.presentationml.presentation"));
}
private void assertMetacard(Metacard mc, Map<String, Object> expectedValues) {
assertThat(mc.getId(), equalTo((String) expectedValues.get(Metacard.ID)));
assertListStringAttribute(
mc,
CswRecordMetacardType.CSW_IDENTIFIER,
(String[]) expectedValues.get(CswRecordMetacardType.CSW_IDENTIFIER));
assertThat(mc.getTitle(), equalTo((String) expectedValues.get(Metacard.TITLE)));
assertListStringAttribute(
mc,
CswRecordMetacardType.CSW_TITLE,
(String[]) expectedValues.get(CswRecordMetacardType.CSW_TITLE));
assertThat(mc.getModifiedDate(), equalTo((Date) expectedValues.get(Metacard.MODIFIED)));
assertListStringAttribute(
mc,
CswRecordMetacardType.CSW_MODIFIED,
(String[]) expectedValues.get(CswRecordMetacardType.CSW_MODIFIED));
assertListStringAttribute(
mc,
CswRecordMetacardType.CSW_SUBJECT,
(String[]) expectedValues.get(CswRecordMetacardType.CSW_SUBJECT));
assertListStringAttribute(
mc,
CswRecordMetacardType.CSW_ABSTRACT,
(String[]) expectedValues.get(CswRecordMetacardType.CSW_ABSTRACT));
assertListStringAttribute(
mc,
CswRecordMetacardType.CSW_RIGHTS,
(String[]) expectedValues.get(CswRecordMetacardType.CSW_RIGHTS));
assertListStringAttribute(
mc,
CswRecordMetacardType.CSW_LANGUAGE,
(String[]) expectedValues.get(CswRecordMetacardType.CSW_LANGUAGE));
assertThat(
(String) mc.getAttribute(CswRecordMetacardType.CSW_TYPE).getValue(),
equalTo((String) expectedValues.get(CswRecordMetacardType.CSW_TYPE)));
assertListStringAttribute(
mc,
CswRecordMetacardType.CSW_FORMAT,
(String[]) expectedValues.get(CswRecordMetacardType.CSW_FORMAT));
assertThat(mc.getLocation(), equalTo((String) expectedValues.get(Metacard.GEOGRAPHY)));
assertListStringAttribute(
mc,
CswRecordMetacardType.OWS_BOUNDING_BOX,
(String[]) expectedValues.get(CswRecordMetacardType.OWS_BOUNDING_BOX));
}
protected void applyCreatedOperationRules(
CreateRequest createRequest, Metacard aMetacard, Date now) {
LOGGER.debug("Applying standard rules on CreateRequest");
aMetacard.setAttribute(
new AttributeImpl(Metacard.ID, UUID.randomUUID().toString().replaceAll("-", "")));
if (aMetacard.getCreatedDate() == null) {
aMetacard.setAttribute(new AttributeImpl(Metacard.CREATED, now));
}
if (aMetacard.getModifiedDate() == null) {
aMetacard.setAttribute(new AttributeImpl(Metacard.MODIFIED, now));
}
if (aMetacard.getEffectiveDate() == null) {
aMetacard.setAttribute(new AttributeImpl(Metacard.EFFECTIVE, now));
}
}
@Override
protected Object doExecute() throws Exception {
String formatString =
"%1$-33s %2$-26s %3$-" + TITLE_MAX_LENGTH + "s %4$-" + EXCERPT_MAX_LENGTH + "s%n";
CatalogFacade catalogProvider = getCatalog();
Filter filter = null;
if (cqlFilter != null) {
filter = CQL.toFilter(cqlFilter);
} else {
if (searchPhrase == null) {
searchPhrase = "*";
}
if (caseSensitive) {
filter =
getFilterBuilder()
.attribute(Metacard.ANY_TEXT)
.is()
.like()
.caseSensitiveText(searchPhrase);
} else {
filter = getFilterBuilder().attribute(Metacard.ANY_TEXT).is().like().text(searchPhrase);
}
}
QueryImpl query = new QueryImpl(filter);
query.setRequestsTotalResultsCount(true);
if (numberOfItems > -1) {
query.setPageSize(numberOfItems);
}
long start = System.currentTimeMillis();
SourceResponse response = catalogProvider.query(new QueryRequestImpl(query));
long end = System.currentTimeMillis();
int size = 0;
if (response.getResults() != null) {
size = response.getResults().size();
}
console.println();
console.printf(
" %d result(s) out of %s%d%s in %3.3f seconds",
(size),
Ansi.ansi().fg(Ansi.Color.CYAN).toString(),
response.getHits(),
Ansi.ansi().reset().toString(),
(end - start) / MILLISECONDS_PER_SECOND);
console.printf(formatString, "", "", "", "");
printHeaderMessage(String.format(formatString, ID, DATE, TITLE, EXCERPT));
for (Result result : response.getResults()) {
Metacard metacard = result.getMetacard();
String title = (metacard.getTitle() != null ? metacard.getTitle() : "N/A");
String excerpt = "N/A";
String modifiedDate = "";
if (searchPhrase != null) {
if (metacard.getMetadata() != null) {
XPathHelper helper = new XPathHelper(metacard.getMetadata());
String indexedText = helper.getDocument().getDocumentElement().getTextContent();
indexedText = indexedText.replaceAll("\\r\\n|\\r|\\n", " ");
String normalizedSearchPhrase = searchPhrase.replaceAll("\\*", "");
int index = -1;
if (caseSensitive) {
index = indexedText.indexOf(normalizedSearchPhrase);
} else {
index = indexedText.toLowerCase().indexOf(normalizedSearchPhrase.toLowerCase());
}
if (index != -1) {
int contextLength = (EXCERPT_MAX_LENGTH - normalizedSearchPhrase.length() - 8) / 2;
excerpt = "..." + indexedText.substring(Math.max(index - contextLength, 0), index);
excerpt = excerpt + Ansi.ansi().fg(Ansi.Color.GREEN).toString();
excerpt =
excerpt + indexedText.substring(index, index + normalizedSearchPhrase.length());
excerpt = excerpt + Ansi.ansi().reset().toString();
excerpt =
excerpt
+ indexedText.substring(
index + normalizedSearchPhrase.length(),
Math.min(
indexedText.length(),
index + normalizedSearchPhrase.length() + contextLength))
+ "...";
}
}
}
if (metacard.getModifiedDate() != null) {
modifiedDate =
new DateTime(metacard.getModifiedDate().getTime()).toString(DATETIME_FORMATTER);
}
console.printf(
formatString,
metacard.getId(),
modifiedDate,
title.substring(0, Math.min(title.length(), TITLE_MAX_LENGTH)),
excerpt);
}
return null;
}