Skip to content

Commit c7a152b

Browse files
petrpodsedpetrpodsed
committed
Refactor: RepositoryAccess facade
1 parent d6bf8c3 commit c7a152b

File tree

3 files changed

+206
-102
lines changed

3 files changed

+206
-102
lines changed

‎shared/common/src/main/java/cz/incad/kramerius/fedora/impl/RepositoryAccessImpl.java

Lines changed: 174 additions & 102 deletions
Original file line numberDiff line numberDiff line change
@@ -6,6 +6,8 @@
66
import com.qbizm.kramerius.imp.jaxb.DigitalObject;
77
import cz.incad.kramerius.StreamHeadersObserver;
88
import cz.incad.kramerius.fedora.RepositoryAccess;
9+
import cz.incad.kramerius.fedora.impl.tmp.ContentFormat;
10+
import cz.incad.kramerius.fedora.impl.tmp.UnsupportedContentFormatException;
911
import cz.incad.kramerius.fedora.om.repository.AkubraRepository;
1012
import cz.incad.kramerius.fedora.om.repository.RepositoryDatastream;
1113
import cz.incad.kramerius.fedora.om.repository.RepositoryException;
@@ -33,12 +35,14 @@
3335

3436
import javax.annotation.Nullable;
3537
import javax.xml.bind.JAXBException;
38+
import javax.xml.parsers.DocumentBuilderFactory;
3639
import javax.xml.xpath.XPathExpressionException;
3740
import java.io.ByteArrayInputStream;
3841
import java.io.IOException;
3942
import java.io.InputStream;
4043
import java.io.StringReader;
4144
import java.nio.charset.Charset;
45+
import java.nio.charset.StandardCharsets;
4246
import java.time.LocalDateTime;
4347
import java.time.format.DateTimeParseException;
4448
import java.util.*;
@@ -91,30 +95,6 @@ public boolean isPidAvailable(String pid) throws IOException, RepositoryExceptio
9195
boolean exists = this.repositoryApi.objectExists(pid);
9296
return exists;
9397
}
94-
@Override
95-
public InputStream getFoxml(String pid, boolean archive) throws IOException {
96-
try {
97-
if (archive){
98-
DigitalObject obj = manager.readObjectCloneFromStorage(pid);
99-
manager.resolveArchivedDatastreams(obj);
100-
return this.manager.marshallObject(obj);
101-
}else {
102-
return this.manager.retrieveObject(pid);
103-
}
104-
} catch (Exception e) {
105-
throw new IOException(e);
106-
}
107-
}
108-
@Override
109-
public org.dom4j.Document getFoxml(String pid) throws RepositoryException, IOException {
110-
Lock readLock = AkubraDOManager.getReadLock(pid);
111-
try {
112-
RepositoryObject object = akubraRepositoryImpl.getObject(pid);
113-
return Utils.inputstreamToDocument(object.getFoxml(), true);
114-
} finally {
115-
readLock.unlock();
116-
}
117-
}
11898
//-------- get object property
11999
@Override
120100
public String getProperty(String pid, String propertyName) throws IOException, RepositoryException {
@@ -187,70 +167,6 @@ public List<String> getDatastreamNames(String pid) throws RepositoryException, I
187167
readLock.unlock();
188168
}
189169
}
190-
// TODO here we always use AkubraUtils.getStreamContent but we have also AkubraObject.AkubraDatastream for fetching stream content
191-
@Override
192-
public InputStream getDataStream(String pid, String datastreamName) throws IOException {
193-
try {
194-
pid = makeSureObjectPid(pid);
195-
if (this.accessLog != null && this.accessLog.isReportingAccess(pid, datastreamName)) {
196-
reportAccess(pid, datastreamName);
197-
}
198-
DigitalObject object = manager.readObjectFromStorage(pid);
199-
if (object != null) {
200-
DatastreamVersionType stream = AkubraUtils.getLastStreamVersion(object, datastreamName);
201-
if (stream != null) {
202-
return AkubraUtils.getStreamContent(stream, manager);
203-
} else {
204-
throw new IOException("cannot find stream '" + datastreamName + "' for pid '" + pid + "'");
205-
}
206-
} else {
207-
throw new IOException("cannot find pid '" + pid + "'");
208-
}
209-
} catch (Exception e) {
210-
throw new IOException(e);
211-
}
212-
}
213-
// XML data stream
214-
@Override
215-
public Document getStream(String pid, String streamName) throws IOException {
216-
DigitalObject object = manager.readObjectFromStorage(pid);
217-
if (object != null) {
218-
DatastreamVersionType stream = AkubraUtils.getLastStreamVersion(object, streamName);
219-
if (stream != null) {
220-
if (stream.getXmlContent() != null) {
221-
List<Element> elementList = stream.getXmlContent().getAny();
222-
if (!elementList.isEmpty()) {
223-
return elementList.get(0).getOwnerDocument();
224-
} else {
225-
throw new IOException("Datastream not found: " + pid + " - " + streamName);
226-
}
227-
} else {
228-
throw new IOException("Expected XML datastream: " + pid + " - " + streamName);
229-
}
230-
}
231-
throw new IOException("Datastream not found: " + pid + " - " + streamName);
232-
}
233-
throw new IOException("Object not found: " + pid);
234-
}
235-
@Override
236-
public org.dom4j.Document getDatastreamXml(String pid, String dsId) throws RepositoryException, IOException {
237-
Lock readLock = AkubraDOManager.getReadLock(pid);
238-
try {
239-
RepositoryObject object = akubraRepositoryImpl.getObject(pid);
240-
if (object.streamExists(dsId)) {
241-
org.dom4j.Document foxml = Utils.inputstreamToDocument(object.getFoxml(), true);
242-
org.dom4j.Element dcEl = (org.dom4j.Element) Dom4jUtils.buildXpath(String.format("/foxml:digitalObject/foxml:datastream[@ID='%s']", dsId)).selectSingleNode(foxml);
243-
org.dom4j.Element detached = (org.dom4j.Element) dcEl.detach();
244-
org.dom4j.Document result = DocumentHelper.createDocument();
245-
result.add(detached);
246-
return result;
247-
} else {
248-
return null;
249-
}
250-
} finally {
251-
readLock.unlock();
252-
}
253-
}
254170
@Override
255171
public List<Map<String, String>> getStreamsOfObject(String pid) throws IOException {
256172
try {
@@ -320,9 +236,120 @@ public Date getStreamLastmodifiedFlag(String pid, String streamName) throws IOEx
320236
}
321237
throw new IOException("Object not found: " + pid);
322238
}
323-
//------ get stream CONTENT
324-
// input data stream
239+
//------------------------------------------------------------------------------------------------------------
240+
// NEW !!!!!!!!!!!!!!!!
241+
public <T> T getStreamContent(String pid, KnownDatastreams dsId, Class<T> returnType) throws IOException, UnsupportedContentFormatException {
242+
// Determine supported formats for the content
243+
ContentFormat supportedFormat = determineSupportedFormat(id);
244+
// Validate the requested format
245+
if ((contentType == String.class && !supportedFormat.supportsString()) ||
246+
(contentType == InputStream.class && !supportedFormat.supportsStream()) ||
247+
(contentType == Document.class && !supportedFormat.supportsXml())) {
248+
throw new UnsupportedContentFormatException("Format not supported for content ID: " + id);
249+
}
250+
// Retrieve content as bytes
251+
byte[] rawContent = fetchContentFromStorage(id);
252+
// Convert content to the requested format
253+
if (contentType == String.class) {
254+
return contentType.cast(new String(rawContent, StandardCharsets.UTF_8));
255+
} else if (contentType == InputStream.class) {
256+
return contentType.cast(new ByteArrayInputStream(rawContent));
257+
} else if (contentType == Document.class) {
258+
return contentType.cast(parseXml(rawContent));
259+
}
260+
throw new IllegalArgumentException("Unsupported content type: " + contentType);
261+
}
262+
public <T> T getFoxml(String pid, KnownDatastreams dsId, Class<T> returnType) throws IOException, UnsupportedContentFormatException {
263+
// Determine supported formats for the content
264+
ContentFormat supportedFormat = determineSupportedFormat(id);
265+
// Validate the requested format
266+
if ((contentType == String.class && !supportedFormat.supportsString()) ||
267+
(contentType == InputStream.class && !supportedFormat.supportsStream()) ||
268+
(contentType == Document.class && !supportedFormat.supportsXml())) {
269+
throw new UnsupportedContentFormatException("Format not supported for content ID: " + id);
270+
}
271+
// Retrieve content as bytes
272+
byte[] rawContent = fetchContentFromStorage(id);
273+
// Convert content to the requested format
274+
if (contentType == String.class) {
275+
return contentType.cast(new String(rawContent, StandardCharsets.UTF_8));
276+
} else if (contentType == InputStream.class) {
277+
return contentType.cast(new ByteArrayInputStream(rawContent));
278+
} else if (contentType == Document.class) {
279+
return contentType.cast(parseXml(rawContent));
280+
}
281+
throw new IllegalArgumentException("Unsupported content type: " + contentType);
282+
}
283+
StreamContentHelper getStreamContentHelper();
284+
285+
// TODO here we always use AkubraUtils.getStreamContent but we have also AkubraObject.AkubraDatastream for fetching stream content
286+
private ContentFormat determineSupportedFormat(String id) {
287+
// Example logic to determine supported formats
288+
if (id.startsWith("streamOnly")) {
289+
return new ContentFormat(false, true, false);
290+
} else {
291+
return new ContentFormat(true, true, true);
292+
}
293+
}
294+
private byte[] fetchContentFromStorage(String id) {
295+
// Mock: Fetch content as bytes from your storage
296+
return ("<xml>Content for ID: " + id + "</xml>").getBytes(StandardCharsets.UTF_8);
297+
}
298+
private Document parseXml(byte[] content) throws IOException {
299+
try {
300+
DocumentBuilderFactory factory = DocumentBuilderFactory.newInstance();
301+
return factory.newDocumentBuilder().parse(new ByteArrayInputStream(content));
302+
} catch (Exception e) {
303+
throw new IOException("Failed to parse XML", e);
304+
}
305+
}
306+
307+
// TODO just one method rendering 2 content types; also we can add new par dsId and if null the whole foxml will be returned
308+
// <--- AkubraObject.getFoXml
309+
public org.dom4j.Document getDatastreamXml(String pid, String dsId) throws RepositoryException, IOException {
310+
Lock readLock = AkubraDOManager.getReadLock(pid);
311+
try {
312+
RepositoryObject object = akubraRepositoryImpl.getObject(pid);
313+
if (object.streamExists(dsId)) {
314+
org.dom4j.Document foxml = Utils.inputstreamToDocument(object.getFoxml(), true);
315+
org.dom4j.Element dcEl = (org.dom4j.Element) Dom4jUtils.buildXpath(String.format("/foxml:digitalObject/foxml:datastream[@ID='%s']", dsId)).selectSingleNode(foxml);
316+
org.dom4j.Element detached = (org.dom4j.Element) dcEl.detach();
317+
org.dom4j.Document result = DocumentHelper.createDocument();
318+
result.add(detached);
319+
return result;
320+
} else {
321+
return null;
322+
}
323+
} finally {
324+
readLock.unlock();
325+
}
326+
}
327+
@Override
328+
public InputStream getFoxml(String pid, boolean archive) throws IOException {
329+
try {
330+
if (archive){
331+
DigitalObject obj = manager.readObjectCloneFromStorage(pid);
332+
manager.resolveArchivedDatastreams(obj);
333+
return this.manager.marshallObject(obj);
334+
}else {
335+
return this.manager.retrieveObject(pid);
336+
}
337+
} catch (Exception e) {
338+
throw new IOException(e);
339+
}
340+
}
325341
@Override
342+
public org.dom4j.Document getFoxml(String pid) throws RepositoryException, IOException {
343+
Lock readLock = AkubraDOManager.getReadLock(pid);
344+
try {
345+
RepositoryObject object = akubraRepositoryImpl.getObject(pid);
346+
return Utils.inputstreamToDocument(object.getFoxml(), true);
347+
} finally {
348+
readLock.unlock();
349+
}
350+
}
351+
352+
// <--- AkubraObject.getStream.getContent (6x)
326353
public InputStream getLatestVersionOfDatastream(String pid, String dsId) throws RepositoryException, IOException {
327354
Lock readLock = AkubraDOManager.getReadLock(pid);
328355
try {
@@ -337,6 +364,62 @@ public InputStream getLatestVersionOfDatastream(String pid, String dsId) throws
337364
readLock.unlock();
338365
}
339366
}
367+
// <-- DigitalObject, AkubraUtils.getLastStreamVersion (3x)
368+
public InputStream getDataStream(String pid, String datastreamName) throws IOException {
369+
try {
370+
pid = makeSureObjectPid(pid);
371+
if (this.accessLog != null && this.accessLog.isReportingAccess(pid, datastreamName)) {
372+
reportAccess(pid, datastreamName);
373+
}
374+
DigitalObject object = manager.readObjectFromStorage(pid);
375+
if (object != null) {
376+
DatastreamVersionType stream = AkubraUtils.getLastStreamVersion(object, datastreamName);
377+
if (stream != null) {
378+
return AkubraUtils.getStreamContent(stream, manager);
379+
} else {
380+
throw new IOException("cannot find stream '" + datastreamName + "' for pid '" + pid + "'");
381+
}
382+
} else {
383+
throw new IOException("cannot find pid '" + pid + "'");
384+
}
385+
} catch (Exception e) {
386+
throw new IOException(e);
387+
}
388+
}
389+
390+
// getLatestVersionOfDatastream (4x)
391+
public org.dom4j.Document getLatestVersionOfInlineXmlDatastream(String pid, String dsId) throws RepositoryException, IOException {
392+
InputStream is = getLatestVersionOfDatastream(pid, dsId);
393+
return is == null ? null : Utils.inputstreamToDocument(is, true);
394+
}
395+
// getLatestVersionOfDatastream (1x)
396+
public String getLatestVersionOfManagedTextDatastream(String pid, String dsId) throws RepositoryException, IOException {
397+
InputStream is = getLatestVersionOfDatastream(pid, dsId);
398+
return is == null ? null : Utils.inputstreamToString(is);
399+
}
400+
401+
// <-- DigitalObject, AkubraUtils.getLastStreamVersion (3x)
402+
public Document getStream(String pid, String streamName) throws IOException {
403+
DigitalObject object = manager.readObjectFromStorage(pid);
404+
if (object != null) {
405+
DatastreamVersionType stream = AkubraUtils.getLastStreamVersion(object, streamName);
406+
if (stream != null) {
407+
if (stream.getXmlContent() != null) {
408+
List<Element> elementList = stream.getXmlContent().getAny();
409+
if (!elementList.isEmpty()) {
410+
return elementList.get(0).getOwnerDocument();
411+
} else {
412+
throw new IOException("Datastream not found: " + pid + " - " + streamName);
413+
}
414+
} else {
415+
throw new IOException("Expected XML datastream: " + pid + " - " + streamName);
416+
}
417+
}
418+
throw new IOException("Datastream not found: " + pid + " - " + streamName);
419+
}
420+
throw new IOException("Object not found: " + pid);
421+
}
422+
340423
@Override
341424
public InputStream getImgFull(String pid) throws IOException, RepositoryException {
342425
this.accessLog.reportAccess(pid, KnownDatastreams.IMG_FULL.toString());
@@ -385,22 +468,10 @@ public InputStream getSmallThumbnail(String pid) throws IOException {
385468
public InputStream getImageFULL(String pid) throws IOException {
386469
return getDataStream(pid, FedoraUtils.IMG_FULL_STREAM);
387470
}
388-
// text data stream
389-
@Override
390-
public String getLatestVersionOfManagedTextDatastream(String pid, String dsId) throws RepositoryException, IOException {
391-
InputStream is = getLatestVersionOfDatastream(pid, dsId);
392-
return is == null ? null : Utils.inputstreamToString(is);
393-
}
394471
@Override
395472
public String getOcrText(String pid) throws IOException, RepositoryException {
396473
return getLatestVersionOfManagedTextDatastream(pid, KnownDatastreams.OCR_TEXT.toString());
397474
}
398-
// XML data stream
399-
@Override
400-
public org.dom4j.Document getLatestVersionOfInlineXmlDatastream(String pid, String dsId) throws RepositoryException, IOException {
401-
InputStream is = getLatestVersionOfDatastream(pid, dsId);
402-
return is == null ? null : Utils.inputstreamToDocument(is, true);
403-
}
404475
@Override
405476
public org.dom4j.Document getRelsExt(String pid, boolean namespaceAware) throws IOException, RepositoryException {
406477
org.dom4j.Document doc = getLatestVersionOfInlineXmlDatastream(pid, KnownDatastreams.RELS_EXT.toString());
@@ -458,6 +529,7 @@ public org.dom4j.Document getOcrAlto(String pid, boolean namespaceAware) throws
458529
}
459530
return doc;
460531
}
532+
//----------------------------------------------------------------------------------------------------------------
461533
// --- check stream exists
462534
@Override
463535
public boolean isStreamAvailable(String pid, String dsId) throws IOException, RepositoryException {

‎shared/common/src/main/java/cz/incad/kramerius/fedora/impl/tmp/ContentFormat.java

Lines changed: 25 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,25 @@
1+
package cz.incad.kramerius.fedora.impl.tmp;
2+
3+
public class ContentFormat {
4+
private final boolean supportsString;
5+
private final boolean supportsStream;
6+
private final boolean supportsXml;
7+
8+
public ContentFormat(boolean supportsString, boolean supportsStream, boolean supportsXml) {
9+
this.supportsString = supportsString;
10+
this.supportsStream = supportsStream;
11+
this.supportsXml = supportsXml;
12+
}
13+
14+
public boolean supportsString() {
15+
return supportsString;
16+
}
17+
18+
public boolean supportsStream() {
19+
return supportsStream;
20+
}
21+
22+
public boolean supportsXml() {
23+
return supportsXml;
24+
}
25+
}

‎shared/common/src/main/java/cz/incad/kramerius/fedora/impl/tmp/UnsupportedContentFormatException.java

Lines changed: 7 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,7 @@
1+
package cz.incad.kramerius.fedora.impl.tmp;
2+
3+
public class UnsupportedContentFormatException extends Exception {
4+
public UnsupportedContentFormatException(String message) {
5+
super(message);
6+
}
7+
}

0 commit comments

Comments
 (0)