001/**
002 * Copyright 2014 DuraSpace, Inc.
003 *
004 * Licensed under the Apache License, Version 2.0 (the "License");
005 * you may not use this file except in compliance with the License.
006 * You may obtain a copy of the License at
007 *
008 *     http://www.apache.org/licenses/LICENSE-2.0
009 *
010 * Unless required by applicable law or agreed to in writing, software
011 * distributed under the License is distributed on an "AS IS" BASIS,
012 * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
013 * See the License for the specific language governing permissions and
014 * limitations under the License.
015 */
016package org.fcrepo.http.api;
017
018
019import static com.google.common.base.Predicates.alwaysTrue;
020import static com.google.common.base.Predicates.and;
021import static com.google.common.base.Predicates.not;
022import static com.google.common.collect.Iterators.concat;
023import static com.google.common.collect.Iterators.filter;
024import static com.google.common.collect.Iterators.transform;
025import static com.hp.hpl.jena.rdf.model.ModelFactory.createDefaultModel;
026import static javax.ws.rs.core.HttpHeaders.CACHE_CONTROL;
027import static javax.ws.rs.core.MediaType.APPLICATION_OCTET_STREAM_TYPE;
028import static javax.ws.rs.core.Response.ok;
029import static javax.ws.rs.core.Response.status;
030import static javax.ws.rs.core.Response.Status.PARTIAL_CONTENT;
031import static javax.ws.rs.core.Response.Status.REQUESTED_RANGE_NOT_SATISFIABLE;
032import static javax.ws.rs.core.Response.temporaryRedirect;
033import static org.apache.commons.lang.StringUtils.isBlank;
034import static org.apache.jena.riot.RDFLanguages.contentTypeToLang;
035import static org.fcrepo.kernel.FedoraJcrTypes.LDP_BASIC_CONTAINER;
036import static org.fcrepo.kernel.FedoraJcrTypes.LDP_DIRECT_CONTAINER;
037import static org.fcrepo.kernel.FedoraJcrTypes.LDP_INDIRECT_CONTAINER;
038import static org.fcrepo.kernel.RdfLexicon.BASIC_CONTAINER;
039import static org.fcrepo.kernel.RdfLexicon.CONTAINER;
040import static org.fcrepo.kernel.RdfLexicon.DIRECT_CONTAINER;
041import static org.fcrepo.kernel.RdfLexicon.INDIRECT_CONTAINER;
042import static org.fcrepo.kernel.RdfLexicon.LDP_NAMESPACE;
043import static org.fcrepo.kernel.RdfLexicon.isManagedNamespace;
044import static org.slf4j.LoggerFactory.getLogger;
045
046import java.io.IOException;
047import java.io.InputStream;
048import java.net.URI;
049import java.net.URISyntaxException;
050import java.util.Date;
051import java.util.Iterator;
052
053import javax.inject.Inject;
054import javax.jcr.Binary;
055import javax.jcr.RepositoryException;
056import javax.jcr.Session;
057import javax.servlet.http.HttpServletResponse;
058import javax.ws.rs.BadRequestException;
059import javax.ws.rs.BeanParam;
060import javax.ws.rs.WebApplicationException;
061import javax.ws.rs.core.CacheControl;
062import javax.ws.rs.core.Context;
063import javax.ws.rs.core.EntityTag;
064import javax.ws.rs.core.MediaType;
065import javax.ws.rs.core.Request;
066import javax.ws.rs.core.Response;
067
068import org.apache.jena.riot.Lang;
069import org.fcrepo.http.commons.api.rdf.HttpTripleUtil;
070import org.fcrepo.http.commons.domain.MultiPrefer;
071import org.fcrepo.http.commons.domain.PreferTag;
072import org.fcrepo.http.commons.domain.Range;
073import org.fcrepo.http.commons.domain.ldp.LdpPreferTag;
074import org.fcrepo.http.commons.responses.RangeRequestInputStream;
075import org.fcrepo.kernel.exception.InvalidChecksumException;
076import org.fcrepo.kernel.exception.MalformedRdfException;
077import org.fcrepo.kernel.exception.RepositoryRuntimeException;
078import org.fcrepo.kernel.impl.rdf.ManagedRdf;
079import org.fcrepo.kernel.impl.rdf.impl.AclRdfContext;
080import org.fcrepo.kernel.impl.rdf.impl.BlankNodeRdfContext;
081import org.fcrepo.kernel.impl.rdf.impl.ChildrenRdfContext;
082import org.fcrepo.kernel.impl.rdf.impl.ContentRdfContext;
083import org.fcrepo.kernel.impl.rdf.impl.HashRdfContext;
084import org.fcrepo.kernel.impl.rdf.impl.LdpContainerRdfContext;
085import org.fcrepo.kernel.impl.rdf.impl.LdpIsMemberOfRdfContext;
086import org.fcrepo.kernel.impl.rdf.impl.LdpRdfContext;
087import org.fcrepo.kernel.impl.rdf.impl.ParentRdfContext;
088import org.fcrepo.kernel.impl.rdf.impl.PropertiesRdfContext;
089import org.fcrepo.kernel.impl.rdf.impl.ReferencesRdfContext;
090import org.fcrepo.kernel.impl.rdf.impl.RootRdfContext;
091import org.fcrepo.kernel.impl.rdf.impl.TypeRdfContext;
092import org.fcrepo.kernel.impl.services.TransactionServiceImpl;
093import org.fcrepo.kernel.models.Container;
094import org.fcrepo.kernel.models.FedoraBinary;
095import org.fcrepo.kernel.models.FedoraResource;
096import org.fcrepo.kernel.models.NonRdfSource;
097import org.fcrepo.kernel.models.NonRdfSourceDescription;
098import org.fcrepo.kernel.services.policy.StoragePolicyDecisionPoint;
099import org.fcrepo.kernel.utils.iterators.RdfStream;
100import org.glassfish.jersey.media.multipart.ContentDisposition;
101import org.jvnet.hk2.annotations.Optional;
102import org.slf4j.Logger;
103
104import com.google.common.base.Function;
105import com.google.common.base.Predicate;
106import com.google.common.collect.ImmutableList;
107import com.google.common.collect.Iterators;
108import com.hp.hpl.jena.graph.Triple;
109import com.hp.hpl.jena.rdf.model.Model;
110import com.hp.hpl.jena.rdf.model.Statement;
111import com.hp.hpl.jena.vocabulary.RDF;
112
113/**
114 * An abstract class that sits between AbstractResource and any resource that
115 * wishes to share the routines for building responses containing binary
116 * content.
117 *
118 * @author Mike Durbin
119 */
120public abstract class ContentExposingResource extends FedoraBaseResource {
121
122    private static final Logger LOGGER = getLogger(ContentExposingResource.class);
123    public static final MediaType MESSAGE_EXTERNAL_BODY = MediaType.valueOf("message/external-body");
124
125    @Context protected Request request;
126    @Context protected HttpServletResponse servletResponse;
127
128    @Inject
129    @Optional
130    private HttpTripleUtil httpTripleUtil;
131
132    @BeanParam
133    protected MultiPrefer prefer;
134
135    @Inject
136    @Optional
137    StoragePolicyDecisionPoint storagePolicyDecisionPoint;
138
139    protected FedoraResource resource;
140
141    private static long MAX_BUFFER_SIZE = 10240000;
142
143    protected abstract String externalPath();
144
145    protected Response getContent(final String rangeValue,
146                                  final RdfStream rdfStream) throws IOException {
147        if (resource() instanceof FedoraBinary) {
148
149            final String contentTypeString = ((FedoraBinary) resource()).getMimeType();
150
151            final Lang lang = contentTypeToLang(contentTypeString);
152
153            if (!contentTypeString.equals("text/plain") && lang != null) {
154
155                final String format = lang.getName().toUpperCase();
156
157                final InputStream content = ((FedoraBinary) resource()).getContent();
158
159                final Model inputModel = createDefaultModel()
160                        .read(content,  (resource()).toString(), format);
161
162                rdfStream.concat(Iterators.transform(inputModel.listStatements(),
163                        new Function<Statement, Triple>() {
164
165                            @Override
166                            public Triple apply(final Statement input) {
167                                return input.asTriple();
168                            }
169                        }));
170            } else {
171
172                final MediaType mediaType = MediaType.valueOf(contentTypeString);
173                if (MESSAGE_EXTERNAL_BODY.isCompatible(mediaType)
174                        && mediaType.getParameters().containsKey("access-type")
175                        && mediaType.getParameters().get("access-type").equals("URL")
176                        && mediaType.getParameters().containsKey("URL") ) {
177                    try {
178                        return temporaryRedirect(new URI(mediaType.getParameters().get("URL"))).build();
179                    } catch (final URISyntaxException e) {
180                        throw new RepositoryRuntimeException(e);
181                    }
182                } else {
183                    return getBinaryContent(rangeValue);
184                }
185            }
186
187        } else {
188            rdfStream.concat(getResourceTriples());
189
190            if (prefer != null) {
191                prefer.getReturn().addResponseHeaders(servletResponse);
192            }
193
194        }
195        servletResponse.addHeader("Vary", "Accept, Range, Accept-Encoding, Accept-Language");
196
197        return Response.ok(rdfStream).build();
198    }
199
200    protected RdfStream getResourceTriples() {
201
202        final PreferTag returnPreference;
203
204        if (prefer != null && prefer.hasReturn()) {
205            returnPreference = prefer.getReturn();
206        } else if (prefer != null && prefer.hasHandling()) {
207            returnPreference = prefer.getHandling();
208        } else {
209            returnPreference = PreferTag.emptyTag();
210        }
211
212        final LdpPreferTag ldpPreferences = new LdpPreferTag(returnPreference);
213
214        final RdfStream rdfStream = new RdfStream();
215
216        final Predicate<Triple> tripleFilter;
217        if (ldpPreferences.prefersServerManaged()) {
218            tripleFilter = alwaysTrue();
219        } else {
220            tripleFilter = and(not(ManagedRdf.isManagedTriple), not(new Predicate<Triple>() {
221                @Override
222                public boolean apply(final Triple input) {
223                    return input.getPredicate().equals(RDF.type.asNode())
224                            && isManagedNamespace.apply(input.getObject().getNameSpace());
225                }
226            }));
227        }
228
229        if (ldpPreferences.prefersServerManaged()) {
230            rdfStream.concat(getTriples(LdpRdfContext.class));
231        }
232
233        rdfStream.concat(filter(getTriples(TypeRdfContext.class), tripleFilter));
234
235        rdfStream.concat(filter(getTriples(PropertiesRdfContext.class), tripleFilter));
236
237        if (!returnPreference.getValue().equals("minimal")) {
238
239            // Additional server-managed triples about this resource
240            if (ldpPreferences.prefersServerManaged()) {
241                rdfStream.concat(getTriples(AclRdfContext.class));
242                rdfStream.concat(getTriples(RootRdfContext.class));
243                rdfStream.concat(getTriples(ContentRdfContext.class));
244                rdfStream.concat(getTriples(ParentRdfContext.class));
245            }
246
247            // containment triples about this resource
248            if (ldpPreferences.prefersContainment()) {
249                rdfStream.concat(getTriples(ChildrenRdfContext.class));
250            }
251
252            // LDP container membership triples for this resource
253            if (ldpPreferences.prefersMembership()) {
254                rdfStream.concat(getTriples(LdpContainerRdfContext.class));
255                rdfStream.concat(getTriples(LdpIsMemberOfRdfContext.class));
256            }
257
258            // Include binary properties if this is a binary description
259            if (resource() instanceof NonRdfSourceDescription) {
260                final FedoraResource described = ((NonRdfSourceDescription) resource()).getDescribedResource();
261                rdfStream.concat(filter(described.getTriples(translator(), ImmutableList.of(TypeRdfContext.class,
262                        PropertiesRdfContext.class,
263                        ContentRdfContext.class)), tripleFilter));
264            }
265
266            // Embed all hash and blank nodes
267            rdfStream.concat(filter(getTriples(HashRdfContext.class), tripleFilter));
268            rdfStream.concat(filter(getTriples(BlankNodeRdfContext.class), tripleFilter));
269
270            // Include inbound references to this object
271            if (ldpPreferences.prefersReferences()) {
272                rdfStream.concat(getTriples(ReferencesRdfContext.class));
273            }
274
275            // Embed the children of this object
276            if (ldpPreferences.prefersEmbed()) {
277
278                final Iterator<FedoraResource> children = resource().getChildren();
279
280                rdfStream.concat(filter(concat(transform(children,
281                        new Function<FedoraResource, RdfStream>() {
282
283                            @Override
284                            public RdfStream apply(final FedoraResource child) {
285                                return child.getTriples(translator(), ImmutableList.of(
286                                        TypeRdfContext.class,
287                                        PropertiesRdfContext.class,
288                                        BlankNodeRdfContext.class));
289                            }
290                        })), tripleFilter));
291
292            }
293        }
294
295        if (httpTripleUtil != null && ldpPreferences.prefersServerManaged()) {
296            httpTripleUtil.addHttpComponentModelsForResourceToStream(rdfStream, resource(), uriInfo, translator());
297        }
298
299
300        return rdfStream;
301    }
302
303    /**
304     * Get the binary content of a datastream
305     *
306     * @return Binary blob
307     * @throws RepositoryException
308     */
309    protected Response getBinaryContent(final String rangeValue)
310            throws IOException {
311            final FedoraBinary binary = (FedoraBinary)resource();
312
313            // we include an explicit etag, because the default behavior is to use the JCR node's etag, not
314            // the jcr:content node digest. The etag is only included if we are not within a transaction.
315            final String txId = TransactionServiceImpl.getCurrentTransactionId(session());
316            if (txId == null) {
317                checkCacheControlHeaders(request, servletResponse, binary, session());
318            }
319            final CacheControl cc = new CacheControl();
320            cc.setMaxAge(0);
321            cc.setMustRevalidate(true);
322            Response.ResponseBuilder builder;
323
324            if (rangeValue != null && rangeValue.startsWith("bytes")) {
325
326                final Range range = Range.convert(rangeValue);
327
328                final long contentSize = binary.getContentSize();
329
330                final String endAsString;
331
332                if (range.end() == -1) {
333                    endAsString = Long.toString(contentSize - 1);
334                } else {
335                    endAsString = Long.toString(range.end());
336                }
337
338                final String contentRangeValue =
339                        String.format("bytes %s-%s/%s", range.start(),
340                                endAsString, contentSize);
341
342                if (range.end() > contentSize ||
343                        (range.end() == -1 && range.start() > contentSize)) {
344
345                    builder = status(REQUESTED_RANGE_NOT_SATISFIABLE)
346                            .header("Content-Range", contentRangeValue);
347                } else {
348                    final long maxBufferSize = MAX_BUFFER_SIZE; // 10MB max buffer size?
349                    final long rangeStart = range.start();
350                    final long rangeSize = range.size() == -1 ? contentSize - rangeStart : range.size();
351                    final long remainingBytes = contentSize - rangeStart;
352                    final long bufSize = rangeSize < remainingBytes ? rangeSize : remainingBytes;
353
354                    if (bufSize < maxBufferSize) {
355                        // Small size range content retrieval use javax.jcr.Binary to improve performance
356                        final byte[] buf = new byte[(int) bufSize];
357
358                        final Binary binaryContent = binary.getBinaryContent();
359                        try {
360                            binaryContent.read(buf, rangeStart);
361                        } catch (RepositoryException e1) {
362                            throw new RepositoryRuntimeException(e1);
363                        }
364                        binaryContent.dispose();
365
366                        builder = status(PARTIAL_CONTENT).entity(buf)
367                                .header("Content-Range", contentRangeValue);
368                    } else {
369                        // For large range content retrieval, go with the InputStream class to balance
370                        // the memory usage, though this is a rare case in range content retrieval.
371                        final InputStream content = binary.getContent();
372                        final RangeRequestInputStream rangeInputStream =
373                                new RangeRequestInputStream(content, range.start(), range.size());
374
375                        builder = status(PARTIAL_CONTENT).entity(rangeInputStream)
376                                .header("Content-Range", contentRangeValue);
377                    }
378                }
379
380            } else {
381                final InputStream content = binary.getContent();
382                builder = ok(content);
383            }
384
385
386            // we set the content-type explicitly to avoid content-negotiation from getting in the way
387            return builder.type(binary.getMimeType())
388                    .cacheControl(cc)
389                    .build();
390
391        }
392
393    protected RdfStream getTriples(final Class<? extends RdfStream> x) {
394        return getTriples(resource(), x);
395    }
396
397    protected RdfStream getTriples(final FedoraResource resource, final Class<? extends RdfStream> x) {
398        return resource.getTriples(translator(), x);
399    }
400
401    protected URI getUri(final FedoraResource resource) {
402        try {
403            final String uri = translator().reverse().convert(resource).getURI();
404            return new URI(uri);
405        } catch (final URISyntaxException e) {
406            throw new BadRequestException(e);
407        }
408    }
409
410    protected FedoraResource resource() {
411        if (resource == null) {
412            resource = getResourceFromPath(externalPath());
413        }
414
415        return resource;
416    }
417
418
419    /**
420     * Add any resource-specific headers to the response
421     * @param resource
422     */
423    protected void addResourceHttpHeaders(final FedoraResource resource) {
424        if (resource instanceof FedoraBinary) {
425
426            final FedoraBinary binary = (FedoraBinary)resource;
427            final ContentDisposition contentDisposition = ContentDisposition.type("attachment")
428                    .fileName(binary.getFilename())
429                    .creationDate(binary.getCreatedDate())
430                    .modificationDate(binary.getLastModifiedDate())
431                    .size(binary.getContentSize())
432                    .build();
433
434            servletResponse.addHeader("Content-Type", binary.getMimeType());
435            servletResponse.addHeader("Content-Length", String.valueOf(binary.getContentSize()));
436            servletResponse.addHeader("Accept-Ranges", "bytes");
437            servletResponse.addHeader("Content-Disposition", contentDisposition.toString());
438        }
439
440        servletResponse.addHeader("Link", "<" + LDP_NAMESPACE + "Resource>;rel=\"type\"");
441
442        if (resource instanceof NonRdfSource) {
443            servletResponse.addHeader("Link", "<" + LDP_NAMESPACE + "NonRDFSource>;rel=\"type\"");
444        } else if (resource instanceof Container) {
445            servletResponse.addHeader("Link", "<" + CONTAINER.getURI() + ">;rel=\"type\"");
446            if (resource.hasType(LDP_BASIC_CONTAINER)) {
447                servletResponse.addHeader("Link", "<" + BASIC_CONTAINER.getURI() + ">;rel=\"type\"");
448            } else if (resource.hasType(LDP_DIRECT_CONTAINER)) {
449                servletResponse.addHeader("Link", "<" + DIRECT_CONTAINER.getURI() + ">;rel=\"type\"");
450            } else if (resource.hasType(LDP_INDIRECT_CONTAINER)) {
451                servletResponse.addHeader("Link", "<" + INDIRECT_CONTAINER.getURI() + ">;rel=\"type\"");
452            } else {
453                servletResponse.addHeader("Link", "<" + BASIC_CONTAINER.getURI() + ">;rel=\"type\"");
454            }
455        } else {
456            servletResponse.addHeader("Link", "<" + LDP_NAMESPACE + "RDFSource>;rel=\"type\"");
457        }
458
459    }
460
461    /**
462     * Evaluate the cache control headers for the request to see if it can be served from
463     * the cache.
464     *
465     * @param request
466     * @param servletResponse
467     * @param resource
468     * @param session
469     * @throws javax.jcr.RepositoryException
470     */
471    protected static void checkCacheControlHeaders(final Request request,
472                                                   final HttpServletResponse servletResponse,
473                                                   final FedoraResource resource,
474                                                   final Session session) {
475        evaluateRequestPreconditions(request, servletResponse, resource, session, true);
476        addCacheControlHeaders(servletResponse, resource, session);
477    }
478
479    /**
480     * Add ETag and Last-Modified cache control headers to the response
481     * @param servletResponse
482     * @param resource
483     */
484    protected static void addCacheControlHeaders(final HttpServletResponse servletResponse,
485                                                 final FedoraResource resource,
486                                                 final Session session) {
487
488        final String txId = TransactionServiceImpl.getCurrentTransactionId(session);
489        if (txId != null) {
490            // Do not add caching headers if in a transaction
491            return;
492        }
493
494        final EntityTag etag = new EntityTag(resource.getEtagValue());
495        final Date date = resource.getLastModifiedDate();
496
497        if (!etag.getValue().isEmpty()) {
498            servletResponse.addHeader("ETag", etag.toString());
499        }
500
501        if (date != null) {
502            servletResponse.addDateHeader("Last-Modified", date.getTime());
503        }
504    }
505
506    /**
507     * Evaluate request preconditions to ensure the resource is the expected state
508     * @param request
509     * @param resource
510     */
511    protected static void evaluateRequestPreconditions(final Request request,
512                                                       final HttpServletResponse servletResponse,
513                                                       final FedoraResource resource,
514                                                       final Session session) {
515        evaluateRequestPreconditions(request, servletResponse, resource, session, false);
516    }
517
518    private static void evaluateRequestPreconditions(final Request request,
519                                                     final HttpServletResponse servletResponse,
520                                                     final FedoraResource resource,
521                                                     final Session session,
522                                                     final boolean cacheControl) {
523
524        final String txId = TransactionServiceImpl.getCurrentTransactionId(session);
525        if (txId != null) {
526            // Force cache revalidation if in a transaction
527            servletResponse.addHeader(CACHE_CONTROL, "must-revalidate");
528            servletResponse.addHeader(CACHE_CONTROL, "max-age=0");
529            return;
530        }
531
532        final EntityTag etag = new EntityTag(resource.getEtagValue());
533        final Date date = resource.getLastModifiedDate();
534        final Date roundedDate = new Date();
535
536        if (date != null) {
537            roundedDate.setTime(date.getTime() - date.getTime() % 1000);
538        }
539
540        Response.ResponseBuilder builder = request.evaluatePreconditions(etag);
541        if ( builder != null ) {
542            builder = builder.entity("ETag mismatch");
543        } else {
544            builder = request.evaluatePreconditions(roundedDate);
545            if ( builder != null ) {
546                builder = builder.entity("Date mismatch");
547            }
548        }
549
550        if (builder != null && cacheControl ) {
551            final CacheControl cc = new CacheControl();
552            cc.setMaxAge(0);
553            cc.setMustRevalidate(true);
554            // here we are implicitly emitting a 304
555            // the exception is not an error, it's genuinely
556            // an exceptional condition
557            builder = builder.cacheControl(cc).lastModified(date).tag(etag);
558        }
559        if (builder != null) {
560            throw new WebApplicationException(builder.build());
561        }
562    }
563
564    protected static MediaType getSimpleContentType(final MediaType requestContentType) {
565        return requestContentType != null ? new MediaType(requestContentType.getType(), requestContentType.getSubtype())
566                : APPLICATION_OCTET_STREAM_TYPE;
567    }
568
569    protected static boolean isRdfContentType(final String contentTypeString) {
570        return contentTypeToLang(contentTypeString) != null;
571    }
572
573    protected void replaceResourceBinaryWithStream(final FedoraBinary result,
574                                                   final InputStream requestBodyStream,
575                                                   final ContentDisposition contentDisposition,
576                                                   final MediaType contentType,
577                                                   final String checksum) throws InvalidChecksumException {
578        final URI checksumURI = checksumURI(checksum);
579        final String originalFileName = contentDisposition != null ? contentDisposition.getFileName() : "";
580        final String originalContentType = contentType != null ? contentType.toString() : "";
581
582        result.setContent(requestBodyStream,
583                originalContentType,
584                checksumURI,
585                originalFileName,
586                storagePolicyDecisionPoint);
587    }
588
589    protected void replaceResourceWithStream(final FedoraResource resource,
590                                             final InputStream requestBodyStream,
591                                             final MediaType contentType,
592                                             final RdfStream resourceTriples) throws MalformedRdfException {
593        final Lang format = contentTypeToLang(contentType.toString());
594
595        final Model inputModel = createDefaultModel()
596                .read(requestBodyStream, getUri(resource).toString(), format.getName().toUpperCase());
597
598        resource.replaceProperties(translator(), inputModel, resourceTriples);
599    }
600
601    protected void patchResourcewithSparql(final FedoraResource resource,
602                                           final String requestBody,
603                                           final RdfStream resourceTriples) throws MalformedRdfException {
604        resource.updateProperties(translator(), requestBody, resourceTriples);
605    }
606
607    /**
608     * Create a checksum URI object.
609     **/
610    private static URI checksumURI( final String checksum ) {
611        if (!isBlank(checksum)) {
612            return URI.create(checksum);
613        }
614        return null;
615    }
616}