Source code

001/*
002 * The contents of this file are subject to the license and copyright
003 * detailed in the LICENSE and NOTICE files at the root of the source
004 * tree.
005 */
006package org.fcrepo.persistence.ocfl.impl;
007
008import static org.apache.jena.graph.NodeFactory.createURI;
009import static org.apache.jena.rdf.model.ModelFactory.createDefaultModel;
010import static org.fcrepo.kernel.api.RdfLexicon.NON_RDF_SOURCE;
011import static org.fcrepo.persistence.ocfl.impl.OcflPersistentStorageUtils.getRdfFormat;
012import static org.slf4j.LoggerFactory.getLogger;
013
014import java.io.IOException;
015import java.io.InputStream;
016import java.util.ArrayList;
017import java.util.List;
018import java.util.Optional;
019import java.util.concurrent.Callable;
020import java.util.concurrent.atomic.AtomicReference;
021
022import javax.inject.Inject;
023import javax.validation.constraints.NotNull;
024
025import edu.wisc.library.ocfl.api.OcflRepository;
026import org.fcrepo.config.FedoraPropsConfig;
027import org.fcrepo.kernel.api.ContainmentIndex;
028import org.fcrepo.kernel.api.RdfLexicon;
029import org.fcrepo.kernel.api.RdfStream;
030import org.fcrepo.kernel.api.Transaction;
031import org.fcrepo.kernel.api.exception.RepositoryRuntimeException;
032import org.fcrepo.kernel.api.identifiers.FedoraId;
033import org.fcrepo.kernel.api.models.ResourceHeaders;
034import org.fcrepo.kernel.api.rdf.DefaultRdfStream;
035import org.fcrepo.kernel.api.services.MembershipService;
036import org.fcrepo.kernel.api.services.ReferenceService;
037import org.fcrepo.persistence.api.PersistentStorageSessionManager;
038import org.fcrepo.persistence.ocfl.api.FedoraToOcflObjectIndex;
039import org.fcrepo.search.api.Condition;
040import org.fcrepo.search.api.InvalidQueryException;
041import org.fcrepo.search.api.SearchIndex;
042import org.fcrepo.search.api.SearchParameters;
043import org.fcrepo.storage.ocfl.OcflObjectSessionFactory;
044import org.fcrepo.storage.ocfl.validation.ObjectValidator;
045
046import org.apache.jena.rdf.model.Model;
047import org.apache.jena.riot.RDFDataMgr;
048import org.slf4j.Logger;
049import org.springframework.beans.factory.annotation.Autowired;
050import org.springframework.beans.factory.annotation.Qualifier;
051import org.springframework.stereotype.Component;
052
053/**
054 * Service that does the reindexing for one OCFL object.
055 * @author whikloj
056 */
057@Component
058public class ReindexService {
059
060    @Inject
061    private PersistentStorageSessionManager persistentStorageSessionManager;
062
063    @Inject
064    private OcflObjectSessionFactory ocflObjectSessionFactory;
065
066    @Autowired
067    @Qualifier("ocflIndex")
068    private FedoraToOcflObjectIndex ocflIndex;
069
070    @Autowired
071    private OcflRepository ocflRepository;
072
073    @Autowired
074    @Qualifier("containmentIndex")
075    private ContainmentIndex containmentIndex;
076
077    @Autowired
078    @Qualifier("searchIndex")
079    private SearchIndex searchIndex;
080
081    @Autowired
082    @Qualifier("referenceService")
083    private ReferenceService referenceService;
084
085    @Inject
086    private MembershipService membershipService;
087
088    @Inject
089    private ObjectValidator objectValidator;
090
091    @Inject
092    private FedoraPropsConfig config;
093
094    private static final Logger LOGGER = getLogger(ReindexService.class);
095
096    private int membershipPageSize = 500;
097
098    public void indexOcflObject(final Transaction tx, final String ocflId) {
099        LOGGER.debug("Indexing ocflId {} in transaction {}", ocflId, tx.getId());
100
101        ocflRepository.invalidateCache(ocflId);
102        objectValidator.validate(ocflId, config.isRebuildFixityCheck());
103
104        try (final var session = ocflObjectSessionFactory.newSession(ocflId)) {
105            final var rootId = new AtomicReference<FedoraId>();
106            final var fedoraIds = new ArrayList<FedoraId>();
107            final var headersList = new ArrayList<ResourceHeaders>();
108
109            session.invalidateCache(ocflId);
110            session.streamResourceHeaders().forEach(storageHeaders -> {
111                final var headers = new ResourceHeadersAdapter(storageHeaders);
112
113                final var fedoraId = headers.getId();
114                fedoraIds.add(fedoraId);
115                if (headers.isArchivalGroup() || headers.isObjectRoot()) {
116                    rootId.set(fedoraId);
117                }
118
119                if (!fedoraId.isRepositoryRoot()) {
120                    var parentId = headers.getParent();
121
122                    if (headers.getParent() == null) {
123                        if (headers.isObjectRoot()) {
124                            parentId = FedoraId.getRepositoryRootId();
125                        } else {
126                            throw new IllegalStateException(
127                                    String.format("Resource %s must have a parent defined", fedoraId.getFullId()));
128                        }
129                    }
130                    final var created = headers.getCreatedDate();
131                    if (!headers.isDeleted()) {
132                        if (!headers.getInteractionModel().equals(NON_RDF_SOURCE.toString())) {
133                            final Optional<InputStream> content = session.readContent(fedoraId.getFullId())
134                                    .getContentStream();
135                            if (content.isPresent()) {
136                                try (final var stream = content.get()) {
137                                    final RdfStream rdf = parseRdf(fedoraId, stream);
138                                    this.referenceService.updateReferences(tx, fedoraId, null, rdf);
139                                } catch (final IOException e) {
140                                    LOGGER.warn("Content stream for {} closed prematurely, inbound references skipped.",
141                                            fedoraId.getFullId());
142                                    throw new RepositoryRuntimeException(e.getMessage(), e);
143                                }
144                            }
145                        }
146
147                        this.containmentIndex.addContainedBy(tx, parentId, fedoraId, created, null);
148                        headersList.add(headers.asKernelHeaders());
149                    } else {
150                        final var deleted = headers.getLastModifiedDate();
151                        this.containmentIndex.addContainedBy(tx, parentId, fedoraId, created, deleted);
152                    }
153                }
154            });
155
156            if (rootId.get() == null) {
157                throw new IllegalStateException(String.format("Failed to find the root resource in object " +
158                        "identified by %s. Please ensure that the object ID you are attempting to index " +
159                        "refers to a corresponding valid Fedora-flavored object in the OCFL repository. Additionally " +
160                        "be sure that the object ID corresponds with the object root resource (as opposed to child " +
161                        "resources within the object).", ocflId));
162            }
163
164            fedoraIds.forEach(fedoraIdentifier -> {
165                final var rootFedoraIdentifier = rootId.get();
166                ocflIndex.addMapping(tx, fedoraIdentifier, rootFedoraIdentifier, ocflId);
167                LOGGER.debug("Rebuilt fedora-to-ocfl object index entry for {}", fedoraIdentifier);
168            });
169
170            headersList.forEach(headers -> {
171                searchIndex.addUpdateIndex(tx, headers);
172                LOGGER.debug("Rebuilt searchIndex for {}", headers.getId());
173            });
174        }
175    }
176
177    /**
178     * Remove persistent sessions for a transaction to avoid memory leaks.
179     * @param transactionId the transaction id.
180     */
181    public void cleanupSession(final String transactionId) {
182        persistentStorageSessionManager.removeSession(transactionId);
183    }
184
185    /**
186     * Set the membership page size.
187     * @param pageSize the new page size.
188     */
189    public void setMembershipPageSize(final int pageSize) {
190        membershipPageSize = pageSize;
191    }
192
193    /**
194     * Reset all the indexes.
195     */
196    public void reset() {
197        ocflIndex.reset();
198        containmentIndex.reset();
199        searchIndex.reset();
200        referenceService.reset();
201        membershipService.reset();
202    }
203
204    /**
205     * Index all membership properties by querying for Direct containers, and then
206     * trying population of the membership index for each one
207     * @param transaction the transaction id.
208     */
209    public void indexMembership(final Transaction transaction) {
210        LOGGER.debug("Starting indexMembership for transaction {}", transaction);
211        final var fields = List.of(Condition.Field.FEDORA_ID);
212        final var conditions = List.of(Condition.fromEnums(Condition.Field.RDF_TYPE, Condition.Operator.EQ,
213                RdfLexicon.DIRECT_CONTAINER.getURI()));
214        int offset = 0;
215
216        try {
217            int numResults;
218            do {
219                final var params = new SearchParameters(fields, conditions, membershipPageSize,
220                        offset, Condition.Field.FEDORA_ID, "asc", false);
221
222                final var searchResult = searchIndex.doSearch(params);
223                final var resultList = searchResult.getItems();
224                numResults = resultList.size();
225
226                resultList.stream()
227                        .map(entry -> FedoraId.create((String) entry.get(Condition.Field.FEDORA_ID.toString())))
228                        .forEach(containerId -> membershipService.populateMembershipHistory(transaction, containerId));
229
230                // Results are paged, so step through pages until we reach the last one
231                offset += membershipPageSize;
232            } while (numResults == membershipPageSize);
233
234        } catch (final InvalidQueryException e) {
235            throw new RepositoryRuntimeException("Failed to repopulate membership history", e);
236        }
237        LOGGER.debug("Finished indexMembership for transaction {}", transaction);
238    }
239
240    /**
241     * Rollback changes in the transaction.
242     * @param tx the transaction
243     */
244    public void rollbackMembership(@NotNull final Transaction tx) {
245        execQuietly("Failed to rollback membership index transaction " + tx.getId(), () -> {
246            membershipService.rollbackTransaction(tx);
247            return null;
248        });
249    }
250
251    /**
252     * Executes the closure, capturing all exceptions, and logging them as errors.
253     *
254     * @param failureMessage what to print if the closure fails
255     * @param callable closure to execute
256     */
257    private void execQuietly(final String failureMessage, final Callable<Void> callable) {
258        try {
259            callable.call();
260        } catch (final Exception e) {
261            LOGGER.error(failureMessage, e);
262        }
263    }
264
265    /**
266     * Parse the inputstream from a Rdf resource to a RDFstream.
267     *
268     * @param fedoraIdentifier the resource identifier.
269     * @param inputStream the inputstream.
270     * @return an RdfStream of the resource triples.
271     */
272    private static RdfStream parseRdf(final FedoraId fedoraIdentifier, final InputStream inputStream) {
273        final Model model = createDefaultModel();
274        RDFDataMgr.read(model, inputStream, getRdfFormat().getLang());
275        final FedoraId topic = (fedoraIdentifier.isDescription() ? fedoraIdentifier.asBaseId() : fedoraIdentifier);
276        return DefaultRdfStream.fromModel(createURI(topic.getFullId()), model);
277    }
278}