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