315 lines
10 KiB
Java
315 lines
10 KiB
Java
package eu.dnetlib.data.mdstore.modular.mongodb;
|
|
|
|
import java.util.*;
|
|
import java.util.concurrent.*;
|
|
import java.util.regex.Pattern;
|
|
|
|
import com.google.common.base.Function;
|
|
import com.google.common.collect.Iterables;
|
|
import com.google.common.collect.Iterators;
|
|
import com.google.common.collect.Lists;
|
|
import com.google.common.collect.Sets;
|
|
import com.mongodb.BasicDBObject;
|
|
import com.mongodb.DBObject;
|
|
import com.mongodb.QueryBuilder;
|
|
import com.mongodb.client.FindIterable;
|
|
import com.mongodb.client.ListIndexesIterable;
|
|
import com.mongodb.client.MongoCollection;
|
|
import com.mongodb.client.MongoDatabase;
|
|
import com.mongodb.client.model.IndexOptions;
|
|
import eu.dnetlib.data.mdstore.DocumentNotFoundException;
|
|
import eu.dnetlib.data.mdstore.MDStoreServiceException;
|
|
import eu.dnetlib.data.mdstore.modular.MDFormatDescription;
|
|
import eu.dnetlib.data.mdstore.modular.RecordParser;
|
|
import eu.dnetlib.data.mdstore.modular.connector.MDStore;
|
|
import eu.dnetlib.enabling.resultset.ResultSetListener;
|
|
import org.apache.commons.lang3.StringUtils;
|
|
import org.apache.commons.logging.Log;
|
|
import org.apache.commons.logging.LogFactory;
|
|
import org.bson.conversions.Bson;
|
|
import org.springframework.beans.factory.annotation.Required;
|
|
|
|
public class MongoMDStore implements MDStore {
|
|
|
|
private static final int BULK_SIZE = 500;
|
|
private static final Log log = LogFactory.getLog(MongoMDStore.class);
|
|
|
|
private final boolean discardRecords;
|
|
private String id;
|
|
private MongoDatabase mongoDatabase;
|
|
private MongoCollection<DBObject> collection;
|
|
private MongoCollection<DBObject> discardedCollection;
|
|
private List<MDFormatDescription> mdformats;
|
|
|
|
private RecordParser recordParser;
|
|
|
|
private static List<String> indices = Lists.newArrayList("id", "timestamp", "originalId");
|
|
private final IndexOptions options = new IndexOptions().background(true);
|
|
|
|
public MongoMDStore(final String id,
|
|
final MongoCollection<DBObject> collection,
|
|
final RecordParser recordParser,
|
|
final boolean discardRecords,
|
|
final MongoDatabase mongoDatabase) {
|
|
this.id = id;
|
|
this.mongoDatabase = mongoDatabase;
|
|
this.collection = collection;
|
|
this.discardedCollection = this.mongoDatabase.getCollection("discarded-" + StringUtils.substringBefore(id, "_"), DBObject.class);
|
|
this.recordParser = recordParser;
|
|
this.discardRecords = discardRecords;
|
|
}
|
|
|
|
@Override
|
|
public int feed(final Iterable<String> records, final boolean incremental, final List<MDFormatDescription> mdformats) {
|
|
this.mdformats = mdformats;
|
|
return feed(records, incremental);
|
|
}
|
|
|
|
@Override
|
|
public int feed(final Iterable<String> records, final boolean incremental) {
|
|
// TODO: remove incremental from MDStore API. It is used in MDStoreModular. Useless here.
|
|
|
|
ensureIndices();
|
|
|
|
final BlockingQueue<Object> queue = new ArrayBlockingQueue<>(100);
|
|
final Object sentinel = new Object();
|
|
int countStored = 0;
|
|
final Callable<Integer> writer = () -> {
|
|
final MongoBulkWritesManager bulkWritesManager =
|
|
new MongoBulkWritesManager(collection, discardedCollection, mdformats, BULK_SIZE, recordParser, discardRecords);
|
|
int count = 0;
|
|
while (true) {
|
|
try {
|
|
final Object record = queue.take();
|
|
if (record == sentinel) {
|
|
bulkWritesManager.flushBulks();
|
|
break;
|
|
}
|
|
count++;
|
|
bulkWritesManager.insert((String) record);
|
|
} catch (final InterruptedException e) {
|
|
log.fatal("got exception in background thread", e);
|
|
throw new IllegalStateException(e);
|
|
}
|
|
}
|
|
log.debug(String.format("extracted %s records from feeder queue", count));
|
|
return count;
|
|
};
|
|
final ExecutorService executorService = Executors.newSingleThreadExecutor();
|
|
Future<Integer> storedCountInt = executorService.submit(writer);
|
|
|
|
try {
|
|
log.info("feeding mdstore " + id);
|
|
if (records != null) {
|
|
for (final String record : records) {
|
|
queue.put(record);
|
|
}
|
|
}
|
|
queue.put(sentinel);
|
|
countStored = storedCountInt.get().intValue();
|
|
} catch (final InterruptedException e) {
|
|
log.error("Error on feeding mdstore with id:" + id, e);
|
|
throw new IllegalStateException(e);
|
|
} catch (ExecutionException e) {
|
|
log.error("Error on feeding mdstore with id:" + id, e);
|
|
throw new IllegalStateException(e);
|
|
}
|
|
log.info("finished feeding mdstore " + id);
|
|
return countStored;
|
|
}
|
|
|
|
public void ensureIndices() {
|
|
for (final String key : indices) {
|
|
collection.createIndex(new BasicDBObject(key, 1), options);
|
|
}
|
|
if (mdformats != null) {
|
|
for (final MDFormatDescription description : mdformats) {
|
|
collection.createIndex(new BasicDBObject(description.getName(), 1), options);
|
|
}
|
|
}
|
|
}
|
|
|
|
public boolean isIndexed() {
|
|
final ListIndexesIterable<DBObject> found = collection.listIndexes(DBObject.class);
|
|
return Sets.newHashSet(Iterables.transform(found, dbo -> {
|
|
final Set<String> keyset = ((DBObject) dbo.get("key")).toMap().keySet();
|
|
return Iterables.getFirst(keyset, "");
|
|
})).containsAll(indices);
|
|
}
|
|
|
|
/**
|
|
* Method searches for the given string grep into this collection and replaces it with the given replacement.
|
|
*
|
|
* @param grep
|
|
* the string to search
|
|
* @param replace
|
|
* the replacement
|
|
*/
|
|
public void replace(final String grep, final String replace) {
|
|
final Pattern regex = Pattern.compile(grep, Pattern.MULTILINE);
|
|
BasicDBObject query = (BasicDBObject) QueryBuilder.start("body").regex(regex).get();
|
|
final FindIterable<DBObject> matches = collection.find(query, DBObject.class);
|
|
//final DBCursor matches = collection.find(QueryBuilder.start("body").regex(regex).get());
|
|
if (log.isDebugEnabled())
|
|
log.debug("FOUND: " + Lists.newArrayList(matches).size());
|
|
|
|
for (final DBObject match : matches) {
|
|
final DBObject o = new BasicDBObject(match.toMap());
|
|
o.put("body", regex.matcher((String) match.get("body")).replaceAll(replace));
|
|
collection.findOneAndReplace(new BasicDBObject("_id", o.get("_id")), o);
|
|
}
|
|
}
|
|
|
|
@Override
|
|
public ResultSetListener deliver(final String from, final String until, final String recordFilter) throws MDStoreServiceException {
|
|
return deliver(from, until, recordFilter, new SerializeMongoRecord());
|
|
}
|
|
|
|
@Override
|
|
public ResultSetListener deliverIds(final String from, final String until, final String recordFilter) throws MDStoreServiceException {
|
|
return deliver(from, until, recordFilter, new SerializeMongoRecordId());
|
|
}
|
|
|
|
public ResultSetListener deliver(final String from, final String until, final String recordFilter, final Function<DBObject, String> serializer)
|
|
throws MDStoreServiceException {
|
|
final Pattern filter = (recordFilter != null) && (recordFilter.length() > 0) ? Pattern.compile(recordFilter, Pattern.MULTILINE) : null;
|
|
|
|
return new MongoResultSetListener(collection, parseLong(from), parseLong(until), filter, serializer); }
|
|
|
|
private Long parseLong(final String s) throws MDStoreServiceException {
|
|
if (StringUtils.isBlank(s)) {
|
|
return null;
|
|
}
|
|
try {
|
|
return Long.valueOf(s);
|
|
} catch (NumberFormatException e) {
|
|
throw new MDStoreServiceException("Invalid date, expected java.lang.Long, or null", e);
|
|
}
|
|
}
|
|
|
|
@Override
|
|
public Iterable<String> iterate() {
|
|
return () -> Iterators.transform(collection.find().iterator(), arg -> (String) arg.get("body"));
|
|
}
|
|
|
|
@Override
|
|
public void deleteRecord(final String recordId) {
|
|
collection.deleteOne(new BasicDBObject("id", recordId));
|
|
}
|
|
|
|
@Override
|
|
public String getRecord(final String recordId) throws DocumentNotFoundException {
|
|
final DBObject obj = collection.find(new BasicDBObject("id", recordId)).first();
|
|
if (obj == null || !obj.containsField("body")) throw new DocumentNotFoundException(String.format(
|
|
"The document with id '%s' does not exist in mdstore: '%s'", recordId, id));
|
|
final String body = (String) obj.get("body");
|
|
if (body.trim().length() == 0) throw new DocumentNotFoundException(String.format("The document with id '%s' does not exist in mdstore: '%s'",
|
|
recordId, id));
|
|
return new SerializeMongoRecord().apply(obj);
|
|
}
|
|
|
|
@Override
|
|
public List<String> deliver(final String mdId, final int pageSize, final int offset, final Map<String, String> queryParam) {
|
|
final QueryBuilder query = QueryBuilder.start();
|
|
|
|
for (String key : queryParam.keySet()) {
|
|
query.and(key).regex(Pattern.compile(queryParam.get(key), Pattern.LITERAL));
|
|
}
|
|
|
|
FindIterable<DBObject> dbObjects = offset > 0
|
|
? collection.find((Bson) query.get()).limit(pageSize).skip(offset)
|
|
: collection.find((Bson) query.get()).limit(pageSize);
|
|
|
|
|
|
queryParam.put("count", "" + collection.count((Bson) query.get()));
|
|
|
|
final List<String> result = new ArrayList<>();
|
|
for (final DBObject item : dbObjects) {
|
|
result.add(item.get("body").toString());
|
|
}
|
|
|
|
return result;
|
|
}
|
|
|
|
@Override
|
|
public void truncate() {
|
|
collection.drop();
|
|
discardedCollection.drop();
|
|
}
|
|
|
|
public DBObject getMDStoreMetadata() {
|
|
return mongoDatabase.getCollection("metadata", DBObject.class).find(new BasicDBObject("mdId", getId())).first();
|
|
}
|
|
|
|
@Override
|
|
public String getFormat() {
|
|
return (String) getMDStoreMetadata().get("format");
|
|
}
|
|
|
|
@Override
|
|
public String getInterpretation() {
|
|
return (String) getMDStoreMetadata().get("interpretation");
|
|
}
|
|
|
|
@Override
|
|
public String getLayout() {
|
|
return (String) getMDStoreMetadata().get("layout");
|
|
}
|
|
|
|
@Override
|
|
public String getId() {
|
|
return id;
|
|
}
|
|
|
|
public void setId(final String id) {
|
|
this.id = id;
|
|
}
|
|
|
|
public MongoCollection<DBObject> getCollection() {
|
|
return collection;
|
|
}
|
|
|
|
public void setCollection(final MongoCollection<DBObject> collection) {
|
|
this.collection = collection;
|
|
}
|
|
|
|
public RecordParser getRecordParser() {
|
|
return recordParser;
|
|
}
|
|
|
|
@Required
|
|
public void setRecordParser(final RecordParser recordParser) {
|
|
this.recordParser = recordParser;
|
|
}
|
|
|
|
@Override
|
|
public int getSize() {
|
|
return (int) collection.count();
|
|
}
|
|
|
|
public MongoCollection<DBObject> getDiscardedCollection() {
|
|
return discardedCollection;
|
|
}
|
|
|
|
public void setDiscardedCollection(final MongoCollection<DBObject> discardedCollection) {
|
|
this.discardedCollection = discardedCollection;
|
|
}
|
|
|
|
private class SerializeMongoRecord implements Function<DBObject, String> {
|
|
|
|
@Override
|
|
public String apply(final DBObject arg) {
|
|
return (String) arg.get("body");
|
|
}
|
|
}
|
|
|
|
private class SerializeMongoRecordId implements Function<DBObject, String> {
|
|
|
|
@Override
|
|
public String apply(final DBObject arg) {
|
|
return (String) arg.get("id");
|
|
}
|
|
}
|
|
|
|
}
|