dnet-core/dnet-data-services/src/main/java/eu/dnetlib/data/mdstore/modular/mongodb/MDStoreDaoImpl.java

328 lines
11 KiB
Java

package eu.dnetlib.data.mdstore.modular.mongodb;
import java.util.List;
import com.google.common.collect.Lists;
import com.google.common.primitives.Ints;
import com.mongodb.BasicDBObject;
import com.mongodb.BasicDBObjectBuilder;
import com.mongodb.DBObject;
import com.mongodb.client.AggregateIterable;
import com.mongodb.client.MongoCollection;
import com.mongodb.client.MongoDatabase;
import com.mongodb.client.model.CreateCollectionOptions;
import eu.dnetlib.data.mdstore.MDStoreServiceException;
import eu.dnetlib.data.mdstore.modular.MDStoreDescription;
import eu.dnetlib.data.mdstore.modular.RecordParser;
import eu.dnetlib.data.mdstore.modular.RecordParserFactory;
import eu.dnetlib.data.mdstore.modular.connector.MDStore;
import eu.dnetlib.data.mdstore.modular.connector.MDStoreDBStatus;
import eu.dnetlib.data.mdstore.modular.connector.MDStoreDao;
import eu.dnetlib.data.mdstore.modular.connector.MDStoreTransactionManager;
import eu.dnetlib.data.mdstore.modular.mongodb.utils.MDStoreUtils;
import eu.dnetlib.miscutils.collections.FilteredCollection;
import eu.dnetlib.miscutils.collections.MappedCollection;
import eu.dnetlib.miscutils.datetime.DateUtils;
import org.apache.commons.logging.Log;
import org.apache.commons.logging.LogFactory;
import org.bson.Document;
import org.springframework.beans.factory.annotation.Autowired;
import org.springframework.beans.factory.annotation.Required;
/**
* Factory bean for MongoMDStore instances.
*
* @author marko
*/
public class MDStoreDaoImpl implements MDStoreDao {
public static final String MD_ID = "mdId";
public static final String FORMAT = "format";
public static final String INTERPRETATION = "interpretation";
public static final String LAYOUT = "layout";
public static final String SIZE = "size";
public static final String METADATA_NAME = "metadata";
private static final Log log = LogFactory.getLog(MDStoreDaoImpl.class);
private MongoDatabase db;
private RecordParserFactory recordParserFactory;
private boolean discardRecords = true;
@Autowired
private MDStoreTransactionManager transactionManager;
/**
* {@inheritDoc}
*/
@Override
public void createMDStore(final String mdId, final String format, final String interpretation, final String layout) throws MDStoreServiceException {
transactionManager.createMDStore(mdId);
final String internalId = transactionManager.getMDStoreCollection(mdId);
if (!Lists.newArrayList(getDb().listCollectionNames()).contains(internalId)) {
log.info(String.format("creating collection %s", internalId));
getDb().createCollection(internalId, new CreateCollectionOptions());
}
final MongoCollection<DBObject> coll = getDb().getCollection(METADATA_NAME, DBObject.class);
final BasicDBObject obj = new BasicDBObject();
obj.put(MD_ID, mdId);
obj.put(FORMAT, format);
obj.put(INTERPRETATION, interpretation);
obj.put(LAYOUT, layout);
obj.put(SIZE, 0);
coll.insertOne(obj);
}
/**
* {@inheritDoc}
*/
@Override
public void deleteMDStore(final String mdId) throws MDStoreServiceException {
final MongoCollection<DBObject> metadata = getDb().getCollection(METADATA_NAME, DBObject.class);
if (metadata == null) throw new MDStoreServiceException("cannot find metadata collection");
transactionManager.dropMDStore(mdId);
metadata.deleteOne(new BasicDBObject(MD_ID, mdId));
log.info("deleted mdId: " + mdId);
}
/**
* {@inheritDoc}
*/
@Override
public MDStore getMDStore(final String mdId) throws MDStoreServiceException {
final String internalId = transactionManager.getMDStoreCollection(mdId);
return new MongoMDStore(mdId, getDb().getCollection(internalId, DBObject.class), getRecordParser(), isDiscardRecords(), getDb());
}
/**
* {@inheritDoc}
*/
@Override
public MDStore readMDStore(final String mdId) throws MDStoreServiceException {
final String internalId = transactionManager.readMdStore(mdId);
return new MongoMDStore(mdId, getDb().getCollection(internalId, DBObject.class), getRecordParser(), isDiscardRecords(), getDb());
}
/**
* {@inheritDoc}
*/
@Override
public MDStore startTransaction(final String mdId, final boolean refresh) throws MDStoreServiceException {
final String transactionId = transactionManager.startTransaction(mdId, refresh);
return new MongoMDStore(transactionId, getDb().getCollection(transactionId, DBObject.class), getRecordParser(), isDiscardRecords(),
getDb());
}
private RecordParser getRecordParser() {
final RecordParser parser = getRecordParserFactory().newInstance();
parser.setTimestamp(DateUtils.now());
return parser;
}
/**
* {@inheritDoc}
*/
@Override
public List<MDStoreDescription> listMDStores() {
return MappedCollection.listMap(getDb().getCollection(METADATA_NAME, DBObject.class).find(), input -> {
final String mdId = (String) input.get(MD_ID);
log.debug("Getting info for " + mdId);
final String format = (String) input.get(FORMAT);
final String layout = (String) input.get(LAYOUT);
final String interpretation = (String) input.get(INTERPRETATION);
MongoMDStore currentMDStore = null;
final MDStoreDescription description = new MDStoreDescription();
try {
currentMDStore = (MongoMDStore) getMDStore(mdId);
} catch (final MDStoreServiceException e) {
log.error("Error on retrieving mdstore for getting info mdId " + mdId);
}
int size = 0;
if (input.containsField(SIZE)) {
log.debug("Size retrieved from metadata for mdId :" + mdId);
size = (Integer) input.get(SIZE);
} else {
if (currentMDStore != null) {
log.debug("Size not Found in metadata for mdId :" + mdId + " calling getCount ");
size = currentMDStore.getSize();
input.put("size", size);
getDb().getCollection(METADATA_NAME, DBObject.class).findOneAndReplace(new BasicDBObject(MD_ID, mdId), input);
}
}
if (currentMDStore != null) {
description.setIndexed(currentMDStore.isIndexed());
}
description.setId(mdId);
description.setFormat(format);
description.setLayout(layout);
description.setInterpretation(interpretation);
description.setSize(size);
return description;
});
}
/**
* {@inheritDoc}
*/
@Override
public List<String> listMDStores(final String format, final String layout, final String interpretation) {
return MappedCollection.listMap(
FilteredCollection.listFilter(getDb().getCollection(METADATA_NAME, DBObject.class).find(), MDStoreUtils.dboFilter(format, layout, interpretation)),
MDStoreUtils.mdId());
}
/**
* {@inheritDoc}
*/
@Override
public int getCachedSize(final String id) throws MDStoreServiceException {
log.debug("retrieve cached size for mdstore: " + id);
final DBObject desc = getDb().getCollection(METADATA_NAME, DBObject.class).find(new BasicDBObject(MD_ID, id)).first();
if (!desc.containsField(SIZE)) {
desc.put(SIZE, getMDStore(id).getSize());
}
final Object oSize = desc.get(SIZE);
return (Integer) oSize;
}
/**
* {@inheritDoc}
*/
@Override
public void refreshSizes() throws MDStoreServiceException {
for (final MDStoreDescription mdStoreId : listMDStores()) {
refreshSize(mdStoreId.getId());
}
}
/**
* {@inheritDoc}
*/
@Override
public int refreshSize(final String mdStoreId) throws MDStoreServiceException {
final int size = (int) getDb().getCollection(transactionManager.getMDStoreCollection(mdStoreId)).count();
final MongoCollection<DBObject> metadata = getDb().getCollection(METADATA_NAME, DBObject.class);
metadata.updateOne(new BasicDBObject(MD_ID, mdStoreId), new BasicDBObject("$set", new BasicDBObject(SIZE, size)));
return size;
}
@Override
public int getSumOfSizes(final String format, final String layout, final String interpretation) throws MDStoreServiceException {
final MongoCollection<DBObject> metadata = getDb().getCollection(METADATA_NAME, DBObject.class);
BasicDBObject matchObj = (BasicDBObject) BasicDBObjectBuilder.start("$match",
BasicDBObjectBuilder.start("format", format).add("layout", layout).add("interpretation", interpretation).get()).get();
BasicDBObject groupObj = (BasicDBObject) BasicDBObjectBuilder.start("$group",
BasicDBObjectBuilder.start("_id", "").add("total", new BasicDBObject("$sum", "$" + SIZE)).get()).get();
BasicDBObject projectObj = new BasicDBObject("$project", new BasicDBObject("_id", 0).append("total", 1));
List<BasicDBObject> pipeline = Lists.newArrayList(matchObj, groupObj, projectObj);
AggregateIterable<DBObject> output = metadata.aggregate(pipeline, DBObject.class);
DBObject result = output.first();
if (result == null || !result.containsField("total")) {
log.debug("No total found");
return 0;
} else return (Integer) result.get("total");
}
/**
* {@inheritDoc}
*/
@Override
public void commit(final String transactionId, final String mdId) throws MDStoreServiceException {
transactionManager.commit(transactionId, mdId, getMDStore(mdId));
}
/**
* Getter for property 'db'.
*
* @return Value for property 'db'.
*/
public MongoDatabase getDb() {
return db;
}
/**
* Setter for property 'db'.
*
* @param db
* Value to set for property 'db'.
*/
@Required
public void setDb(final MongoDatabase db) {
this.db = db;
}
/**
* Getter for property 'recordParser'.
*
* @return Value for property 'recordParser'.
*/
public RecordParserFactory getRecordParserFactory() {
return recordParserFactory;
}
@Required
public void setRecordParserFactory(final RecordParserFactory recordParserFactory) {
this.recordParserFactory = recordParserFactory;
}
/**
* Getter for property 'discardRecords'.
*
* @return Value for property 'discardRecords'.
*/
public boolean isDiscardRecords() {
return discardRecords;
}
/**
* Setter for property 'discardRecords'.
*
* @param discardRecords
* Value to set for property 'discardRecords'.
*/
public void setDiscardRecords(final boolean discardRecords) {
this.discardRecords = discardRecords;
}
@Override
public MDStoreDBStatus getDBStatus() {
final int handledDatastructures = Ints.saturatedCast(getDb().getCollection(METADATA_NAME).count());
//final int usedDiskSpace = Ints.saturatedCast(getDb().getStats().getLong("storageSize") / (1024 * 1024)); // in MB
//{ dbStats: 1, scale: 1 }
BasicDBObject statsQuery = new BasicDBObject("dbStats", 1);
statsQuery.put("scale", 1024 * 1024); //storageSize in MB
final Document statsRes = getDb().runCommand(statsQuery);
log.debug("DBStatus -- " + statsRes.toJson());
int usedDiskSpace;
//trying to handle different versions of the mongo server: old version returns storage size as long, new version as double
//TODO: simplify this when dev, beta, production are aligned with our local, latest, mongo version
String usedDiskSpaceStr = statsRes.get("storageSize").toString();
try {
Long usedDiskSpaceLong = Long.parseLong(usedDiskSpaceStr);
usedDiskSpace = Ints.saturatedCast(usedDiskSpaceLong);
} catch (NumberFormatException nfe) {
Double usedDiskSpaceDbl = Double.parseDouble(usedDiskSpaceStr);
usedDiskSpace = usedDiskSpaceDbl.intValue();
}
final String date = DateUtils.now_ISO8601();
return new MDStoreDBStatus(handledDatastructures, usedDiskSpace, date);
}
@Override
public void startGarbage() throws MDStoreServiceException {
this.transactionManager.garbage();
}
@Override
public void invalidTransaction(final String transactionId, final String mdId) throws MDStoreServiceException {
transactionManager.dropTransaction(mdId, transactionId);
}
}