Project

General

Profile

1
package eu.dnetlib.data.mdstore.modular.mongodb;
2

    
3
import java.util.List;
4
import java.util.Map;
5

    
6
import com.google.common.collect.Lists;
7
import com.mongodb.BasicDBObject;
8
import com.mongodb.DBObject;
9
import com.mongodb.WriteConcern;
10
import com.mongodb.client.MongoCollection;
11
import com.mongodb.client.model.*;
12
import eu.dnetlib.data.mdstore.MDStoreServiceException;
13
import eu.dnetlib.data.mdstore.modular.MDFormatDescription;
14
import eu.dnetlib.data.mdstore.modular.RecordParser;
15
import eu.dnetlib.data.mdstore.modular.mongodb.utils.IndexFieldRecordParser;
16
import eu.dnetlib.data.mdstore.modular.mongodb.utils.IndexFieldRecordParserException;
17
import org.apache.commons.logging.Log;
18
import org.apache.commons.logging.LogFactory;
19

    
20
import static eu.dnetlib.data.mdstore.modular.MDStoreConstants.*;
21

    
22
public class MongoBulkWritesManager {
23

    
24
	private static final Log log = LogFactory.getLog(MongoBulkWritesManager.class);
25
	private final boolean discardRecords;
26
	private final boolean indexRecords;
27
	private final IndexFieldRecordParser indexFieldRecordParser = new IndexFieldRecordParser();
28
	private final List<MDFormatDescription> mdref;
29
	private RecordParser recordParser;
30
	private MongoCollection<DBObject> validCollection;
31
	private List<WriteModel<DBObject>> validBulkOperationList;
32
	private BulkWriteOptions writeOptions;
33
	private MongoCollection<DBObject> discardedCollection;
34
	private List<WriteModel<DBObject>> discardedBulkOperationList;
35
	private int bulkSize;
36

    
37
	public MongoBulkWritesManager(final MongoCollection<DBObject> collection,
38
			final MongoCollection<DBObject> discardedCollection,
39
			final List<MDFormatDescription> mdref,
40
			final int bulkSize,
41
			final RecordParser parser,
42
			final boolean discardRecords) {
43
		this.validCollection = collection.withWriteConcern(WriteConcern.ACKNOWLEDGED);
44
		this.validBulkOperationList = Lists.newArrayList();
45

    
46
		this.discardedCollection = discardedCollection.withWriteConcern(WriteConcern.ACKNOWLEDGED);
47
		this.discardedBulkOperationList = Lists.newArrayList();
48

    
49
		this.bulkSize = bulkSize;
50
		this.recordParser = parser;
51
		this.discardRecords = discardRecords;
52
		this.mdref = mdref;
53

    
54
		this.indexRecords = this.mdref != null;
55
		this.writeOptions = new BulkWriteOptions().ordered(false);
56
	}
57

    
58
	public void insert(final String record) throws MDStoreServiceException {
59
		Map<String, String> recordProperties = null;
60
		try {
61
			recordProperties = recordParser.parseRecord(record);
62
		} catch (Throwable e) {
63
			if (discardRecords) {
64
				log.debug("unhandled exception: " + e.getMessage());
65
				discardRecord(record);
66
			}
67
		}
68
		Map<String, List<String>> indexRecordField = null;
69
		try {
70
			if (indexRecords) {
71
				indexRecordField = indexFieldRecordParser.parseRecord(record, mdref);
72
			}
73
		} catch (IndexFieldRecordParserException e) {
74
			// could not index record fields
75
			throw new MDStoreServiceException("Are you using the correct type of store / index definition for the records in " + validCollection.getNamespace() + " ?", e);
76
		}
77

    
78
		log.debug("found props: " + recordProperties);
79
		if (recordProperties.containsKey(ID)) {
80
			final DBObject obj = buildDBObject(record, recordProperties, indexRecordField);
81
			if (log.isDebugEnabled()) {
82
				log.debug("Saving object" + obj);
83
			}
84
			validBulkOperationList.add(new ReplaceOneModel(new BasicDBObject(ID, obj.get(ID)), obj, new UpdateOptions().upsert(true)));
85
			if (((validBulkOperationList.size() % bulkSize) == 0) && !validBulkOperationList.isEmpty()) {
86
				validCollection.bulkWrite(validBulkOperationList, writeOptions);
87
				validBulkOperationList.clear();
88
			}
89
		} else {
90
			if (discardRecords) {
91
				log.debug("parsed record seems invalid");
92
				discardRecord(record);
93
			}
94
		}
95
	}
96

    
97
	private void discardRecord(final String record) {
98
		discardedBulkOperationList.add(new InsertOneModel(new BasicDBObject(BODY, record)));
99

    
100
		if (((discardedBulkOperationList.size() % bulkSize) == 0) && !discardedBulkOperationList.isEmpty()) {
101
			discardedCollection.bulkWrite(discardedBulkOperationList, writeOptions);
102
			discardedBulkOperationList.clear();
103
		}
104
	}
105

    
106
	public void flushBulks() {
107
		//setting to journaled write concern to be sure that when the write returns everything has been flushed to disk (https://docs.mongodb.org/manual/faq/developers/#when-does-mongodb-write-updates-to-disk)
108
		//the explicit fsync command can't be run anymore: 'Command failed with error 13: 'fsync may only be run against the admin database.'
109
		if (!validBulkOperationList.isEmpty()) {
110
			validCollection = getCollectionWithWriteConcern(validCollection, WriteConcern.JOURNALED);
111
			validCollection.bulkWrite(validBulkOperationList, writeOptions);
112
		}
113
		if (!discardedBulkOperationList.isEmpty()) {
114
			discardedCollection = getCollectionWithWriteConcern(discardedCollection, WriteConcern.JOURNALED);
115
			discardedCollection.bulkWrite(discardedBulkOperationList, writeOptions);
116
		}
117
		//setting write concern back to ACKNOWLEDGE to avoid the execution of future writes all in Journaled mode
118
		validCollection = getCollectionWithWriteConcern(validCollection, WriteConcern.ACKNOWLEDGED);
119
		discardedCollection = getCollectionWithWriteConcern(discardedCollection, WriteConcern.ACKNOWLEDGED);
120
	}
121

    
122
	private DBObject buildDBObject(final String record, final Map<String, String> recordProperties, final Map<String, List<String>> indexFieldProperties) {
123
		final DBObject obj = new BasicDBObject();
124
		obj.put(ID, recordProperties.get(ID));
125
		obj.put(ORIGINALID, recordProperties.get(ORIGINALID));
126
		obj.put(BODY, record);
127
		obj.put(TIMESTAMP, Long.valueOf(recordProperties.get(TIMESTAMP)));
128
		if (indexFieldProperties != null)
129
			obj.putAll(indexFieldProperties);
130
		return obj;
131
	}
132

    
133
	private MongoCollection<DBObject> getCollectionWithWriteConcern(MongoCollection<DBObject> collection, WriteConcern writeConcern) {
134
		return collection.withWriteConcern(writeConcern);
135
	}
136

    
137
}
(3-3/5)