1
|
package eu.dnetlib.oai.parser;
|
2
|
|
3
|
import java.io.IOException;
|
4
|
import java.util.List;
|
5
|
|
6
|
import com.google.common.collect.Lists;
|
7
|
import com.mongodb.BasicDBObject;
|
8
|
import com.mongodb.BasicDBObjectBuilder;
|
9
|
import eu.dnetlib.data.information.oai.publisher.OaiPublisherRuntimeException;
|
10
|
import eu.dnetlib.data.information.oai.publisher.conf.OAIConfigurationReader;
|
11
|
import eu.dnetlib.functionality.index.parse.Relation;
|
12
|
import eu.dnetlib.functionality.index.parse.Relations;
|
13
|
import org.apache.commons.lang.StringUtils;
|
14
|
import org.apache.commons.logging.Log;
|
15
|
import org.apache.commons.logging.LogFactory;
|
16
|
import org.bson.conversions.Bson;
|
17
|
import org.bson.types.ObjectId;
|
18
|
import org.joda.time.DateTime;
|
19
|
import org.joda.time.format.DateTimeFormat;
|
20
|
import org.joda.time.format.DateTimeFormatter;
|
21
|
import org.joda.time.format.ISODateTimeFormat;
|
22
|
import org.z3950.zing.cql.*;
|
23
|
|
24
|
/**
|
25
|
* Instances of this class parse query string into mongo DBObject query.
|
26
|
*
|
27
|
* @author alessia
|
28
|
*/
|
29
|
public class MongoQueryParser {
|
30
|
|
31
|
private static final Log log = LogFactory.getLog(MongoQueryParser.class); // NOPMD by marko on 11/24/08 5:02 PM
|
32
|
|
33
|
/**
|
34
|
* Parses the given query string into a mongo DBObject.
|
35
|
*
|
36
|
* @param query String to parse
|
37
|
* @return DBObject corresponding to the query string
|
38
|
*/
|
39
|
public Bson parse(final String query) {
|
40
|
log.debug("PARSING: " + query);
|
41
|
if (StringUtils.isBlank(query)) return new BasicDBObject();
|
42
|
Bson parsed = toMongo(query);
|
43
|
log.debug(parsed);
|
44
|
return parsed;
|
45
|
}
|
46
|
|
47
|
private Bson toMongo(final String query) {
|
48
|
CQLParser parser = new CQLParser();
|
49
|
CQLNode root;
|
50
|
try {
|
51
|
root = parser.parse(query);
|
52
|
return this.toMongo(root);
|
53
|
} catch (CQLParseException e) {
|
54
|
throw new OaiPublisherRuntimeException(e);
|
55
|
} catch (IOException e) {
|
56
|
throw new OaiPublisherRuntimeException(e);
|
57
|
}
|
58
|
}
|
59
|
|
60
|
private Bson toMongo(final CQLNode node) {
|
61
|
if (node instanceof CQLTermNode) return doTranslate((CQLTermNode) node);
|
62
|
if (node instanceof CQLBooleanNode) return doTranslate((CQLBooleanNode) node);
|
63
|
|
64
|
throw new RuntimeException("error choice for CQLNode " + node.getClass());
|
65
|
}
|
66
|
|
67
|
private Bson doTranslate(final CQLTermNode termNode) {
|
68
|
if (termNode.getTerm().equals("*")) return new BasicDBObject();
|
69
|
String relation = termNode.getRelation().getBase();
|
70
|
Relation rel = Relations.get(relation);
|
71
|
return this.handleRelationNode(rel, termNode);
|
72
|
}
|
73
|
|
74
|
private Bson handleRelationNode(final Relation rel, final CQLTermNode termNode) {
|
75
|
BasicDBObject mongoQueryObject = new BasicDBObject();
|
76
|
String term = termNode.getTerm();
|
77
|
String indexName = termNode.getIndex();
|
78
|
Object termObj = term;
|
79
|
if (indexName.equals("_id")) {
|
80
|
termObj = new ObjectId(term);
|
81
|
} else {
|
82
|
if (indexName.equals(OAIConfigurationReader.DATESTAMP_FIELD) || indexName.equals(OAIConfigurationReader.LAST_COLLECTION_DATE_FIELD)) {
|
83
|
// termObj = this.parseDate(term);
|
84
|
OAIDate termDate = this.parseDate(term);
|
85
|
return handleDateRelationNode(indexName, rel, termDate);
|
86
|
}
|
87
|
}
|
88
|
switch (rel) {
|
89
|
case EQUAL:
|
90
|
case EXACT:
|
91
|
mongoQueryObject.put(indexName, termObj);
|
92
|
break;
|
93
|
case NOT:
|
94
|
mongoQueryObject.put(indexName, new BasicDBObject("$ne", termObj));
|
95
|
break;
|
96
|
case GT:
|
97
|
mongoQueryObject.put(indexName, new BasicDBObject("$gt", termObj));
|
98
|
break;
|
99
|
case GTE:
|
100
|
mongoQueryObject.put(indexName, new BasicDBObject("$gte", termObj));
|
101
|
break;
|
102
|
case LT:
|
103
|
mongoQueryObject.put(indexName, new BasicDBObject("$lt", termObj));
|
104
|
break;
|
105
|
case LTE:
|
106
|
mongoQueryObject.put(indexName, new BasicDBObject("$lte", termObj));
|
107
|
break;
|
108
|
default:
|
109
|
throw new OaiPublisherRuntimeException("Can't parse query: relation " + rel + " not supported!");
|
110
|
}
|
111
|
return mongoQueryObject;
|
112
|
}
|
113
|
|
114
|
/**
|
115
|
* The construction of the query changes based on the granularity of the date to handle.
|
116
|
* <p>
|
117
|
* If the date has yyyy-MM-ddThh:mm:ssZ granularity we have to create a range query because in mongo we have milliseconds, hence an
|
118
|
* exact match will return nothing.
|
119
|
* </p>
|
120
|
* <p>
|
121
|
* If the date has yyyy-MM-dd granularity then we have to trick the query. If we are interested in the date 2013-10-28, the date has
|
122
|
* been converted into 2013-10-28T00:00:00Z : if we ask for datestamp = 2013-10-28T00:00:00Z, we'll get nothing: we have to ask for
|
123
|
* records whose day is the one specified by the date.
|
124
|
* <p>
|
125
|
* </p>
|
126
|
*
|
127
|
* @param indexName
|
128
|
* @param rel
|
129
|
* @param date
|
130
|
* @return
|
131
|
*/
|
132
|
private Bson handleDateRelationNode(final String indexName, final Relation rel, final OAIDate date) {
|
133
|
BasicDBObject mongoQueryObject = new BasicDBObject();
|
134
|
DateTime fromDate = date.date;
|
135
|
switch (rel) {
|
136
|
case EQUAL:
|
137
|
case EXACT:
|
138
|
if (date.onlyDate) {
|
139
|
DateTime endDate = date.date.plusDays(1);
|
140
|
mongoQueryObject.put(indexName, BasicDBObjectBuilder.start("$gte", fromDate.toDate()).append("$lt", endDate.toDate()).get());
|
141
|
} else {
|
142
|
DateTime endDate = date.date.plusSeconds(1);
|
143
|
mongoQueryObject.put(indexName, BasicDBObjectBuilder.start("$gte", fromDate.toDate()).append("$lt", endDate.toDate()).get());
|
144
|
}
|
145
|
break;
|
146
|
case NOT:
|
147
|
mongoQueryObject.put(indexName, new BasicDBObject("$ne", fromDate.toDate()));
|
148
|
break;
|
149
|
case GT:
|
150
|
mongoQueryObject.put(indexName, new BasicDBObject("$gt", fromDate.toDate()));
|
151
|
break;
|
152
|
case GTE:
|
153
|
mongoQueryObject.put(indexName, new BasicDBObject("$gte", fromDate.toDate()));
|
154
|
break;
|
155
|
case LT:
|
156
|
mongoQueryObject.put(indexName, new BasicDBObject("$lt", fromDate.toDate()));
|
157
|
break;
|
158
|
case LTE:
|
159
|
/*
|
160
|
If the request is date <= YYYY-MM-DD then we need to change the date.
|
161
|
The parseDate returned YYYY-MM-DDT00:00:00Z, but we need YYYY-MM-DDT23:59:59Z.
|
162
|
To simplify we can add one day and perform < instead of <=.
|
163
|
*/
|
164
|
if (date.onlyDate) {
|
165
|
fromDate = date.date.plusDays(1);
|
166
|
mongoQueryObject.put(indexName, new BasicDBObject("$lt", fromDate.toDate()));
|
167
|
} else {
|
168
|
mongoQueryObject.put(indexName, new BasicDBObject("$lte", fromDate.toDate()));
|
169
|
}
|
170
|
break;
|
171
|
default:
|
172
|
throw new OaiPublisherRuntimeException("Can't parse query: relation " + rel + " not supported!");
|
173
|
}
|
174
|
return mongoQueryObject;
|
175
|
}
|
176
|
|
177
|
private Bson doTranslate(final CQLBooleanNode node) {
|
178
|
if (node instanceof CQLAndNode) return getBooleanQuery("$and", node);
|
179
|
if (node instanceof CQLOrNode) return getBooleanQuery("$or", node);
|
180
|
if (node instanceof CQLNotNode) return getNotQuery((CQLNotNode) node);
|
181
|
throw new RuntimeException("error choice for CQLBooleanNode " + node.getClass());
|
182
|
}
|
183
|
|
184
|
private Bson getBooleanQuery(final String mongoOperator, final CQLBooleanNode node) {
|
185
|
Bson left = this.toMongo(node.left);
|
186
|
Bson right = this.toMongo(node.right);
|
187
|
BasicDBObject opQuery = new BasicDBObject();
|
188
|
List<Bson> termList = Lists.newArrayList(left, right);
|
189
|
opQuery.put(mongoOperator, termList);
|
190
|
return opQuery;
|
191
|
}
|
192
|
|
193
|
private Bson getNotQuery(final CQLNotNode node) {
|
194
|
Bson left = this.toMongo(node.left);
|
195
|
Bson right = this.toMongo(node.right);
|
196
|
Bson notRight = new BasicDBObject("$not", right);
|
197
|
BasicDBObject andQuery = new BasicDBObject();
|
198
|
List<Bson> termList = Lists.newArrayList(left, notRight);
|
199
|
andQuery.put("$and", termList);
|
200
|
return andQuery;
|
201
|
}
|
202
|
|
203
|
private OAIDate parseDate(final String date) {
|
204
|
DateTimeFormatter dateNoTimeFormatter = DateTimeFormat.forPattern("yyyy-MM-dd").withZoneUTC();
|
205
|
DateTimeFormatter iso8601NoMsTimeFormatter = DateTimeFormat.forPattern("yyyy-MM-dd'T'HH:mm:ssZ").withZoneUTC();
|
206
|
DateTimeFormatter iso8601Formatter = ISODateTimeFormat.dateTime().withZoneUTC();
|
207
|
OAIDate res = null;
|
208
|
try {
|
209
|
log.debug("Using default " + iso8601Formatter.getClass());
|
210
|
DateTime dt = iso8601Formatter.parseDateTime(date);
|
211
|
res = new OAIDate(dt, false);
|
212
|
} catch (Exception e) {
|
213
|
try {
|
214
|
log.debug("Switching to ISO with no millisecond date formatter: yyyy-MM-dd'T'HH:mm:ssZ");
|
215
|
DateTime dt = iso8601NoMsTimeFormatter.parseDateTime(date);
|
216
|
res = new OAIDate(dt, false);
|
217
|
} catch (Exception ex) {
|
218
|
log.debug("Switching to simple date formatter: yyyy-MM-dd");
|
219
|
DateTime dt = dateNoTimeFormatter.parseDateTime(date);
|
220
|
res = new OAIDate(dt, true);
|
221
|
}
|
222
|
}
|
223
|
return res;
|
224
|
}
|
225
|
|
226
|
class OAIDate {
|
227
|
|
228
|
DateTime date;
|
229
|
boolean onlyDate;
|
230
|
|
231
|
OAIDate(final DateTime date, final boolean onlyDate) {
|
232
|
this.date = date;
|
233
|
this.onlyDate = onlyDate;
|
234
|
}
|
235
|
|
236
|
}
|
237
|
|
238
|
}
|