Mercurial > hg > openmind

annotate src/main/java/org/mpi/openmind/repository/utils/OM4StreamWriter.java @ 103:1149eb948036

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
add isodate tag to XML dump format 4.9.
author Robert Casties <casties@mpiwg-berlin.mpg.de>
date Wed, 14 Nov 2018 18:31:55 +0100
parents 734c0d8c7369
children 93c7dbfaf062
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
1
jurzua
parents:
diff changeset
1 package org.mpi.openmind.repository.utils;
jurzua
parents:
diff changeset
2
jurzua
parents:
diff changeset
3 import java.io.FileOutputStream;
jurzua
parents:
diff changeset
4 import java.io.IOException;
jurzua
parents:
diff changeset
5 import java.io.OutputStreamWriter;
jurzua
parents:
diff changeset
6 import java.text.DecimalFormat;
jurzua
parents:
diff changeset
7 import java.util.ArrayList;
79
b0aebac0780a put statistics about number of entities, relations and attributes in xml dump.
casties
parents: 78
diff changeset
8 import java.util.HashMap;
1
jurzua
parents:
diff changeset
9 import java.util.List;
79
b0aebac0780a put statistics about number of entities, relations and attributes in xml dump.
casties
parents: 78
diff changeset
10 import java.util.Map;
1
jurzua
parents:
diff changeset
11
29
5786aa6caeb3 new XML export and test script.
casties
parents: 1
diff changeset
12 import javax.xml.stream.XMLOutputFactory;
5786aa6caeb3 new XML export and test script.
casties
parents: 1
diff changeset
13 import javax.xml.stream.XMLStreamException;
5786aa6caeb3 new XML export and test script.
casties
parents: 1
diff changeset
14 import javax.xml.stream.XMLStreamWriter;
1
jurzua
parents:
diff changeset
15
jurzua
parents:
diff changeset
16 import org.apache.commons.lang.StringUtils;
jurzua
parents:
diff changeset
17 import org.apache.log4j.Logger;
103
1149eb948036 add isodate tag to XML dump format 4.9.
Robert Casties <casties@mpiwg-berlin.mpg.de>
parents: 100
diff changeset
18 import org.joda.time.DateTime;
1149eb948036 add isodate tag to XML dump format 4.9.
Robert Casties <casties@mpiwg-berlin.mpg.de>
parents: 100
diff changeset
19 import org.joda.time.format.DateTimeFormatter;
1149eb948036 add isodate tag to XML dump format 4.9.
Robert Casties <casties@mpiwg-berlin.mpg.de>
parents: 100
diff changeset
20 import org.joda.time.format.ISODateTimeFormat;
1149eb948036 add isodate tag to XML dump format 4.9.
Robert Casties <casties@mpiwg-berlin.mpg.de>
parents: 100
diff changeset
21 import org.json.JSONException;
1149eb948036 add isodate tag to XML dump format 4.9.
Robert Casties <casties@mpiwg-berlin.mpg.de>
parents: 100
diff changeset
22 import org.json.JSONObject;
1
jurzua
parents:
diff changeset
23 import org.mpi.openmind.repository.bo.Attribute;
jurzua
parents:
diff changeset
24 import org.mpi.openmind.repository.bo.Entity;
jurzua
parents:
diff changeset
25 import org.mpi.openmind.repository.bo.Node;
jurzua
parents:
diff changeset
26 import org.mpi.openmind.repository.bo.Relation;
jurzua
parents:
diff changeset
27 import org.mpi.openmind.repository.services.PersistenceService;
jurzua
parents:
diff changeset
28
31
7d8ebe8ac8a2 create reader and check script for XML dumps.
casties
parents: 29
diff changeset
29 /**
7d8ebe8ac8a2 create reader and check script for XML dumps.
casties
parents: 29
diff changeset
30 * Export all entities and relations and definitions to XML.
7d8ebe8ac8a2 create reader and check script for XML dumps.
casties
parents: 29
diff changeset
31 *
75
e0be7c0030f5 cleanup and better comments.
casties
parents: 31
diff changeset
32 * Saves (content) entities and relations (i.e. assertions) and definitions
31
7d8ebe8ac8a2 create reader and check script for XML dumps.
casties
parents: 29
diff changeset
33 * (i.e. definition entities and relations) in separate files.
7d8ebe8ac8a2 create reader and check script for XML dumps.
casties
parents: 29
diff changeset
34 *
7d8ebe8ac8a2 create reader and check script for XML dumps.
casties
parents: 29
diff changeset
35 * @author jurzua, casties
7d8ebe8ac8a2 create reader and check script for XML dumps.
casties
parents: 29
diff changeset
36 *
7d8ebe8ac8a2 create reader and check script for XML dumps.
casties
parents: 29
diff changeset
37 */
1
jurzua
parents:
diff changeset
38 public class OM4StreamWriter {
jurzua
parents:
diff changeset
39
103
1149eb948036 add isodate tag to XML dump format 4.9.
Robert Casties <casties@mpiwg-berlin.mpg.de>
parents: 100
diff changeset
40 protected static final String FORMAT_VERSION = "4.9";
87
8005f7011975 update version number in xml.
Robert Casties <casties@mpiwg-berlin.mpg.de>
parents: 86
diff changeset
41
29
5786aa6caeb3 new XML export and test script.
casties
parents: 1
diff changeset
42 private static Logger logger = Logger.getLogger(OM4StreamWriter.class);
5786aa6caeb3 new XML export and test script.
casties
parents: 1
diff changeset
43
79
b0aebac0780a put statistics about number of entities, relations and attributes in xml dump.
casties
parents: 78
diff changeset
44 private static final int itemsPerPage = 500;
77
a59984fd3c3f add normalized own-values to xml dump.
casties
parents: 75
diff changeset
45
a59984fd3c3f add normalized own-values to xml dump.
casties
parents: 75
diff changeset
46 /** Include normalized own-values. */
78
b32b176a8aad make normalizations configurable (static).
casties
parents: 77
diff changeset
47 public static boolean includeNormalizations = true;
79
b0aebac0780a put statistics about number of entities, relations and attributes in xml dump.
casties
parents: 78
diff changeset
48
b0aebac0780a put statistics about number of entities, relations and attributes in xml dump.
casties
parents: 78
diff changeset
49 /** key for entity count in attribute counts map */
b0aebac0780a put statistics about number of entities, relations and attributes in xml dump.
casties
parents: 78
diff changeset
50 private static final String ENT_KEY = "<entity-count>";
29
5786aa6caeb3 new XML export and test script.
casties
parents: 1
diff changeset
51
103
1149eb948036 add isodate tag to XML dump format 4.9.
Robert Casties <casties@mpiwg-berlin.mpg.de>
parents: 100
diff changeset
52 /** formatter for isodate tag */
1149eb948036 add isodate tag to XML dump format 4.9.
Robert Casties <casties@mpiwg-berlin.mpg.de>
parents: 100
diff changeset
53 public static DateTimeFormatter dateFormatter = ISODateTimeFormat.date();
1149eb948036 add isodate tag to XML dump format 4.9.
Robert Casties <casties@mpiwg-berlin.mpg.de>
parents: 100
diff changeset
54
29
5786aa6caeb3 new XML export and test script.
casties
parents: 1
diff changeset
55 /**
5786aa6caeb3 new XML export and test script.
casties
parents: 1
diff changeset
56 * Return the object's string representation or "null" if its null.
5786aa6caeb3 new XML export and test script.
casties
parents: 1
diff changeset
57 *
5786aa6caeb3 new XML export and test script.
casties
parents: 1
diff changeset
58 * @param s
5786aa6caeb3 new XML export and test script.
casties
parents: 1
diff changeset
59 * @return
5786aa6caeb3 new XML export and test script.
casties
parents: 1
diff changeset
60 */
31
7d8ebe8ac8a2 create reader and check script for XML dumps.
casties
parents: 29
diff changeset
61 private static String defaultString(Object s) {
29
5786aa6caeb3 new XML export and test script.
casties
parents: 1
diff changeset
62 if (s == null) {
5786aa6caeb3 new XML export and test script.
casties
parents: 1
diff changeset
63 return "null";
5786aa6caeb3 new XML export and test script.
casties
parents: 1
diff changeset
64 } else {
5786aa6caeb3 new XML export and test script.
casties
parents: 1
diff changeset
65 return s.toString();
5786aa6caeb3 new XML export and test script.
casties
parents: 1
diff changeset
66 }
5786aa6caeb3 new XML export and test script.
casties
parents: 1
diff changeset
67 }
5786aa6caeb3 new XML export and test script.
casties
parents: 1
diff changeset
68
5786aa6caeb3 new XML export and test script.
casties
parents: 1
diff changeset
69
75
e0be7c0030f5 cleanup and better comments.
casties
parents: 31
diff changeset
70 /**
e0be7c0030f5 cleanup and better comments.
casties
parents: 31
diff changeset
71 * Saves all content Entities with their Attributes and Relations in a XML file with the given fileName.
e0be7c0030f5 cleanup and better comments.
casties
parents: 31
diff changeset
72 *
e0be7c0030f5 cleanup and better comments.
casties
parents: 31
diff changeset
73 * @param fileName
e0be7c0030f5 cleanup and better comments.
casties
parents: 31
diff changeset
74 * @param ps
e0be7c0030f5 cleanup and better comments.
casties
parents: 31
diff changeset
75 */
29
5786aa6caeb3 new XML export and test script.
casties
parents: 1
diff changeset
76 public static void backupEntities(String fileName, PersistenceService ps) {
78
b32b176a8aad make normalizations configurable (static).
casties
parents: 77
diff changeset
77 writeEntsAndRels(fileName, ps, Node.TYPE_ABOX, includeNormalizations);
29
5786aa6caeb3 new XML export and test script.
casties
parents: 1
diff changeset
78 }
5786aa6caeb3 new XML export and test script.
casties
parents: 1
diff changeset
79
75
e0be7c0030f5 cleanup and better comments.
casties
parents: 31
diff changeset
80 /**
e0be7c0030f5 cleanup and better comments.
casties
parents: 31
diff changeset
81 * Saves all definitions in a XML file with the given fileName.
e0be7c0030f5 cleanup and better comments.
casties
parents: 31
diff changeset
82 *
e0be7c0030f5 cleanup and better comments.
casties
parents: 31
diff changeset
83 * @param fileName
e0be7c0030f5 cleanup and better comments.
casties
parents: 31
diff changeset
84 * @param ps
e0be7c0030f5 cleanup and better comments.
casties
parents: 31
diff changeset
85 */
29
5786aa6caeb3 new XML export and test script.
casties
parents: 1
diff changeset
86 public static void backupDefinitions(String fileName, PersistenceService ps) {
78
b32b176a8aad make normalizations configurable (static).
casties
parents: 77
diff changeset
87 writeEntsAndRels(fileName, ps, Node.TYPE_TBOX, false);
29
5786aa6caeb3 new XML export and test script.
casties
parents: 1
diff changeset
88 }
5786aa6caeb3 new XML export and test script.
casties
parents: 1
diff changeset
89
5786aa6caeb3 new XML export and test script.
casties
parents: 1
diff changeset
90 /**
5786aa6caeb3 new XML export and test script.
casties
parents: 1
diff changeset
91 * Writes all entities of the given type and their relations to the XML file at fileName.
5786aa6caeb3 new XML export and test script.
casties
parents: 1
diff changeset
92 *
75
e0be7c0030f5 cleanup and better comments.
casties
parents: 31
diff changeset
93 * Type is either TYPE_TBOX or TYPE_ABOX.
e0be7c0030f5 cleanup and better comments.
casties
parents: 31
diff changeset
94 *
29
5786aa6caeb3 new XML export and test script.
casties
parents: 1
diff changeset
95 * @param fileName
5786aa6caeb3 new XML export and test script.
casties
parents: 1
diff changeset
96 * @param ps
75
e0be7c0030f5 cleanup and better comments.
casties
parents: 31
diff changeset
97 * @param type
29
5786aa6caeb3 new XML export and test script.
casties
parents: 1
diff changeset
98 */
78
b32b176a8aad make normalizations configurable (static).
casties
parents: 77
diff changeset
99 private static void writeEntsAndRels(String fileName, PersistenceService ps, String type, boolean includeNorm) {
29
5786aa6caeb3 new XML export and test script.
casties
parents: 1
diff changeset
100 OutputStreamWriter out;
5786aa6caeb3 new XML export and test script.
casties
parents: 1
diff changeset
101 try {
79
b0aebac0780a put statistics about number of entities, relations and attributes in xml dump.
casties
parents: 78
diff changeset
102 // statistics collection Maps
b0aebac0780a put statistics about number of entities, relations and attributes in xml dump.
casties
parents: 78
diff changeset
103 Map<String, Map<String, Long>> entStats = new HashMap<String, Map<String, Long>>();
b0aebac0780a put statistics about number of entities, relations and attributes in xml dump.
casties
parents: 78
diff changeset
104 Map<String, Map<String, Long>> relStats = new HashMap<String, Map<String, Long>>();
b0aebac0780a put statistics about number of entities, relations and attributes in xml dump.
casties
parents: 78
diff changeset
105
b0aebac0780a put statistics about number of entities, relations and attributes in xml dump.
casties
parents: 78
diff changeset
106 // setup xml writer
29
5786aa6caeb3 new XML export and test script.
casties
parents: 1
diff changeset
107 FileOutputStream fileOut = new FileOutputStream(fileName);
5786aa6caeb3 new XML export and test script.
casties
parents: 1
diff changeset
108 out = new OutputStreamWriter(fileOut, "UTF-8");
5786aa6caeb3 new XML export and test script.
casties
parents: 1
diff changeset
109 XMLOutputFactory factory = XMLOutputFactory.newInstance();
5786aa6caeb3 new XML export and test script.
casties
parents: 1
diff changeset
110 XMLStreamWriter writer = factory.createXMLStreamWriter(out);
5786aa6caeb3 new XML export and test script.
casties
parents: 1
diff changeset
111
5786aa6caeb3 new XML export and test script.
casties
parents: 1
diff changeset
112 int entitiesCount = 0;
5786aa6caeb3 new XML export and test script.
casties
parents: 1
diff changeset
113
5786aa6caeb3 new XML export and test script.
casties
parents: 1
diff changeset
114 writer.writeStartDocument("UTF-8", "1.0");
5786aa6caeb3 new XML export and test script.
casties
parents: 1
diff changeset
115
5786aa6caeb3 new XML export and test script.
casties
parents: 1
diff changeset
116 if (type.equals(Node.TYPE_ABOX)) {
5786aa6caeb3 new XML export and test script.
casties
parents: 1
diff changeset
117 writer.writeStartElement(XMLUtil.OPENMIND_DATA);
87
8005f7011975 update version number in xml.
Robert Casties <casties@mpiwg-berlin.mpg.de>
parents: 86
diff changeset
118 writer.writeAttribute("version", FORMAT_VERSION);
75
e0be7c0030f5 cleanup and better comments.
casties
parents: 31
diff changeset
119 // get number of content Entities
29
5786aa6caeb3 new XML export and test script.
casties
parents: 1
diff changeset
120 entitiesCount = ps.getEntityCount(null).intValue();
5786aa6caeb3 new XML export and test script.
casties
parents: 1
diff changeset
121 } else {
5786aa6caeb3 new XML export and test script.
casties
parents: 1
diff changeset
122 writer.writeStartElement(XMLUtil.META_DATA);
87
8005f7011975 update version number in xml.
Robert Casties <casties@mpiwg-berlin.mpg.de>
parents: 86
diff changeset
123 writer.writeAttribute("version", FORMAT_VERSION);
75
e0be7c0030f5 cleanup and better comments.
casties
parents: 31
diff changeset
124 // get number of definition Entities
29
5786aa6caeb3 new XML export and test script.
casties
parents: 1
diff changeset
125 entitiesCount = ps.getEntityCount(Node.TYPE_TBOX).intValue();
5786aa6caeb3 new XML export and test script.
casties
parents: 1
diff changeset
126 }
5786aa6caeb3 new XML export and test script.
casties
parents: 1
diff changeset
127
5786aa6caeb3 new XML export and test script.
casties
parents: 1
diff changeset
128 int numberOfPages = entitiesCount / itemsPerPage;
5786aa6caeb3 new XML export and test script.
casties
parents: 1
diff changeset
129 // debug: int numberOfPages = 1;
5786aa6caeb3 new XML export and test script.
casties
parents: 1
diff changeset
130 int counter = 0;
5786aa6caeb3 new XML export and test script.
casties
parents: 1
diff changeset
131 long start = System.currentTimeMillis();
5786aa6caeb3 new XML export and test script.
casties
parents: 1
diff changeset
132 DecimalFormat df = new DecimalFormat("#.##");
1
jurzua
parents:
diff changeset
133
75
e0be7c0030f5 cleanup and better comments.
casties
parents: 31
diff changeset
134 // list of Relations (filled from Entities)
29
5786aa6caeb3 new XML export and test script.
casties
parents: 1
diff changeset
135 List<Relation> relList = new ArrayList<Relation>();
5786aa6caeb3 new XML export and test script.
casties
parents: 1
diff changeset
136
5786aa6caeb3 new XML export and test script.
casties
parents: 1
diff changeset
137 /*
5786aa6caeb3 new XML export and test script.
casties
parents: 1
diff changeset
138 * write entities
5786aa6caeb3 new XML export and test script.
casties
parents: 1
diff changeset
139 */
5786aa6caeb3 new XML export and test script.
casties
parents: 1
diff changeset
140 writer.writeStartElement((type.equals(Node.TYPE_TBOX)) ? XMLUtil.DEFINITIONS : XMLUtil.ENTITIES);
82
90f9a1c45b15 small change to xml format.
Robert Casties <casties@mpiwg-berlin.mpg.de>
parents: 79
diff changeset
141 writer.writeAttribute("count", Integer.toString(entitiesCount));
79
b0aebac0780a put statistics about number of entities, relations and attributes in xml dump.
casties
parents: 78
diff changeset
142 // iterate database by pages
29
5786aa6caeb3 new XML export and test script.
casties
parents: 1
diff changeset
143 for (int currentPage = 0; currentPage <= numberOfPages; currentPage++) {
5786aa6caeb3 new XML export and test script.
casties
parents: 1
diff changeset
144 int startRecord = currentPage * itemsPerPage;
5786aa6caeb3 new XML export and test script.
casties
parents: 1
diff changeset
145 List<Entity> entities;
5786aa6caeb3 new XML export and test script.
casties
parents: 1
diff changeset
146
5786aa6caeb3 new XML export and test script.
casties
parents: 1
diff changeset
147 if (type.equals(Node.TYPE_ABOX)) {
75
e0be7c0030f5 cleanup and better comments.
casties
parents: 31
diff changeset
148 // get page of content Entities
29
5786aa6caeb3 new XML export and test script.
casties
parents: 1
diff changeset
149 entities = ps.getEntityPage(null, startRecord, itemsPerPage);
5786aa6caeb3 new XML export and test script.
casties
parents: 1
diff changeset
150 } else {
75
e0be7c0030f5 cleanup and better comments.
casties
parents: 31
diff changeset
151 // get page of definition Entities
29
5786aa6caeb3 new XML export and test script.
casties
parents: 1
diff changeset
152 entities = ps.getEntityPage(Node.TYPE_TBOX, startRecord, itemsPerPage);
5786aa6caeb3 new XML export and test script.
casties
parents: 1
diff changeset
153 }
5786aa6caeb3 new XML export and test script.
casties
parents: 1
diff changeset
154
79
b0aebac0780a put statistics about number of entities, relations and attributes in xml dump.
casties
parents: 78
diff changeset
155 // iterate entities
29
5786aa6caeb3 new XML export and test script.
casties
parents: 1
diff changeset
156 for (Entity ent : entities) {
75
e0be7c0030f5 cleanup and better comments.
casties
parents: 31
diff changeset
157 // write entity to XML
79
b0aebac0780a put statistics about number of entities, relations and attributes in xml dump.
casties
parents: 78
diff changeset
158 writeEntity(ent, writer, ps, includeNorm, entStats);
29
5786aa6caeb3 new XML export and test script.
casties
parents: 1
diff changeset
159 // add (source)relations to list
86
d4b456623d43 Updated XML export. Saves relation source-type and target-type. Expanded statistics with per-entity-type relation counts.
Robert Casties <casties@mpiwg-berlin.mpg.de>
parents: 82
diff changeset
160 List<Relation> rels = ent.getSourceRelations();
d4b456623d43 Updated XML export. Saves relation source-type and target-type. Expanded statistics with per-entity-type relation counts.
Robert Casties <casties@mpiwg-berlin.mpg.de>
parents: 82
diff changeset
161 relList.addAll(rels);
d4b456623d43 Updated XML export. Saves relation source-type and target-type. Expanded statistics with per-entity-type relation counts.
Robert Casties <casties@mpiwg-berlin.mpg.de>
parents: 82
diff changeset
162 // update stats for relations
d4b456623d43 Updated XML export. Saves relation source-type and target-type. Expanded statistics with per-entity-type relation counts.
Robert Casties <casties@mpiwg-berlin.mpg.de>
parents: 82
diff changeset
163 Map<String, Long> entRelStats = entStats.get(ent.getObjectClass());
d4b456623d43 Updated XML export. Saves relation source-type and target-type. Expanded statistics with per-entity-type relation counts.
Robert Casties <casties@mpiwg-berlin.mpg.de>
parents: 82
diff changeset
164 for (Relation rel: rels) {
d4b456623d43 Updated XML export. Saves relation source-type and target-type. Expanded statistics with per-entity-type relation counts.
Robert Casties <casties@mpiwg-berlin.mpg.de>
parents: 82
diff changeset
165 // update source relations
d4b456623d43 Updated XML export. Saves relation source-type and target-type. Expanded statistics with per-entity-type relation counts.
Robert Casties <casties@mpiwg-berlin.mpg.de>
parents: 82
diff changeset
166 updateRelStats(rel, true, entRelStats);
d4b456623d43 Updated XML export. Saves relation source-type and target-type. Expanded statistics with per-entity-type relation counts.
Robert Casties <casties@mpiwg-berlin.mpg.de>
parents: 82
diff changeset
167 }
d4b456623d43 Updated XML export. Saves relation source-type and target-type. Expanded statistics with per-entity-type relation counts.
Robert Casties <casties@mpiwg-berlin.mpg.de>
parents: 82
diff changeset
168 for (Relation rel: ent.getTargetRelations()) {
d4b456623d43 Updated XML export. Saves relation source-type and target-type. Expanded statistics with per-entity-type relation counts.
Robert Casties <casties@mpiwg-berlin.mpg.de>
parents: 82
diff changeset
169 // update target relations
d4b456623d43 Updated XML export. Saves relation source-type and target-type. Expanded statistics with per-entity-type relation counts.
Robert Casties <casties@mpiwg-berlin.mpg.de>
parents: 82
diff changeset
170 updateRelStats(rel, false, entRelStats);
d4b456623d43 Updated XML export. Saves relation source-type and target-type. Expanded statistics with per-entity-type relation counts.
Robert Casties <casties@mpiwg-berlin.mpg.de>
parents: 82
diff changeset
171 }
d4b456623d43 Updated XML export. Saves relation source-type and target-type. Expanded statistics with per-entity-type relation counts.
Robert Casties <casties@mpiwg-berlin.mpg.de>
parents: 82
diff changeset
172 // count entities
29
5786aa6caeb3 new XML export and test script.
casties
parents: 1
diff changeset
173 counter++;
5786aa6caeb3 new XML export and test script.
casties
parents: 1
diff changeset
174 }
5786aa6caeb3 new XML export and test script.
casties
parents: 1
diff changeset
175
5786aa6caeb3 new XML export and test script.
casties
parents: 1
diff changeset
176 long runtime = System.currentTimeMillis() - start;
5786aa6caeb3 new XML export and test script.
casties
parents: 1
diff changeset
177 double percent = ((double) counter / (double) entitiesCount) * 100.0;
5786aa6caeb3 new XML export and test script.
casties
parents: 1
diff changeset
178 logger.debug("(" + df.format(percent) + "%) \t[" + counter + "/" + entitiesCount + "]\t");
5786aa6caeb3 new XML export and test script.
casties
parents: 1
diff changeset
179 logger.debug("Speed[ents/s]: " + df.format((double) counter / ((double) runtime / 1000)));
5786aa6caeb3 new XML export and test script.
casties
parents: 1
diff changeset
180 writer.flush();
5786aa6caeb3 new XML export and test script.
casties
parents: 1
diff changeset
181 }
5786aa6caeb3 new XML export and test script.
casties
parents: 1
diff changeset
182 writer.writeEndElement();
5786aa6caeb3 new XML export and test script.
casties
parents: 1
diff changeset
183
5786aa6caeb3 new XML export and test script.
casties
parents: 1
diff changeset
184 /*
5786aa6caeb3 new XML export and test script.
casties
parents: 1
diff changeset
185 * write relations (from list)
5786aa6caeb3 new XML export and test script.
casties
parents: 1
diff changeset
186 */
5786aa6caeb3 new XML export and test script.
casties
parents: 1
diff changeset
187 writer.writeStartElement(XMLUtil.RELATIONS);
82
90f9a1c45b15 small change to xml format.
Robert Casties <casties@mpiwg-berlin.mpg.de>
parents: 79
diff changeset
188 writer.writeAttribute("count", Integer.toString(relList.size()));
29
5786aa6caeb3 new XML export and test script.
casties
parents: 1
diff changeset
189 for (Relation rel : relList) {
79
b0aebac0780a put statistics about number of entities, relations and attributes in xml dump.
casties
parents: 78
diff changeset
190 writeRelation(rel, writer, includeNorm, relStats);
29
5786aa6caeb3 new XML export and test script.
casties
parents: 1
diff changeset
191 }
5786aa6caeb3 new XML export and test script.
casties
parents: 1
diff changeset
192 writer.writeEndElement();
5786aa6caeb3 new XML export and test script.
casties
parents: 1
diff changeset
193
79
b0aebac0780a put statistics about number of entities, relations and attributes in xml dump.
casties
parents: 78
diff changeset
194 /*
b0aebac0780a put statistics about number of entities, relations and attributes in xml dump.
casties
parents: 78
diff changeset
195 * write statistics
b0aebac0780a put statistics about number of entities, relations and attributes in xml dump.
casties
parents: 78
diff changeset
196 */
b0aebac0780a put statistics about number of entities, relations and attributes in xml dump.
casties
parents: 78
diff changeset
197 // entity stats
b0aebac0780a put statistics about number of entities, relations and attributes in xml dump.
casties
parents: 78
diff changeset
198 writeStats(XMLUtil.ENTITY_STATS, XMLUtil.ENTITY, entStats, writer);
b0aebac0780a put statistics about number of entities, relations and attributes in xml dump.
casties
parents: 78
diff changeset
199 // relation stats
b0aebac0780a put statistics about number of entities, relations and attributes in xml dump.
casties
parents: 78
diff changeset
200 writeStats(XMLUtil.RELATION_STATS, XMLUtil.RELATION, relStats, writer);
b0aebac0780a put statistics about number of entities, relations and attributes in xml dump.
casties
parents: 78
diff changeset
201
29
5786aa6caeb3 new XML export and test script.
casties
parents: 1
diff changeset
202 // end file.
5786aa6caeb3 new XML export and test script.
casties
parents: 1
diff changeset
203 writer.writeEndElement();
5786aa6caeb3 new XML export and test script.
casties
parents: 1
diff changeset
204
5786aa6caeb3 new XML export and test script.
casties
parents: 1
diff changeset
205 writer.flush();
5786aa6caeb3 new XML export and test script.
casties
parents: 1
diff changeset
206 writer.close();
5786aa6caeb3 new XML export and test script.
casties
parents: 1
diff changeset
207
5786aa6caeb3 new XML export and test script.
casties
parents: 1
diff changeset
208 logger.info("END Stream Writer");
5786aa6caeb3 new XML export and test script.
casties
parents: 1
diff changeset
209 } catch (IOException e) {
75
e0be7c0030f5 cleanup and better comments.
casties
parents: 31
diff changeset
210 logger.error(e);
29
5786aa6caeb3 new XML export and test script.
casties
parents: 1
diff changeset
211 } catch (XMLStreamException e) {
75
e0be7c0030f5 cleanup and better comments.
casties
parents: 31
diff changeset
212 logger.error(e);
29
5786aa6caeb3 new XML export and test script.
casties
parents: 1
diff changeset
213 }
5786aa6caeb3 new XML export and test script.
casties
parents: 1
diff changeset
214 }
1
jurzua
parents:
diff changeset
215
29
5786aa6caeb3 new XML export and test script.
casties
parents: 1
diff changeset
216 /**
5786aa6caeb3 new XML export and test script.
casties
parents: 1
diff changeset
217 * Write OpenMind relation to XML.
5786aa6caeb3 new XML export and test script.
casties
parents: 1
diff changeset
218 *
5786aa6caeb3 new XML export and test script.
casties
parents: 1
diff changeset
219 * @param rel
5786aa6caeb3 new XML export and test script.
casties
parents: 1
diff changeset
220 * @param writer
79
b0aebac0780a put statistics about number of entities, relations and attributes in xml dump.
casties
parents: 78
diff changeset
221 * @param relStats
29
5786aa6caeb3 new XML export and test script.
casties
parents: 1
diff changeset
222 * @throws XMLStreamException
5786aa6caeb3 new XML export and test script.
casties
parents: 1
diff changeset
223 */
79
b0aebac0780a put statistics about number of entities, relations and attributes in xml dump.
casties
parents: 78
diff changeset
224 private static void writeRelation(Relation rel, XMLStreamWriter writer, boolean includeNorm,
b0aebac0780a put statistics about number of entities, relations and attributes in xml dump.
casties
parents: 78
diff changeset
225 Map<String, Map<String, Long>> relStats) throws XMLStreamException {
29
5786aa6caeb3 new XML export and test script.
casties
parents: 1
diff changeset
226 writer.writeStartElement(XMLUtil.RELATION);
5786aa6caeb3 new XML export and test script.
casties
parents: 1
diff changeset
227
79
b0aebac0780a put statistics about number of entities, relations and attributes in xml dump.
casties
parents: 78
diff changeset
228 // update stats
100
734c0d8c7369 add relations-tag with source relations for each entity to XML dump format 4.8.
Robert Casties <casties@mpiwg-berlin.mpg.de>
parents: 87
diff changeset
229 Map<String, Long> attStats = null;
734c0d8c7369 add relations-tag with source relations for each entity to XML dump format 4.8.
Robert Casties <casties@mpiwg-berlin.mpg.de>
parents: 87
diff changeset
230 if (relStats != null) {
734c0d8c7369 add relations-tag with source relations for each entity to XML dump format 4.8.
Robert Casties <casties@mpiwg-berlin.mpg.de>
parents: 87
diff changeset
231 attStats = updateNodeStats(rel, relStats);
734c0d8c7369 add relations-tag with source relations for each entity to XML dump format 4.8.
Robert Casties <casties@mpiwg-berlin.mpg.de>
parents: 87
diff changeset
232 }
79
b0aebac0780a put statistics about number of entities, relations and attributes in xml dump.
casties
parents: 78
diff changeset
233
29
5786aa6caeb3 new XML export and test script.
casties
parents: 1
diff changeset
234 /*
5786aa6caeb3 new XML export and test script.
casties
parents: 1
diff changeset
235 * write XML-attributes
5786aa6caeb3 new XML export and test script.
casties
parents: 1
diff changeset
236 */
5786aa6caeb3 new XML export and test script.
casties
parents: 1
diff changeset
237 writer.writeAttribute(XMLUtil.OBJECT_CLASS, defaultString(rel.getObjectClass()));
5786aa6caeb3 new XML export and test script.
casties
parents: 1
diff changeset
238 writer.writeAttribute(XMLUtil.ID, defaultString(rel.getId()));
5786aa6caeb3 new XML export and test script.
casties
parents: 1
diff changeset
239 writer.writeAttribute(XMLUtil.ROW_ID, defaultString(rel.getRowId()));
75
e0be7c0030f5 cleanup and better comments.
casties
parents: 31
diff changeset
240 if (StringUtils.isNotEmpty(rel.getContentType())) {
29
5786aa6caeb3 new XML export and test script.
casties
parents: 1
diff changeset
241 writer.writeAttribute(XMLUtil.CONTENT_TYPE, rel.getContentType());
75
e0be7c0030f5 cleanup and better comments.
casties
parents: 31
diff changeset
242 }
29
5786aa6caeb3 new XML export and test script.
casties
parents: 1
diff changeset
243 writer.writeAttribute(XMLUtil.RELATION_SOURCE_ID, defaultString(rel.getSourceId()));
86
d4b456623d43 Updated XML export. Saves relation source-type and target-type. Expanded statistics with per-entity-type relation counts.
Robert Casties <casties@mpiwg-berlin.mpg.de>
parents: 82
diff changeset
244 writer.writeAttribute(XMLUtil.RELATION_SOURCE, defaultString(rel.getSourceObjectClass()));
29
5786aa6caeb3 new XML export and test script.
casties
parents: 1
diff changeset
245 writer.writeAttribute(XMLUtil.RELATION_TARGET_ID, defaultString(rel.getTargetId()));
86
d4b456623d43 Updated XML export. Saves relation source-type and target-type. Expanded statistics with per-entity-type relation counts.
Robert Casties <casties@mpiwg-berlin.mpg.de>
parents: 82
diff changeset
246 writer.writeAttribute(XMLUtil.RELATION_TARGET, defaultString(rel.getTargetObjectClass()));
29
5786aa6caeb3 new XML export and test script.
casties
parents: 1
diff changeset
247 writer.writeAttribute(XMLUtil.VERSION, defaultString(rel.getVersion()));
5786aa6caeb3 new XML export and test script.
casties
parents: 1
diff changeset
248 writer.writeAttribute(XMLUtil.MODIFICATION_TIME, defaultString(rel.getModificationTime()));
75
e0be7c0030f5 cleanup and better comments.
casties
parents: 31
diff changeset
249 if (rel.getUser() != null) {
29
5786aa6caeb3 new XML export and test script.
casties
parents: 1
diff changeset
250 writer.writeAttribute(XMLUtil.USER, rel.getUser());
75
e0be7c0030f5 cleanup and better comments.
casties
parents: 31
diff changeset
251 }
e0be7c0030f5 cleanup and better comments.
casties
parents: 31
diff changeset
252 if (rel.getIsPublic()) {
29
5786aa6caeb3 new XML export and test script.
casties
parents: 1
diff changeset
253 writer.writeAttribute(XMLUtil.PUBLIC, "true");
75
e0be7c0030f5 cleanup and better comments.
casties
parents: 31
diff changeset
254 }
29
5786aa6caeb3 new XML export and test script.
casties
parents: 1
diff changeset
255
5786aa6caeb3 new XML export and test script.
casties
parents: 1
diff changeset
256 /*
5786aa6caeb3 new XML export and test script.
casties
parents: 1
diff changeset
257 * write OpenMind attributes of this relation as XML tags
5786aa6caeb3 new XML export and test script.
casties
parents: 1
diff changeset
258 */
5786aa6caeb3 new XML export and test script.
casties
parents: 1
diff changeset
259 if (rel.getAttributes().size() > 0) {
5786aa6caeb3 new XML export and test script.
casties
parents: 1
diff changeset
260 writer.writeStartElement(XMLUtil.ATTRIBUTES);
79
b0aebac0780a put statistics about number of entities, relations and attributes in xml dump.
casties
parents: 78
diff changeset
261 for (Attribute att : rel.getAttributes()) {
100
734c0d8c7369 add relations-tag with source relations for each entity to XML dump format 4.8.
Robert Casties <casties@mpiwg-berlin.mpg.de>
parents: 87
diff changeset
262 if (attStats != null) {
734c0d8c7369 add relations-tag with source relations for each entity to XML dump format 4.8.
Robert Casties <casties@mpiwg-berlin.mpg.de>
parents: 87
diff changeset
263 // update stats
734c0d8c7369 add relations-tag with source relations for each entity to XML dump format 4.8.
Robert Casties <casties@mpiwg-berlin.mpg.de>
parents: 87
diff changeset
264 updateAttStats(att, attStats);
734c0d8c7369 add relations-tag with source relations for each entity to XML dump format 4.8.
Robert Casties <casties@mpiwg-berlin.mpg.de>
parents: 87
diff changeset
265 }
79
b0aebac0780a put statistics about number of entities, relations and attributes in xml dump.
casties
parents: 78
diff changeset
266 // write xml
b0aebac0780a put statistics about number of entities, relations and attributes in xml dump.
casties
parents: 78
diff changeset
267 writeAttribute(att, writer, includeNorm);
b0aebac0780a put statistics about number of entities, relations and attributes in xml dump.
casties
parents: 78
diff changeset
268 }
29
5786aa6caeb3 new XML export and test script.
casties
parents: 1
diff changeset
269 writer.writeEndElement();
5786aa6caeb3 new XML export and test script.
casties
parents: 1
diff changeset
270 }
1
jurzua
parents:
diff changeset
271
29
5786aa6caeb3 new XML export and test script.
casties
parents: 1
diff changeset
272 /*
5786aa6caeb3 new XML export and test script.
casties
parents: 1
diff changeset
273 * write own value as content
5786aa6caeb3 new XML export and test script.
casties
parents: 1
diff changeset
274 */
75
e0be7c0030f5 cleanup and better comments.
casties
parents: 31
diff changeset
275 if (StringUtils.isNotEmpty(rel.getOwnValue())) {
29
5786aa6caeb3 new XML export and test script.
casties
parents: 1
diff changeset
276 writer.writeCharacters(rel.getOwnValue());
75
e0be7c0030f5 cleanup and better comments.
casties
parents: 31
diff changeset
277 }
29
5786aa6caeb3 new XML export and test script.
casties
parents: 1
diff changeset
278
5786aa6caeb3 new XML export and test script.
casties
parents: 1
diff changeset
279 writer.writeEndElement();
5786aa6caeb3 new XML export and test script.
casties
parents: 1
diff changeset
280 }
5786aa6caeb3 new XML export and test script.
casties
parents: 1
diff changeset
281
5786aa6caeb3 new XML export and test script.
casties
parents: 1
diff changeset
282 /**
5786aa6caeb3 new XML export and test script.
casties
parents: 1
diff changeset
283 * Write OpenMind entity to XML.
5786aa6caeb3 new XML export and test script.
casties
parents: 1
diff changeset
284 *
5786aa6caeb3 new XML export and test script.
casties
parents: 1
diff changeset
285 * @param entity
5786aa6caeb3 new XML export and test script.
casties
parents: 1
diff changeset
286 * @param writer
5786aa6caeb3 new XML export and test script.
casties
parents: 1
diff changeset
287 * @param ps
79
b0aebac0780a put statistics about number of entities, relations and attributes in xml dump.
casties
parents: 78
diff changeset
288 * @param entStats
29
5786aa6caeb3 new XML export and test script.
casties
parents: 1
diff changeset
289 * @throws XMLStreamException
5786aa6caeb3 new XML export and test script.
casties
parents: 1
diff changeset
290 */
79
b0aebac0780a put statistics about number of entities, relations and attributes in xml dump.
casties
parents: 78
diff changeset
291 private static void writeEntity(Entity entity, XMLStreamWriter writer, PersistenceService ps, boolean includeNorm,
b0aebac0780a put statistics about number of entities, relations and attributes in xml dump.
casties
parents: 78
diff changeset
292 Map<String, Map<String, Long>> entStats)
29
5786aa6caeb3 new XML export and test script.
casties
parents: 1
diff changeset
293 throws XMLStreamException {
1
jurzua
parents:
diff changeset
294
29
5786aa6caeb3 new XML export and test script.
casties
parents: 1
diff changeset
295 writer.writeStartElement((entity.getType().equals(Node.TYPE_TBOX)) ? XMLUtil.DEFINITION : XMLUtil.ENTITY);
5786aa6caeb3 new XML export and test script.
casties
parents: 1
diff changeset
296
5786aa6caeb3 new XML export and test script.
casties
parents: 1
diff changeset
297 if (entity.isLightweight()) {
86
d4b456623d43 Updated XML export. Saves relation source-type and target-type. Expanded statistics with per-entity-type relation counts.
Robert Casties <casties@mpiwg-berlin.mpg.de>
parents: 82
diff changeset
298 // make sure we have all attributes and relations
29
5786aa6caeb3 new XML export and test script.
casties
parents: 1
diff changeset
299 entity = ps.getEntityContent(entity);
5786aa6caeb3 new XML export and test script.
casties
parents: 1
diff changeset
300 }
79
b0aebac0780a put statistics about number of entities, relations and attributes in xml dump.
casties
parents: 78
diff changeset
301
b0aebac0780a put statistics about number of entities, relations and attributes in xml dump.
casties
parents: 78
diff changeset
302 // update stats
b0aebac0780a put statistics about number of entities, relations and attributes in xml dump.
casties
parents: 78
diff changeset
303 Map<String, Long> attStats = updateNodeStats(entity, entStats);
29
5786aa6caeb3 new XML export and test script.
casties
parents: 1
diff changeset
304
5786aa6caeb3 new XML export and test script.
casties
parents: 1
diff changeset
305 /*
5786aa6caeb3 new XML export and test script.
casties
parents: 1
diff changeset
306 * write XML attributes
5786aa6caeb3 new XML export and test script.
casties
parents: 1
diff changeset
307 */
79
b0aebac0780a put statistics about number of entities, relations and attributes in xml dump.
casties
parents: 78
diff changeset
308 writer.writeAttribute(XMLUtil.OBJECT_CLASS, defaultString(entity.getObjectClass()));
29
5786aa6caeb3 new XML export and test script.
casties
parents: 1
diff changeset
309 writer.writeAttribute(XMLUtil.ID, defaultString(entity.getId()));
5786aa6caeb3 new XML export and test script.
casties
parents: 1
diff changeset
310 writer.writeAttribute(XMLUtil.ROW_ID, defaultString(entity.getRowId()));
75
e0be7c0030f5 cleanup and better comments.
casties
parents: 31
diff changeset
311 if (StringUtils.isNotEmpty(entity.getContentType())) {
29
5786aa6caeb3 new XML export and test script.
casties
parents: 1
diff changeset
312 writer.writeAttribute(XMLUtil.CONTENT_TYPE, entity.getContentType());
75
e0be7c0030f5 cleanup and better comments.
casties
parents: 31
diff changeset
313 }
29
5786aa6caeb3 new XML export and test script.
casties
parents: 1
diff changeset
314 writer.writeAttribute(XMLUtil.VERSION, defaultString(entity.getVersion()));
5786aa6caeb3 new XML export and test script.
casties
parents: 1
diff changeset
315 writer.writeAttribute(XMLUtil.MODIFICATION_TIME, defaultString(entity.getModificationTime()));
75
e0be7c0030f5 cleanup and better comments.
casties
parents: 31
diff changeset
316 if (entity.getUser() != null) {
29
5786aa6caeb3 new XML export and test script.
casties
parents: 1
diff changeset
317 writer.writeAttribute(XMLUtil.USER, entity.getUser());
75
e0be7c0030f5 cleanup and better comments.
casties
parents: 31
diff changeset
318 }
e0be7c0030f5 cleanup and better comments.
casties
parents: 31
diff changeset
319 if (entity.getIsPublic()) {
29
5786aa6caeb3 new XML export and test script.
casties
parents: 1
diff changeset
320 writer.writeAttribute(XMLUtil.PUBLIC, "true");
75
e0be7c0030f5 cleanup and better comments.
casties
parents: 31
diff changeset
321 }
1
jurzua
parents:
diff changeset
322
29
5786aa6caeb3 new XML export and test script.
casties
parents: 1
diff changeset
323 /*
5786aa6caeb3 new XML export and test script.
casties
parents: 1
diff changeset
324 * write OpenMind attributes of this entity as XML tags
5786aa6caeb3 new XML export and test script.
casties
parents: 1
diff changeset
325 */
5786aa6caeb3 new XML export and test script.
casties
parents: 1
diff changeset
326 if (entity.getAttributes().size() > 0) {
5786aa6caeb3 new XML export and test script.
casties
parents: 1
diff changeset
327 writer.writeStartElement(XMLUtil.ATTRIBUTES);
5786aa6caeb3 new XML export and test script.
casties
parents: 1
diff changeset
328 for (Attribute att : entity.getAttributes()) {
86
d4b456623d43 Updated XML export. Saves relation source-type and target-type. Expanded statistics with per-entity-type relation counts.
Robert Casties <casties@mpiwg-berlin.mpg.de>
parents: 82
diff changeset
329 // update stats
d4b456623d43 Updated XML export. Saves relation source-type and target-type. Expanded statistics with per-entity-type relation counts.
Robert Casties <casties@mpiwg-berlin.mpg.de>
parents: 82
diff changeset
330 updateAttStats(att, attStats);
d4b456623d43 Updated XML export. Saves relation source-type and target-type. Expanded statistics with per-entity-type relation counts.
Robert Casties <casties@mpiwg-berlin.mpg.de>
parents: 82
diff changeset
331 // write xml
78
b32b176a8aad make normalizations configurable (static).
casties
parents: 77
diff changeset
332 writeAttribute(att, writer, includeNorm);
29
5786aa6caeb3 new XML export and test script.
casties
parents: 1
diff changeset
333 }
5786aa6caeb3 new XML export and test script.
casties
parents: 1
diff changeset
334 writer.writeEndElement();
5786aa6caeb3 new XML export and test script.
casties
parents: 1
diff changeset
335 }
5786aa6caeb3 new XML export and test script.
casties
parents: 1
diff changeset
336
5786aa6caeb3 new XML export and test script.
casties
parents: 1
diff changeset
337 /*
100
734c0d8c7369 add relations-tag with source relations for each entity to XML dump format 4.8.
Robert Casties <casties@mpiwg-berlin.mpg.de>
parents: 87
diff changeset
338 * write outgoing relations of this entity as XML tags
734c0d8c7369 add relations-tag with source relations for each entity to XML dump format 4.8.
Robert Casties <casties@mpiwg-berlin.mpg.de>
parents: 87
diff changeset
339 */
734c0d8c7369 add relations-tag with source relations for each entity to XML dump format 4.8.
Robert Casties <casties@mpiwg-berlin.mpg.de>
parents: 87
diff changeset
340 if (entity.getSourceRelations().size() > 0) {
734c0d8c7369 add relations-tag with source relations for each entity to XML dump format 4.8.
Robert Casties <casties@mpiwg-berlin.mpg.de>
parents: 87
diff changeset
341 writer.writeStartElement(XMLUtil.RELATIONS);
734c0d8c7369 add relations-tag with source relations for each entity to XML dump format 4.8.
Robert Casties <casties@mpiwg-berlin.mpg.de>
parents: 87
diff changeset
342 for (Relation rel : entity.getSourceRelations()) {
734c0d8c7369 add relations-tag with source relations for each entity to XML dump format 4.8.
Robert Casties <casties@mpiwg-berlin.mpg.de>
parents: 87
diff changeset
343 // write xml (without stats)
734c0d8c7369 add relations-tag with source relations for each entity to XML dump format 4.8.
Robert Casties <casties@mpiwg-berlin.mpg.de>
parents: 87
diff changeset
344 writeRelation(rel, writer, includeNorm, null);
734c0d8c7369 add relations-tag with source relations for each entity to XML dump format 4.8.
Robert Casties <casties@mpiwg-berlin.mpg.de>
parents: 87
diff changeset
345 }
734c0d8c7369 add relations-tag with source relations for each entity to XML dump format 4.8.
Robert Casties <casties@mpiwg-berlin.mpg.de>
parents: 87
diff changeset
346 writer.writeEndElement();
734c0d8c7369 add relations-tag with source relations for each entity to XML dump format 4.8.
Robert Casties <casties@mpiwg-berlin.mpg.de>
parents: 87
diff changeset
347 }
734c0d8c7369 add relations-tag with source relations for each entity to XML dump format 4.8.
Robert Casties <casties@mpiwg-berlin.mpg.de>
parents: 87
diff changeset
348
734c0d8c7369 add relations-tag with source relations for each entity to XML dump format 4.8.
Robert Casties <casties@mpiwg-berlin.mpg.de>
parents: 87
diff changeset
349 /*
29
5786aa6caeb3 new XML export and test script.
casties
parents: 1
diff changeset
350 * write own value
5786aa6caeb3 new XML export and test script.
casties
parents: 1
diff changeset
351 */
77
a59984fd3c3f add normalized own-values to xml dump.
casties
parents: 75
diff changeset
352 String ov = entity.getOwnValue();
a59984fd3c3f add normalized own-values to xml dump.
casties
parents: 75
diff changeset
353 if (StringUtils.isNotEmpty(ov)) {
a59984fd3c3f add normalized own-values to xml dump.
casties
parents: 75
diff changeset
354 writer.writeCharacters(ov);
a59984fd3c3f add normalized own-values to xml dump.
casties
parents: 75
diff changeset
355 String nov = entity.getNormalizedOwnValue();
a59984fd3c3f add normalized own-values to xml dump.
casties
parents: 75
diff changeset
356 if (includeNorm && StringUtils.isNotEmpty(nov) && !ov.equals(nov)) {
a59984fd3c3f add normalized own-values to xml dump.
casties
parents: 75
diff changeset
357 // write normalized value
a59984fd3c3f add normalized own-values to xml dump.
casties
parents: 75
diff changeset
358 writer.writeStartElement(XMLUtil.NORMALIZED);
a59984fd3c3f add normalized own-values to xml dump.
casties
parents: 75
diff changeset
359 writer.writeCharacters(nov);
a59984fd3c3f add normalized own-values to xml dump.
casties
parents: 75
diff changeset
360 writer.writeEndElement();
a59984fd3c3f add normalized own-values to xml dump.
casties
parents: 75
diff changeset
361 }
75
e0be7c0030f5 cleanup and better comments.
casties
parents: 31
diff changeset
362 }
29
5786aa6caeb3 new XML export and test script.
casties
parents: 1
diff changeset
363
5786aa6caeb3 new XML export and test script.
casties
parents: 1
diff changeset
364 writer.writeEndElement();
5786aa6caeb3 new XML export and test script.
casties
parents: 1
diff changeset
365 }
5786aa6caeb3 new XML export and test script.
casties
parents: 1
diff changeset
366
79
b0aebac0780a put statistics about number of entities, relations and attributes in xml dump.
casties
parents: 78
diff changeset
367
b0aebac0780a put statistics about number of entities, relations and attributes in xml dump.
casties
parents: 78
diff changeset
368 private static void writeAttribute(Attribute att, XMLStreamWriter writer, boolean includeNorm) throws XMLStreamException {
29
5786aa6caeb3 new XML export and test script.
casties
parents: 1
diff changeset
369 writer.writeStartElement(XMLUtil.ATTRIBUTE);
5786aa6caeb3 new XML export and test script.
casties
parents: 1
diff changeset
370
5786aa6caeb3 new XML export and test script.
casties
parents: 1
diff changeset
371 /*
5786aa6caeb3 new XML export and test script.
casties
parents: 1
diff changeset
372 * write XML attributes
5786aa6caeb3 new XML export and test script.
casties
parents: 1
diff changeset
373 */
5786aa6caeb3 new XML export and test script.
casties
parents: 1
diff changeset
374 writer.writeAttribute(XMLUtil.ATTRIBUTE_NAME, defaultString(att.getName()));
5786aa6caeb3 new XML export and test script.
casties
parents: 1
diff changeset
375 writer.writeAttribute(XMLUtil.ID, defaultString(att.getId()));
5786aa6caeb3 new XML export and test script.
casties
parents: 1
diff changeset
376 writer.writeAttribute(XMLUtil.ROW_ID, defaultString(att.getRowId()));
5786aa6caeb3 new XML export and test script.
casties
parents: 1
diff changeset
377 writer.writeAttribute(XMLUtil.CONTENT_TYPE, defaultString(att.getContentType()));
5786aa6caeb3 new XML export and test script.
casties
parents: 1
diff changeset
378 writer.writeAttribute(XMLUtil.VERSION, defaultString(att.getVersion()));
5786aa6caeb3 new XML export and test script.
casties
parents: 1
diff changeset
379 writer.writeAttribute(XMLUtil.MODIFICATION_TIME, defaultString(att.getModificationTime()));
75
e0be7c0030f5 cleanup and better comments.
casties
parents: 31
diff changeset
380 if (att.getUser() != null) {
29
5786aa6caeb3 new XML export and test script.
casties
parents: 1
diff changeset
381 writer.writeAttribute(XMLUtil.USER, att.getUser());
75
e0be7c0030f5 cleanup and better comments.
casties
parents: 31
diff changeset
382 }
e0be7c0030f5 cleanup and better comments.
casties
parents: 31
diff changeset
383 if (att.getIsPublic()) {
29
5786aa6caeb3 new XML export and test script.
casties
parents: 1
diff changeset
384 writer.writeAttribute(XMLUtil.PUBLIC, "true");
75
e0be7c0030f5 cleanup and better comments.
casties
parents: 31
diff changeset
385 }
e0be7c0030f5 cleanup and better comments.
casties
parents: 31
diff changeset
386
29
5786aa6caeb3 new XML export and test script.
casties
parents: 1
diff changeset
387 /*
5786aa6caeb3 new XML export and test script.
casties
parents: 1
diff changeset
388 * write value as content
5786aa6caeb3 new XML export and test script.
casties
parents: 1
diff changeset
389 */
77
a59984fd3c3f add normalized own-values to xml dump.
casties
parents: 75
diff changeset
390 String ov = att.getValue();
a59984fd3c3f add normalized own-values to xml dump.
casties
parents: 75
diff changeset
391 if (StringUtils.isNotEmpty(ov)) {
a59984fd3c3f add normalized own-values to xml dump.
casties
parents: 75
diff changeset
392 writer.writeCharacters(ov);
a59984fd3c3f add normalized own-values to xml dump.
casties
parents: 75
diff changeset
393 String nov = att.getNormalizedOwnValue();
a59984fd3c3f add normalized own-values to xml dump.
casties
parents: 75
diff changeset
394 if (includeNorm && StringUtils.isNotEmpty(nov) && !ov.equals(nov)) {
a59984fd3c3f add normalized own-values to xml dump.
casties
parents: 75
diff changeset
395 // write normalized value
a59984fd3c3f add normalized own-values to xml dump.
casties
parents: 75
diff changeset
396 writer.writeStartElement(XMLUtil.NORMALIZED);
a59984fd3c3f add normalized own-values to xml dump.
casties
parents: 75
diff changeset
397 writer.writeCharacters(nov);
a59984fd3c3f add normalized own-values to xml dump.
casties
parents: 75
diff changeset
398 writer.writeEndElement();
a59984fd3c3f add normalized own-values to xml dump.
casties
parents: 75
diff changeset
399 }
103
1149eb948036 add isodate tag to XML dump format 4.9.
Robert Casties <casties@mpiwg-berlin.mpg.de>
parents: 100
diff changeset
400 // convert date JSON into additional isodate
1149eb948036 add isodate tag to XML dump format 4.9.
Robert Casties <casties@mpiwg-berlin.mpg.de>
parents: 100
diff changeset
401 if (ov.startsWith("{")) {
1149eb948036 add isodate tag to XML dump format 4.9.
Robert Casties <casties@mpiwg-berlin.mpg.de>
parents: 100
diff changeset
402 try {
1149eb948036 add isodate tag to XML dump format 4.9.
Robert Casties <casties@mpiwg-berlin.mpg.de>
parents: 100
diff changeset
403 JSONObject json = new JSONObject(ov);
1149eb948036 add isodate tag to XML dump format 4.9.
Robert Casties <casties@mpiwg-berlin.mpg.de>
parents: 100
diff changeset
404 JSONObject date = null;
1149eb948036 add isodate tag to XML dump format 4.9.
Robert Casties <casties@mpiwg-berlin.mpg.de>
parents: 100
diff changeset
405 if (json.has("date")) {
1149eb948036 add isodate tag to XML dump format 4.9.
Robert Casties <casties@mpiwg-berlin.mpg.de>
parents: 100
diff changeset
406 date = json.getJSONObject("date");
1149eb948036 add isodate tag to XML dump format 4.9.
Robert Casties <casties@mpiwg-berlin.mpg.de>
parents: 100
diff changeset
407 } else if (json.has("from")) {
1149eb948036 add isodate tag to XML dump format 4.9.
Robert Casties <casties@mpiwg-berlin.mpg.de>
parents: 100
diff changeset
408 date = json.getJSONObject("from");
1149eb948036 add isodate tag to XML dump format 4.9.
Robert Casties <casties@mpiwg-berlin.mpg.de>
parents: 100
diff changeset
409 }
1149eb948036 add isodate tag to XML dump format 4.9.
Robert Casties <casties@mpiwg-berlin.mpg.de>
parents: 100
diff changeset
410 if (date != null) {
1149eb948036 add isodate tag to XML dump format 4.9.
Robert Casties <casties@mpiwg-berlin.mpg.de>
parents: 100
diff changeset
411 int year = date.getInt("year");
1149eb948036 add isodate tag to XML dump format 4.9.
Robert Casties <casties@mpiwg-berlin.mpg.de>
parents: 100
diff changeset
412 int month = date.getInt("month");
1149eb948036 add isodate tag to XML dump format 4.9.
Robert Casties <casties@mpiwg-berlin.mpg.de>
parents: 100
diff changeset
413 int day = date.getInt("dayOfMonth");
1149eb948036 add isodate tag to XML dump format 4.9.
Robert Casties <casties@mpiwg-berlin.mpg.de>
parents: 100
diff changeset
414 DateTime dt = new DateTime(year, month, day, 0, 0);
1149eb948036 add isodate tag to XML dump format 4.9.
Robert Casties <casties@mpiwg-berlin.mpg.de>
parents: 100
diff changeset
415 writer.writeStartElement(XMLUtil.ISODATE);
1149eb948036 add isodate tag to XML dump format 4.9.
Robert Casties <casties@mpiwg-berlin.mpg.de>
parents: 100
diff changeset
416 writer.writeCharacters(dateFormatter.print(dt));
1149eb948036 add isodate tag to XML dump format 4.9.
Robert Casties <casties@mpiwg-berlin.mpg.de>
parents: 100
diff changeset
417 writer.writeEndElement();
1149eb948036 add isodate tag to XML dump format 4.9.
Robert Casties <casties@mpiwg-berlin.mpg.de>
parents: 100
diff changeset
418 }
1149eb948036 add isodate tag to XML dump format 4.9.
Robert Casties <casties@mpiwg-berlin.mpg.de>
parents: 100
diff changeset
419 } catch (JSONException e) {
1149eb948036 add isodate tag to XML dump format 4.9.
Robert Casties <casties@mpiwg-berlin.mpg.de>
parents: 100
diff changeset
420 // maybe not JSON...
1149eb948036 add isodate tag to XML dump format 4.9.
Robert Casties <casties@mpiwg-berlin.mpg.de>
parents: 100
diff changeset
421 }
1149eb948036 add isodate tag to XML dump format 4.9.
Robert Casties <casties@mpiwg-berlin.mpg.de>
parents: 100
diff changeset
422 }
75
e0be7c0030f5 cleanup and better comments.
casties
parents: 31
diff changeset
423 }
1
jurzua
parents:
diff changeset
424
29
5786aa6caeb3 new XML export and test script.
casties
parents: 1
diff changeset
425 writer.writeEndElement();
5786aa6caeb3 new XML export and test script.
casties
parents: 1
diff changeset
426 }
79
b0aebac0780a put statistics about number of entities, relations and attributes in xml dump.
casties
parents: 78
diff changeset
427
b0aebac0780a put statistics about number of entities, relations and attributes in xml dump.
casties
parents: 78
diff changeset
428
86
d4b456623d43 Updated XML export. Saves relation source-type and target-type. Expanded statistics with per-entity-type relation counts.
Robert Casties <casties@mpiwg-berlin.mpg.de>
parents: 82
diff changeset
429 private static void writeStats(String statsTag, String entryTag, Map<String, Map<String, Long>> nodeStats,
d4b456623d43 Updated XML export. Saves relation source-type and target-type. Expanded statistics with per-entity-type relation counts.
Robert Casties <casties@mpiwg-berlin.mpg.de>
parents: 82
diff changeset
430 XMLStreamWriter writer) throws XMLStreamException {
d4b456623d43 Updated XML export. Saves relation source-type and target-type. Expanded statistics with per-entity-type relation counts.
Robert Casties <casties@mpiwg-berlin.mpg.de>
parents: 82
diff changeset
431 // write stats tag
79
b0aebac0780a put statistics about number of entities, relations and attributes in xml dump.
casties
parents: 78
diff changeset
432 writer.writeStartElement(statsTag);
86
d4b456623d43 Updated XML export. Saves relation source-type and target-type. Expanded statistics with per-entity-type relation counts.
Robert Casties <casties@mpiwg-berlin.mpg.de>
parents: 82
diff changeset
433
79
b0aebac0780a put statistics about number of entities, relations and attributes in xml dump.
casties
parents: 78
diff changeset
434 for (String nodeType : nodeStats.keySet()) {
86
d4b456623d43 Updated XML export. Saves relation source-type and target-type. Expanded statistics with per-entity-type relation counts.
Robert Casties <casties@mpiwg-berlin.mpg.de>
parents: 82
diff changeset
435 Map<String, Long> attStats = nodeStats.get(nodeType);
d4b456623d43 Updated XML export. Saves relation source-type and target-type. Expanded statistics with per-entity-type relation counts.
Robert Casties <casties@mpiwg-berlin.mpg.de>
parents: 82
diff changeset
436 Long nodeCnt = attStats.get(ENT_KEY);
d4b456623d43 Updated XML export. Saves relation source-type and target-type. Expanded statistics with per-entity-type relation counts.
Robert Casties <casties@mpiwg-berlin.mpg.de>
parents: 82
diff changeset
437 // write tag for entity/attribute
d4b456623d43 Updated XML export. Saves relation source-type and target-type. Expanded statistics with per-entity-type relation counts.
Robert Casties <casties@mpiwg-berlin.mpg.de>
parents: 82
diff changeset
438 writer.writeStartElement(entryTag);
d4b456623d43 Updated XML export. Saves relation source-type and target-type. Expanded statistics with per-entity-type relation counts.
Robert Casties <casties@mpiwg-berlin.mpg.de>
parents: 82
diff changeset
439 writer.writeAttribute(XMLUtil.OBJECT_CLASS, (nodeType == null) ? "null" : nodeType);
d4b456623d43 Updated XML export. Saves relation source-type and target-type. Expanded statistics with per-entity-type relation counts.
Robert Casties <casties@mpiwg-berlin.mpg.de>
parents: 82
diff changeset
440 writer.writeAttribute(XMLUtil.COUNT, nodeCnt.toString());
d4b456623d43 Updated XML export. Saves relation source-type and target-type. Expanded statistics with per-entity-type relation counts.
Robert Casties <casties@mpiwg-berlin.mpg.de>
parents: 82
diff changeset
441
d4b456623d43 Updated XML export. Saves relation source-type and target-type. Expanded statistics with per-entity-type relation counts.
Robert Casties <casties@mpiwg-berlin.mpg.de>
parents: 82
diff changeset
442 // write attributes
d4b456623d43 Updated XML export. Saves relation source-type and target-type. Expanded statistics with per-entity-type relation counts.
Robert Casties <casties@mpiwg-berlin.mpg.de>
parents: 82
diff changeset
443 for (String attName : attStats.keySet()) {
d4b456623d43 Updated XML export. Saves relation source-type and target-type. Expanded statistics with per-entity-type relation counts.
Robert Casties <casties@mpiwg-berlin.mpg.de>
parents: 82
diff changeset
444 // skip ENT_KEY
d4b456623d43 Updated XML export. Saves relation source-type and target-type. Expanded statistics with per-entity-type relation counts.
Robert Casties <casties@mpiwg-berlin.mpg.de>
parents: 82
diff changeset
445 if (attName.equals(ENT_KEY))
d4b456623d43 Updated XML export. Saves relation source-type and target-type. Expanded statistics with per-entity-type relation counts.
Robert Casties <casties@mpiwg-berlin.mpg.de>
parents: 82
diff changeset
446 continue;
d4b456623d43 Updated XML export. Saves relation source-type and target-type. Expanded statistics with per-entity-type relation counts.
Robert Casties <casties@mpiwg-berlin.mpg.de>
parents: 82
diff changeset
447 if (attName.contains("[")) {
d4b456623d43 Updated XML export. Saves relation source-type and target-type. Expanded statistics with per-entity-type relation counts.
Robert Casties <casties@mpiwg-berlin.mpg.de>
parents: 82
diff changeset
448 // write relation tag
d4b456623d43 Updated XML export. Saves relation source-type and target-type. Expanded statistics with per-entity-type relation counts.
Robert Casties <casties@mpiwg-berlin.mpg.de>
parents: 82
diff changeset
449 writer.writeStartElement(XMLUtil.RELATION);
d4b456623d43 Updated XML export. Saves relation source-type and target-type. Expanded statistics with per-entity-type relation counts.
Robert Casties <casties@mpiwg-berlin.mpg.de>
parents: 82
diff changeset
450 } else {
d4b456623d43 Updated XML export. Saves relation source-type and target-type. Expanded statistics with per-entity-type relation counts.
Robert Casties <casties@mpiwg-berlin.mpg.de>
parents: 82
diff changeset
451 // write attribute tag
d4b456623d43 Updated XML export. Saves relation source-type and target-type. Expanded statistics with per-entity-type relation counts.
Robert Casties <casties@mpiwg-berlin.mpg.de>
parents: 82
diff changeset
452 writer.writeStartElement(XMLUtil.ATTRIBUTE);
d4b456623d43 Updated XML export. Saves relation source-type and target-type. Expanded statistics with per-entity-type relation counts.
Robert Casties <casties@mpiwg-berlin.mpg.de>
parents: 82
diff changeset
453 }
d4b456623d43 Updated XML export. Saves relation source-type and target-type. Expanded statistics with per-entity-type relation counts.
Robert Casties <casties@mpiwg-berlin.mpg.de>
parents: 82
diff changeset
454 writer.writeAttribute(XMLUtil.ATTRIBUTE_NAME, attName);
d4b456623d43 Updated XML export. Saves relation source-type and target-type. Expanded statistics with per-entity-type relation counts.
Robert Casties <casties@mpiwg-berlin.mpg.de>
parents: 82
diff changeset
455 Long attCnt = attStats.get(attName);
d4b456623d43 Updated XML export. Saves relation source-type and target-type. Expanded statistics with per-entity-type relation counts.
Robert Casties <casties@mpiwg-berlin.mpg.de>
parents: 82
diff changeset
456 writer.writeAttribute(XMLUtil.COUNT, attCnt.toString());
d4b456623d43 Updated XML export. Saves relation source-type and target-type. Expanded statistics with per-entity-type relation counts.
Robert Casties <casties@mpiwg-berlin.mpg.de>
parents: 82
diff changeset
457 writer.writeEndElement();
d4b456623d43 Updated XML export. Saves relation source-type and target-type. Expanded statistics with per-entity-type relation counts.
Robert Casties <casties@mpiwg-berlin.mpg.de>
parents: 82
diff changeset
458 }
d4b456623d43 Updated XML export. Saves relation source-type and target-type. Expanded statistics with per-entity-type relation counts.
Robert Casties <casties@mpiwg-berlin.mpg.de>
parents: 82
diff changeset
459 // end of entity/attribute tag
d4b456623d43 Updated XML export. Saves relation source-type and target-type. Expanded statistics with per-entity-type relation counts.
Robert Casties <casties@mpiwg-berlin.mpg.de>
parents: 82
diff changeset
460 writer.writeEndElement();
79
b0aebac0780a put statistics about number of entities, relations and attributes in xml dump.
casties
parents: 78
diff changeset
461 }
b0aebac0780a put statistics about number of entities, relations and attributes in xml dump.
casties
parents: 78
diff changeset
462 // end of stats tag
b0aebac0780a put statistics about number of entities, relations and attributes in xml dump.
casties
parents: 78
diff changeset
463 writer.writeEndElement();
86
d4b456623d43 Updated XML export. Saves relation source-type and target-type. Expanded statistics with per-entity-type relation counts.
Robert Casties <casties@mpiwg-berlin.mpg.de>
parents: 82
diff changeset
464 }
79
b0aebac0780a put statistics about number of entities, relations and attributes in xml dump.
casties
parents: 78
diff changeset
465
b0aebac0780a put statistics about number of entities, relations and attributes in xml dump.
casties
parents: 78
diff changeset
466 /**
b0aebac0780a put statistics about number of entities, relations and attributes in xml dump.
casties
parents: 78
diff changeset
467 * @param objectClass
b0aebac0780a put statistics about number of entities, relations and attributes in xml dump.
casties
parents: 78
diff changeset
468 * @param entStats
b0aebac0780a put statistics about number of entities, relations and attributes in xml dump.
casties
parents: 78
diff changeset
469 * @return
b0aebac0780a put statistics about number of entities, relations and attributes in xml dump.
casties
parents: 78
diff changeset
470 */
86
d4b456623d43 Updated XML export. Saves relation source-type and target-type. Expanded statistics with per-entity-type relation counts.
Robert Casties <casties@mpiwg-berlin.mpg.de>
parents: 82
diff changeset
471 protected static Map<String, Long> updateNodeStats(Node ent, Map<String, Map<String, Long>> entStats) {
d4b456623d43 Updated XML export. Saves relation source-type and target-type. Expanded statistics with per-entity-type relation counts.
Robert Casties <casties@mpiwg-berlin.mpg.de>
parents: 82
diff changeset
472 String objectClass = ent.getObjectClass();
79
b0aebac0780a put statistics about number of entities, relations and attributes in xml dump.
casties
parents: 78
diff changeset
473 Map<String, Long> attStats = entStats.get(objectClass);
b0aebac0780a put statistics about number of entities, relations and attributes in xml dump.
casties
parents: 78
diff changeset
474 if (attStats == null) {
86
d4b456623d43 Updated XML export. Saves relation source-type and target-type. Expanded statistics with per-entity-type relation counts.
Robert Casties <casties@mpiwg-berlin.mpg.de>
parents: 82
diff changeset
475 // create new attribute stats entry
d4b456623d43 Updated XML export. Saves relation source-type and target-type. Expanded statistics with per-entity-type relation counts.
Robert Casties <casties@mpiwg-berlin.mpg.de>
parents: 82
diff changeset
476 attStats = new HashMap<String, Long>();
d4b456623d43 Updated XML export. Saves relation source-type and target-type. Expanded statistics with per-entity-type relation counts.
Robert Casties <casties@mpiwg-berlin.mpg.de>
parents: 82
diff changeset
477 // add key to count entities
d4b456623d43 Updated XML export. Saves relation source-type and target-type. Expanded statistics with per-entity-type relation counts.
Robert Casties <casties@mpiwg-berlin.mpg.de>
parents: 82
diff changeset
478 attStats.put(ENT_KEY, 1l);
d4b456623d43 Updated XML export. Saves relation source-type and target-type. Expanded statistics with per-entity-type relation counts.
Robert Casties <casties@mpiwg-berlin.mpg.de>
parents: 82
diff changeset
479 // add to map
d4b456623d43 Updated XML export. Saves relation source-type and target-type. Expanded statistics with per-entity-type relation counts.
Robert Casties <casties@mpiwg-berlin.mpg.de>
parents: 82
diff changeset
480 entStats.put(objectClass, attStats);
79
b0aebac0780a put statistics about number of entities, relations and attributes in xml dump.
casties
parents: 78
diff changeset
481 } else {
86
d4b456623d43 Updated XML export. Saves relation source-type and target-type. Expanded statistics with per-entity-type relation counts.
Robert Casties <casties@mpiwg-berlin.mpg.de>
parents: 82
diff changeset
482 // increment entity count
d4b456623d43 Updated XML export. Saves relation source-type and target-type. Expanded statistics with per-entity-type relation counts.
Robert Casties <casties@mpiwg-berlin.mpg.de>
parents: 82
diff changeset
483 Long entCnt = attStats.get(ENT_KEY);
d4b456623d43 Updated XML export. Saves relation source-type and target-type. Expanded statistics with per-entity-type relation counts.
Robert Casties <casties@mpiwg-berlin.mpg.de>
parents: 82
diff changeset
484 attStats.put(ENT_KEY, entCnt + 1);
79
b0aebac0780a put statistics about number of entities, relations and attributes in xml dump.
casties
parents: 78
diff changeset
485 }
86
d4b456623d43 Updated XML export. Saves relation source-type and target-type. Expanded statistics with per-entity-type relation counts.
Robert Casties <casties@mpiwg-berlin.mpg.de>
parents: 82
diff changeset
486 return attStats;
d4b456623d43 Updated XML export. Saves relation source-type and target-type. Expanded statistics with per-entity-type relation counts.
Robert Casties <casties@mpiwg-berlin.mpg.de>
parents: 82
diff changeset
487 }
79
b0aebac0780a put statistics about number of entities, relations and attributes in xml dump.
casties
parents: 78
diff changeset
488
b0aebac0780a put statistics about number of entities, relations and attributes in xml dump.
casties
parents: 78
diff changeset
489 /**
b0aebac0780a put statistics about number of entities, relations and attributes in xml dump.
casties
parents: 78
diff changeset
490 * @param att
b0aebac0780a put statistics about number of entities, relations and attributes in xml dump.
casties
parents: 78
diff changeset
491 * @param attStats
b0aebac0780a put statistics about number of entities, relations and attributes in xml dump.
casties
parents: 78
diff changeset
492 */
b0aebac0780a put statistics about number of entities, relations and attributes in xml dump.
casties
parents: 78
diff changeset
493 protected static void updateAttStats(Attribute att, Map<String, Long> attStats) {
b0aebac0780a put statistics about number of entities, relations and attributes in xml dump.
casties
parents: 78
diff changeset
494 String attName = att.getName();
b0aebac0780a put statistics about number of entities, relations and attributes in xml dump.
casties
parents: 78
diff changeset
495 Long cnt = attStats.get(attName);
b0aebac0780a put statistics about number of entities, relations and attributes in xml dump.
casties
parents: 78
diff changeset
496 if (cnt == null) {
b0aebac0780a put statistics about number of entities, relations and attributes in xml dump.
casties
parents: 78
diff changeset
497 attStats.put(attName, 1l);
b0aebac0780a put statistics about number of entities, relations and attributes in xml dump.
casties
parents: 78
diff changeset
498 } else {
b0aebac0780a put statistics about number of entities, relations and attributes in xml dump.
casties
parents: 78
diff changeset
499 attStats.put(attName, cnt + 1);
b0aebac0780a put statistics about number of entities, relations and attributes in xml dump.
casties
parents: 78
diff changeset
500 }
b0aebac0780a put statistics about number of entities, relations and attributes in xml dump.
casties
parents: 78
diff changeset
501 }
b0aebac0780a put statistics about number of entities, relations and attributes in xml dump.
casties
parents: 78
diff changeset
502
86
d4b456623d43 Updated XML export. Saves relation source-type and target-type. Expanded statistics with per-entity-type relation counts.
Robert Casties <casties@mpiwg-berlin.mpg.de>
parents: 82
diff changeset
503 /**
d4b456623d43 Updated XML export. Saves relation source-type and target-type. Expanded statistics with per-entity-type relation counts.
Robert Casties <casties@mpiwg-berlin.mpg.de>
parents: 82
diff changeset
504 * Update relation statistics.
d4b456623d43 Updated XML export. Saves relation source-type and target-type. Expanded statistics with per-entity-type relation counts.
Robert Casties <casties@mpiwg-berlin.mpg.de>
parents: 82
diff changeset
505 *
d4b456623d43 Updated XML export. Saves relation source-type and target-type. Expanded statistics with per-entity-type relation counts.
Robert Casties <casties@mpiwg-berlin.mpg.de>
parents: 82
diff changeset
506 * Relation stats are saved like attribute stats but with "[entity-type]" before
d4b456623d43 Updated XML export. Saves relation source-type and target-type. Expanded statistics with per-entity-type relation counts.
Robert Casties <casties@mpiwg-berlin.mpg.de>
parents: 82
diff changeset
507 * or after the relation name.
d4b456623d43 Updated XML export. Saves relation source-type and target-type. Expanded statistics with per-entity-type relation counts.
Robert Casties <casties@mpiwg-berlin.mpg.de>
parents: 82
diff changeset
508 *
d4b456623d43 Updated XML export. Saves relation source-type and target-type. Expanded statistics with per-entity-type relation counts.
Robert Casties <casties@mpiwg-berlin.mpg.de>
parents: 82
diff changeset
509 * @param rel
d4b456623d43 Updated XML export. Saves relation source-type and target-type. Expanded statistics with per-entity-type relation counts.
Robert Casties <casties@mpiwg-berlin.mpg.de>
parents: 82
diff changeset
510 * @param relStats
d4b456623d43 Updated XML export. Saves relation source-type and target-type. Expanded statistics with per-entity-type relation counts.
Robert Casties <casties@mpiwg-berlin.mpg.de>
parents: 82
diff changeset
511 */
d4b456623d43 Updated XML export. Saves relation source-type and target-type. Expanded statistics with per-entity-type relation counts.
Robert Casties <casties@mpiwg-berlin.mpg.de>
parents: 82
diff changeset
512 protected static void updateRelStats(Relation rel, boolean isSrcRel, Map<String, Long> relStats) {
d4b456623d43 Updated XML export. Saves relation source-type and target-type. Expanded statistics with per-entity-type relation counts.
Robert Casties <casties@mpiwg-berlin.mpg.de>
parents: 82
diff changeset
513 String relName = rel.getObjectClass();
d4b456623d43 Updated XML export. Saves relation source-type and target-type. Expanded statistics with per-entity-type relation counts.
Robert Casties <casties@mpiwg-berlin.mpg.de>
parents: 82
diff changeset
514 if (isSrcRel) {
d4b456623d43 Updated XML export. Saves relation source-type and target-type. Expanded statistics with per-entity-type relation counts.
Robert Casties <casties@mpiwg-berlin.mpg.de>
parents: 82
diff changeset
515 relName = relName + "[" + rel.getTargetObjectClass() + "]";
d4b456623d43 Updated XML export. Saves relation source-type and target-type. Expanded statistics with per-entity-type relation counts.
Robert Casties <casties@mpiwg-berlin.mpg.de>
parents: 82
diff changeset
516 } else {
d4b456623d43 Updated XML export. Saves relation source-type and target-type. Expanded statistics with per-entity-type relation counts.
Robert Casties <casties@mpiwg-berlin.mpg.de>
parents: 82
diff changeset
517 relName = "[" + rel.getSourceObjectClass() + "]" + relName;
d4b456623d43 Updated XML export. Saves relation source-type and target-type. Expanded statistics with per-entity-type relation counts.
Robert Casties <casties@mpiwg-berlin.mpg.de>
parents: 82
diff changeset
518 }
d4b456623d43 Updated XML export. Saves relation source-type and target-type. Expanded statistics with per-entity-type relation counts.
Robert Casties <casties@mpiwg-berlin.mpg.de>
parents: 82
diff changeset
519 Long cnt = relStats.get(relName);
d4b456623d43 Updated XML export. Saves relation source-type and target-type. Expanded statistics with per-entity-type relation counts.
Robert Casties <casties@mpiwg-berlin.mpg.de>
parents: 82
diff changeset
520 if (cnt == null) {
d4b456623d43 Updated XML export. Saves relation source-type and target-type. Expanded statistics with per-entity-type relation counts.
Robert Casties <casties@mpiwg-berlin.mpg.de>
parents: 82
diff changeset
521 relStats.put(relName, 1l);
d4b456623d43 Updated XML export. Saves relation source-type and target-type. Expanded statistics with per-entity-type relation counts.
Robert Casties <casties@mpiwg-berlin.mpg.de>
parents: 82
diff changeset
522 } else {
d4b456623d43 Updated XML export. Saves relation source-type and target-type. Expanded statistics with per-entity-type relation counts.
Robert Casties <casties@mpiwg-berlin.mpg.de>
parents: 82
diff changeset
523 relStats.put(relName, cnt + 1);
d4b456623d43 Updated XML export. Saves relation source-type and target-type. Expanded statistics with per-entity-type relation counts.
Robert Casties <casties@mpiwg-berlin.mpg.de>
parents: 82
diff changeset
524 }
d4b456623d43 Updated XML export. Saves relation source-type and target-type. Expanded statistics with per-entity-type relation counts.
Robert Casties <casties@mpiwg-berlin.mpg.de>
parents: 82
diff changeset
525 }
d4b456623d43 Updated XML export. Saves relation source-type and target-type. Expanded statistics with per-entity-type relation counts.
Robert Casties <casties@mpiwg-berlin.mpg.de>
parents: 82
diff changeset
526
79
b0aebac0780a put statistics about number of entities, relations and attributes in xml dump.
casties
parents: 78
diff changeset
527
b0aebac0780a put statistics about number of entities, relations and attributes in xml dump.
casties
parents: 78
diff changeset
528
1
jurzua
parents:
diff changeset
529 }