openmind: src/main/java/org/mpi/openmind/repository/utils/OM4StreamWriter.java annotate

annotate src/main/java/org/mpi/openmind/repository/utils/OM4StreamWriter.java @ 86:d4b456623d43

Updated XML export. Saves relation source-type and target-type. Expanded statistics with per-entity-type relation counts.

author	Robert Casties <casties@mpiwg-berlin.mpg.de>
date	Mon, 05 Feb 2018 20:06:38 +0100
parents	90f9a1c45b15
children	8005f7011975

rev	line source
1 615d27dce9b3 jurzua parents: diff changeset	1 package org.mpi.openmind.repository.utils;
615d27dce9b3 jurzua parents: diff changeset	2
615d27dce9b3 jurzua parents: diff changeset	3 import java.io.FileOutputStream;
615d27dce9b3 jurzua parents: diff changeset	4 import java.io.IOException;
615d27dce9b3 jurzua parents: diff changeset	5 import java.io.OutputStreamWriter;
615d27dce9b3 jurzua parents: diff changeset	6 import java.text.DecimalFormat;
615d27dce9b3 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 615d27dce9b3 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 615d27dce9b3 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 615d27dce9b3 jurzua parents: diff changeset	15
615d27dce9b3 jurzua parents: diff changeset	16 import org.apache.commons.lang.StringUtils;
615d27dce9b3 jurzua parents: diff changeset	17 import org.apache.log4j.Logger;
615d27dce9b3 jurzua parents: diff changeset	18 import org.mpi.openmind.repository.bo.Attribute;
615d27dce9b3 jurzua parents: diff changeset	19 import org.mpi.openmind.repository.bo.Entity;
615d27dce9b3 jurzua parents: diff changeset	20 import org.mpi.openmind.repository.bo.Node;
615d27dce9b3 jurzua parents: diff changeset	21 import org.mpi.openmind.repository.bo.Relation;
615d27dce9b3 jurzua parents: diff changeset	22 import org.mpi.openmind.repository.services.PersistenceService;
615d27dce9b3 jurzua parents: diff changeset	23
31 7d8ebe8ac8a2 create reader and check script for XML dumps. casties parents: 29 diff changeset	24 /**
7d8ebe8ac8a2 create reader and check script for XML dumps. casties parents: 29 diff changeset	25 * Export all entities and relations and definitions to XML.
7d8ebe8ac8a2 create reader and check script for XML dumps. casties parents: 29 diff changeset	26 *
75 e0be7c0030f5 cleanup and better comments. casties parents: 31 diff changeset	27 * 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	28 * (i.e. definition entities and relations) in separate files.
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 * @author jurzua, casties
7d8ebe8ac8a2 create reader and check script for XML dumps. casties parents: 29 diff changeset	31 *
7d8ebe8ac8a2 create reader and check script for XML dumps. casties parents: 29 diff changeset	32 */
1 615d27dce9b3 jurzua parents: diff changeset	33 public class OM4StreamWriter {
615d27dce9b3 jurzua parents: diff changeset	34
29 5786aa6caeb3 new XML export and test script. casties parents: 1 diff changeset	35 private static Logger logger = Logger.getLogger(OM4StreamWriter.class);
5786aa6caeb3 new XML export and test script. casties parents: 1 diff changeset	36
79 b0aebac0780a put statistics about number of entities, relations and attributes in xml dump. casties parents: 78 diff changeset	37 private static final int itemsPerPage = 500;
77 a59984fd3c3f add normalized own-values to xml dump. casties parents: 75 diff changeset	38
a59984fd3c3f add normalized own-values to xml dump. casties parents: 75 diff changeset	39 /** Include normalized own-values. */
78 b32b176a8aad make normalizations configurable (static). casties parents: 77 diff changeset	40 public static boolean includeNormalizations = true;
79 b0aebac0780a put statistics about number of entities, relations and attributes in xml dump. casties parents: 78 diff changeset	41
b0aebac0780a put statistics about number of entities, relations and attributes in xml dump. casties parents: 78 diff changeset	42 /** 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	43 private static final String ENT_KEY = "<entity-count>";
29 5786aa6caeb3 new XML export and test script. casties parents: 1 diff changeset	44
5786aa6caeb3 new XML export and test script. casties parents: 1 diff changeset	45 /**
5786aa6caeb3 new XML export and test script. casties parents: 1 diff changeset	46 * Return the object's string representation or "null" if its null.
5786aa6caeb3 new XML export and test script. casties parents: 1 diff changeset	47 *
5786aa6caeb3 new XML export and test script. casties parents: 1 diff changeset	48 * @param s
5786aa6caeb3 new XML export and test script. casties parents: 1 diff changeset	49 * @return
5786aa6caeb3 new XML export and test script. casties parents: 1 diff changeset	50 */
31 7d8ebe8ac8a2 create reader and check script for XML dumps. casties parents: 29 diff changeset	51 private static String defaultString(Object s) {
29 5786aa6caeb3 new XML export and test script. casties parents: 1 diff changeset	52 if (s == null) {
5786aa6caeb3 new XML export and test script. casties parents: 1 diff changeset	53 return "null";
5786aa6caeb3 new XML export and test script. casties parents: 1 diff changeset	54 } else {
5786aa6caeb3 new XML export and test script. casties parents: 1 diff changeset	55 return s.toString();
5786aa6caeb3 new XML export and test script. casties parents: 1 diff changeset	56 }
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
5786aa6caeb3 new XML export and test script. casties parents: 1 diff changeset	59
75 e0be7c0030f5 cleanup and better comments. casties parents: 31 diff changeset	60 /**
e0be7c0030f5 cleanup and better comments. casties parents: 31 diff changeset	61 * 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	62 *
e0be7c0030f5 cleanup and better comments. casties parents: 31 diff changeset	63 * @param fileName
e0be7c0030f5 cleanup and better comments. casties parents: 31 diff changeset	64 * @param ps
e0be7c0030f5 cleanup and better comments. casties parents: 31 diff changeset	65 */
29 5786aa6caeb3 new XML export and test script. casties parents: 1 diff changeset	66 public static void backupEntities(String fileName, PersistenceService ps) {
78 b32b176a8aad make normalizations configurable (static). casties parents: 77 diff changeset	67 writeEntsAndRels(fileName, ps, Node.TYPE_ABOX, includeNormalizations);
29 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 definitions 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 backupDefinitions(String fileName, PersistenceService ps) {
78 b32b176a8aad make normalizations configurable (static). casties parents: 77 diff changeset	77 writeEntsAndRels(fileName, ps, Node.TYPE_TBOX, false);
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
5786aa6caeb3 new XML export and test script. casties parents: 1 diff changeset	80 /**
5786aa6caeb3 new XML export and test script. casties parents: 1 diff changeset	81 * 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	82 *
75 e0be7c0030f5 cleanup and better comments. casties parents: 31 diff changeset	83 * Type is either TYPE_TBOX or TYPE_ABOX.
e0be7c0030f5 cleanup and better comments. casties parents: 31 diff changeset	84 *
29 5786aa6caeb3 new XML export and test script. casties parents: 1 diff changeset	85 * @param fileName
5786aa6caeb3 new XML export and test script. casties parents: 1 diff changeset	86 * @param ps
75 e0be7c0030f5 cleanup and better comments. casties parents: 31 diff changeset	87 * @param type
29 5786aa6caeb3 new XML export and test script. casties parents: 1 diff changeset	88 */
78 b32b176a8aad make normalizations configurable (static). casties parents: 77 diff changeset	89 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	90 OutputStreamWriter out;
5786aa6caeb3 new XML export and test script. casties parents: 1 diff changeset	91 try {
79 b0aebac0780a put statistics about number of entities, relations and attributes in xml dump. casties parents: 78 diff changeset	92 // statistics collection Maps
b0aebac0780a put statistics about number of entities, relations and attributes in xml dump. casties parents: 78 diff changeset	93 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	94 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	95
b0aebac0780a put statistics about number of entities, relations and attributes in xml dump. casties parents: 78 diff changeset	96 // setup xml writer
29 5786aa6caeb3 new XML export and test script. casties parents: 1 diff changeset	97 FileOutputStream fileOut = new FileOutputStream(fileName);
5786aa6caeb3 new XML export and test script. casties parents: 1 diff changeset	98 out = new OutputStreamWriter(fileOut, "UTF-8");
5786aa6caeb3 new XML export and test script. casties parents: 1 diff changeset	99 XMLOutputFactory factory = XMLOutputFactory.newInstance();
5786aa6caeb3 new XML export and test script. casties parents: 1 diff changeset	100 XMLStreamWriter writer = factory.createXMLStreamWriter(out);
5786aa6caeb3 new XML export and test script. casties parents: 1 diff changeset	101
5786aa6caeb3 new XML export and test script. casties parents: 1 diff changeset	102 int entitiesCount = 0;
5786aa6caeb3 new XML export and test script. casties parents: 1 diff changeset	103
5786aa6caeb3 new XML export and test script. casties parents: 1 diff changeset	104 writer.writeStartDocument("UTF-8", "1.0");
5786aa6caeb3 new XML export and test script. casties parents: 1 diff changeset	105
5786aa6caeb3 new XML export and test script. casties parents: 1 diff changeset	106 if (type.equals(Node.TYPE_ABOX)) {
5786aa6caeb3 new XML export and test script. casties parents: 1 diff changeset	107 writer.writeStartElement(XMLUtil.OPENMIND_DATA);
82 90f9a1c45b15 small change to xml format. Robert Casties <casties@mpiwg-berlin.mpg.de> parents: 79 diff changeset	108 writer.writeAttribute("version", "4.6");
75 e0be7c0030f5 cleanup and better comments. casties parents: 31 diff changeset	109 // get number of content Entities
29 5786aa6caeb3 new XML export and test script. casties parents: 1 diff changeset	110 entitiesCount = ps.getEntityCount(null).intValue();
5786aa6caeb3 new XML export and test script. casties parents: 1 diff changeset	111 } else {
5786aa6caeb3 new XML export and test script. casties parents: 1 diff changeset	112 writer.writeStartElement(XMLUtil.META_DATA);
82 90f9a1c45b15 small change to xml format. Robert Casties <casties@mpiwg-berlin.mpg.de> parents: 79 diff changeset	113 writer.writeAttribute("version", "4.6");
75 e0be7c0030f5 cleanup and better comments. casties parents: 31 diff changeset	114 // get number of definition Entities
29 5786aa6caeb3 new XML export and test script. casties parents: 1 diff changeset	115 entitiesCount = ps.getEntityCount(Node.TYPE_TBOX).intValue();
5786aa6caeb3 new XML export and test script. casties parents: 1 diff changeset	116 }
5786aa6caeb3 new XML export and test script. casties parents: 1 diff changeset	117
5786aa6caeb3 new XML export and test script. casties parents: 1 diff changeset	118 int numberOfPages = entitiesCount / itemsPerPage;
5786aa6caeb3 new XML export and test script. casties parents: 1 diff changeset	119 // debug: int numberOfPages = 1;
5786aa6caeb3 new XML export and test script. casties parents: 1 diff changeset	120 int counter = 0;
5786aa6caeb3 new XML export and test script. casties parents: 1 diff changeset	121 long start = System.currentTimeMillis();
5786aa6caeb3 new XML export and test script. casties parents: 1 diff changeset	122 DecimalFormat df = new DecimalFormat("#.##");
1 615d27dce9b3 jurzua parents: diff changeset	123
75 e0be7c0030f5 cleanup and better comments. casties parents: 31 diff changeset	124 // list of Relations (filled from Entities)
29 5786aa6caeb3 new XML export and test script. casties parents: 1 diff changeset	125 List<Relation> relList = new ArrayList<Relation>();
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 * write entities
5786aa6caeb3 new XML export and test script. casties parents: 1 diff changeset	129 */
5786aa6caeb3 new XML export and test script. casties parents: 1 diff changeset	130 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	131 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	132 // iterate database by pages
29 5786aa6caeb3 new XML export and test script. casties parents: 1 diff changeset	133 for (int currentPage = 0; currentPage <= numberOfPages; currentPage++) {
5786aa6caeb3 new XML export and test script. casties parents: 1 diff changeset	134 int startRecord = currentPage * itemsPerPage;
5786aa6caeb3 new XML export and test script. casties parents: 1 diff changeset	135 List<Entity> entities;
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 if (type.equals(Node.TYPE_ABOX)) {
75 e0be7c0030f5 cleanup and better comments. casties parents: 31 diff changeset	138 // get page of content Entities
29 5786aa6caeb3 new XML export and test script. casties parents: 1 diff changeset	139 entities = ps.getEntityPage(null, startRecord, itemsPerPage);
5786aa6caeb3 new XML export and test script. casties parents: 1 diff changeset	140 } else {
75 e0be7c0030f5 cleanup and better comments. casties parents: 31 diff changeset	141 // get page of definition Entities
29 5786aa6caeb3 new XML export and test script. casties parents: 1 diff changeset	142 entities = ps.getEntityPage(Node.TYPE_TBOX, startRecord, itemsPerPage);
5786aa6caeb3 new XML export and test script. casties parents: 1 diff changeset	143 }
5786aa6caeb3 new XML export and test script. casties parents: 1 diff changeset	144
79 b0aebac0780a put statistics about number of entities, relations and attributes in xml dump. casties parents: 78 diff changeset	145 // iterate entities
29 5786aa6caeb3 new XML export and test script. casties parents: 1 diff changeset	146 for (Entity ent : entities) {
75 e0be7c0030f5 cleanup and better comments. casties parents: 31 diff changeset	147 // write entity to XML
79 b0aebac0780a put statistics about number of entities, relations and attributes in xml dump. casties parents: 78 diff changeset	148 writeEntity(ent, writer, ps, includeNorm, entStats);
29 5786aa6caeb3 new XML export and test script. casties parents: 1 diff changeset	149 // 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	150 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	151 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	152 // 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	153 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	154 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	155 // 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	156 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	157 }
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	158 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	159 // 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	160 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	161 }
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 // count entities
29 5786aa6caeb3 new XML export and test script. casties parents: 1 diff changeset	163 counter++;
5786aa6caeb3 new XML export and test script. casties parents: 1 diff changeset	164 }
5786aa6caeb3 new XML export and test script. casties parents: 1 diff changeset	165
5786aa6caeb3 new XML export and test script. casties parents: 1 diff changeset	166 long runtime = System.currentTimeMillis() - start;
5786aa6caeb3 new XML export and test script. casties parents: 1 diff changeset	167 double percent = ((double) counter / (double) entitiesCount) * 100.0;
5786aa6caeb3 new XML export and test script. casties parents: 1 diff changeset	168 logger.debug("(" + df.format(percent) + "%) \t[" + counter + "/" + entitiesCount + "]\t");
5786aa6caeb3 new XML export and test script. casties parents: 1 diff changeset	169 logger.debug("Speed[ents/s]: " + df.format((double) counter / ((double) runtime / 1000)));
5786aa6caeb3 new XML export and test script. casties parents: 1 diff changeset	170 writer.flush();
5786aa6caeb3 new XML export and test script. casties parents: 1 diff changeset	171 }
5786aa6caeb3 new XML export and test script. casties parents: 1 diff changeset	172 writer.writeEndElement();
5786aa6caeb3 new XML export and test script. casties parents: 1 diff changeset	173
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 * write relations (from list)
5786aa6caeb3 new XML export and test script. casties parents: 1 diff changeset	176 */
5786aa6caeb3 new XML export and test script. casties parents: 1 diff changeset	177 writer.writeStartElement(XMLUtil.RELATIONS);
82 90f9a1c45b15 small change to xml format. Robert Casties <casties@mpiwg-berlin.mpg.de> parents: 79 diff changeset	178 writer.writeAttribute("count", Integer.toString(relList.size()));
29 5786aa6caeb3 new XML export and test script. casties parents: 1 diff changeset	179 for (Relation rel : relList) {
79 b0aebac0780a put statistics about number of entities, relations and attributes in xml dump. casties parents: 78 diff changeset	180 writeRelation(rel, writer, includeNorm, relStats);
29 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
79 b0aebac0780a put statistics about number of entities, relations and attributes in xml dump. casties parents: 78 diff changeset	184 /*
b0aebac0780a put statistics about number of entities, relations and attributes in xml dump. casties parents: 78 diff changeset	185 * write statistics
b0aebac0780a put statistics about number of entities, relations and attributes in xml dump. casties parents: 78 diff changeset	186 */
b0aebac0780a put statistics about number of entities, relations and attributes in xml dump. casties parents: 78 diff changeset	187 // entity stats
b0aebac0780a put statistics about number of entities, relations and attributes in xml dump. casties parents: 78 diff changeset	188 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	189 // relation stats
b0aebac0780a put statistics about number of entities, relations and attributes in xml dump. casties parents: 78 diff changeset	190 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	191
29 5786aa6caeb3 new XML export and test script. casties parents: 1 diff changeset	192 // end file.
5786aa6caeb3 new XML export and test script. casties parents: 1 diff changeset	193 writer.writeEndElement();
5786aa6caeb3 new XML export and test script. casties parents: 1 diff changeset	194
5786aa6caeb3 new XML export and test script. casties parents: 1 diff changeset	195 writer.flush();
5786aa6caeb3 new XML export and test script. casties parents: 1 diff changeset	196 writer.close();
5786aa6caeb3 new XML export and test script. casties parents: 1 diff changeset	197
5786aa6caeb3 new XML export and test script. casties parents: 1 diff changeset	198 logger.info("END Stream Writer");
5786aa6caeb3 new XML export and test script. casties parents: 1 diff changeset	199 } catch (IOException e) {
75 e0be7c0030f5 cleanup and better comments. casties parents: 31 diff changeset	200 logger.error(e);
29 5786aa6caeb3 new XML export and test script. casties parents: 1 diff changeset	201 } catch (XMLStreamException e) {
75 e0be7c0030f5 cleanup and better comments. casties parents: 31 diff changeset	202 logger.error(e);
29 5786aa6caeb3 new XML export and test script. casties parents: 1 diff changeset	203 }
5786aa6caeb3 new XML export and test script. casties parents: 1 diff changeset	204 }
1 615d27dce9b3 jurzua parents: diff changeset	205
29 5786aa6caeb3 new XML export and test script. casties parents: 1 diff changeset	206 /**
5786aa6caeb3 new XML export and test script. casties parents: 1 diff changeset	207 * Write OpenMind relation to XML.
5786aa6caeb3 new XML export and test script. casties parents: 1 diff changeset	208 *
5786aa6caeb3 new XML export and test script. casties parents: 1 diff changeset	209 * @param rel
5786aa6caeb3 new XML export and test script. casties parents: 1 diff changeset	210 * @param writer
79 b0aebac0780a put statistics about number of entities, relations and attributes in xml dump. casties parents: 78 diff changeset	211 * @param relStats
29 5786aa6caeb3 new XML export and test script. casties parents: 1 diff changeset	212 * @throws XMLStreamException
5786aa6caeb3 new XML export and test script. casties parents: 1 diff changeset	213 */
79 b0aebac0780a put statistics about number of entities, relations and attributes in xml dump. casties parents: 78 diff changeset	214 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	215 Map<String, Map<String, Long>> relStats) throws XMLStreamException {
29 5786aa6caeb3 new XML export and test script. casties parents: 1 diff changeset	216 writer.writeStartElement(XMLUtil.RELATION);
5786aa6caeb3 new XML export and test script. casties parents: 1 diff changeset	217
79 b0aebac0780a put statistics about number of entities, relations and attributes in xml dump. casties parents: 78 diff changeset	218 // update stats
b0aebac0780a put statistics about number of entities, relations and attributes in xml dump. casties parents: 78 diff changeset	219 Map<String, Long> attStats = updateNodeStats(rel, relStats);
b0aebac0780a put statistics about number of entities, relations and attributes in xml dump. casties parents: 78 diff changeset	220
29 5786aa6caeb3 new XML export and test script. casties parents: 1 diff changeset	221 /*
5786aa6caeb3 new XML export and test script. casties parents: 1 diff changeset	222 * write XML-attributes
5786aa6caeb3 new XML export and test script. casties parents: 1 diff changeset	223 */
5786aa6caeb3 new XML export and test script. casties parents: 1 diff changeset	224 writer.writeAttribute(XMLUtil.OBJECT_CLASS, defaultString(rel.getObjectClass()));
5786aa6caeb3 new XML export and test script. casties parents: 1 diff changeset	225 writer.writeAttribute(XMLUtil.ID, defaultString(rel.getId()));
5786aa6caeb3 new XML export and test script. casties parents: 1 diff changeset	226 writer.writeAttribute(XMLUtil.ROW_ID, defaultString(rel.getRowId()));
75 e0be7c0030f5 cleanup and better comments. casties parents: 31 diff changeset	227 if (StringUtils.isNotEmpty(rel.getContentType())) {
29 5786aa6caeb3 new XML export and test script. casties parents: 1 diff changeset	228 writer.writeAttribute(XMLUtil.CONTENT_TYPE, rel.getContentType());
75 e0be7c0030f5 cleanup and better comments. casties parents: 31 diff changeset	229 }
29 5786aa6caeb3 new XML export and test script. casties parents: 1 diff changeset	230 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	231 writer.writeAttribute(XMLUtil.RELATION_SOURCE, defaultString(rel.getSourceObjectClass()));
29 5786aa6caeb3 new XML export and test script. casties parents: 1 diff changeset	232 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	233 writer.writeAttribute(XMLUtil.RELATION_TARGET, defaultString(rel.getTargetObjectClass()));
29 5786aa6caeb3 new XML export and test script. casties parents: 1 diff changeset	234 writer.writeAttribute(XMLUtil.VERSION, defaultString(rel.getVersion()));
5786aa6caeb3 new XML export and test script. casties parents: 1 diff changeset	235 writer.writeAttribute(XMLUtil.MODIFICATION_TIME, defaultString(rel.getModificationTime()));
75 e0be7c0030f5 cleanup and better comments. casties parents: 31 diff changeset	236 if (rel.getUser() != null) {
29 5786aa6caeb3 new XML export and test script. casties parents: 1 diff changeset	237 writer.writeAttribute(XMLUtil.USER, rel.getUser());
75 e0be7c0030f5 cleanup and better comments. casties parents: 31 diff changeset	238 }
e0be7c0030f5 cleanup and better comments. casties parents: 31 diff changeset	239 if (rel.getIsPublic()) {
29 5786aa6caeb3 new XML export and test script. casties parents: 1 diff changeset	240 writer.writeAttribute(XMLUtil.PUBLIC, "true");
75 e0be7c0030f5 cleanup and better comments. casties parents: 31 diff changeset	241 }
29 5786aa6caeb3 new XML export and test script. casties parents: 1 diff changeset	242
5786aa6caeb3 new XML export and test script. casties parents: 1 diff changeset	243 /*
5786aa6caeb3 new XML export and test script. casties parents: 1 diff changeset	244 * write OpenMind attributes of this relation as XML tags
5786aa6caeb3 new XML export and test script. casties parents: 1 diff changeset	245 */
5786aa6caeb3 new XML export and test script. casties parents: 1 diff changeset	246 if (rel.getAttributes().size() > 0) {
5786aa6caeb3 new XML export and test script. casties parents: 1 diff changeset	247 writer.writeStartElement(XMLUtil.ATTRIBUTES);
79 b0aebac0780a put statistics about number of entities, relations and attributes in xml dump. casties parents: 78 diff changeset	248 for (Attribute att : rel.getAttributes()) {
b0aebac0780a put statistics about number of entities, relations and attributes in xml dump. casties parents: 78 diff changeset	249 // update stats
b0aebac0780a put statistics about number of entities, relations and attributes in xml dump. casties parents: 78 diff changeset	250 updateAttStats(att, attStats);
b0aebac0780a put statistics about number of entities, relations and attributes in xml dump. casties parents: 78 diff changeset	251 // write xml
b0aebac0780a put statistics about number of entities, relations and attributes in xml dump. casties parents: 78 diff changeset	252 writeAttribute(att, writer, includeNorm);
b0aebac0780a put statistics about number of entities, relations and attributes in xml dump. casties parents: 78 diff changeset	253 }
29 5786aa6caeb3 new XML export and test script. casties parents: 1 diff changeset	254 writer.writeEndElement();
5786aa6caeb3 new XML export and test script. casties parents: 1 diff changeset	255 }
1 615d27dce9b3 jurzua parents: diff changeset	256
29 5786aa6caeb3 new XML export and test script. casties parents: 1 diff changeset	257 /*
5786aa6caeb3 new XML export and test script. casties parents: 1 diff changeset	258 * write own value as content
5786aa6caeb3 new XML export and test script. casties parents: 1 diff changeset	259 */
75 e0be7c0030f5 cleanup and better comments. casties parents: 31 diff changeset	260 if (StringUtils.isNotEmpty(rel.getOwnValue())) {
29 5786aa6caeb3 new XML export and test script. casties parents: 1 diff changeset	261 writer.writeCharacters(rel.getOwnValue());
75 e0be7c0030f5 cleanup and better comments. casties parents: 31 diff changeset	262 }
29 5786aa6caeb3 new XML export and test script. casties parents: 1 diff changeset	263
5786aa6caeb3 new XML export and test script. casties parents: 1 diff changeset	264 writer.writeEndElement();
5786aa6caeb3 new XML export and test script. casties parents: 1 diff changeset	265 }
5786aa6caeb3 new XML export and test script. casties parents: 1 diff changeset	266
5786aa6caeb3 new XML export and test script. casties parents: 1 diff changeset	267 /**
5786aa6caeb3 new XML export and test script. casties parents: 1 diff changeset	268 * Write OpenMind entity to XML.
5786aa6caeb3 new XML export and test script. casties parents: 1 diff changeset	269 *
5786aa6caeb3 new XML export and test script. casties parents: 1 diff changeset	270 * @param entity
5786aa6caeb3 new XML export and test script. casties parents: 1 diff changeset	271 * @param writer
5786aa6caeb3 new XML export and test script. casties parents: 1 diff changeset	272 * @param ps
79 b0aebac0780a put statistics about number of entities, relations and attributes in xml dump. casties parents: 78 diff changeset	273 * @param entStats
29 5786aa6caeb3 new XML export and test script. casties parents: 1 diff changeset	274 * @throws XMLStreamException
5786aa6caeb3 new XML export and test script. casties parents: 1 diff changeset	275 */
79 b0aebac0780a put statistics about number of entities, relations and attributes in xml dump. casties parents: 78 diff changeset	276 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	277 Map<String, Map<String, Long>> entStats)
29 5786aa6caeb3 new XML export and test script. casties parents: 1 diff changeset	278 throws XMLStreamException {
1 615d27dce9b3 jurzua parents: diff changeset	279
29 5786aa6caeb3 new XML export and test script. casties parents: 1 diff changeset	280 writer.writeStartElement((entity.getType().equals(Node.TYPE_TBOX)) ? XMLUtil.DEFINITION : XMLUtil.ENTITY);
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 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	283 // make sure we have all attributes and relations
29 5786aa6caeb3 new XML export and test script. casties parents: 1 diff changeset	284 entity = ps.getEntityContent(entity);
5786aa6caeb3 new XML export and test script. casties parents: 1 diff changeset	285 }
79 b0aebac0780a put statistics about number of entities, relations and attributes in xml dump. casties parents: 78 diff changeset	286
b0aebac0780a put statistics about number of entities, relations and attributes in xml dump. casties parents: 78 diff changeset	287 // update stats
b0aebac0780a put statistics about number of entities, relations and attributes in xml dump. casties parents: 78 diff changeset	288 Map<String, Long> attStats = updateNodeStats(entity, entStats);
29 5786aa6caeb3 new XML export and test script. casties parents: 1 diff changeset	289
5786aa6caeb3 new XML export and test script. casties parents: 1 diff changeset	290 /*
5786aa6caeb3 new XML export and test script. casties parents: 1 diff changeset	291 * write XML attributes
5786aa6caeb3 new XML export and test script. casties parents: 1 diff changeset	292 */
79 b0aebac0780a put statistics about number of entities, relations and attributes in xml dump. casties parents: 78 diff changeset	293 writer.writeAttribute(XMLUtil.OBJECT_CLASS, defaultString(entity.getObjectClass()));
29 5786aa6caeb3 new XML export and test script. casties parents: 1 diff changeset	294 writer.writeAttribute(XMLUtil.ID, defaultString(entity.getId()));
5786aa6caeb3 new XML export and test script. casties parents: 1 diff changeset	295 writer.writeAttribute(XMLUtil.ROW_ID, defaultString(entity.getRowId()));
75 e0be7c0030f5 cleanup and better comments. casties parents: 31 diff changeset	296 if (StringUtils.isNotEmpty(entity.getContentType())) {
29 5786aa6caeb3 new XML export and test script. casties parents: 1 diff changeset	297 writer.writeAttribute(XMLUtil.CONTENT_TYPE, entity.getContentType());
75 e0be7c0030f5 cleanup and better comments. casties parents: 31 diff changeset	298 }
29 5786aa6caeb3 new XML export and test script. casties parents: 1 diff changeset	299 writer.writeAttribute(XMLUtil.VERSION, defaultString(entity.getVersion()));
5786aa6caeb3 new XML export and test script. casties parents: 1 diff changeset	300 writer.writeAttribute(XMLUtil.MODIFICATION_TIME, defaultString(entity.getModificationTime()));
75 e0be7c0030f5 cleanup and better comments. casties parents: 31 diff changeset	301 if (entity.getUser() != null) {
29 5786aa6caeb3 new XML export and test script. casties parents: 1 diff changeset	302 writer.writeAttribute(XMLUtil.USER, entity.getUser());
75 e0be7c0030f5 cleanup and better comments. casties parents: 31 diff changeset	303 }
e0be7c0030f5 cleanup and better comments. casties parents: 31 diff changeset	304 if (entity.getIsPublic()) {
29 5786aa6caeb3 new XML export and test script. casties parents: 1 diff changeset	305 writer.writeAttribute(XMLUtil.PUBLIC, "true");
75 e0be7c0030f5 cleanup and better comments. casties parents: 31 diff changeset	306 }
1 615d27dce9b3 jurzua parents: diff changeset	307
29 5786aa6caeb3 new XML export and test script. casties parents: 1 diff changeset	308 /*
5786aa6caeb3 new XML export and test script. casties parents: 1 diff changeset	309 * write OpenMind attributes of this entity as XML tags
5786aa6caeb3 new XML export and test script. casties parents: 1 diff changeset	310 */
5786aa6caeb3 new XML export and test script. casties parents: 1 diff changeset	311 if (entity.getAttributes().size() > 0) {
5786aa6caeb3 new XML export and test script. casties parents: 1 diff changeset	312 writer.writeStartElement(XMLUtil.ATTRIBUTES);
5786aa6caeb3 new XML export and test script. casties parents: 1 diff changeset	313 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	314 // 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	315 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	316 // write xml
78 b32b176a8aad make normalizations configurable (static). casties parents: 77 diff changeset	317 writeAttribute(att, writer, includeNorm);
29 5786aa6caeb3 new XML export and test script. casties parents: 1 diff changeset	318 }
5786aa6caeb3 new XML export and test script. casties parents: 1 diff changeset	319 writer.writeEndElement();
5786aa6caeb3 new XML export and test script. casties parents: 1 diff changeset	320 }
5786aa6caeb3 new XML export and test script. casties parents: 1 diff changeset	321
5786aa6caeb3 new XML export and test script. casties parents: 1 diff changeset	322 /*
5786aa6caeb3 new XML export and test script. casties parents: 1 diff changeset	323 * write own value
5786aa6caeb3 new XML export and test script. casties parents: 1 diff changeset	324 */
77 a59984fd3c3f add normalized own-values to xml dump. casties parents: 75 diff changeset	325 String ov = entity.getOwnValue();
a59984fd3c3f add normalized own-values to xml dump. casties parents: 75 diff changeset	326 if (StringUtils.isNotEmpty(ov)) {
a59984fd3c3f add normalized own-values to xml dump. casties parents: 75 diff changeset	327 writer.writeCharacters(ov);
a59984fd3c3f add normalized own-values to xml dump. casties parents: 75 diff changeset	328 String nov = entity.getNormalizedOwnValue();
a59984fd3c3f add normalized own-values to xml dump. casties parents: 75 diff changeset	329 if (includeNorm && StringUtils.isNotEmpty(nov) && !ov.equals(nov)) {
a59984fd3c3f add normalized own-values to xml dump. casties parents: 75 diff changeset	330 // write normalized value
a59984fd3c3f add normalized own-values to xml dump. casties parents: 75 diff changeset	331 writer.writeStartElement(XMLUtil.NORMALIZED);
a59984fd3c3f add normalized own-values to xml dump. casties parents: 75 diff changeset	332 writer.writeCharacters(nov);
a59984fd3c3f add normalized own-values to xml dump. casties parents: 75 diff changeset	333 writer.writeEndElement();
a59984fd3c3f add normalized own-values to xml dump. casties parents: 75 diff changeset	334 }
75 e0be7c0030f5 cleanup and better comments. casties parents: 31 diff changeset	335 }
29 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 writer.writeEndElement();
5786aa6caeb3 new XML export and test script. casties parents: 1 diff changeset	338 }
5786aa6caeb3 new XML export and test script. casties parents: 1 diff changeset	339
79 b0aebac0780a put statistics about number of entities, relations and attributes in xml dump. casties parents: 78 diff changeset	340
b0aebac0780a put statistics about number of entities, relations and attributes in xml dump. casties parents: 78 diff changeset	341 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	342 writer.writeStartElement(XMLUtil.ATTRIBUTE);
5786aa6caeb3 new XML export and test script. casties parents: 1 diff changeset	343
5786aa6caeb3 new XML export and test script. casties parents: 1 diff changeset	344 /*
5786aa6caeb3 new XML export and test script. casties parents: 1 diff changeset	345 * write XML attributes
5786aa6caeb3 new XML export and test script. casties parents: 1 diff changeset	346 */
5786aa6caeb3 new XML export and test script. casties parents: 1 diff changeset	347 writer.writeAttribute(XMLUtil.ATTRIBUTE_NAME, defaultString(att.getName()));
5786aa6caeb3 new XML export and test script. casties parents: 1 diff changeset	348 writer.writeAttribute(XMLUtil.ID, defaultString(att.getId()));
5786aa6caeb3 new XML export and test script. casties parents: 1 diff changeset	349 writer.writeAttribute(XMLUtil.ROW_ID, defaultString(att.getRowId()));
5786aa6caeb3 new XML export and test script. casties parents: 1 diff changeset	350 writer.writeAttribute(XMLUtil.CONTENT_TYPE, defaultString(att.getContentType()));
5786aa6caeb3 new XML export and test script. casties parents: 1 diff changeset	351 writer.writeAttribute(XMLUtil.VERSION, defaultString(att.getVersion()));
5786aa6caeb3 new XML export and test script. casties parents: 1 diff changeset	352 writer.writeAttribute(XMLUtil.MODIFICATION_TIME, defaultString(att.getModificationTime()));
75 e0be7c0030f5 cleanup and better comments. casties parents: 31 diff changeset	353 if (att.getUser() != null) {
29 5786aa6caeb3 new XML export and test script. casties parents: 1 diff changeset	354 writer.writeAttribute(XMLUtil.USER, att.getUser());
75 e0be7c0030f5 cleanup and better comments. casties parents: 31 diff changeset	355 }
e0be7c0030f5 cleanup and better comments. casties parents: 31 diff changeset	356 if (att.getIsPublic()) {
29 5786aa6caeb3 new XML export and test script. casties parents: 1 diff changeset	357 writer.writeAttribute(XMLUtil.PUBLIC, "true");
75 e0be7c0030f5 cleanup and better comments. casties parents: 31 diff changeset	358 }
e0be7c0030f5 cleanup and better comments. casties parents: 31 diff changeset	359
29 5786aa6caeb3 new XML export and test script. casties parents: 1 diff changeset	360 /*
5786aa6caeb3 new XML export and test script. casties parents: 1 diff changeset	361 * write value as content
5786aa6caeb3 new XML export and test script. casties parents: 1 diff changeset	362 */
77 a59984fd3c3f add normalized own-values to xml dump. casties parents: 75 diff changeset	363 String ov = att.getValue();
a59984fd3c3f add normalized own-values to xml dump. casties parents: 75 diff changeset	364 if (StringUtils.isNotEmpty(ov)) {
a59984fd3c3f add normalized own-values to xml dump. casties parents: 75 diff changeset	365 writer.writeCharacters(ov);
a59984fd3c3f add normalized own-values to xml dump. casties parents: 75 diff changeset	366 String nov = att.getNormalizedOwnValue();
a59984fd3c3f add normalized own-values to xml dump. casties parents: 75 diff changeset	367 if (includeNorm && StringUtils.isNotEmpty(nov) && !ov.equals(nov)) {
a59984fd3c3f add normalized own-values to xml dump. casties parents: 75 diff changeset	368 // write normalized value
a59984fd3c3f add normalized own-values to xml dump. casties parents: 75 diff changeset	369 writer.writeStartElement(XMLUtil.NORMALIZED);
a59984fd3c3f add normalized own-values to xml dump. casties parents: 75 diff changeset	370 writer.writeCharacters(nov);
a59984fd3c3f add normalized own-values to xml dump. casties parents: 75 diff changeset	371 writer.writeEndElement();
a59984fd3c3f add normalized own-values to xml dump. casties parents: 75 diff changeset	372 }
75 e0be7c0030f5 cleanup and better comments. casties parents: 31 diff changeset	373 }
1 615d27dce9b3 jurzua parents: diff changeset	374
29 5786aa6caeb3 new XML export and test script. casties parents: 1 diff changeset	375 writer.writeEndElement();
5786aa6caeb3 new XML export and test script. casties parents: 1 diff changeset	376 }
79 b0aebac0780a put statistics about number of entities, relations and attributes in xml dump. casties parents: 78 diff changeset	377
b0aebac0780a put statistics about number of entities, relations and attributes in xml dump. casties parents: 78 diff changeset	378
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	379 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	380 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	381 // write stats tag
79 b0aebac0780a put statistics about number of entities, relations and attributes in xml dump. casties parents: 78 diff changeset	382 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	383
79 b0aebac0780a put statistics about number of entities, relations and attributes in xml dump. casties parents: 78 diff changeset	384 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	385 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	386 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	387 // 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	388 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	389 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	390 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	391
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	392 // 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	393 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	394 // 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	395 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	396 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	397 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	398 // 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	399 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	400 } 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	401 // 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	402 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	403 }
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	404 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	405 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	406 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	407 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	408 }
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	409 // 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	410 writer.writeEndElement();
79 b0aebac0780a put statistics about number of entities, relations and attributes in xml dump. casties parents: 78 diff changeset	411 }
b0aebac0780a put statistics about number of entities, relations and attributes in xml dump. casties parents: 78 diff changeset	412 // end of stats tag
b0aebac0780a put statistics about number of entities, relations and attributes in xml dump. casties parents: 78 diff changeset	413 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	414 }
79 b0aebac0780a put statistics about number of entities, relations and attributes in xml dump. casties parents: 78 diff changeset	415
b0aebac0780a put statistics about number of entities, relations and attributes in xml dump. casties parents: 78 diff changeset	416 /**
b0aebac0780a put statistics about number of entities, relations and attributes in xml dump. casties parents: 78 diff changeset	417 * @param objectClass
b0aebac0780a put statistics about number of entities, relations and attributes in xml dump. casties parents: 78 diff changeset	418 * @param entStats
b0aebac0780a put statistics about number of entities, relations and attributes in xml dump. casties parents: 78 diff changeset	419 * @return
b0aebac0780a put statistics about number of entities, relations and attributes in xml dump. casties parents: 78 diff changeset	420 */
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	421 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	422 String objectClass = ent.getObjectClass();
79 b0aebac0780a put statistics about number of entities, relations and attributes in xml dump. casties parents: 78 diff changeset	423 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	424 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	425 // 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	426 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	427 // 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	428 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	429 // 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	430 entStats.put(objectClass, attStats);
79 b0aebac0780a put statistics about number of entities, relations and attributes in xml dump. casties parents: 78 diff changeset	431 } 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	432 // 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	433 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	434 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	435 }
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	436 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	437 }
79 b0aebac0780a put statistics about number of entities, relations and attributes in xml dump. casties parents: 78 diff changeset	438
b0aebac0780a put statistics about number of entities, relations and attributes in xml dump. casties parents: 78 diff changeset	439 /**
b0aebac0780a put statistics about number of entities, relations and attributes in xml dump. casties parents: 78 diff changeset	440 * @param att
b0aebac0780a put statistics about number of entities, relations and attributes in xml dump. casties parents: 78 diff changeset	441 * @param attStats
b0aebac0780a put statistics about number of entities, relations and attributes in xml dump. casties parents: 78 diff changeset	442 */
b0aebac0780a put statistics about number of entities, relations and attributes in xml dump. casties parents: 78 diff changeset	443 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	444 String attName = att.getName();
b0aebac0780a put statistics about number of entities, relations and attributes in xml dump. casties parents: 78 diff changeset	445 Long cnt = attStats.get(attName);
b0aebac0780a put statistics about number of entities, relations and attributes in xml dump. casties parents: 78 diff changeset	446 if (cnt == null) {
b0aebac0780a put statistics about number of entities, relations and attributes in xml dump. casties parents: 78 diff changeset	447 attStats.put(attName, 1l);
b0aebac0780a put statistics about number of entities, relations and attributes in xml dump. casties parents: 78 diff changeset	448 } else {
b0aebac0780a put statistics about number of entities, relations and attributes in xml dump. casties parents: 78 diff changeset	449 attStats.put(attName, cnt + 1);
b0aebac0780a put statistics about number of entities, relations and attributes in xml dump. casties parents: 78 diff changeset	450 }
b0aebac0780a put statistics about number of entities, relations and attributes in xml dump. casties parents: 78 diff changeset	451 }
b0aebac0780a put statistics about number of entities, relations and attributes in xml dump. casties parents: 78 diff changeset	452
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	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 * 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	455 *
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 * 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	457 * 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	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 * @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	460 * @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	461 */
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	462 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	463 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	464 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	465 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	466 } 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	467 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	468 }
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	469 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	470 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	471 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	472 } 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	473 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	474 }
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 }
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
79 b0aebac0780a put statistics about number of entities, relations and attributes in xml dump. casties parents: 78 diff changeset	477
b0aebac0780a put statistics about number of entities, relations and attributes in xml dump. casties parents: 78 diff changeset	478
1 615d27dce9b3 jurzua parents: diff changeset	479 }

Mercurial > hg > openmind

annotate src/main/java/org/mpi/openmind/repository/utils/OM4StreamWriter.java @ 86:d4b456623d43