|
| 1 | +package ldbc.snb.datagen.serializer.snb.interactive.num_date; |
| 2 | + |
| 3 | +import ldbc.snb.datagen.dictionary.Dictionaries; |
| 4 | +import ldbc.snb.datagen.objects.Organization; |
| 5 | +import ldbc.snb.datagen.objects.Place; |
| 6 | +import ldbc.snb.datagen.objects.Tag; |
| 7 | +import ldbc.snb.datagen.objects.TagClass; |
| 8 | +import ldbc.snb.datagen.serializer.HDFSCSVWriter; |
| 9 | +import ldbc.snb.datagen.serializer.InvariantSerializer; |
| 10 | +import ldbc.snb.datagen.vocabulary.DBP; |
| 11 | +import ldbc.snb.datagen.vocabulary.DBPOWL; |
| 12 | +import org.apache.hadoop.conf.Configuration; |
| 13 | + |
| 14 | +import java.util.ArrayList; |
| 15 | + |
| 16 | +/** |
| 17 | + * Created by aprat on 12/17/14. |
| 18 | + */ |
| 19 | +public class CSVInvariantSerializer extends InvariantSerializer { |
| 20 | + |
| 21 | + private HDFSCSVWriter[] writers; |
| 22 | + |
| 23 | + private enum FileNames { |
| 24 | + TAG ("tag"), |
| 25 | + TAG_HAS_TYPE_TAGCLASS("tag_hasType_tagclass"), |
| 26 | + TAGCLASS ("tagclass"), |
| 27 | + TAGCLASS_IS_SUBCLASS_OF_TAGCLASS ("tagclass_isSubclassOf_tagclass"), |
| 28 | + PLACE ("place"), |
| 29 | + PLACE_IS_PART_OF_PLACE ("place_isPartOf_place"), |
| 30 | + ORGANIZATION ("organisation"), |
| 31 | + ORGANIZATION_IS_LOCATED_IN_PLACE ("organisation_isLocatedIn_place"); |
| 32 | + |
| 33 | + private final String name; |
| 34 | + |
| 35 | + private FileNames( String name ) { |
| 36 | + this.name = name; |
| 37 | + } |
| 38 | + public String toString() { |
| 39 | + return name; |
| 40 | + } |
| 41 | + } |
| 42 | + |
| 43 | + public void initialize(Configuration conf, int reducerId) { |
| 44 | + int numFiles = FileNames.values().length; |
| 45 | + writers = new HDFSCSVWriter[numFiles]; |
| 46 | + for( int i = 0; i < numFiles; ++i) { |
| 47 | + writers[i] = new HDFSCSVWriter(conf.get("ldbc.snb.datagen.serializer.socialNetworkDir"), |
| 48 | + FileNames.values()[i].toString() + "_" + reducerId,conf.getInt("ldbc.snb.datagen.numPartitions",1),conf.getBoolean("ldbc.snb.datagen.serializer.compressed",false),"|",false); |
| 49 | + } |
| 50 | + |
| 51 | + ArrayList<String> arguments = new ArrayList<String>(); |
| 52 | + arguments.add("id"); |
| 53 | + arguments.add("name"); |
| 54 | + arguments.add("url"); |
| 55 | + writers[FileNames.TAG.ordinal()].writeEntry(arguments); |
| 56 | + |
| 57 | + arguments.clear(); |
| 58 | + arguments.add("Tag.id"); |
| 59 | + arguments.add("TagClass.id"); |
| 60 | + writers[FileNames.TAG_HAS_TYPE_TAGCLASS.ordinal()].writeEntry(arguments); |
| 61 | + |
| 62 | + arguments.clear(); |
| 63 | + arguments.add("id"); |
| 64 | + arguments.add("name"); |
| 65 | + arguments.add("url"); |
| 66 | + writers[FileNames.TAGCLASS.ordinal()].writeEntry(arguments); |
| 67 | + |
| 68 | + arguments.clear(); |
| 69 | + arguments.add("TagClass.id"); |
| 70 | + arguments.add("TagClass.id"); |
| 71 | + writers[FileNames.TAGCLASS_IS_SUBCLASS_OF_TAGCLASS.ordinal()].writeEntry(arguments); |
| 72 | + |
| 73 | + arguments.clear(); |
| 74 | + arguments.add("id"); |
| 75 | + arguments.add("name"); |
| 76 | + arguments.add("url"); |
| 77 | + arguments.add("type"); |
| 78 | + writers[FileNames.PLACE.ordinal()].writeEntry(arguments); |
| 79 | + |
| 80 | + arguments.clear(); |
| 81 | + arguments.add("id"); |
| 82 | + arguments.add("type"); |
| 83 | + arguments.add("name"); |
| 84 | + arguments.add("url"); |
| 85 | + writers[FileNames.ORGANIZATION.ordinal()].writeEntry(arguments); |
| 86 | + |
| 87 | + arguments.clear(); |
| 88 | + arguments.add("Organisation.id"); |
| 89 | + arguments.add("Place.id"); |
| 90 | + writers[FileNames.ORGANIZATION_IS_LOCATED_IN_PLACE.ordinal()].writeEntry(arguments); |
| 91 | + |
| 92 | + arguments.clear(); |
| 93 | + arguments.add("Place.id"); |
| 94 | + arguments.add("Place.id"); |
| 95 | + writers[FileNames.PLACE_IS_PART_OF_PLACE.ordinal()].writeEntry(arguments); |
| 96 | + } |
| 97 | + |
| 98 | + public void close() { |
| 99 | + int numFiles = FileNames.values().length; |
| 100 | + for(int i = 0; i < numFiles; ++i) { |
| 101 | + writers[i].close(); |
| 102 | + } |
| 103 | + } |
| 104 | + |
| 105 | + protected void serialize(Place place) { |
| 106 | + ArrayList<String> arguments = new ArrayList<String>(); |
| 107 | + arguments.add(Integer.toString(place.getId())); |
| 108 | + arguments.add(place.getName()); |
| 109 | + arguments.add(DBP.getUrl(place.getName())); |
| 110 | + arguments.add(place.getType()); |
| 111 | + writers[FileNames.PLACE.ordinal()].writeEntry(arguments); |
| 112 | + |
| 113 | + if (place.getType() == Place.CITY || |
| 114 | + place.getType() == Place.COUNTRY) { |
| 115 | + arguments.clear(); |
| 116 | + arguments.add(Integer.toString(place.getId())); |
| 117 | + arguments.add(Integer.toString(Dictionaries.places.belongsTo(place.getId()))); |
| 118 | + writers[FileNames.PLACE_IS_PART_OF_PLACE.ordinal()].writeEntry(arguments); |
| 119 | + } |
| 120 | + } |
| 121 | + |
| 122 | + protected void serialize(Organization organization) { |
| 123 | + ArrayList<String> arguments = new ArrayList<String>(); |
| 124 | + arguments.add(Long.toString(organization.id)); |
| 125 | + arguments.add(organization.type.toString()); |
| 126 | + arguments.add(organization.name); |
| 127 | + arguments.add(DBP.getUrl(organization.name)); |
| 128 | + writers[FileNames.ORGANIZATION.ordinal()].writeEntry(arguments); |
| 129 | + |
| 130 | + arguments.clear(); |
| 131 | + arguments.add(Long.toString(organization.id)); |
| 132 | + arguments.add(Integer.toString(organization.location)); |
| 133 | + writers[FileNames.ORGANIZATION_IS_LOCATED_IN_PLACE.ordinal()].writeEntry(arguments); |
| 134 | + } |
| 135 | + |
| 136 | + protected void serialize(TagClass tagClass) { |
| 137 | + ArrayList<String> arguments = new ArrayList<String>(); |
| 138 | + arguments.add(Integer.toString(tagClass.id)); |
| 139 | + arguments.add(tagClass.name); |
| 140 | + if (tagClass.name.equals("Thing")) { |
| 141 | + arguments.add("http://www.w3.org/2002/07/owl#Thing"); |
| 142 | + } else { |
| 143 | + arguments.add(DBPOWL.getUrl(tagClass.name)); |
| 144 | + } |
| 145 | + writers[FileNames.TAGCLASS.ordinal()].writeEntry(arguments); |
| 146 | + |
| 147 | + if (tagClass.parent != -1) { |
| 148 | + arguments.clear(); |
| 149 | + arguments.add(Integer.toString(tagClass.id)); |
| 150 | + arguments.add(Integer.toString(tagClass.parent)); |
| 151 | + writers[FileNames.TAGCLASS_IS_SUBCLASS_OF_TAGCLASS.ordinal()].writeEntry(arguments); |
| 152 | + } |
| 153 | + } |
| 154 | + |
| 155 | + protected void serialize(Tag tag) { |
| 156 | + ArrayList<String> arguments = new ArrayList<String>(); |
| 157 | + arguments.add(Integer.toString(tag.id)); |
| 158 | + arguments.add(tag.name); |
| 159 | + arguments.add(DBP.getUrl(tag.name)); |
| 160 | + writers[FileNames.TAG.ordinal()].writeEntry(arguments); |
| 161 | + |
| 162 | + arguments.clear(); |
| 163 | + arguments.add(Integer.toString(tag.id)); |
| 164 | + arguments.add(Integer.toString(tag.tagClass)); |
| 165 | + writers[FileNames.TAG_HAS_TYPE_TAGCLASS.ordinal()].writeEntry(arguments); |
| 166 | + } |
| 167 | + public void reset() { |
| 168 | + |
| 169 | + } |
| 170 | +} |
0 commit comments