|
| 1 | +/* |
| 2 | + * Copyright 2018-2022 contributors to the Marquez project |
| 3 | + * SPDX-License-Identifier: Apache-2.0 |
| 4 | + */ |
| 5 | + |
| 6 | +package marquez.db; |
| 7 | + |
| 8 | +import com.fasterxml.jackson.databind.JsonNode; |
| 9 | +import java.time.Instant; |
| 10 | +import java.util.Arrays; |
| 11 | +import java.util.Spliterator; |
| 12 | +import java.util.Spliterators; |
| 13 | +import java.util.UUID; |
| 14 | +import java.util.stream.StreamSupport; |
| 15 | +import lombok.NonNull; |
| 16 | +import marquez.common.Utils; |
| 17 | +import marquez.service.models.LineageEvent; |
| 18 | +import org.jdbi.v3.sqlobject.statement.SqlUpdate; |
| 19 | +import org.jdbi.v3.sqlobject.transaction.Transaction; |
| 20 | +import org.postgresql.util.PGobject; |
| 21 | + |
| 22 | +/** The DAO for {@code dataset} facets. */ |
| 23 | +public interface DatasetFacetsDao { |
| 24 | + /* An {@code enum} used ... */ |
| 25 | + enum Type { |
| 26 | + DATASET, |
| 27 | + INPUT, |
| 28 | + OUTPUT, |
| 29 | + UNKNOWN; |
| 30 | + } |
| 31 | + |
| 32 | + /* An {@code enum} used to determine the dataset facet. */ |
| 33 | + enum DatasetFacet { |
| 34 | + DOCUMENTATION(Type.DATASET, "documentation"), |
| 35 | + DESCRIPTION(Type.DATASET, "description"), |
| 36 | + SCHEMA(Type.DATASET, "schema"), |
| 37 | + DATASOURCE(Type.DATASET, "dataSource"), |
| 38 | + LIFECYCLE_STATE_CHANGE(Type.DATASET, "lifecycleStateChange"), |
| 39 | + VERSION(Type.DATASET, "version"), |
| 40 | + COLUMN_LINEAGE(Type.DATASET, "columnLineage"), |
| 41 | + OWNERSHIP(Type.DATASET, "ownership"), |
| 42 | + DATA_QUALITY_METRICS(Type.INPUT, "dataQualityMetrics"), |
| 43 | + DATA_QUALITY_ASSERTIONS(Type.INPUT, "dataQualityAssertions"), |
| 44 | + OUTPUT_STATISTICS(Type.OUTPUT, "outputStatistics"); |
| 45 | + |
| 46 | + final Type type; |
| 47 | + final String name; |
| 48 | + |
| 49 | + DatasetFacet(@NonNull final Type type, @NonNull final String name) { |
| 50 | + this.type = type; |
| 51 | + this.name = name; |
| 52 | + } |
| 53 | + |
| 54 | + Type getType() { |
| 55 | + return type; |
| 56 | + } |
| 57 | + |
| 58 | + String getName() { |
| 59 | + return name; |
| 60 | + } |
| 61 | + |
| 62 | + /** ... */ |
| 63 | + public static Type typeFromName(@NonNull final String name) { |
| 64 | + return Arrays.stream(DatasetFacet.values()) |
| 65 | + .filter(facet -> facet.getName().equalsIgnoreCase(name)) |
| 66 | + .map(facet -> facet.getType()) |
| 67 | + .findFirst() |
| 68 | + .orElse(Type.UNKNOWN); |
| 69 | + } |
| 70 | + } |
| 71 | + |
| 72 | + /** |
| 73 | + * @param uuid |
| 74 | + * @param createdAt |
| 75 | + * @param datasetUuid |
| 76 | + * @param runUuid |
| 77 | + * @param lineageEventTime |
| 78 | + * @param lineageEventType |
| 79 | + * @param type |
| 80 | + * @param name |
| 81 | + * @param facet |
| 82 | + */ |
| 83 | + @SqlUpdate( |
| 84 | + """ |
| 85 | + INSERT INTO dataset_facets ( |
| 86 | + uuid, |
| 87 | + created_at, |
| 88 | + dataset_uuid, |
| 89 | + run_uuid, |
| 90 | + lineage_event_time, |
| 91 | + lineage_event_type, |
| 92 | + type, |
| 93 | + name, |
| 94 | + facet |
| 95 | + ) VALUES ( |
| 96 | + :uuid, |
| 97 | + :createdAt, |
| 98 | + :datasetUuid, |
| 99 | + :runUuid, |
| 100 | + :lineageEventTime, |
| 101 | + :lineageEventType, |
| 102 | + :type, |
| 103 | + :name, |
| 104 | + :facet |
| 105 | + ) |
| 106 | + """) |
| 107 | + void insertDatasetFacet( |
| 108 | + UUID uuid, |
| 109 | + Instant createdAt, |
| 110 | + UUID datasetUuid, |
| 111 | + UUID runUuid, |
| 112 | + Instant lineageEventTime, |
| 113 | + String lineageEventType, |
| 114 | + Type type, |
| 115 | + String name, |
| 116 | + PGobject facet); |
| 117 | + |
| 118 | + /** |
| 119 | + * @param datasetUuid |
| 120 | + * @param runUuid |
| 121 | + * @param lineageEventTime |
| 122 | + * @param lineageEventType |
| 123 | + * @param datasetFacets |
| 124 | + */ |
| 125 | + @Transaction |
| 126 | + default void insertDatasetFacetsFor( |
| 127 | + @NonNull UUID datasetUuid, |
| 128 | + @NonNull UUID runUuid, |
| 129 | + @NonNull Instant lineageEventTime, |
| 130 | + @NonNull String lineageEventType, |
| 131 | + @NonNull LineageEvent.DatasetFacets datasetFacets) { |
| 132 | + final Instant now = Instant.now(); |
| 133 | + |
| 134 | + JsonNode jsonNode = Utils.getMapper().valueToTree(datasetFacets); |
| 135 | + StreamSupport.stream( |
| 136 | + Spliterators.spliteratorUnknownSize(jsonNode.fieldNames(), Spliterator.DISTINCT), false) |
| 137 | + .forEach( |
| 138 | + fieldName -> |
| 139 | + insertDatasetFacet( |
| 140 | + UUID.randomUUID(), |
| 141 | + now, |
| 142 | + datasetUuid, |
| 143 | + runUuid, |
| 144 | + lineageEventTime, |
| 145 | + lineageEventType, |
| 146 | + DatasetFacet.typeFromName(fieldName), |
| 147 | + fieldName, |
| 148 | + FacetUtils.toPgObject(fieldName, jsonNode.get(fieldName)))); |
| 149 | + } |
| 150 | + |
| 151 | + record DatasetFacetRow( |
| 152 | + UUID uuid, |
| 153 | + Instant createdAt, |
| 154 | + UUID datasetUuid, |
| 155 | + UUID runUuid, |
| 156 | + Instant lineageEventTime, |
| 157 | + String lineageEventType, |
| 158 | + DatasetFacetsDao.Type type, |
| 159 | + String name, |
| 160 | + PGobject facet) {} |
| 161 | +} |
0 commit comments