|
| 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 static marquez.db.Columns.toPgObject; |
| 9 | + |
| 10 | +import com.fasterxml.jackson.databind.node.ObjectNode; |
| 11 | +import java.time.Instant; |
| 12 | +import java.util.Optional; |
| 13 | +import java.util.UUID; |
| 14 | +import lombok.NonNull; |
| 15 | +import marquez.common.Utils; |
| 16 | +import marquez.service.models.LineageEvent; |
| 17 | +import org.jdbi.v3.sqlobject.statement.SqlUpdate; |
| 18 | +import org.jdbi.v3.sqlobject.transaction.Transaction; |
| 19 | +import org.postgresql.util.PGobject; |
| 20 | + |
| 21 | +/** The DAO for {@code dataset} facets. */ |
| 22 | +public interface DatasetFacetsDao { |
| 23 | + /* An {@code enum} used ... */ |
| 24 | + enum Type { |
| 25 | + DATASET, |
| 26 | + INPUT, |
| 27 | + OUTPUT, |
| 28 | + UNKNOWN; |
| 29 | + } |
| 30 | + |
| 31 | + /* An {@code enum} used to determine the dataset facet. */ |
| 32 | + enum Facet { |
| 33 | + DOCUMENTATION(Type.DATASET, "documentation"), |
| 34 | + SCHEMA(Type.DATASET, "schema"), |
| 35 | + DATASOURCE(Type.DATASET, "dataSource"), |
| 36 | + DESCRIPTION(Type.DATASET, "description"), |
| 37 | + LIFECYCLE_STATE_CHANGE(Type.DATASET, "lifecycleStateChange"), |
| 38 | + VERSION(Type.DATASET, "version"), |
| 39 | + COLUMN_LINEAGE(Type.DATASET, "columnLineage"), |
| 40 | + OWNERSHIP(Type.DATASET, "ownership"), |
| 41 | + DATA_QUALITY_METRICS(Type.INPUT, "dataQualityMetrics"), |
| 42 | + DATA_QUALITY_ASSERTIONS(Type.INPUT, "dataQualityAssertions"), |
| 43 | + OUTPUT_STATISTICS(Type.OUTPUT, "outputStatistics"); |
| 44 | + |
| 45 | + // whenever adding new types, please mind migration script in |
| 46 | + // marquez.db.migrations.V55_5__BackfillFacets |
| 47 | + |
| 48 | + final Type type; |
| 49 | + final String name; |
| 50 | + |
| 51 | + Facet(@NonNull final Type type, @NonNull final String name) { |
| 52 | + this.type = type; |
| 53 | + this.name = name; |
| 54 | + } |
| 55 | + |
| 56 | + Type getType() { |
| 57 | + return type; |
| 58 | + } |
| 59 | + |
| 60 | + String getName() { |
| 61 | + return name; |
| 62 | + } |
| 63 | + |
| 64 | + /** ... */ |
| 65 | + public static Facet fromName(@NonNull final String name) { |
| 66 | + for (final Facet facet : Facet.values()) { |
| 67 | + if (facet.getName().equalsIgnoreCase(name)) { |
| 68 | + return facet; |
| 69 | + } |
| 70 | + } |
| 71 | + return null; |
| 72 | + } |
| 73 | + |
| 74 | + /** ... */ |
| 75 | + public ObjectNode asJson(@NonNull Object facetValue) { |
| 76 | + return asJson(name, facetValue); |
| 77 | + } |
| 78 | + |
| 79 | + /** ... */ |
| 80 | + public static ObjectNode asJson(@NonNull final String facetName, @NonNull Object facetValue) { |
| 81 | + final ObjectNode facetAsJson = Utils.getMapper().createObjectNode(); |
| 82 | + facetAsJson.putPOJO(facetName, facetValue); |
| 83 | + return facetAsJson; |
| 84 | + } |
| 85 | + } |
| 86 | + |
| 87 | + /** |
| 88 | + * @param uuid |
| 89 | + * @param createdAt |
| 90 | + * @param datasetUuid |
| 91 | + * @param runUuid |
| 92 | + * @param lineageEventTime |
| 93 | + * @param lineageEventType |
| 94 | + * @param type |
| 95 | + * @param name |
| 96 | + * @param facet |
| 97 | + */ |
| 98 | + @SqlUpdate( |
| 99 | + """ |
| 100 | + INSERT INTO dataset_facets ( |
| 101 | + uuid, |
| 102 | + created_at, |
| 103 | + dataset_uuid, |
| 104 | + run_uuid, |
| 105 | + lineage_event_time, |
| 106 | + lineage_event_type, |
| 107 | + type, |
| 108 | + name, |
| 109 | + facet |
| 110 | + ) VALUES ( |
| 111 | + :uuid, |
| 112 | + :createdAt, |
| 113 | + :datasetUuid, |
| 114 | + :runUuid, |
| 115 | + :lineageEventTime, |
| 116 | + :lineageEventType, |
| 117 | + :type, |
| 118 | + :name, |
| 119 | + :facet |
| 120 | + ) |
| 121 | + """) |
| 122 | + void insertDatasetFacet( |
| 123 | + UUID uuid, |
| 124 | + Instant createdAt, |
| 125 | + UUID datasetUuid, |
| 126 | + UUID runUuid, |
| 127 | + Instant lineageEventTime, |
| 128 | + String lineageEventType, |
| 129 | + Type type, |
| 130 | + String name, |
| 131 | + PGobject facet); |
| 132 | + |
| 133 | + /** |
| 134 | + * @param datasetUuid |
| 135 | + * @param runUuid |
| 136 | + * @param lineageEventTime |
| 137 | + * @param lineageEventType |
| 138 | + * @param datasetFacets |
| 139 | + */ |
| 140 | + @Transaction |
| 141 | + default void insertDatasetFacetsFor( |
| 142 | + @NonNull UUID datasetUuid, |
| 143 | + @NonNull UUID runUuid, |
| 144 | + @NonNull Instant lineageEventTime, |
| 145 | + @NonNull String lineageEventType, |
| 146 | + @NonNull LineageEvent.DatasetFacets datasetFacets) { |
| 147 | + final Instant now = Instant.now(); |
| 148 | + |
| 149 | + // Add ... |
| 150 | + Optional.ofNullable(datasetFacets.getDocumentation()) |
| 151 | + .ifPresent( |
| 152 | + documentation -> |
| 153 | + insertDatasetFacet( |
| 154 | + UUID.randomUUID(), |
| 155 | + now, |
| 156 | + datasetUuid, |
| 157 | + runUuid, |
| 158 | + lineageEventTime, |
| 159 | + lineageEventType, |
| 160 | + Facet.DOCUMENTATION.getType(), |
| 161 | + Facet.DOCUMENTATION.getName(), |
| 162 | + toPgObject(Facet.DOCUMENTATION.asJson(documentation)))); |
| 163 | + |
| 164 | + // Add ... |
| 165 | + Optional.ofNullable(datasetFacets.getSchema()) |
| 166 | + .ifPresent( |
| 167 | + schema -> |
| 168 | + insertDatasetFacet( |
| 169 | + UUID.randomUUID(), |
| 170 | + now, |
| 171 | + datasetUuid, |
| 172 | + runUuid, |
| 173 | + lineageEventTime, |
| 174 | + lineageEventType, |
| 175 | + Facet.SCHEMA.getType(), |
| 176 | + Facet.SCHEMA.getName(), |
| 177 | + toPgObject(Facet.SCHEMA.asJson(schema)))); |
| 178 | + |
| 179 | + // Add ... |
| 180 | + Optional.ofNullable(datasetFacets.getDataSource()) |
| 181 | + .ifPresent( |
| 182 | + datasource -> |
| 183 | + insertDatasetFacet( |
| 184 | + UUID.randomUUID(), |
| 185 | + now, |
| 186 | + datasetUuid, |
| 187 | + runUuid, |
| 188 | + lineageEventTime, |
| 189 | + lineageEventType, |
| 190 | + Facet.DATASOURCE.getType(), |
| 191 | + Facet.DATASOURCE.getName(), |
| 192 | + toPgObject(Facet.DATASOURCE.asJson(datasource)))); |
| 193 | + |
| 194 | + // Add ... |
| 195 | + Optional.ofNullable(datasetFacets.getDescription()) |
| 196 | + .ifPresent( |
| 197 | + description -> |
| 198 | + insertDatasetFacet( |
| 199 | + UUID.randomUUID(), |
| 200 | + now, |
| 201 | + datasetUuid, |
| 202 | + runUuid, |
| 203 | + lineageEventTime, |
| 204 | + lineageEventType, |
| 205 | + Facet.DESCRIPTION.getType(), |
| 206 | + Facet.DESCRIPTION.getName(), |
| 207 | + toPgObject(Facet.DESCRIPTION.asJson(description)))); |
| 208 | + |
| 209 | + // Add ... |
| 210 | + Optional.ofNullable(datasetFacets.getLifecycleStateChange()) |
| 211 | + .ifPresent( |
| 212 | + lifecycleStateChange -> |
| 213 | + insertDatasetFacet( |
| 214 | + UUID.randomUUID(), |
| 215 | + now, |
| 216 | + datasetUuid, |
| 217 | + runUuid, |
| 218 | + lineageEventTime, |
| 219 | + lineageEventType, |
| 220 | + Facet.LIFECYCLE_STATE_CHANGE.getType(), |
| 221 | + Facet.LIFECYCLE_STATE_CHANGE.getName(), |
| 222 | + toPgObject(Facet.LIFECYCLE_STATE_CHANGE.asJson(lifecycleStateChange)))); |
| 223 | + |
| 224 | + // Add ... |
| 225 | + Optional.ofNullable(datasetFacets.getColumnLineage()) |
| 226 | + .ifPresent( |
| 227 | + lifecycleStateChange -> |
| 228 | + insertDatasetFacet( |
| 229 | + UUID.randomUUID(), |
| 230 | + now, |
| 231 | + datasetUuid, |
| 232 | + runUuid, |
| 233 | + lineageEventTime, |
| 234 | + lineageEventType, |
| 235 | + Facet.COLUMN_LINEAGE.getType(), |
| 236 | + Facet.COLUMN_LINEAGE.getName(), |
| 237 | + toPgObject(Facet.COLUMN_LINEAGE.asJson(lifecycleStateChange)))); |
| 238 | + |
| 239 | + // Add .. |
| 240 | + Optional.ofNullable(datasetFacets.getAdditionalFacets()) |
| 241 | + .ifPresent( |
| 242 | + additional -> |
| 243 | + additional.forEach( |
| 244 | + (name, facet) -> { |
| 245 | + Optional.ofNullable(Facet.fromName(name)) |
| 246 | + .ifPresentOrElse( |
| 247 | + (x) -> { |
| 248 | + insertDatasetFacet( |
| 249 | + UUID.randomUUID(), |
| 250 | + now, |
| 251 | + datasetUuid, |
| 252 | + runUuid, |
| 253 | + lineageEventTime, |
| 254 | + lineageEventType, |
| 255 | + x.getType(), |
| 256 | + x.getName(), |
| 257 | + toPgObject(x.asJson(facet))); |
| 258 | + }, |
| 259 | + () -> { |
| 260 | + insertDatasetFacet( |
| 261 | + UUID.randomUUID(), |
| 262 | + now, |
| 263 | + datasetUuid, |
| 264 | + runUuid, |
| 265 | + lineageEventTime, |
| 266 | + lineageEventType, |
| 267 | + Type.UNKNOWN, |
| 268 | + name, |
| 269 | + toPgObject(Facet.asJson(name, facet))); |
| 270 | + }); |
| 271 | + })); |
| 272 | + } |
| 273 | + |
| 274 | + record DatasetFacetRow( |
| 275 | + UUID uuid, |
| 276 | + Instant createdAt, |
| 277 | + UUID datasetUuid, |
| 278 | + UUID runUuid, |
| 279 | + Instant lineageEventTime, |
| 280 | + String lineageEventType, |
| 281 | + DatasetFacetsDao.Type type, |
| 282 | + String name, |
| 283 | + PGobject facet) {} |
| 284 | +} |
0 commit comments