| 1 | import * as s3tables from "@distilled.cloud/aws/s3tables"; |
| 2 | import * as Effect from "effect/Effect"; |
| 3 | import * as Schedule from "effect/Schedule"; |
| 4 | import { isResolved } from "../../Diff.ts"; |
| 5 | import { createPhysicalName } from "../../PhysicalName.ts"; |
| 6 | import * as Provider from "../../Provider.ts"; |
| 7 | import { Resource } from "../../Resource.ts"; |
| 8 | import type { Providers } from "../Providers.ts"; |
| 9 | import type { TableBucketArn } from "./TableBucket.ts"; |
| 10 | |
| 11 | /** |
| 12 | * A field in an Apache Iceberg table schema. |
| 13 | */ |
| 14 | export interface IcebergSchemaField { |
| 15 | /** |
| 16 | * Name of the field. |
| 17 | */ |
| 18 | name: string; |
| 19 | /** |
| 20 | * Iceberg field type, e.g. `int`, `long`, `string`, `boolean`, |
| 21 | * `timestamp`, `date`, `decimal(10,2)`. |
| 22 | */ |
| 23 | type: string; |
| 24 | /** |
| 25 | * Whether the field is required (non-nullable). |
| 26 | * @default false |
| 27 | */ |
| 28 | required?: boolean; |
| 29 | } |
| 30 | |
| 31 | export interface TableProps { |
| 32 | /** |
| 33 | * ARN of the table bucket that owns the table. Changing it replaces the |
| 34 | * table. |
| 35 | */ |
| 36 | tableBucket: TableBucketArn | string; |
| 37 | /** |
| 38 | * Name of the namespace that owns the table. Changing it replaces the |
| 39 | * table. |
| 40 | */ |
| 41 | namespace: string; |
| 42 | /** |
| 43 | * Name of the table. Must be 1-255 characters of lowercase letters, |
| 44 | * numbers, and underscores, beginning with a letter or number. Changing |
| 45 | * the name replaces the table. |
| 46 | * @default a deterministic name derived from the app, stage, and logical ID |
| 47 | */ |
| 48 | name?: string; |
| 49 | /** |
| 50 | * Open table format. Only `ICEBERG` is supported. |
| 51 | * @default "ICEBERG" |
| 52 | */ |
| 53 | format?: "ICEBERG"; |
| 54 | /** |
| 55 | * Iceberg schema for the table, applied at create time. Schema evolution |
| 56 | * after creation is a data-plane concern; changing this replaces the |
| 57 | * table. |
| 58 | */ |
| 59 | schema?: { |
| 60 | /** |
| 61 | * Ordered list of schema fields. |
| 62 | */ |
| 63 | fields: IcebergSchemaField[]; |
| 64 | }; |
| 65 | } |
| 66 | |
| 67 | export interface Table extends Resource< |
| 68 | "AWS.S3Tables.Table", |
| 69 | TableProps, |
| 70 | { |
| 71 | tableArn: string; |
| 72 | name: string; |
| 73 | namespace: string; |
| 74 | tableBucketArn: string; |
| 75 | versionToken: string; |
| 76 | metadataLocation: string | undefined; |
| 77 | warehouseLocation: string; |
| 78 | format: string; |
| 79 | type: string; |
| 80 | createdAt: Date; |
| 81 | }, |
| 82 | never, |
| 83 | Providers |
| 84 | > {} |
| 85 | |
| 86 | /** |
| 87 | * A fully-managed Apache Iceberg table within an S3 Tables {@link Namespace}. |
| 88 | * |
| 89 | * S3 Tables manages the table's storage, metadata, and maintenance |
| 90 | * (compaction, snapshot expiration). Query it through engines like Amazon |
| 91 | * Athena, Amazon EMR, or Apache Spark via the S3 Tables Iceberg catalog. |
| 92 | * ### Creating Tables |
| 93 | * **Example:** Table with an Iceberg schema |
| 94 | * ```typescript |
| 95 | * import * as S3Tables from "alchemy/AWS/S3Tables"; |
| 96 | * |
| 97 | * const bucket = yield* S3Tables.TableBucket("Analytics"); |
| 98 | * const ns = yield* S3Tables.Namespace("Events", { |
| 99 | * tableBucket: bucket.tableBucketArn, |
| 100 | * }); |
| 101 | * const table = yield* S3Tables.Table("PageViews", { |
| 102 | * tableBucket: bucket.tableBucketArn, |
| 103 | * namespace: ns.namespace, |
| 104 | * schema: { |
| 105 | * fields: [ |
| 106 | * { name: "id", type: "long", required: true }, |
| 107 | * { name: "url", type: "string" }, |
| 108 | * { name: "ts", type: "timestamp" }, |
| 109 | * ], |
| 110 | * }, |
| 111 | * }); |
| 112 | * ``` |
| 113 | * |
| 114 | * @resource |
| 115 | */ |
| 116 | export const Table = Resource<Table>("AWS.S3Tables.Table"); |
| 117 | |
| 118 | const createTableName = (id: string, props: { name?: string | undefined }) => |
| 119 | Effect.gen(function* () { |
| 120 | if (props.name) { |
| 121 | return props.name; |
| 122 | } |
| 123 | // Table names allow lowercase letters, numbers, and underscores only. |
| 124 | const base = yield* createPhysicalName({ |
| 125 | id, |
| 126 | maxLength: 60, |
| 127 | lowercase: true, |
| 128 | }); |
| 129 | return base.replaceAll("-", "_"); |
| 130 | }); |
| 131 | |
| 132 | const buildMetadata = ( |
| 133 | props: TableProps, |
| 134 | ): s3tables.TableMetadata | undefined => |
| 135 | props.schema |
| 136 | ? { |
| 137 | iceberg: { |
| 138 | schema: { |
| 139 | fields: props.schema.fields.map((f) => ({ |
| 140 | name: f.name, |
| 141 | type: f.type, |
| 142 | required: f.required, |
| 143 | })), |
| 144 | }, |
| 145 | }, |
| 146 | } |
| 147 | : undefined; |
| 148 | |
| 149 | export const TableProvider = () => |
| 150 | Provider.succeed(Table, { |
| 151 | stables: ["tableArn", "name", "namespace", "tableBucketArn"], |
| 152 | // Tables are scoped to a parent namespace; the engine drives lifecycle |
| 153 | // from state rather than ambient enumeration. |
| 154 | list: () => Effect.succeed([]), |
| 155 | read: Effect.fn(function* ({ id, olds, output }) { |
| 156 | const tableBucketArn = output?.tableBucketArn ?? olds?.tableBucket; |
| 157 | const namespace = output?.namespace ?? olds?.namespace; |
| 158 | if (typeof tableBucketArn !== "string" || typeof namespace !== "string") { |
| 159 | return undefined; |
| 160 | } |
| 161 | const name = output?.name ?? (yield* createTableName(id, olds ?? {})); |
| 162 | return yield* s3tables |
| 163 | .getTable({ tableBucketARN: tableBucketArn, namespace, name }) |
| 164 | .pipe( |
| 165 | Effect.map((t): Table["Attributes"] => ({ |
| 166 | tableArn: t.tableARN, |
| 167 | name: t.name, |
| 168 | namespace: t.namespace[0] ?? namespace, |
| 169 | tableBucketArn, |
| 170 | versionToken: t.versionToken, |
| 171 | metadataLocation: t.metadataLocation, |
| 172 | warehouseLocation: t.warehouseLocation, |
| 173 | format: t.format, |
| 174 | type: t.type, |
| 175 | createdAt: t.createdAt, |
| 176 | })), |
| 177 | Effect.catchTag("NotFoundException", () => Effect.succeed(undefined)), |
| 178 | ); |
| 179 | }), |
| 180 | diff: Effect.fn(function* ({ id, news, olds }) { |
| 181 | if (!isResolved(news)) return; |
| 182 | if ( |
| 183 | news.tableBucket !== olds?.tableBucket || |
| 184 | news.namespace !== olds?.namespace |
| 185 | ) { |
| 186 | return { action: "replace" } as const; |
| 187 | } |
| 188 | const oldName = yield* createTableName(id, olds ?? {}); |
| 189 | const newName = yield* createTableName(id, news); |
| 190 | if (oldName !== newName) { |
| 191 | return { action: "replace" } as const; |
| 192 | } |
| 193 | if ((news.format ?? "ICEBERG") !== (olds?.format ?? "ICEBERG")) { |
| 194 | return { action: "replace" } as const; |
| 195 | } |
| 196 | // Schema is fixed at create time; schema evolution is out of band. |
| 197 | const oldSchema = JSON.stringify(olds?.schema ?? null); |
| 198 | const newSchema = JSON.stringify(news.schema ?? null); |
| 199 | if (oldSchema !== newSchema) { |
| 200 | return { action: "replace" } as const; |
| 201 | } |
| 202 | }), |
| 203 | reconcile: Effect.fn(function* ({ id, news, output, session }) { |
| 204 | const tableBucketArn = news.tableBucket as string; |
| 205 | const namespace = news.namespace as string; |
| 206 | const name = output?.name ?? (yield* createTableName(id, news)); |
| 207 | |
| 208 | // Observe — read live state; the table may have been deleted |
| 209 | // out-of-band even if `output` cached its ARN. |
| 210 | let table = yield* s3tables |
| 211 | .getTable({ tableBucketARN: tableBucketArn, namespace, name }) |
| 212 | .pipe( |
| 213 | Effect.map((t) => t), |
| 214 | Effect.catchTag("NotFoundException", () => Effect.succeed(undefined)), |
| 215 | ); |
| 216 | |
| 217 | // Ensure — create if missing, tolerating a concurrent create. |
| 218 | if (table === undefined) { |
| 219 | yield* s3tables |
| 220 | .createTable({ |
| 221 | tableBucketARN: tableBucketArn, |
| 222 | namespace, |
| 223 | name, |
| 224 | format: news.format ?? "ICEBERG", |
| 225 | metadata: buildMetadata(news), |
| 226 | }) |
| 227 | .pipe( |
| 228 | Effect.asVoid, |
| 229 | Effect.catchTag("ConflictException", () => Effect.void), |
| 230 | ); |
| 231 | // Eventual consistency: getTable can briefly 404 a table that |
| 232 | // createTable just returned. |
| 233 | table = yield* s3tables |
| 234 | .getTable({ tableBucketARN: tableBucketArn, namespace, name }) |
| 235 | .pipe( |
| 236 | Effect.retry({ |
| 237 | while: (e) => e._tag === "NotFoundException", |
| 238 | schedule: Schedule.max([ |
| 239 | Schedule.exponential(500), |
| 240 | Schedule.recurs(8), |
| 241 | ]), |
| 242 | }), |
| 243 | ); |
| 244 | } |
| 245 | |
| 246 | yield* session.note(table.tableARN); |
| 247 | return { |
| 248 | tableArn: table.tableARN, |
| 249 | name: table.name, |
| 250 | namespace: table.namespace[0] ?? namespace, |
| 251 | tableBucketArn, |
| 252 | versionToken: table.versionToken, |
| 253 | metadataLocation: table.metadataLocation, |
| 254 | warehouseLocation: table.warehouseLocation, |
| 255 | format: table.format, |
| 256 | type: table.type, |
| 257 | createdAt: table.createdAt, |
| 258 | }; |
| 259 | }), |
| 260 | delete: Effect.fn(function* ({ output }) { |
| 261 | // Deliberately NO versionToken: destroy is unconditional. Runtime |
| 262 | // commits (UpdateTableMetadataLocation) rotate the token, so the |
| 263 | // persisted one may be stale and would fail with ConflictException. |
| 264 | yield* s3tables |
| 265 | .deleteTable({ |
| 266 | tableBucketARN: output.tableBucketArn, |
| 267 | namespace: output.namespace, |
| 268 | name: output.name, |
| 269 | }) |
| 270 | .pipe(Effect.catchTag("NotFoundException", () => Effect.void)); |
| 271 | }), |
| 272 | }); |