@apibara/plugin-drizzle 2.1.0-beta.4 → 2.1.0-beta.40

package/dist/index.mjs

@@ -1,10 +1,11 @@
 import { useIndexerContext } from '@apibara/indexer';
-import { defineIndexerPlugin } from '@apibara/indexer/plugins';
+import { defineIndexerPlugin, useLogger } from '@apibara/indexer/plugins';
 import { generateIndexerId } from '@apibara/indexer/internal';
 import { useInternalContext } from '@apibara/indexer/internal/plugins';
+import { S as SCHEMA_NAME, D as DRIZZLE_PROPERTY, a as DRIZZLE_STORAGE_DB_PROPERTY } from './shared/plugin-drizzle.2d226351.mjs';
+import { entityKind, sql, eq, and, isNull, gt, lt } from 'drizzle-orm';
 import { normalizeCursor } from '@apibara/protocol';
-import { eq, and, isNull, gt, lt, sql } from 'drizzle-orm';
-import { pgTable, text, integer, primaryKey, serial, char, jsonb } from 'drizzle-orm/pg-core';
+import { pgSchema, text, integer, primaryKey, serial, char, jsonb } from 'drizzle-orm/pg-core';
 
 class DrizzleStorageError extends Error {
   constructor(message, options) {
@@ -33,16 +34,89 @@ function serialize(obj) {
 function sleep(ms) {
   return new Promise((resolve) => setTimeout(resolve, ms));
 }
+const getIdColumnForTable = (tableName, idColumn) => {
+  if (idColumn[tableName]) {
+    return idColumn[tableName];
+  }
+  return idColumn["*"];
+};
 
-const CHECKPOINTS_TABLE_NAME = "__indexer_checkpoints";
-const FILTERS_TABLE_NAME = "__indexer_filters";
-const SCHEMA_VERSION_TABLE_NAME = "__indexer_schema_version";
-const checkpoints = pgTable(CHECKPOINTS_TABLE_NAME, {
+function drizzle(options) {
+  const {
+    connectionString = process.env["POSTGRES_CONNECTION_STRING"] ?? "memory://",
+    schema,
+    type = "pglite",
+    config,
+    poolConfig
+  } = options ?? {};
+  if (isPgliteConnectionString(connectionString) && type === "pglite") {
+    const { drizzle: drizzlePGLite } = require("drizzle-orm/pglite");
+    return drizzlePGLite({
+      schema,
+      connection: {
+        dataDir: connectionString || "memory://pglite"
+      },
+      ...config || {}
+    });
+  }
+  const { Pool } = require("pg");
+  const { drizzle: drizzleNode } = require("drizzle-orm/node-postgres");
+  const pool = new Pool({
+    connectionString,
+    ...poolConfig || {}
+  });
+  return drizzleNode(pool, { schema, ...config || {} });
+}
+async function migrate(db, options) {
+  const isPglite = isDrizzleKind(db, "PgliteDatabase");
+  try {
+    if (isPglite) {
+      const { migrate: migratePGLite } = require("drizzle-orm/pglite/migrator");
+      await migratePGLite(db, options);
+    } else {
+      const {
+        migrate: migrateNode
+      } = require("drizzle-orm/node-postgres/migrator");
+      await migrateNode(db, options);
+    }
+  } catch (error) {
+    throw new DrizzleStorageError(
+      "Failed to apply migrations! Please check if you have generated migrations using drizzle:generate",
+      {
+        cause: error
+      }
+    );
+  }
+}
+function isPgliteConnectionString(conn) {
+  return conn.startsWith("memory://") || conn.startsWith("file://") || conn.startsWith("idb://");
+}
+function isDrizzleKind(value, entityKindValue) {
+  if (!value || typeof value !== "object") {
+    return false;
+  }
+  let cls = Object.getPrototypeOf(value).constructor;
+  if (cls) {
+    while (cls) {
+      if (entityKind in cls && cls[entityKind] === entityKindValue) {
+        return true;
+      }
+      cls = Object.getPrototypeOf(cls);
+    }
+  }
+  return false;
+}
+
+const CHECKPOINTS_TABLE_NAME = "checkpoints";
+const FILTERS_TABLE_NAME = "filters";
+const SCHEMA_VERSION_TABLE_NAME = "schema_version";
+const schema$1 = pgSchema(SCHEMA_NAME);
+const checkpoints = schema$1.table(CHECKPOINTS_TABLE_NAME, {
   id: text("id").notNull().primaryKey(),
   orderKey: integer("order_key").notNull(),
   uniqueKey: text("unique_key")
 });
-const filters = pgTable(
+const filters = schema$1.table(
   FILTERS_TABLE_NAME,
   {
     id: text("id").notNull(),
@@ -56,7 +130,7 @@ const filters = pgTable(
     }
   ]
 );
-const schemaVersion = pgTable(SCHEMA_VERSION_TABLE_NAME, {
+const schemaVersion = schema$1.table(SCHEMA_VERSION_TABLE_NAME, {
   k: integer("k").notNull().primaryKey(),
   version: integer("version").notNull()
 });
@@ -67,12 +141,19 @@ const MIGRATIONS = [
   // Add more migration arrays for future versions
 ];
 async function initializePersistentState(tx) {
-  await tx.execute(`
-    CREATE TABLE IF NOT EXISTS ${SCHEMA_VERSION_TABLE_NAME} (
+  await tx.execute(
+    sql.raw(`
+      CREATE SCHEMA IF NOT EXISTS ${SCHEMA_NAME};
+  `)
+  );
+  await tx.execute(
+    sql.raw(`
+    CREATE TABLE IF NOT EXISTS ${SCHEMA_NAME}.${SCHEMA_VERSION_TABLE_NAME} (
       k INTEGER PRIMARY KEY,
       version INTEGER NOT NULL
     );
-  `);
+  `)
+  );
   const versionRows = await tx.select().from(schemaVersion).where(eq(schemaVersion.k, 0));
   const storedVersion = versionRows[0]?.version ?? -1;
   if (storedVersion > CURRENT_SCHEMA_VERSION) {
@@ -82,22 +163,26 @@ async function initializePersistentState(tx) {
   }
   try {
     if (storedVersion === -1) {
-      await tx.execute(`
-        CREATE TABLE IF NOT EXISTS ${CHECKPOINTS_TABLE_NAME} (
+      await tx.execute(
+        sql.raw(`
+        CREATE TABLE IF NOT EXISTS ${SCHEMA_NAME}.${CHECKPOINTS_TABLE_NAME} (
           id TEXT PRIMARY KEY,
           order_key INTEGER NOT NULL,
           unique_key TEXT
         );
-      `);
-      await tx.execute(`
-        CREATE TABLE IF NOT EXISTS ${FILTERS_TABLE_NAME} (
+      `)
+      );
+      await tx.execute(
+        sql.raw(`
+        CREATE TABLE IF NOT EXISTS ${SCHEMA_NAME}.${FILTERS_TABLE_NAME} (
           id TEXT NOT NULL,
           filter TEXT NOT NULL,
           from_block INTEGER NOT NULL,
           to_block INTEGER DEFAULT NULL,
           PRIMARY KEY (id, from_block)
         );
-      `);
+      `)
+      );
       await tx.insert(schemaVersion).values({
         k: 0,
         version: CURRENT_SCHEMA_VERSION
@@ -132,7 +217,9 @@ async function persistState(props) {
         target: checkpoints.id,
         set: {
           orderKey: Number(endCursor.orderKey),
-          uniqueKey: endCursor.uniqueKey
+          // Explicitly set the unique key to `null` to indicate that it has been deleted
+          // Otherwise drizzle will not update its value.
+          uniqueKey: endCursor.uniqueKey ? endCursor.uniqueKey : null
         }
       });
       if (filter) {
@@ -211,11 +298,24 @@ async function finalizeState(props) {
     });
   }
 }
+async function resetPersistence(props) {
+  const { tx, indexerId } = props;
+  try {
+    await tx.delete(checkpoints).where(eq(checkpoints.id, indexerId));
+    await tx.delete(filters).where(eq(filters.id, indexerId));
+  } catch (error) {
+    throw new DrizzleStorageError("Failed to reset persistence state", {
+      cause: error
+    });
+  }
+}
 
+const ROLLBACK_TABLE_NAME = "reorg_rollback";
+const schema = pgSchema(SCHEMA_NAME);
 function getReorgTriggerName(table, indexerId) {
   return `${table}_reorg_${indexerId}`;
 }
-pgTable("__reorg_rollback", {
+schema.table(ROLLBACK_TABLE_NAME, {
   n: serial("n").primaryKey(),
   op: char("op", { length: 1 }).$type().notNull(),
   table_name: text("table_name").notNull(),
@@ -226,9 +326,12 @@ pgTable("__reorg_rollback", {
 });
 async function initializeReorgRollbackTable(tx, indexerId) {
   try {
+    await tx.execute(`
+    CREATE SCHEMA IF NOT EXISTS ${SCHEMA_NAME};
+    `);
     await tx.execute(
       sql.raw(`
-        CREATE TABLE IF NOT EXISTS __reorg_rollback(
+        CREATE TABLE IF NOT EXISTS ${SCHEMA_NAME}.${ROLLBACK_TABLE_NAME}(
           n SERIAL PRIMARY KEY,
           op CHAR(1) NOT NULL,
           table_name TEXT NOT NULL,
@@ -241,7 +344,7 @@ async function initializeReorgRollbackTable(tx, indexerId) {
     );
     await tx.execute(
       sql.raw(`
-        CREATE INDEX IF NOT EXISTS idx_reorg_rollback_indexer_id_cursor ON __reorg_rollback(indexer_id, cursor);
+        CREATE INDEX IF NOT EXISTS idx_reorg_rollback_indexer_id_cursor ON ${SCHEMA_NAME}.${ROLLBACK_TABLE_NAME}(indexer_id, cursor);
       `)
     );
   } catch (error) {
@@ -252,24 +355,25 @@ async function initializeReorgRollbackTable(tx, indexerId) {
   try {
     await tx.execute(
       sql.raw(`
-      CREATE OR REPLACE FUNCTION reorg_checkpoint()
+      CREATE OR REPLACE FUNCTION ${SCHEMA_NAME}.reorg_checkpoint()
       RETURNS TRIGGER AS $$
       DECLARE
-        id_col TEXT := TG_ARGV[0]::TEXT;
-        order_key INTEGER := TG_ARGV[1]::INTEGER;
-        indexer_id TEXT := TG_ARGV[2]::TEXT;
+        table_name TEXT := TG_ARGV[0]::TEXT;
+        id_col TEXT := TG_ARGV[1]::TEXT;
+        order_key INTEGER := TG_ARGV[2]::INTEGER;
+        indexer_id TEXT := TG_ARGV[3]::TEXT;
         new_id_value TEXT := row_to_json(NEW.*)->>id_col;
         old_id_value TEXT := row_to_json(OLD.*)->>id_col;
       BEGIN
         IF (TG_OP = 'DELETE') THEN
-          INSERT INTO __reorg_rollback(op, table_name, cursor, row_id, row_value, indexer_id)
-            SELECT 'D', TG_TABLE_NAME, order_key, old_id_value, row_to_json(OLD.*), indexer_id;
+          INSERT INTO ${SCHEMA_NAME}.${ROLLBACK_TABLE_NAME}(op, table_name, cursor, row_id, row_value, indexer_id)
+            SELECT 'D', table_name, order_key, old_id_value, row_to_json(OLD.*), indexer_id;
         ELSIF (TG_OP = 'UPDATE') THEN
-          INSERT INTO __reorg_rollback(op, table_name, cursor, row_id, row_value, indexer_id)
-            SELECT 'U', TG_TABLE_NAME, order_key, new_id_value, row_to_json(OLD.*), indexer_id;
+          INSERT INTO ${SCHEMA_NAME}.${ROLLBACK_TABLE_NAME}(op, table_name, cursor, row_id, row_value, indexer_id)
+            SELECT 'U', table_name, order_key, new_id_value, row_to_json(OLD.*), indexer_id;
         ELSIF (TG_OP = 'INSERT') THEN
-          INSERT INTO __reorg_rollback(op, table_name, cursor, row_id, row_value, indexer_id)
-            SELECT 'I', TG_TABLE_NAME, order_key, new_id_value, null, indexer_id;
+          INSERT INTO ${SCHEMA_NAME}.${ROLLBACK_TABLE_NAME}(op, table_name, cursor, row_id, row_value, indexer_id)
+            SELECT 'I', table_name, order_key, new_id_value, null, indexer_id;
         END IF;
         RETURN NULL;
       END;
@@ -285,9 +389,10 @@ async function initializeReorgRollbackTable(tx, indexerId) {
     );
   }
 }
-async function registerTriggers(tx, tables, endCursor, idColumn, indexerId) {
+async function registerTriggers(tx, tables, endCursor, idColumnMap, indexerId) {
   try {
     for (const table of tables) {
+      const tableIdColumn = getIdColumnForTable(table, idColumnMap);
       await tx.execute(
         sql.raw(
           `DROP TRIGGER IF EXISTS ${getReorgTriggerName(table, indexerId)} ON ${table};`
@@ -298,7 +403,7 @@ async function registerTriggers(tx, tables, endCursor, idColumn, indexerId) {
           CREATE CONSTRAINT TRIGGER ${getReorgTriggerName(table, indexerId)}
           AFTER INSERT OR UPDATE OR DELETE ON ${table}
           DEFERRABLE INITIALLY DEFERRED
-          FOR EACH ROW EXECUTE FUNCTION reorg_checkpoint('${idColumn}', ${`${Number(endCursor.orderKey)}`}, '${indexerId}');
+          FOR EACH ROW EXECUTE FUNCTION ${SCHEMA_NAME}.reorg_checkpoint('${table}', '${tableIdColumn}', ${Number(endCursor.orderKey)}, '${indexerId}');
         `)
       );
     }
@@ -323,11 +428,11 @@ async function removeTriggers(db, tables, indexerId) {
     });
   }
 }
-async function invalidate(tx, cursor, idColumn, indexerId) {
+async function invalidate(tx, cursor, idColumnMap, indexerId) {
   const { rows: result } = await tx.execute(
     sql.raw(`
       WITH deleted AS (
-        DELETE FROM __reorg_rollback
+        DELETE FROM ${SCHEMA_NAME}.${ROLLBACK_TABLE_NAME}
         WHERE cursor > ${Number(cursor.orderKey)}
         AND indexer_id = '${indexerId}'
         RETURNING *
@@ -341,6 +446,7 @@ async function invalidate(tx, cursor, idColumn, indexerId) {
     );
   }
   for (const op of result) {
+    const tableIdColumn = getIdColumnForTable(op.table_name, idColumnMap);
     switch (op.op) {
       case "I":
         try {
@@ -350,7 +456,7 @@ async function invalidate(tx, cursor, idColumn, indexerId) {
           await tx.execute(
             sql.raw(`
                 DELETE FROM ${op.table_name}
-                WHERE ${idColumn} = '${op.row_id}'
+                WHERE ${tableIdColumn} = '${op.row_id}'
               `)
           );
         } catch (error) {
@@ -390,7 +496,9 @@ async function invalidate(tx, cursor, idColumn, indexerId) {
             );
           }
           const rowValue = typeof op.row_value === "string" ? JSON.parse(op.row_value) : op.row_value;
-          const nonIdKeys = Object.keys(rowValue).filter((k) => k !== idColumn);
+          const nonIdKeys = Object.keys(rowValue).filter(
+            (k) => k !== tableIdColumn
+          );
           const fields = nonIdKeys.map((c) => `${c} = prev.${c}`).join(", ");
           const query = sql.raw(`
               UPDATE ${op.table_name}
@@ -398,7 +506,7 @@ async function invalidate(tx, cursor, idColumn, indexerId) {
               FROM (
                 SELECT * FROM json_populate_record(null::${op.table_name}, '${JSON.stringify(op.row_value)}'::json)
               ) as prev
-              WHERE ${op.table_name}.${idColumn} = '${op.row_id}'
+              WHERE ${op.table_name}.${tableIdColumn} = '${op.row_id}'
               `);
           await tx.execute(query);
         } catch (error) {
@@ -420,7 +528,7 @@ async function finalize(tx, cursor, indexerId) {
   try {
     await tx.execute(
       sql.raw(`
-      DELETE FROM __reorg_rollback
+      DELETE FROM ${SCHEMA_NAME}.${ROLLBACK_TABLE_NAME}
       WHERE cursor <= ${Number(cursor.orderKey)}
       AND indexer_id = '${indexerId}'
     `)
@@ -431,8 +539,37 @@ async function finalize(tx, cursor, indexerId) {
     });
   }
 }
+async function cleanupStorage(tx, tables, indexerId) {
+  try {
+    for (const table of tables) {
+      await tx.execute(
+        sql.raw(
+          `DROP TRIGGER IF EXISTS ${getReorgTriggerName(table, indexerId)} ON ${table};`
+        )
+      );
+    }
+    await tx.execute(
+      sql.raw(`
+        DELETE FROM ${SCHEMA_NAME}.${ROLLBACK_TABLE_NAME}
+        WHERE indexer_id = '${indexerId}'
+      `)
+    );
+    for (const table of tables) {
+      try {
+        await tx.execute(sql.raw(`TRUNCATE TABLE ${table} CASCADE;`));
+      } catch (error) {
+        throw new DrizzleStorageError(`Failed to truncate table ${table}`, {
+          cause: error
+        });
+      }
+    }
+  } catch (error) {
+    throw new DrizzleStorageError("Failed to clean up storage", {
+      cause: error
+    });
+  }
+}
 
-const DRIZZLE_PROPERTY = "_drizzle";
 const MAX_RETRIES = 5;
 function useDrizzleStorage(_db) {
   const context = useIndexerContext();
@@ -443,40 +580,92 @@ function useDrizzleStorage(_db) {
   }
   return context[DRIZZLE_PROPERTY];
 }
+function useTestDrizzleStorage() {
+  const context = useIndexerContext();
+  if (!context[DRIZZLE_STORAGE_DB_PROPERTY]) {
+    throw new DrizzleStorageError(
+      "drizzle storage db is not available. Did you register the plugin?"
+    );
+  }
+  return context[DRIZZLE_STORAGE_DB_PROPERTY];
+}
 function drizzleStorage({
   db,
   persistState: enablePersistence = true,
   indexerName: identifier = "default",
-  schema,
-  idColumn = "id"
+  schema: _schema,
+  idColumn,
+  migrate: migrateOptions
 }) {
   return defineIndexerPlugin((indexer) => {
     let tableNames = [];
     let indexerId = "";
+    const alwaysReindex = process.env["APIBARA_ALWAYS_REINDEX"] === "true";
+    let prevFinality;
+    const schema = _schema ?? db._.schema ?? {};
+    const idColumnMap = {
+      "*": typeof idColumn === "string" ? idColumn : "id",
+      ...typeof idColumn === "object" ? idColumn : {}
+    };
     try {
-      tableNames = Object.values(schema ?? db._.schema ?? {}).map(
-        (table) => table.dbName
-      );
+      tableNames = Object.values(schema).map((table) => table.dbName);
     } catch (error) {
       throw new DrizzleStorageError("Failed to get table names from schema", {
         cause: error
       });
     }
-    indexer.hooks.hook("run:before", async () => {
-      const { indexerName: indexerFileName, availableIndexers } = useInternalContext();
+    for (const table of Object.values(schema)) {
+      const columns = table.columns;
+      const tableIdColumn = getIdColumnForTable(table.dbName, idColumnMap);
+      const columnExists = Object.values(columns).some(
+        (column) => column.name === tableIdColumn
+      );
+      if (!columnExists) {
+        throw new DrizzleStorageError(
+          `Column \`"${tableIdColumn}"\` does not exist in table \`"${table.dbName}"\`. Make sure the table has the specified column or provide a valid \`idColumn\` mapping to \`drizzleStorage\`.`
+        );
+      }
+    }
+    indexer.hooks.hook("plugins:init", async () => {
+      const internalContext = useInternalContext();
+      const context = useIndexerContext();
+      const logger = useLogger();
+      context[DRIZZLE_STORAGE_DB_PROPERTY] = db;
+      const { indexerName: indexerFileName, availableIndexers } = internalContext;
       indexerId = generateIndexerId(indexerFileName, identifier);
       let retries = 0;
+      let migrationsApplied = false;
+      let cleanupApplied = false;
       while (retries <= MAX_RETRIES) {
         try {
+          if (migrateOptions && !migrationsApplied) {
+            await migrate(db, migrateOptions);
+            migrationsApplied = true;
+            logger.success("Migrations applied");
+          }
           await withTransaction(db, async (tx) => {
             await initializeReorgRollbackTable(tx, indexerId);
             if (enablePersistence) {
               await initializePersistentState(tx);
             }
+            if (alwaysReindex && !cleanupApplied) {
+              logger.warn(
+                `Reindexing: Deleting all data from tables - ${tableNames.join(", ")}`
+              );
+              await cleanupStorage(tx, tableNames, indexerId);
+              if (enablePersistence) {
+                await resetPersistence({ tx, indexerId });
+              }
+              cleanupApplied = true;
+              logger.success("Tables have been cleaned up for reindexing");
+            }
           });
           break;
         } catch (error) {
           if (retries === MAX_RETRIES) {
+            if (error instanceof DrizzleStorageError) {
+              throw error;
+            }
             throw new DrizzleStorageError(
               "Initialization failed after 5 retries",
               {
@@ -512,7 +701,7 @@ function drizzleStorage({
         return;
       }
       await withTransaction(db, async (tx) => {
-        await invalidate(tx, cursor, idColumn, indexerId);
+        await invalidate(tx, cursor, idColumnMap, indexerId);
         if (enablePersistence) {
           await invalidateState({ tx, cursor, indexerId });
         }
@@ -533,7 +722,7 @@ function drizzleStorage({
       }
     });
     indexer.hooks.hook("message:finalize", async ({ message }) => {
-      const { cursor } = message.finalize;
+      const { cursor } = message;
       if (!cursor) {
         throw new DrizzleStorageError("Finalized Cursor is undefined");
       }
@@ -545,12 +734,12 @@ function drizzleStorage({
       });
     });
     indexer.hooks.hook("message:invalidate", async ({ message }) => {
-      const { cursor } = message.invalidate;
+      const { cursor } = message;
       if (!cursor) {
         throw new DrizzleStorageError("Invalidate Cursor is undefined");
       }
       await withTransaction(db, async (tx) => {
-        await invalidate(tx, cursor, idColumn, indexerId);
+        await invalidate(tx, cursor, idColumnMap, indexerId);
         if (enablePersistence) {
           await invalidateState({ tx, cursor, indexerId });
         }
@@ -559,30 +748,34 @@ function drizzleStorage({
     indexer.hooks.hook("handler:middleware", async ({ use }) => {
       use(async (context, next) => {
         try {
-          const { endCursor, finality } = context;
+          const { endCursor, finality, cursor } = context;
           if (!endCursor) {
             throw new DrizzleStorageError("End Cursor is undefined");
           }
           await withTransaction(db, async (tx) => {
             context[DRIZZLE_PROPERTY] = { db: tx };
+            if (prevFinality === "pending") {
+              await invalidate(tx, cursor, idColumnMap, indexerId);
+            }
             if (finality !== "finalized") {
               await registerTriggers(
                 tx,
                 tableNames,
                 endCursor,
-                idColumn,
+                idColumnMap,
                 indexerId
               );
             }
             await next();
             delete context[DRIZZLE_PROPERTY];
-            if (enablePersistence) {
+            if (enablePersistence && finality !== "pending") {
               await persistState({
                 tx,
                 endCursor,
                 indexerId
               });
             }
+            prevFinality = finality;
           });
           if (finality !== "finalized") {
             await removeTriggers(db, tableNames, indexerId);
@@ -598,4 +791,5 @@ function drizzleStorage({
   });
 }
 
-export { drizzleStorage, useDrizzleStorage };
+export { drizzle, drizzleStorage, migrate, useDrizzleStorage, useTestDrizzleStorage };
+//# sourceMappingURL=index.mjs.map