Adds a small embedder sidecar (Xenova/bge-small-en-v1.5, ONNX, CPU-only) that the web app calls inline on memory.write and memory.update, and on demand from the new memory.search tool. memory.search performs three candidate fetches in parallel — pgvector cosine similarity, Postgres full-text via plainto_tsquery + ts_rank_cd, and tag-set overlap — then fuses them with Reciprocal Rank Fusion (k=60). Each result carries its per-source rank so the model can see *why* a memory surfaced. The migrator boot step gained an idempotent embedding backfill: any row with embedding IS NULL is batched (32 at a time) through the embedder after SQL migrations apply. Safe to run on every boot. New tool memory.update fixes the missing edit path; centralises the re-embed-on-content-change rule alongside write. Stack additions: - apps/embedder/ — Fastify server, persistent /data/models volume so the ~30 MB model only downloads once - apps/web/lib/embedder.ts — typed HTTP client with batched embed + health probe - packages/schemas — MemoryUpdateInput, MemorySearchInput - docker-compose — embedder service, healthcheck, app + migrator both depend_on it healthy; EMBEDDER_URL promoted to a required env var Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
161 lines
5.1 KiB
TypeScript
161 lines
5.1 KiB
TypeScript
/**
|
||
* Run pending SQL migrations from ./drizzle in lexical filename order.
|
||
*
|
||
* Lightweight runner — drizzle-kit's TS migrator doesn't handle the raw SQL
|
||
* features we need (pgvector, generated columns), so we manage migration
|
||
* state ourselves in `_migrations` and apply files as plain SQL.
|
||
*/
|
||
import { existsSync } from "node:fs";
|
||
import { readFile, readdir } from "node:fs/promises";
|
||
import { dirname, join } from "node:path";
|
||
import { fileURLToPath } from "node:url";
|
||
import postgres from "postgres";
|
||
|
||
const __dirname = dirname(fileURLToPath(import.meta.url));
|
||
|
||
// Resolve the migrations directory. Try, in order:
|
||
// 1. $MIGRATIONS_DIR env override (explicit deploy-time control)
|
||
// 2. `<script>/drizzle` — production: migrate.mjs sits alongside drizzle/
|
||
// 3. `<script>/../drizzle` — dev: scripts/migrate.ts has drizzle/ one up
|
||
function findMigrationsDir(): string {
|
||
if (process.env.MIGRATIONS_DIR) return process.env.MIGRATIONS_DIR;
|
||
const sibling = join(__dirname, "drizzle");
|
||
if (existsSync(sibling)) return sibling;
|
||
const parent = join(__dirname, "..", "drizzle");
|
||
if (existsSync(parent)) return parent;
|
||
throw new Error(
|
||
`Couldn't locate migrations directory (tried ${sibling}, ${parent}). ` +
|
||
`Set MIGRATIONS_DIR to override.`,
|
||
);
|
||
}
|
||
|
||
const MIGRATIONS_DIR = findMigrationsDir();
|
||
|
||
async function main() {
|
||
const url = process.env.DATABASE_URL;
|
||
if (!url) {
|
||
console.error("DATABASE_URL is not set");
|
||
process.exit(1);
|
||
}
|
||
|
||
const sql = postgres(url, { max: 1, prepare: false });
|
||
|
||
try {
|
||
await sql`
|
||
CREATE TABLE IF NOT EXISTS "_migrations" (
|
||
"id" serial PRIMARY KEY,
|
||
"name" text NOT NULL UNIQUE,
|
||
"applied_at" timestamptz NOT NULL DEFAULT now()
|
||
)
|
||
`;
|
||
|
||
const files = (await readdir(MIGRATIONS_DIR))
|
||
.filter((f) => f.endsWith(".sql"))
|
||
.sort();
|
||
|
||
const applied = new Set(
|
||
(await sql<{ name: string }[]>`SELECT name FROM "_migrations"`).map((r) => r.name),
|
||
);
|
||
|
||
for (const file of files) {
|
||
if (applied.has(file)) {
|
||
console.log(`✓ ${file} (already applied)`);
|
||
continue;
|
||
}
|
||
const body = await readFile(join(MIGRATIONS_DIR, file), "utf8");
|
||
console.log(`→ ${file} (applying)`);
|
||
await sql.begin(async (tx) => {
|
||
await tx.unsafe(body);
|
||
await tx`INSERT INTO "_migrations" (name) VALUES (${file})`;
|
||
});
|
||
console.log(`✓ ${file}`);
|
||
}
|
||
|
||
console.log("Migrations complete.");
|
||
|
||
if (process.env.EMBEDDER_URL) {
|
||
await backfillEmbeddings(sql);
|
||
} else {
|
||
console.log("EMBEDDER_URL not set — skipping embedding backfill.");
|
||
}
|
||
} finally {
|
||
await sql.end({ timeout: 5 });
|
||
}
|
||
}
|
||
|
||
/**
|
||
* Backfill embedding column for any memory written before embeddings were
|
||
* online. Idempotent: only touches rows where embedding IS NULL. Runs on
|
||
* every migrator boot, so deploying Phase 2 — or recovering from an
|
||
* embedder outage that left fresh rows unembedded — needs no manual step.
|
||
*/
|
||
async function backfillEmbeddings(sql: ReturnType<typeof postgres>) {
|
||
const embedderUrl = process.env.EMBEDDER_URL!.replace(/\/$/, "");
|
||
const BATCH = 32;
|
||
|
||
// Wait for the embedder to report ready — its first boot has to download
|
||
// and load the model, which can take 30–60s on a cold container.
|
||
const waitDeadline = Date.now() + 180_000;
|
||
for (;;) {
|
||
try {
|
||
const res = await fetch(`${embedderUrl}/health`);
|
||
if (res.ok) {
|
||
const body = (await res.json()) as { ready?: boolean };
|
||
if (body.ready) break;
|
||
}
|
||
} catch {
|
||
/* embedder not up yet */
|
||
}
|
||
if (Date.now() > waitDeadline) {
|
||
throw new Error("embedder did not become ready within 180s");
|
||
}
|
||
await new Promise((r) => setTimeout(r, 2000));
|
||
}
|
||
|
||
let total = 0;
|
||
for (;;) {
|
||
const rows = await sql<{ id: string; content: string }[]>`
|
||
SELECT id, content FROM memories
|
||
WHERE embedding IS NULL AND deleted_at IS NULL
|
||
ORDER BY created_at
|
||
LIMIT ${BATCH}
|
||
`;
|
||
if (rows.length === 0) break;
|
||
|
||
const res = await fetch(`${embedderUrl}/embed`, {
|
||
method: "POST",
|
||
headers: { "content-type": "application/json" },
|
||
body: JSON.stringify({ texts: rows.map((r) => r.content) }),
|
||
});
|
||
if (!res.ok) {
|
||
const detail = await res.text().catch(() => "");
|
||
throw new Error(`embedder error ${res.status}: ${detail.slice(0, 200)}`);
|
||
}
|
||
const { vectors } = (await res.json()) as { vectors: number[][] };
|
||
|
||
await sql.begin(async (tx) => {
|
||
for (let i = 0; i < rows.length; i++) {
|
||
const id = rows[i]!.id;
|
||
const vec = vectors[i];
|
||
if (!vec) continue;
|
||
const literal = `[${vec.join(",")}]`;
|
||
await tx`UPDATE memories SET embedding = ${literal}::vector WHERE id = ${id}`;
|
||
}
|
||
});
|
||
|
||
total += rows.length;
|
||
console.log(` embedded ${rows.length} memories (total: ${total})`);
|
||
}
|
||
|
||
if (total === 0) {
|
||
console.log("Embedding backfill: nothing to do.");
|
||
} else {
|
||
console.log(`Embedding backfill complete: ${total} memories embedded.`);
|
||
}
|
||
}
|
||
|
||
main().catch((err) => {
|
||
console.error("Migration failed:", err);
|
||
process.exit(1);
|
||
});
|