dify/knowledge-fs/packages/database/scripts/migration-artifacts.ts

import { readFile, readdir, writeFile } from "node:fs/promises";
import { resolve } from "node:path";

const mode = process.argv[2];
const root = resolve(import.meta.dirname, "../../..");
const migrationsDirectory = resolve(root, "packages/database/migrations");
const generatedRegistryPath = resolve(
  root,
  "packages/database/src/migration-artifacts.generated.ts",
);

if (mode !== "write" && mode !== "check") {
  throw new Error("Usage: tsx packages/database/scripts/migration-artifacts.ts <write|check>");
}

const sourceArtifacts = await readMigrationSourceArtifacts();
const expectedRegistry = renderGeneratedRegistry(sourceArtifacts);

if (mode === "write") {
  // SQL files are the immutable migration sources. Never regenerate or overwrite them from the
  // current schema catalog; only refresh the registry embedded in the production bundle.
  await writeFile(generatedRegistryPath, expectedRegistry, "utf8");
} else {
  const currentRegistry = await readFile(generatedRegistryPath, "utf8").catch(() => undefined);
  if (currentRegistry !== expectedRegistry) {
    throw new Error("Database migration registry is out of date. Run pnpm db:migrations:write.");
  }
}

interface SourceMigrationArtifact {
  readonly content: string;
  readonly filename: string;
  readonly migrationId: string;
  readonly path: string;
}

async function readMigrationSourceArtifacts(): Promise<readonly SourceMigrationArtifact[]> {
  const filenames = (await readdir(migrationsDirectory))
    .filter((filename) => filename.endsWith(".sql"))
    .sort((left, right) => left.localeCompare(right));
  const artifacts = await Promise.all(
    filenames.map(async (filename): Promise<SourceMigrationArtifact> => {
      const match = /^(\d{4}_[a-z0-9_]+)\.(postgres|tidb)\.sql$/u.exec(filename);
      if (!match) {
        throw new Error(`Invalid database migration filename: ${filename}`);
      }

      const migrationId = match[1];
      const dialect = match[2];
      if (!migrationId || !dialect) {
        throw new Error(`Invalid database migration filename: ${filename}`);
      }

      const content = await readFile(resolve(migrationsDirectory, filename), "utf8");
      if (!content.includes(`-- Migration id: ${migrationId}\n`)) {
        throw new Error(`Migration ${filename} has a mismatched migration id header`);
      }
      if (!content.includes(`-- Dialect: ${dialect}\n`)) {
        throw new Error(`Migration ${filename} has a mismatched dialect header`);
      }

      return {
        content,
        filename,
        migrationId,
        path: `packages/database/migrations/${filename}`,
      };
    }),
  );

  const dialectsByMigration = new Map<string, Set<string>>();
  for (const artifact of artifacts) {
    const dialect = artifact.filename.split(".").at(-2);
    const dialects = dialectsByMigration.get(artifact.migrationId) ?? new Set<string>();
    dialects.add(dialect ?? "");
    dialectsByMigration.set(artifact.migrationId, dialects);
  }
  for (const [migrationId, dialects] of dialectsByMigration) {
    if (!dialects.has("postgres") || !dialects.has("tidb") || dialects.size !== 2) {
      throw new Error(`Migration ${migrationId} must provide exactly postgres and tidb artifacts`);
    }
  }

  return artifacts;
}

function renderGeneratedRegistry(artifacts: readonly SourceMigrationArtifact[]): string {
  const rows = artifacts.map(
    (artifact) =>
      `  { content: ${JSON.stringify(artifact.content)}, path: ${JSON.stringify(artifact.path)} },`,
  );

  return [
    "// Generated by packages/database/scripts/migration-artifacts.ts. Do not edit manually.",
    'import type { MigrationArtifact } from "./migration-file";',
    "",
    "// biome-ignore format: one deterministic row per immutable SQL artifact keeps generation cheap.",
    "export const migrationArtifacts = [",
    ...rows,
    "] as const satisfies readonly MigrationArtifact[];",
    "",
  ].join("\n");
}