ThothII/backend/test/workspace-maintenance.test.ts

import { expect, test, vi } from "vitest";
import {
  runWorkspaceMaintenanceCli,
  type WorkspaceMaintenanceIo,
} from "../src/workspace-maintenance.js";
import type { WorkspaceOperationResult } from "../src/workspaces/preprocessing-service.js";

function io(stdin: string): WorkspaceMaintenanceIo & { stdout: string[]; stderr: string[] } {
  const stdout: string[] = [];
  const stderr: string[] = [];
  return {
    stdin,
    stdout,
    stderr,
    writeStdout: (value) => void stdout.push(value),
    writeStderr: (value) => void stderr.push(value),
  };
}

function ok(operation: string): WorkspaceOperationResult {
  return {
    schemaVersion: 1,
    status: "succeeded",
    code: "ok",
    workspaceId: "psd-clinical",
    workspaceRevision: "a".repeat(40),
    descriptorBlob: "b".repeat(40),
    operation,
    completedStages: [],
  };
}

test("entrypoint emits exactly one pristine JSON document and maps success/block/failure exits", async () => {
  const service = {
    inspect: vi.fn(async () => ok("inspect")),
    preprocessDwh: vi.fn(async () => ({ ...ok("preprocess dwh"), status: "blocked", code: "manual_review_required" as const })),
    run: vi.fn(async () => ({ ...ok("preprocess run"), status: "failed", code: "semantic_index_incompatible" as const })),
  } as any;

  const inspectIo = io(JSON.stringify({ schemaVersion: 1, workspaceId: "psd-clinical" }));
  expect(await runWorkspaceMaintenanceCli(["node", "workspace-maintenance", "inspect"], service, inspectIo)).toBe(0);
  expect(JSON.parse(inspectIo.stdout.join(""))).toMatchObject({ operation: "inspect", code: "ok" });
  expect(inspectIo.stderr.join("")).toBe("");

  const blockedIo = io(JSON.stringify({ schemaVersion: 1, workspaceId: "psd-clinical" }));
  expect(await runWorkspaceMaintenanceCli(["node", "workspace-maintenance", "preprocess-dwh"], service, blockedIo)).toBe(3);
  expect(JSON.parse(blockedIo.stdout.join(""))).toMatchObject({ code: "manual_review_required" });

  const failedIo = io(JSON.stringify({ schemaVersion: 1, workspaceId: "psd-clinical" }));
  expect(await runWorkspaceMaintenanceCli(["node", "workspace-maintenance", "preprocess-run"], service, failedIo)).toBe(1);
  expect(JSON.parse(failedIo.stdout.join(""))).toMatchObject({ code: "semantic_index_incompatible" });
});

test("malformed stdin, unknown commands, and extra fields fail with exit 2 but still return bounded JSON", async () => {
  const service = {} as any;

  const malformedIo = io("not-json");
  expect(await runWorkspaceMaintenanceCli(["node", "workspace-maintenance", "inspect"], service, malformedIo)).toBe(2);
  expect(JSON.parse(malformedIo.stdout.join(""))).toMatchObject({ status: "failed", operation: "inspect" });

  const extraFieldIo = io(JSON.stringify({ schemaVersion: 1, workspaceId: "psd-clinical", unexpected: true }));
  expect(await runWorkspaceMaintenanceCli(["node", "workspace-maintenance", "inspect"], service, extraFieldIo)).toBe(2);
  expect(JSON.parse(extraFieldIo.stdout.join(""))).toMatchObject({ status: "failed", operation: "inspect" });

  const unknownIo = io(JSON.stringify({ schemaVersion: 1, workspaceId: "psd-clinical" }));
  expect(await runWorkspaceMaintenanceCli(["node", "workspace-maintenance", "explode"], service, unknownIo)).toBe(2);
  expect(JSON.parse(unknownIo.stdout.join(""))).toMatchObject({ status: "failed", operation: "explode" });
});

test("raw exception text is redacted from stderr and stdout remains within the public result contract", async () => {
  const service = {
    inspect: vi.fn(async () => {
      throw new Error("https://secret.example.invalid?q=token SELECT * FROM sensitive_table");
    }),
  } as any;
  const captured = io(JSON.stringify({ schemaVersion: 1, workspaceId: "psd-clinical" }));

  expect(await runWorkspaceMaintenanceCli(["node", "workspace-maintenance", "inspect"], service, captured)).toBe(1);
  expect(JSON.parse(captured.stdout.join(""))).toMatchObject({
    status: "failed",
    operation: "inspect",
  });
  expect(captured.stderr.join("")).not.toContain("secret.example.invalid");
  expect(captured.stderr.join("")).not.toContain("SELECT *");
});

test("vector-inspect and vector-rebuild dispatch to the service with the exact envelope", async () => {
  const service = {
    vectorInspect: vi.fn(async () => ok("vector inspect")),
    vectorRebuild: vi.fn(async () => ok("vector rebuild")),
  } as any;

  const inspectIo = io(JSON.stringify({ schemaVersion: 1, workspaceId: "psd-clinical" }));
  expect(await runWorkspaceMaintenanceCli(["node", "workspace-maintenance", "vector-inspect"], service, inspectIo)).toBe(0);
  expect(service.vectorInspect).toHaveBeenCalledWith({ workspaceId: "psd-clinical" });
  expect(JSON.parse(inspectIo.stdout.join(""))).toMatchObject({ operation: "vector inspect", code: "ok" });

  const rebuildIo = io(JSON.stringify({ schemaVersion: 1, workspaceId: "psd-clinical", collection: "psd-clinical", confirm: "psd-clinical", destroy: true }));
  expect(await runWorkspaceMaintenanceCli(["node", "workspace-maintenance", "vector-rebuild"], service, rebuildIo)).toBe(0);
  expect(service.vectorRebuild).toHaveBeenCalledWith({
    workspaceId: "psd-clinical",
    collection: "psd-clinical",
    confirm: "psd-clinical",
    destroy: true,
  });
});

test("vector-rebuild without exact confirmation is refused by the service", async () => {
  const service = {
    vectorRebuild: vi.fn(async () => ({ ...ok("vector rebuild"), status: "failed", code: "semantic_index_incompatible" as const })),
  } as any;
  const rebuildIo = io(JSON.stringify({ schemaVersion: 1, workspaceId: "psd-clinical", collection: "other", confirm: "other", destroy: true }));
  expect(await runWorkspaceMaintenanceCli(["node", "workspace-maintenance", "vector-rebuild"], service, rebuildIo)).toBe(1);
});