ThothII/backend/test/workspace-documents-cli.test.ts

import { spawnSync } from "node:child_process";
import { mkdirSync, mkdtempSync, readFileSync, rmSync, writeFileSync, symlinkSync } from "node:fs";
import { tmpdir } from "node:os";
import { join, resolve } from "node:path";
import { afterEach, expect, test } from "vitest";
import { parseWorkspaceCatalogYaml } from "../src/workspaces/catalog.js";
import { parseWorkspaceYaml } from "../src/workspaces/schema.js";

const roots: string[] = [];
function directory() {
  const root = mkdtempSync(join(tmpdir(), "tht-documents-"));
  roots.push(root);
  return join(root, "workspaces");
}
afterEach(() => roots.splice(0).forEach((root) => rmSync(root, { recursive: true, force: true })));

function cli(...args: string[]) {
  const packaged = process.env.THT_WORKSPACE_TEST_CLI;
  const result = spawnSync(packaged ?? process.execPath, [...(packaged ? ["workspace"] : ["--import", "tsx", resolve("src/workspace-documents-cli.ts")]), ...args, "--json"], {
    encoding: "utf8", env: { ...process.env, PATH: "" },
  });
  return { ...result, report: result.stdout.trim() ? JSON.parse(result.stdout) : null };
}

const descriptor = "workspace:\n  schema_version: 4\n  id: practice\n  name: Practice\n  language: en\n";
const catalog = "schema_version: 1\nworkspaces:\n  - id: practice\n    name: Practice\n";
const corpus = [
  { name: "minimal", descriptor, catalog, valid: true },
  { name: "optional Evidence", descriptor: descriptor + "evidence:\n  source:\n    type: http\n    uris: [https://example.com/manual.md]\n", catalog, valid: true },
  { name: "duplicate YAML key", descriptor: descriptor + "  id: practice\n", catalog, valid: false },
  { name: "ambiguous document", descriptor: descriptor + "---\n" + descriptor, catalog, valid: false },
  { name: "unknown workspace field", descriptor: descriptor + "  secret: VERY_SECRET_VALUE\n", catalog, valid: false },
  { name: "database binding in authored descriptor", descriptor: descriptor + "dwh: {password: VERY_SECRET_VALUE}\n", catalog, valid: false },
  { name: "duplicate catalog id", descriptor, catalog: catalog + "  - id: practice\n    name: Practice\n", valid: false },
  { name: "unknown catalog field", descriptor, catalog: catalog + "secret: VERY_SECRET_VALUE\n", valid: false },
  { name: "invalid catalog YAML", descriptor, catalog: catalog + "schema_version: 1\n", valid: false },
  { name: "unsafe Evidence URI", descriptor: descriptor + "evidence:\n  source:\n    type: http\n    uris: [https://user:VERY_SECRET_VALUE@example.com/file]\n", catalog, valid: false },
];

test.each(corpus)("CLI and runtime agree: $name", (fixture) => {
  const root = directory();
  mkdirSync(join(root, "practice"), { recursive: true });
  writeFileSync(join(root, "thoth-workspaces.yaml"), fixture.catalog);
  writeFileSync(join(root, "practice/workspace.yaml"), fixture.descriptor);
  let accepted = true;
  try { parseWorkspaceCatalogYaml(fixture.catalog); parseWorkspaceYaml(fixture.descriptor); }
  catch { accepted = false; }
  expect(accepted).toBe(fixture.valid);
  const checked = cli("validate", "--directory", root);
  expect(checked.status).toBe(fixture.valid ? 0 : 1);
  expect(checked.report.ok).toBe(accepted);
  expect(checked.stdout + checked.stderr).not.toContain("VERY_SECRET_VALUE");
  if (!fixture.valid) {
    expect(checked.report.issues[0].document).not.toBe(".");
    expect(checked.report.issues[0].correction.length).toBeGreaterThan(10);
  }
});

test("prepare refuses existing directories and invalid options without changing documents", () => {
  const root = directory();
  expect(cli("prepare", "--directory", root, "--id", "practice", "--name", "Practice").status).toBe(0);
  const before = readFileSync(join(root, "thoth-workspaces.yaml"), "utf8");
  expect(cli("prepare", "--directory", root, "--id", "other", "--name", "Other").report.issues[0].code).toBe("destination_exists");
  expect(readFileSync(join(root, "thoth-workspaces.yaml"), "utf8")).toBe(before);
  expect(cli("validate", "--directory", root, "--typo", "VERY_SECRET_VALUE").status).toBe(2);
});

test("validation rejects directory mismatches, missing references and symlinks without following them", () => {
  const root = directory();
  cli("prepare", "--directory", root, "--id", "practice", "--name", "Practice");
  mkdirSync(join(root, "unlisted"));
  expect(cli("validate", "--directory", root).report.issues.some((i: {code: string}) => i.code === "unlisted_directory")).toBe(true);
  rmSync(join(root, "unlisted"), { recursive: true });
  writeFileSync(join(root, "practice/workspace.yaml"), descriptor.replace("name: Practice", "name: Different"));
  expect(cli("validate", "--directory", root).report.issues[0].code).toBe("catalog_mismatch");
  writeFileSync(join(root, "practice/workspace.yaml"), descriptor + "evidence:\n  source:\n    type: filesystem\n    uri: practice/evidence\n");
  expect(cli("validate", "--directory", root).report.issues[0].code).toBe("missing_reference");
  symlinkSync(roots[roots.length - 1], join(root, "practice/evidence"), "dir");
  expect(cli("validate", "--directory", root).report.issues[0].code).toBe("unsafe_reference");
});

test("local Evidence checks references and YAML frontmatter, and explicitly defers canonical content validation", () => {
  const root = directory();
  cli("prepare", "--directory", root, "--id", "practice", "--name", "Practice");
  writeFileSync(join(root, "practice/workspace.yaml"), descriptor + "evidence:\n  schema_version: 2\n  source:\n    type: filesystem\n    uri: practice/evidence\n    patterns: ['curated/**/*.md']\n");
  mkdirSync(join(root, "practice/evidence/curated/domain"), { recursive: true });
  const evidence = join(root, "practice/evidence/curated/domain/rule.md");
  writeFileSync(evidence, "---\nschema_version: 4\nid: evidence:rule\nkind: domain\nlanguage: en\npurposes: [sql_generation]\n---\n# Rule\n\n## Rule\nUse the order number.\n");
  const checked = cli("validate", "--directory", root);
  expect(checked.status).toBe(0);
  expect(checked.report.workspaces).toEqual([{ id: "practice", evidence: "local-files" }]);
  expect(checked.report.deferred_checks).toContain("practice:evidence-content-provenance-and-indexing");
  writeFileSync(evidence, "---\nid: first\nid: VERY_SECRET_VALUE\n---\n# Rule\n");
  const invalid = cli("validate", "--directory", root);
  expect(invalid.status).toBe(1);
  expect(invalid.report.issues[0].document).toBe("practice/evidence/curated/domain/rule.md");
  expect(invalid.stdout).not.toContain("VERY_SECRET_VALUE");
  writeFileSync(join(root, "practice/workspace.yaml"), descriptor + "evidence:\n  source:\n    type: filesystem\n    uri: practice/evidence\n    patterns: [missing.md]\n");
  expect(cli("validate", "--directory", root).report.issues[0].code).toBe("missing_reference");
});

test("prepare creates documents accepted by the runtime and validate is repeatable without installation or services", () => {
  const root = directory();
  const prepared = cli("prepare", "--directory", root, "--id", "practice", "--name", "Practice");
  expect(prepared.stderr).toBe("");
  expect(prepared.status).toBe(0);
  expect(prepared.report.ok).toBe(true);
  const catalog = readFileSync(join(root, "thoth-workspaces.yaml"), "utf8");
  const descriptor = readFileSync(join(root, "practice/workspace.yaml"), "utf8");
  expect(parseWorkspaceCatalogYaml(catalog).workspaces[0].id).toBe("practice");
  expect(parseWorkspaceYaml(descriptor).evidence).toBeUndefined();
  for (let index = 0; index < 2; index++) {
    const checked = cli("validate", "--directory", root);
    expect(checked.status).toBe(0);
    expect(checked.report.workspaces).toEqual([{ id: "practice", evidence: "absent" }]);
  }
  expect(readFileSync(join(root, "thoth-workspaces.yaml"), "utf8")).toBe(catalog);
  expect(readFileSync(join(root, "practice/workspace.yaml"), "utf8")).toBe(descriptor);
});