linkedin-studio/scripts/brain/tests/id.test.ts

import { describe, test } from "node:test";
import assert from "node:assert/strict";

import { slugify, mintEntityId, normalizeProvenance } from "../src/id.js";

describe("id module (SC4)", () => {
  describe("slugify", () => {
    test("lowercases, trims, and dashes non-alphanumeric runs", () => {
      assert.equal(slugify("Current Role"), "current-role");
    });

    test("collapses repeated separators and strips edge dashes", () => {
      assert.equal(slugify("  Core Expertise Areas (5 topics)  "), "core-expertise-areas-5-topics");
    });

    test("is stable across case + whitespace variation of the same label", () => {
      assert.equal(slugify("Current Role"), slugify("  current   ROLE "));
    });

    // N24.6 — transliteration. Before this, every non-ASCII letter fell into the
    // `[^a-z0-9]+` run and became a dash, so a Norwegian label was mangled
    // ("Løkkene" → "l-kkene") and distinct labels collapsed onto one slug.
    test("transliterates the non-decomposable Nordic/German letters", () => {
      assert.equal(slugify("Løkkene"), "lokkene");
      assert.equal(slugify("Æresmedlem"), "aeresmedlem");
      assert.equal(slugify("Straße"), "strasse");
    });

    test("strips combining diacritics rather than dashing them", () => {
      assert.equal(slugify("Måte"), "mate");
      assert.equal(slugify("Café résumé"), "cafe-resume");
      assert.equal(slugify("Señor"), "senor");
    });

    test("keeps ø-labels and å-labels distinct (they both collapsed to `-` before)", () => {
      assert.notEqual(slugify("Møte"), slugify("Måte"));
    });

    test("output stays whitelisted to [a-z0-9-] for any input", () => {
      // Load-bearing beyond this repo: the ingestion-guard exposure numbers reported
      // to the guard owner rest on this whitelist being STRUCTURAL (no %-escape and no
      // image URL can survive a slug). Transliteration runs BEFORE the collapse so the
      // collapse stays the final gate.
      for (const raw of ["%2Fetc%2Fpasswd", "https://x.tld/a b.png?q=1#f", "Løkke%C3%B8", "日本語", "ab"]) {
        assert.match(slugify(raw), /^[a-z0-9-]*$/, `slug of ${JSON.stringify(raw)} is whitelisted`);
      }
    });
  });

  describe("mintEntityId", () => {
    test("is deterministic — same seed yields the same id", () => {
      assert.equal(
        mintEntityId({ kind: "profile-field", key: "Current Role" }),
        mintEntityId({ kind: "profile-field", key: "Current Role" }),
      );
    });

    test("is stable across label case/whitespace (id keyed on the slug)", () => {
      assert.equal(
        mintEntityId({ kind: "profile-field", key: "Current Role" }),
        mintEntityId({ kind: "profile-field", key: "  current   ROLE " }),
      );
    });

    test("different key yields a different id", () => {
      assert.notEqual(
        mintEntityId({ kind: "profile-field", key: "Current Role" }),
        mintEntityId({ kind: "profile-field", key: "Organization" }),
      );
    });

    test("different kind yields a different id (kind is part of the seed)", () => {
      assert.notEqual(
        mintEntityId({ kind: "profile-field", key: "role" }),
        mintEntityId({ kind: "expertise-area", key: "role" }),
      );
    });

    // N24.6 id-stability pin. The transliteration change re-mints any id whose key
    // carries a non-ASCII letter. Every label the SHIPPED profile template mints is
    // ASCII (27/27, measured), so for the profile layer the change is a provable
    // no-op — these goldens were computed BEFORE the change and must not move.
    test("ASCII keys keep their pre-transliteration ids (profile-field goldens)", () => {
      const golden: Record<string, string> = {
        Name: "4f321cef28b9",
        "Current Role": "25c03976c855",
        "expertise-area-1": "ad3cac5a825f",
        "Topics to AVOID": "36803b03c55f",
        "90-day growth goal": "e18da9c9e66a",
      };
      for (const [key, id] of Object.entries(golden)) {
        assert.equal(mintEntityId({ kind: "profile-field", key }), id, `id for ${JSON.stringify(key)}`);
      }
    });

    test("id is 12 lowercase hex characters", () => {
      assert.match(mintEntityId({ kind: "profile-field", key: "Name" }), /^[0-9a-f]{12}$/);
    });
  });

  describe("normalizeProvenance", () => {
    test("returns each accepted value unchanged", () => {
      assert.equal(normalizeProvenance("human"), "human");
      assert.equal(normalizeProvenance("published"), "published");
      assert.equal(normalizeProvenance("ai-draft"), "ai-draft");
    });

    test("trims and lowercases before matching", () => {
      assert.equal(normalizeProvenance("  Human "), "human");
      assert.equal(normalizeProvenance("AI-DRAFT"), "ai-draft");
    });

    test("throws on anything outside the provenance vocabulary", () => {
      assert.throws(() => normalizeProvenance("robot"));
      assert.throws(() => normalizeProvenance(""));
      assert.throws(() => normalizeProvenance("humanoid"));
    });
  });
});