linkedin-studio/scripts/analytics/tests/queue-join.test.ts

import { describe, test, afterEach } from "node:test";
import assert from "node:assert/strict";
import { mkdtempSync, rmSync, writeFileSync, existsSync } from "node:fs";
import { join } from "node:path";
import { tmpdir } from "node:os";
import { loadQueueEntries, labelPosts } from "../src/utils/queue-join.js";
import type { QueueEntry } from "../src/utils/queue-join.js";

/**
 * Format and pillar are NOT analytics data — they live in the post queue, which
 * the operator's create surfaces write. The per-format / per-pillar baseline
 * therefore rests on a JOIN, and a join that guesses is worse than no labels at
 * all: a post filed under the wrong format would poison the very baseline it is
 * being measured against. Every ambiguous case here resolves to UNLABELLED.
 */
describe("queue-join", () => {
  let tempDir: string;

  afterEach(() => {
    if (tempDir && existsSync(tempDir)) {
      rmSync(tempDir, { recursive: true, force: true });
    }
  });

  const post = (publishedDate: string, title = "Some post") => ({
    publishedDate,
    title,
    metrics: { impressions: 1000, engagementRate: 5 },
  });

  const entry = (overrides: Partial<QueueEntry> & { scheduled_date: string }): QueueEntry => ({
    id: overrides.id ?? `q-${overrides.scheduled_date}`,
    scheduled_date: overrides.scheduled_date,
    pillar: overrides.pillar ?? "ai-agents",
    format: overrides.format ?? "text",
    hook_preview: overrides.hook_preview,
  });

  describe("loadQueueEntries", () => {
    test("should return an empty list when there is no queue file", () => {
      tempDir = mkdtempSync(join(tmpdir(), "queue-join-"));
      const result = loadQueueEntries(join(tempDir, "queue.json"));
      assert.deepEqual(result, [], "A missing queue is unlabelled data, not an error");
    });

    test("should return an empty list for malformed JSON rather than throwing", () => {
      tempDir = mkdtempSync(join(tmpdir(), "queue-join-"));
      const file = join(tempDir, "queue.json");
      writeFileSync(file, "{ not json", "utf8");
      assert.deepEqual(loadQueueEntries(file), []);
    });

    test("should return an empty list when the file is not an array", () => {
      tempDir = mkdtempSync(join(tmpdir(), "queue-join-"));
      const file = join(tempDir, "queue.json");
      writeFileSync(file, JSON.stringify({ entries: [] }), "utf8");
      assert.deepEqual(loadQueueEntries(file), []);
    });

    test("should read entries and keep those missing pillar or format", () => {
      tempDir = mkdtempSync(join(tmpdir(), "queue-join-"));
      const file = join(tempDir, "queue.json");
      writeFileSync(
        file,
        JSON.stringify([
          { id: "a", scheduled_date: "2026-07-01", pillar: "ai", format: "text" },
          { id: "b", scheduled_date: "2026-07-03" },
        ]),
        "utf8"
      );

      const result = loadQueueEntries(file);
      assert.equal(result.length, 2);
      assert.equal(result[1].pillar, undefined, "A partial entry is data; dropping it would hide posts");
    });

    test("should skip rows with no scheduled_date, which cannot join on anything", () => {
      tempDir = mkdtempSync(join(tmpdir(), "queue-join-"));
      const file = join(tempDir, "queue.json");
      writeFileSync(
        file,
        JSON.stringify([{ id: "a", pillar: "ai", format: "text" }, { id: "b", scheduled_date: "2026-07-03" }]),
        "utf8"
      );
      assert.equal(loadQueueEntries(file).length, 1);
    });
  });

  describe("labelPosts", () => {
    test("should label a post whose publish date matches a queue entry exactly", () => {
      const result = labelPosts([post("2026-07-08")], [entry({ scheduled_date: "2026-07-08", format: "carousel" })]);
      assert.equal(result[0].format, "carousel");
      assert.equal(result[0].pillar, "ai-agents");
    });

    test("should label through a small reschedule gap", () => {
      // A queue entry is a PLAN; the operator publishing a day late is normal.
      const result = labelPosts([post("2026-07-09")], [entry({ scheduled_date: "2026-07-08" })]);
      assert.equal(result[0].format, "text");
    });

    test("should not label across a gap wider than two days", () => {
      const result = labelPosts([post("2026-07-15")], [entry({ scheduled_date: "2026-07-08" })]);
      assert.equal(result[0].format, undefined, "A week apart is a different post, not a reschedule");
      assert.equal(result[0].pillar, undefined);
    });

    test("should prefer an exact date match over a near one", () => {
      const result = labelPosts(
        [post("2026-07-08")],
        [entry({ scheduled_date: "2026-07-07", format: "video" }), entry({ scheduled_date: "2026-07-08", format: "text" })]
      );
      assert.equal(result[0].format, "text");
    });

    test("should leave a post unlabelled when two entries tie on the same date", () => {
      const result = labelPosts(
        [post("2026-07-08")],
        [
          entry({ id: "x", scheduled_date: "2026-07-08", format: "text" }),
          entry({ id: "y", scheduled_date: "2026-07-08", format: "carousel" }),
        ]
      );
      assert.equal(result[0].format, undefined, "A coin-flip label would poison the format baseline");
    });

    test("should disambiguate same-date entries on the hook preview", () => {
      const result = labelPosts(
        [post("2026-07-08", "The thing nobody tells you about agent evals")],
        [
          entry({ id: "x", scheduled_date: "2026-07-08", format: "text", hook_preview: "Three ways to lose a week" }),
          entry({
            id: "y",
            scheduled_date: "2026-07-08",
            format: "carousel",
            hook_preview: "The thing nobody tells you",
          }),
        ]
      );
      assert.equal(result[0].format, "carousel", "The hook is the only honest tiebreaker available");
    });

    test("should leave the post unlabelled when the queue is empty", () => {
      const result = labelPosts([post("2026-07-08")], []);
      assert.equal(result[0].format, undefined);
    });

    test("should never consume one queue entry for two different posts", () => {
      // Two posts, one entry: labelling both would double-count the entry's
      // format and invent a second post of that format that never existed.
      const result = labelPosts([post("2026-07-08"), post("2026-07-09")], [entry({ scheduled_date: "2026-07-08" })]);
      const labelled = result.filter((r) => r.format !== undefined);
      assert.equal(labelled.length, 1, "An entry labels at most one post");
      assert.equal(labelled[0].publishedDate, "2026-07-08", "The exact match wins the entry");
    });

    test("should preserve every post, labelled or not", () => {
      const posts = [post("2026-07-08"), post("2026-07-30")];
      const result = labelPosts(posts, [entry({ scheduled_date: "2026-07-08" })]);
      assert.equal(result.length, 2, "Unlabelled posts stay in the list; they are just not grouped");
    });
  });
});