import { test } from "node:test";
import assert from "node:assert/strict";
import { readFileSync } from "node:fs";
import path from "node:path";
import { fileURLToPath } from "node:url";
import {
  feedDateToTimestamp,
  feedDateToUploadDate,
  htmlToPlainText,
  parseItunesDuration,
  parseRssFeed,
} from "./rssFeed";

// Run with:
//   pnpm --filter yt-dlp-transcript-common exec tsx --test lib/rssFeed.test.ts

const HERE = path.dirname(fileURLToPath(import.meta.url));
const FIXTURE = readFileSync(
  path.join(HERE, "__fixtures__", "demo-podcast.rss"),
  "utf8",
);

test("the fixture feed: channel title, and every item in order", () => {
  const feed = parseRssFeed(FIXTURE);
  assert.equal(feed.title, "Demo & Friends Podcast");
  assert.deepEqual(
    feed.items.map((i) => i.guid),
    ["guid-0001", "guid-0002", "https://example.com/?p=3", "guid-0004", "guid-trailer"],
    "a commented-out item is not an item, and a CDATA </item> ends nothing",
  );
});

test("CDATA is verbatim; entities are decoded once outside it", () => {
  const [one, two] = parseRssFeed(FIXTURE).items;
  assert.equal(one.title, "Episode 1: Cats & <Dogs>");
  assert.equal(two.title, "Two & a half ’quotes’ \"here\"");
  // An attribute's &amp; is a URL's &.
  assert.equal(
    one.enclosureUrl,
    "https://cdn.example.com/audio/abc123.mp3?key=a&updated=1",
  );
});

test("the item's own <title> wins over itunes:title, and a comment is no title", () => {
  const items = parseRssFeed(FIXTURE).items;
  assert.equal(items[1].title?.startsWith("Two"), true);
  assert.equal(items[2].title, "Episode three");
});

test("descriptions become plain text: HTML in CDATA, escaped HTML, itunes:summary", () => {
  const [one, two, three, four, trailer] = parseRssFeed(FIXTURE).items;
  assert.equal(one.description, "First line & more.\nSecond line\nthird");
  assert.equal(two.description, "Hello & goodbye");
  assert.equal(three.description, "Summary only.");
  assert.equal(
    four.description,
    "A CDATA body that mentions and ]] without ending.",
    "the CDATA's literal </item> is markup to the plain-text pass, not the end of the item",
  );
  assert.equal(trailer.description, null);
});

test("itunes tags, links, enclosures in either quote style", () => {
  const [one, two, three, four, trailer] = parseRssFeed(FIXTURE).items;
  assert.equal(one.durationSeconds, 3723);
  assert.equal(two.durationSeconds, 2730);
  assert.equal(three.durationSeconds, 3725);
  assert.equal(four.durationSeconds, null);
  assert.equal(one.link, "https://example.com/episodes/one");
  assert.equal(two.link, null);
  assert.equal(three.enclosureUrl, "https://cdn.example.com/audio/ghi789.mp3");
  assert.equal(
    two.enclosureUrl,
    "https://track.example.net/redirect.mp3/cdn.example.com/audio/def456.mp3?updated=2",
  );
  assert.equal(trailer.enclosureUrl, null);
  assert.equal(one.pubDate, "Tue, 05 Mar 2024 10:00:00 GMT");
});

test("parseItunesDuration: seconds, MM:SS, HH:MM:SS; junk is null", () => {
  assert.equal(parseItunesDuration("3725"), 3725);
  assert.equal(parseItunesDuration("45:30"), 2730);
  assert.equal(parseItunesDuration("1:02:03"), 3723);
  assert.equal(parseItunesDuration(" 90.6 "), 91);
  assert.equal(parseItunesDuration("1 hour"), null);
  assert.equal(parseItunesDuration(""), null);
  assert.equal(parseItunesDuration(null), null);
});

test("upload_date is the UTC day of the publication instant", () => {
  assert.equal(feedDateToUploadDate("Tue, 05 Mar 2024 10:00:00 GMT"), "20240305");
  // 22:30 at -0500 is 03:30 the next day in UTC.
  assert.equal(feedDateToUploadDate("Mon, 04 Mar 2024 22:30:00 -0500"), "20240305");
  assert.equal(feedDateToUploadDate("Mon, 04 Mar 2024 22:30:00 EST"), "20240305");
  assert.equal(feedDateToUploadDate("2024-03-05T23:30:00+02:00"), "20240305");
  // No zone at all reads as UTC, never as this machine's zone.
  assert.equal(feedDateToUploadDate("Tue, 05 Mar 2024 23:30:00"), "20240305");
  assert.equal(feedDateToUploadDate("2024-03-05T23:30:00"), "20240305");
  assert.equal(feedDateToUploadDate("2024-03-05"), "20240305");
  assert.equal(feedDateToUploadDate("not a date"), null);
  assert.equal(feedDateToUploadDate(null), null);
  assert.equal(feedDateToTimestamp("Tue, 05 Mar 2024 10:00:00 GMT"), 1709632800);
});

test("htmlToPlainText: list items, blank-line runs, numeric entities", () => {
  assert.equal(
    htmlToPlainText("<ul><li>one</li><li>two</li></ul><p></p><p></p><p>&#169; x</p>"),
    "- one\n- two\n\n© x",
  );
  assert.equal(htmlToPlainText("plain text"), "plain text");
});

test("a self-closing Atom-style link is read from its href; a BOM is ignored", () => {
  const feed = parseRssFeed(
    "<rss><channel><title>T</title><item><title>A</title>" +
      '<link href="https://example.com/a"/></item></channel></rss>',
  );
  assert.equal(feed.title, "T");
  assert.equal(feed.items[0].link, "https://example.com/a");
});