import { parse } from 'node-html-parser';

// Fandom/MediaWiki cruft that adds no value in a reader UI.
const REMOVE_SELECTORS = [
	'script',
	'.mw-editsection',
	'.mw-empty-elt',
	// ad rails / sidebars occasionally present in parsed output
	'.wds-global-footer',
	'.page-side-tools',
	'[data-ad]',
	'.ad-slot',
	// Fandom-injected FAQ/engagement widgets (trfc161 etc.)
	'[class^="trfc"]',
	// caption info icons reference Fandom's SVG sprite sheet, which we don't serve
	'.info-icon'
];

/**
 * Sanitize and rewrite a MediaWiki `parse` HTML fragment for our reader UI.
 *
 * - Internal `/wiki/X` links are rewritten to `/<slug>/wiki/X`; category links
 *   `/wiki/Category:X` become `/<slug>/category/X`. Query/hash preserved.
 * - External (http/https) links get `target="_blank"` + safe `rel`.
 * - `<script>`, edit sections and known empty/cruft elements are removed.
 *   Inline `<style>` is preserved (infoboxes depend on it).
 * - Images become lazy/async-decoding; Fandom `data-src` is promoted to `src`;
 *   `srcset` is dropped only when it points at `data:` placeholders.
 */
export function transformArticleHtml(html: string, slug: string): string {
	const root = parse(html, {
		comment: false,
		voidTag: { closingSlash: true }
	});

	for (const selector of REMOVE_SELECTORS) {
		for (const el of root.querySelectorAll(selector)) {
			el.remove();
		}
	}

	for (const a of root.querySelectorAll('a')) {
		const href = a.getAttribute('href');
		if (!href) continue;

		if (href.startsWith('/wiki/')) {
			const rest = href.slice('/wiki/'.length);
			if (rest.startsWith('Category:')) {
				a.setAttribute('href', `/${slug}/category/${rest.slice('Category:'.length)}`);
			} else {
				a.setAttribute('href', `/${slug}/wiki/${rest}`);
			}
		} else if (/^https?:\/\//i.test(href)) {
			a.setAttribute('target', '_blank');
			a.setAttribute('rel', 'noopener noreferrer');
		}
	}

	for (const img of root.querySelectorAll('img')) {
		img.setAttribute('loading', 'lazy');
		img.setAttribute('decoding', 'async');
		// Fandom's image CDN 404s hotlinked requests that carry a foreign Referer
		img.setAttribute('referrerpolicy', 'no-referrer');

		const dataSrc = img.getAttribute('data-src');
		if (dataSrc) {
			img.setAttribute('src', dataSrc);
		}

		const srcset = img.getAttribute('srcset');
		if (srcset && srcset.includes('data:')) {
			img.removeAttribute('srcset');
		}
	}

	return root.toString();
}