ffxiv-wiki-edb-id-script/edb-id-bot.mjs

#!/usr/bin/env node

import {execSync} from 'node:child_process';
import makeFetchCookie from 'fetch-cookie';

const fetchWithCookies = makeFetchCookie(fetch);

/**
 * @see https://stackoverflow.com/a/6969486
 * @param {string}
 * @returns {string}
 */
const regExpEscape = str => str.replace(/[.*+?^${}()|[\]\\]/g, '\\$&');

/**
 * Creates a regular expression that matches a link to the named item and
 * captures its EDB ID from the matched link's `href` attribute.
 * @param {string} name
 * @returns {RegExp}
 */
const itemLinkRegExp = name => new RegExp(`<a href="/lodestone/playguide/db/item/(?<id>[a-z0-9]+)[^"]+"[^>]*>(?<name>${regExpEscape(name)})</a>`, 'i');

/**
 * Gets the ID of the named item in Eorzea Database.
 * @param {string} name
 * @returns {Promise<string | undefined>}
 */
async function findItemID (name) {
	// execute a search for the item's name
	const searchURL = `https://na.finalfantasyxiv.com/lodestone/playguide/db/item/?q=${encodeURIComponent(name)}`;
	const response = await fetchWithCookies(searchURL);
	const body = await response.text();
	// find an `<a>` in the HTML response whose text exactly matches the name
	const match = body.match(itemLinkRegExp(name));
	// return the ID parsed from the URL in the `href` attribute
	return match?.groups.id;
}

/**
 * Gets the current contents of the named item's wiki page and returns the
 * contents with the infobox updated to use the given EDB item ID.
 * @param {string} name
 * @returns {Promise<string>}
 */
async function getWikiPageContents (name) {
	const response = await fetchWithCookies(`https://ffxiv.consolegameswiki.com/mediawiki/index.php?action=raw&title=${encodeURIComponent(name)}`);
	const rawContents = await response.text();
	return rawContents;
}

/**
 * Matches an empty `id-edb` infobox parameter which can just have a value
 * inserted after it.
 */
const existingEmptyEDBIDParameter = /^(\s*\|\s*id-edb\s*=)\s*$/m;

/**
 * Matches an `id-gt` infobox parameter above which we can insert an `id-edb`
 * parameter above. Whitespace in capture groups so the inserted parameter can
 * use the exact same formatting.
 */
const existingGTIDParameter = /^(\s*)\|(\s*)id-gt(\s*)=(\s*).*$/m;

/**
 * Inserts the `id-edb` item infobox parameter into the given page contents.
 * @param {string} pageContent
 * @param {string} edbID
 * @returns {string}
 */
function insertInfoboxEDBID (pageContent, edbID) {
	if (pageContent.includes(`edb-id = ${edbID}`)) {
		console.log('what????', pageContent);
		throw new Error('Page seems to already contain its own EDB ID??');
	}
	if (pageContent.match(existingEmptyEDBIDParameter)) {
		console.log('Page has existing empty `id-edb` parameter');
		return pageContent.replace(existingEmptyEDBIDParameter, `$1 ${edbID}`);
	}
	if (pageContent.match(existingGTIDParameter)) {
		console.log('Page has `id-gt` parameter, inserting `id-edb` above that');
		return pageContent.replace(existingGTIDParameter, `$1|$2id-edb$3=$4${edbID}\n$&`);
	}
	console.log('bad', pageContent);
	throw new Error('Dunno how to insert the parameter into this page');
}

const wikiAPI = 'https://ffxiv.consolegameswiki.com/mediawiki/api.php';

/**
 * Gets the list of wiki pages from "Category:Missing EDB ID".
 * @returns {Promise<{pageid: number; title: string}[]>}
 */
async function getItemPagesWithNoEDBID () {
	const response = await fetchWithCookies(`https://ffxiv.consolegameswiki.com/mediawiki/api.php?${new URLSearchParams({
		action: 'query',
		list: 'categorymembers',
		cmlimit: 500,
		cmtitle: 'Category:Missing EDB ID',
		format: 'json',
	})}`);
	const body = await response.json();
	if (body.error) {
		throw new Error(`[${body.error.code}] ${body.error.info}`);
	}
	return body.query.categorymembers;
}

/** terrible terrible terrible string diff helper for debugging */
function diff (a, b) {
	// base64 input strings before passing to shell to avoid escaping issues
	// https://stackoverflow.com/a/60221847
	// also use `|| true` to not throw an error when `diff` returns non-zero
	execSync(`bash -c '
		diff --color -u <(echo ${btoa(a)} | base64 -d) <(echo ${btoa(b)} | base64 -d)
	' || true`, {
		// display result directly in terminal
		stdio: 'inherit',
	});
}

/**
 * Given an item name, looks up its EDB ID and edits its wiki page to include
 * that ID in the item infobox if it doesn't already.
 */
async function processItem (name) {
	console.log('Page:', name);
	const edbID = await findItemID(name);
	if (!edbID) {
		console.log('No EDB ID found for this item, skipping');
		return;
	}
	console.log('EDB ID:', edbID, `(https://na.finalfantasyxiv.com/lodestone/playguide/db/item/${encodeURIComponent(edbID)})`);

	let updatedText;
	try {
		const originalText = await getWikiPageContents(name);
		updatedText = insertInfoboxEDBID(originalText, edbID);
		diff(originalText, updatedText);
	} catch (error) {
		console.log(error);
		console.log('not doing anything with this item');
		return;
	}

	// TODO: actually submit wiki edit
}

const itemPagesWithoutEDBIDs = await getItemPagesWithNoEDBID();
console.log('Looking up EDB IDs of', itemPagesWithoutEDBIDs.length, 'items\n');

for (const {title} of itemPagesWithoutEDBIDs) {
	await processItem(title);

	console.log();

	// this runs serially with an artificial delay between requests to decrease
	// the chance of sqenix sending ninjas to my house
	await new Promise(resolve => setTimeout(resolve, 5000));
}

console.log('done!');