most-gans: fix credits for signatures with HTML entities in them

siddharthvp · May 3, 2024 · d5b8a5e · d5b8a5e
1 parent 15de771
commit d5b8a5e
Show file tree

Hide file tree

Showing 3 changed files with 37 additions and 12 deletions.
diff --git a/most-gans/generate-initial.ts b/most-gans/generate-initial.ts
@@ -1,7 +1,7 @@
-import { bot, log } from '../botbase';
+import { argv, bot, log } from '../botbase';
 import { closeTunnels, createLocalSSHTunnel } from "../utils";
 import { TOOLS_DB_HOST } from "../db";
-import { processArticle, TABLE, db, runManualEdits } from "./model";
+import { processArticle, db, runManualEdits } from "./model";
 import { restartDeployment } from "../k8s";
 
 bot.setOptions({
@@ -13,6 +13,8 @@ bot.setOptions({
 });
 
 (async function () {
+	const TABLE = 'nominators2';
+
 	await createLocalSSHTunnel(TOOLS_DB_HOST);
 	await bot.getSiteInfo();
 
@@ -37,11 +39,13 @@ bot.setOptions({
 						  PRIMARY KEY (article)
 					  ) COLLATE 'utf8_unicode_ci'`);
 
-	// Restart the stream task, so that new GAs are processed
-	restartDeployment('stream').catch(err => {
-		log(`[E] Failed to restart stream task`);
-		log(err);
-	});
+	if (!argv.test) {
+		// Restart the stream task, so that new GAs are processed
+		restartDeployment('stream').catch(err => {
+			log(`[E] Failed to restart stream task`);
+			log(err);
+		});
+	}
 
 	let authorNotFound = [];
 
@@ -69,7 +73,9 @@ ${authorNotFound.map(p => `#[[${p}]]`).join('\n')}
 `);
 	log(`[S] Saved error list on-wiki`);
 
-	await runManualEdits();
+	if (!argv.test) {
+		await runManualEdits();
+	}
 	closeTunnels();
 
 })();
diff --git a/most-gans/model.ts b/most-gans/model.ts
@@ -2,6 +2,7 @@ import { bot, log, toolsdb } from "../botbase";
 import { NS_USER, NS_USER_TALK } from "../namespaces";
 import { createLogStream } from "../utils";
 import * as fs from "fs";
+import { JSDOM } from 'jsdom';
 
 export const db = new toolsdb('goodarticles_p');
 export const TABLE = 'nominators';
@@ -82,11 +83,19 @@ function addToDb(article: string, nom: string, date, fallbackStrategy = false):
 	return Promise.resolve([nom, date_str, fallbackStrategy]);
 }
 
+const { document } = new JSDOM('<html></html>').window;
+
+export function decodeHtmlEntities(str) {
+	const d = document.createElement('div');
+	d.innerHTML = str;
+	return d.innerHTML;
+}
+
 function getUsernameFromSignature(sig: string) {
 	if (typeof sig !== 'string') {
 		return;
 	}
-	let wkt = new bot.wikitext(sig);
+	let wkt = new bot.wikitext(decodeHtmlEntities(sig));
 	wkt.parseLinks();
 	let userPageLinks = [], userTalkPageLinks = [];
 	wkt.links.forEach(link => {
@@ -144,8 +153,8 @@ export async function getCurrentTitle(title: string, date: string): Promise<stri
 export async function getCurrentUsername(username: string, date: string): Promise<string> {
 	// This is called during the stream task as well, avoid API call in such cases
 	const dateParsed = new bot.date(date);
-	// if now - dateParsed < 30 seconds
-	if (new bot.date().subtract(30, 'seconds').isBefore(dateParsed)) {
+	// if now - dateParsed < 2 minutes
+	if (new bot.date().subtract(2, 'minutes').isBefore(dateParsed)) {
 		return username;
 	}
 	const rename = (await bot.query({

diff --git a/most-gans/test.ts b/most-gans/test.ts
@@ -1,6 +1,6 @@
 import assert = require("assert");
 import { bot } from "../botbase";
-import {getCurrentUsername, processArticle, db, getCurrentTitle} from "./model";
+import {getCurrentUsername, processArticle, db, getCurrentTitle, decodeHtmlEntities} from "./model";
 
 describe('most-gans', () => {
 	before(() => {
@@ -15,6 +15,9 @@ describe('most-gans', () => {
 			['Fight for This Love', 'Lil-unique1', '2010-06-25'], // has rev-delled talkpage revs
 			['Norman Finkelstein', 'Giggy'],
 			['Etchmiadzin Cathedral', 'Yerevantsi', '2013-12-27'], // user renamed between nomination and promotion
+			// ['Panagiotis Stamatakis', 'UndercoverClassicist', '2023-02-04'],
+			// ['Serious Sam: The First Encounter', 'IceWelder', '2023-10-30'],
+			['The Wing of Madoola', 'KGRAMR', '2023-12-28'],
 		];
 		for (let [article, nomExpected, dateExpected] of testCases) {
 			const [nom, date] = await processArticle(article);
@@ -50,4 +53,11 @@ describe('most-gans', () => {
 		assert.strictEqual(await getCurrentTitle('Shivers (song)', '2014-11-19'), 'Shivers (The Boys Next Door song)');
 	});
 
+	it('decodeHtmlEntities', function () {
+		assert.strictEqual(
+			decodeHtmlEntities('[[User:NoD&#39;ohnuts|NoD&#39;ohnuts]] ([[User talk:NoD&#39;ohnuts|talk]])'),
+			"[[User:NoD'ohnuts|NoD'ohnuts]] ([[User talk:NoD'ohnuts|talk]])"
+		);
+	});
+
 });