From b004c8ed1aee125163e6bf12ed09618ce0dcd2e7 Mon Sep 17 00:00:00 2001
From: Sam Carlton <sam@sam.lc>
Date: Sun, 16 May 2021 17:09:12 -0500
Subject: [PATCH] Test that local sitemap matches production

---
 test/main.js | 40 ++++++++++++++++++++++++++++++++++++++++
 1 file changed, 40 insertions(+)

diff --git a/test/main.js b/test/main.js
index d342235..7150ca8 100644
--- a/test/main.js
+++ b/test/main.js
@@ -2,6 +2,7 @@ import { promises as fs } from 'fs'
 
 import test from 'ava'
 import parser from 'fast-xml-parser'
+import axios from 'axios'
 import { structuredDataTest } from 'structured-data-testing-tool'
 import { Google, Twitter } from 'structured-data-testing-tool/presets'
 
@@ -77,6 +78,45 @@ test('Sitemap contains no double slashes in paths', (t) => {
     t.pass()
 })
 
+
+test('Sitemap mostly matches production', async (t) => {
+    // console.log('t.context.sitemapUrls', t.context.sitemapUrls)
+
+    const theshold = 1
+
+    const urlsNotOnLive = new Set()
+    // const newLocalUrls = new Set()
+
+    const liveSitemapXml = await axios( 'https://doesitarm.com/sitemap.xml' ).then( response => response.data )
+    const liveSitemap = parser.parse( liveSitemapXml )
+
+    // Store sitemap urls to context
+    const liveSitemapUrls = new Map( liveSitemap.urlset.url.map( tag => [ tag.loc, new URL( tag.loc )] ) )
+
+
+    for ( const localUrl of t.context.sitemapUrls ) {
+        const theoreticalLiveUrl = `https://doesitarm.com${ localUrl.pathname }`
+
+        if ( liveSitemapUrls.has( theoreticalLiveUrl ) ) {
+            liveSitemapUrls.delete( theoreticalLiveUrl )
+            continue
+        }
+
+        // localUrl is either: Missing or New
+        urlsNotOnLive.add( theoreticalLiveUrl )
+
+    }
+
+    const message = `${ urlsNotOnLive.size } new or missing from live and ${ liveSitemapUrls.size } not found locally`
+
+    if ( (urlsNotOnLive.size + liveSitemapUrls.size) >= theshold ) {
+        t.fail( message )
+    }
+
+    t.log( message )
+    t.pass()
+})
+
 test('All Category pages have valid FAQPage structured data', async (t) => {
 
     const categoryUrls = t.context.sitemapUrls.filter( url => url.pathname.startsWith('/kind/') )