From d475e908aaed8d6233f26c121bde4b0152955173 Mon Sep 17 00:00:00 2001
From: Dominik Hayon <dominikhayon@gmail.com>
Date: Mon, 1 Jul 2024 21:48:00 +0200
Subject: [PATCH] poc for id based genios search

---
 src/extractor.ts | 25 ++++++++++++++++++-------
 src/sites.ts     | 31 +++++++++++++++++++++++++++++--
 src/types.ts     |  1 +
 3 files changed, 48 insertions(+), 9 deletions(-)

diff --git a/src/extractor.ts b/src/extractor.ts
index 7733055..5733e2c 100644
--- a/src/extractor.ts
+++ b/src/extractor.ts
@@ -143,16 +143,27 @@ class Extractor implements ExtractorInterface {
   extractArticleInfo (): ArticleInfo {
     const articleInfoSelectors = ['query', 'edition', 'date']
     const articleInfo: RawArticleInfo = {}
-    for (const key of articleInfoSelectors) {
-      if (this.site.selectors[key]) {
-        const selector = this.site.selectors[key]
-        let result = this.runSelectorQuery(selector)
-        if (result instanceof window.HTMLElement) {
-          result = result.innerText
+
+    if (this.site.extractId) {
+      const idFromCustomFunc = this.site.extractId()
+      if (idFromCustomFunc) {
+        articleInfo.query = idFromCustomFunc
+      }
+    }
+
+    if (!articleInfo.query) {
+      for (const key of articleInfoSelectors) {
+        if (this.site.selectors[key]) {
+          const selector = this.site.selectors[key]
+          let result = this.runSelectorQuery(selector)
+          if (result instanceof window.HTMLElement) {
+            result = result.innerText
+          }
+          articleInfo[key] = result
         }
-        articleInfo[key] = result
       }
     }
+
     return {
       query: articleInfo.query,
       edition: articleInfo.edition,
diff --git a/src/sites.ts b/src/sites.ts
index c3d7a34..20de653 100644
--- a/src/sites.ts
+++ b/src/sites.ts
@@ -94,7 +94,7 @@ const sites: Sites = {
       {
         url: 'https://www.spiegel.de/politik/deutschland/klara-geywitz-ueber-sanierungspflicht-von-immobilien-neuen-wohnraum-und-fluechtlinge-a-6aeb319e-fc25-4efa-a0cf-66e10ed49969',
         selectors: {
-          query: 'nicht ohne Ordnungsrecht gehen wenn wir die Klimaziele erreichen wollen«'
+          query: '6aeb319e-fc25-4efa-a0cf-66e10ed49969'
         }
       }
     ],
@@ -104,6 +104,11 @@ const sites: Sites = {
       main: 'article section.relative',
       paywall: "div[data-component='Paywall'], div[data-target-id='paywall']"
     },
+    extractId: () => {
+      const url = window.location.href
+      const match = url.match(/-([a-f0-9]{8}-[a-f0-9]{4}-[a-f0-9]{4}-[a-f0-9]{4}-[a-f0-9]{12})/)
+      return match ? match[1] : null
+    },
     mimic: (content) => {
       return `
       <div class="lg:mt-32 md:mt-32 sm:mt-24 md:mb-48 lg:mb-48 sm:mb-32">
@@ -121,6 +126,14 @@ const sites: Sites = {
     }
   },
   'www.manager-magazin.de': {
+    examples: [
+      {
+        url: 'https://www.manager-magazin.de/unternehmen/fussball-em-2024-martin-kallen-der-unbekannte-milliardenmacher-hinter-der-em-a-f89271c7-d048-49b4-bcb6-a974cc7eff26',
+        selectors: {
+          query: 'f89271c7-d048-49b4-bcb6-a974cc7eff26'
+        }
+      }
+    ],
     selectors: {
       query: makeQueryFunc('header h2~div:nth-of-type(1)'),
       date: 'time',
@@ -128,6 +141,11 @@ const sites: Sites = {
       paywall: '[data-area="paywall"]',
       main: '[data-area="body"]'
     },
+    extractId: () => {
+      const url = window.location.href
+      const match = url.match(/-([a-f0-9]{8}-[a-f0-9]{4}-[a-f0-9]{4}-[a-f0-9]{4}-[a-f0-9]{12})/)
+      return match ? match[1] : null
+    },
     source: 'genios.de',
     sourceParams: {
       dbShortcut: 'MM,MMAG'
@@ -453,10 +471,19 @@ const sites: Sites = {
       {
         url: 'https://www.wiwo.de/my/unternehmen/industrie/mischkonzern-zeppelin-ein-ausschluss-russlands-aus-swift-wuerde-eine-weltwirtschaftskrise-ausloesen/28091946.html',
         selectors: {
-          query: 'Mischkonzern Zeppelin vertreibt unter anderem US-amerikanische Baumaschinen in Russland und der Ukraine Ein'
+          query: 'WW_28091946'
         }
       }
     ],
+    extractId: () => {
+      const url = window.location.href
+      const id = url.match('/[0-9]{10}.html/')
+      if (id[1]) {
+        return `WW_${id[1]}`
+      }
+
+      return null
+    },
     selectors: {
       query: makeQueryFunc('.c-leadtext', false),
       main: '.o-article__content',
diff --git a/src/types.ts b/src/types.ts
index fdb14cd..7500736 100644
--- a/src/types.ts
+++ b/src/types.ts
@@ -80,6 +80,7 @@ export interface PartialSite {
   dateRange?: DateRange
   testSetup?: (page: PlaywrightPage) => Promise<void>
   examples?: TestExample[]
+  extractId?: () => string | null
 }
 
 export interface Site extends PartialSite {