From acdbfd54cf3287b070a414a174c1217a9ac220a1 Mon Sep 17 00:00:00 2001 From: Christian Kellner Date: Sun, 10 Jun 2018 10:50:45 +0200 Subject: [PATCH] fixing immowelt crawler --- lib/provider/immowelt.js | 7 +++---- test/immowelt.test.js | 4 +++- 2 files changed, 6 insertions(+), 5 deletions(-) diff --git a/lib/provider/immowelt.js b/lib/provider/immowelt.js index 5c352e2..5e6f309 100755 --- a/lib/provider/immowelt.js +++ b/lib/provider/immowelt.js @@ -3,7 +3,7 @@ const config = require('../../conf/config.json'); const utils = require('../utils'); function normalize(o) { - const size = o.size.split('Wohnfläche')[0]; + const size = o.size.split('Wohnfläche')[1].replace(' (ca.) ', ''); const address = o.address; return Object.assign(o, { size, address }); @@ -24,11 +24,10 @@ const immowelt = { crawlFields: { id: '@data-estateid | int', price: '.hardfacts_3 strong | removeNewline | trim', - size: '.hardfacts_3 div:nth-child(2):not(.hardfactlabel)| removeNewline | trim', + size: '.js-object.listitem_wrap .hardfacts_3 div:nth-child(2)| removeNewline | trim', title: '.listcontent.clear h2', link: 'a@href', - description: '.listconten_offset .listmerkmale| removeNewline | trim', - address: '.listconten_offset .listlocation| removeNewline | trim' + address: '.listcontent .details .listlocation| removeNewline | trim' }, paginate: '#pnlPaging #nlbPlus@href', normalize: normalize, diff --git a/test/immowelt.test.js b/test/immowelt.test.js index cb70691..f787961 100644 --- a/test/immowelt.test.js +++ b/test/immowelt.test.js @@ -41,7 +41,9 @@ describe('#immowelt testsuite()', () => { immoweltDbContent.immowelt[idx] ); expect(notify.price).that.does.include('€'); - expect(notify.size).that.does.include('m²'); + if(notify.size.trim().toLowerCase() !== 'k.a.') { + expect(notify.size).that.does.include('m²'); + } expect(notify.title).to.be.not.empty; expect(notify.link).that.does.include('https://www.immowelt.de'); expect(notify.address).to.be.not.empty;