From 7aa903883bdcc17793cdc255fc258649ca0b2631 Mon Sep 17 00:00:00 2001 From: alpcentaur Date: Fri, 3 Nov 2023 12:23:04 +0000 Subject: [PATCH] update to config.yaml --- spiders/config.yaml | 13 ++++++++----- spiders/fdb_spider.py | 2 +- 2 files changed, 9 insertions(+), 6 deletions(-) diff --git a/spiders/config.yaml b/spiders/config.yaml index da83c38..fc060a3 100644 --- a/spiders/config.yaml +++ b/spiders/config.yaml @@ -6,11 +6,14 @@ foerderinfo.bund.de: domain: 'http://foerderinfo.bund.de' - entryList: - link: '' - parent: '//html//body//form//table//tr//td//table//tr' - child-name: '//td//a/text()' - child-link: '//td//a/@href' + entry-list: + link1: 'https://www.foerderinfo.bund.de/SiteGlobals/Forms/foerderinfo/expertensuche/Servicesuche_Formular.html#searchResults' + link2: 'https://www.foerderinfo.bund.de/SiteGlobals/Forms/foerderinfo/expertensuche/Servicesuche_Formular.html?gtp=33498_list%253D2#searchResults' + link3: 'https://www.foerderinfo.bund.de/SiteGlobals/Forms/foerderinfo/expertensuche/Servicesuche_Formular.html?gtp=33498_list%253D3#searchResults' + iteration-var-list: [1,1 + parent: '//html//body//form//table//tr//td//table//tr' + child-name: '//td//a/text()' + child-link: '//td//a/@href' member: info-1: parent: '//html//body//form//table' diff --git a/spiders/fdb_spider.py b/spiders/fdb_spider.py index 13e4635..8a5f397 100644 --- a/spiders/fdb_spider.py +++ b/spiders/fdb_spider.py @@ -35,7 +35,7 @@ class fdb_spider(object): ) try: entry_list_link = entry_list.get("link") - except Exception as e: + except Exception as e: print( "No entryListLink defined in config.yaml - the original error message is:", e,