From ac4d114214d820b20e18518a2dbc809337e39043 Mon Sep 17 00:00:00 2001 From: =?UTF-8?q?Nicolas=20L=C5=93uillet?= Date: Fri, 6 Dec 2013 10:13:03 +0100 Subject: [add] new specific configuration files --- .../site_config/standard/elektroniknet.de.txt | 27 ++++++++++++++++++++++ 1 file changed, 27 insertions(+) create mode 100644 inc/3rdparty/site_config/standard/elektroniknet.de.txt (limited to 'inc/3rdparty/site_config/standard/elektroniknet.de.txt') diff --git a/inc/3rdparty/site_config/standard/elektroniknet.de.txt b/inc/3rdparty/site_config/standard/elektroniknet.de.txt new file mode 100644 index 00000000..07664719 --- /dev/null +++ b/inc/3rdparty/site_config/standard/elektroniknet.de.txt @@ -0,0 +1,27 @@ +title: //h1 +date: //div[@class='datum'] +single_page_link: //a[contains(@href, '?type=99')] + +# this hack preserves the intro text, because it would be striped otherwise if the title is set to //h1 +dissolve: //div[@class='artikelMeldung'] + + +strip_id_or_class: anzeige +strip_id_or_class: top_page_navigation +strip_id_or_class: cr_image_container +strip_id_or_class: cr_image_reference +strip_id_or_class: cr_image_icon +strip_id_or_class: _close_txt +strip_id_or_class: _close_ico +strip_id_or_class: clearer + +strip://h1 +strip://h6 +strip://div[contains(@id, 'plista')] +strip://img[contains(@id,'tiny')] +strip://img[@class='cr_image'] + +# strip url at the top +strip: //p[@style='font-size: 10px;'] + +test_url: http://www.elektroniknet.de/automotive/technik-know-how/sicherheitselektronik/article/87717/0/Besser_als_die_Wirklichkeit/ \ No newline at end of file -- cgit v1.2.3