From 4e067ceabd705201a16b4c92cf4b23f3b990326c Mon Sep 17 00:00:00 2001 From: =?UTF-8?q?Nicolas=20L=C5=93uillet?= Date: Sun, 13 Jul 2014 10:15:40 +0200 Subject: updated specific configuration for parsing --- .../site_config/standard/elektroniknet.de.txt | 50 +++++++++++----------- 1 file changed, 25 insertions(+), 25 deletions(-) mode change 100644 => 100755 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 old mode 100644 new mode 100755 index 07664719..56fba5ff --- a/inc/3rdparty/site_config/standard/elektroniknet.de.txt +++ b/inc/3rdparty/site_config/standard/elektroniknet.de.txt @@ -1,27 +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;'] +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