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 --- inc/3rdparty/site_config/standard/xfgjls.com.txt | 11 +++++++++++ 1 file changed, 11 insertions(+) create mode 100755 inc/3rdparty/site_config/standard/xfgjls.com.txt (limited to 'inc/3rdparty/site_config/standard/xfgjls.com.txt') diff --git a/inc/3rdparty/site_config/standard/xfgjls.com.txt b/inc/3rdparty/site_config/standard/xfgjls.com.txt new file mode 100755 index 00000000..2dc247a0 --- /dev/null +++ b/inc/3rdparty/site_config/standard/xfgjls.com.txt @@ -0,0 +1,11 @@ +# This filter is tested on: +# http://www.xfgjls.com/magazine/html/?131.html +# http://www.xfgjls.com/magazine/html/?170.html + +body://h3/following-sibling::div +title: //h3 +date: substring-before(//h3/following-sibling::div/p, ' ') +author: substring-before(substring-after(//h3/following-sibling::div/p, '作者:'), '来源') +wrap_in(strong)://span[contains(@style, "FONT-WEIGHT: bold")] +dissolve://span[@style="FONT-FAMILY: '宋体'; FONT-SIZE: 10.5pt; FONT-WEIGHT: bold; mso-spacerun: 'yes'"] +test_url: http://www.xfgjls.com/magazine/html/?170.html \ No newline at end of file -- cgit v1.2.3