From 4e067ceabd705201a16b4c92cf4b23f3b990326c Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Nicolas=20L=C5=93uillet?= <nicolas@loeuillet.org>
Date: Sun, 13 Jul 2014 10:15:40 +0200
Subject: updated specific configuration for parsing

---
 inc/3rdparty/site_config/standard/faz.net.txt | 66 +++++++++++++++------------
 1 file changed, 36 insertions(+), 30 deletions(-)
 mode change 100644 => 100755 inc/3rdparty/site_config/standard/faz.net.txt

(limited to 'inc/3rdparty/site_config/standard/faz.net.txt')

diff --git a/inc/3rdparty/site_config/standard/faz.net.txt b/inc/3rdparty/site_config/standard/faz.net.txt
old mode 100644
new mode 100755
index 4fe5968b..d087d2aa
--- a/inc/3rdparty/site_config/standard/faz.net.txt
+++ b/inc/3rdparty/site_config/standard/faz.net.txt
@@ -1,30 +1,36 @@
-# Title
-title: //p[@class='Content HeadlineShort']
-
-# Authors
-# some are known and have a link, others don't
-author: substring-after(//span[@class='Autor'], 'Von')
-
-# Date
-date: //span[@class='Datum']
-
-# Body
-body: //div[@class='Artikel']
-
-# Removements before body text
-strip: //div[@class='Breadcrumbs']
-strip: //div[@class='QuickSearchBox']
-strip: //div[@class='FAZArtikelEinleitung']
-strip: //div[@class='FAZArtikelReiter']
-strip: //div[@class='clear']
-
-# General removements
-strip: //span[@class='Bildnachweis']
-
-# Removements after body text
-strip: //div[@class='ArtikelAbbinder']
-strip: //div[@class='ArtikelKommentieren Artikelfuss GETS;tk;boxen.top-lesermeinungen;tp;content']
-strip: //div[@class='FAZArtikelKommentare FAZArtikelContent']
-strip: //div[@class='FAZArtikelFunktionen']
-strip: //div[@id='FAZContentRight']
-test_url: http://www.faz.net/aktuell/gesellschaft/ehe-haltbarkeitsformel-verliebe-dich-oft-verlobe-dich-selten-heirate-vielleicht-11685306.html
\ No newline at end of file
+# Title
+title: //p[@class='Content HeadlineShort']
+
+# Authors
+# some are known and have a link, others don't
+author: substring-after(//span[@class='Autor'], 'Von')
+
+# Date
+date: //span[@class='Datum']
+
+# Body
+body: //div[@class='Artikel']
+
+# Removements before body text
+strip: //div[@class='Breadcrumbs']
+strip: //div[@class='QuickSearchBox']
+strip: //div[@class='FAZArtikelEinleitung']
+strip: //div[@class='FAZArtikelReiter']
+strip: //div[@class='clear']
+
+# General removements
+strip: //span[@class='Bildnachweis']
+strip: //img[@class='MediaIcon']
+strip: //div[@class='ArtikelMediaLink']
+dissolve: //a[img]
+
+# Removements after body text
+strip: //div[@class='ArtikelAbbinder']
+strip: //div[@class='ArtikelKommentieren Artikelfuss GETS;tk;boxen.top-lesermeinungen;tp;content']
+strip: //div[@class='FAZArtikelKommentare FAZArtikelContent']
+strip: //div[@class='FAZArtikelFunktionen']
+strip: //div[@id='FAZContentRight']
+
+# Fix picture captions
+wrap_in(small): //span[@class='Bildunterschrift']/text()
+test_url: http://www.faz.net/aktuell/feuilleton/zum-tod-von-margaret-thatcher-die-reizfigur-12141919.html#Drucken
\ No newline at end of file
-- 
cgit v1.2.3