From d7da5aedc2f7d414713d9c975db51ba8b3d2d6b7 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?J=C3=A9r=C3=A9my=20Benoist?=
 <j0k3r@users.noreply.github.com>
Date: Mon, 23 Oct 2017 14:16:00 +0200
Subject: [PATCH 01/10] Update wsj.com.txt (#351)

Fix author & date
---
 wsj.com.txt | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/wsj.com.txt b/wsj.com.txt
index 75913a0d..08d799f0 100644
--- a/wsj.com.txt
+++ b/wsj.com.txt
@@ -3,10 +3,10 @@ body: //div[@id='wsj-article-wrap']
 # is this still used?
 body: //div[@id='article_story_body']
 
-author: //h3[@class='byline']/a
+author: //meta[@name="author"]/@content
 # for slide show content
 body: //ul[@id='imageSlide']//li[@class='firstSlide']//img | (//div[@class='txt_body']//p)[1]
-date: //li[@class='dateStamp']/small
+date: //meta[@itemprop="dateCreated"]/@content
 
 strip_id_or_class: insetFullBracket
 strip_id_or_class: insettipBox

From b6150980d0b4a60afd09aa6e53d572c75c8330b9 Mon Sep 17 00:00:00 2001
From: Strubbl <github@linux4tw.de>
Date: Sat, 28 Oct 2017 14:14:31 +0200
Subject: [PATCH 02/10] fix heise.de title detection (#353)

Fixes the title detection of heise.de articles. This config used default
<title> tag, but this tag is polluted with the prefix string
"Druckversion - " and sometimes with suffixed e.g with " | heise online"
Therefore we replace those strings with nothing.
---
 heise.de.txt | 6 ++++++
 1 file changed, 6 insertions(+)

diff --git a/heise.de.txt b/heise.de.txt
index cc80203d..b6d36cb3 100644
--- a/heise.de.txt
+++ b/heise.de.txt
@@ -42,6 +42,12 @@ strip_id_or_class: ad_
 # Some optimizations
 replace_string(<h5>): <h2>
 replace_string(</h5>): </h2>
+replace_string(<title>Druckversion - ): <title>
+replace_string( | heise online</title>): </title>
+replace_string( | c't Magazin</title>): </title>
+replace_string( | Telepolis</title>): </title>
+replace_string( | heise Security</title>): </title>
+replace_string( | heise Autos</title>): </title>
 # this line breaks the parser
 #replace_string(<span class="bild_rechts" style="width:): <p "
 replace_string(<div class="heisebox">): <blockquote>

From b320dc5f5ebcfa726ea7e7532c4b035b658c36f5 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?J=C3=A9r=C3=A9my=20Benoist?=
 <j0k3r@users.noreply.github.com>
Date: Sat, 28 Oct 2017 22:42:35 +0200
Subject: [PATCH 03/10] Update lemonde.fr.txt (#354)

Still some crazy stuff from lemonde -_-
---
 lemonde.fr.txt | 5 +++++
 1 file changed, 5 insertions(+)

diff --git a/lemonde.fr.txt b/lemonde.fr.txt
index d5f9a9e9..d38b9b2e 100644
--- a/lemonde.fr.txt
+++ b/lemonde.fr.txt
@@ -17,6 +17,11 @@ body: //div[@id='articleBody']
 find_string: <a target='_blank' onclick='return false;' class='lien_interne conjug'
 replace_string: <input type='hidden' style='display:none;' 
 
+# Remove the insane cross-linking categories "Toute l’actualité"
+find_string: <a class="lien_interne rub"
+replace_string: <input type="hidden" style="display:none;"
+
 prune: no
 
 test_url: http://www.lemonde.fr/economie/article/2011/07/05/moody-s-abaisse-la-note-du-portugal-de-quatre-crans_1545237_3234.html
+test_url: http://www.lemonde.fr/big-browser/article/2017/10/27/assassinat-de-kennedy-ce-qu-on-a-appris-dans-les-documents-declassifies_5207029_4832693.html

From df67d13e3af29610e76e0cfa907d241ce1d01c91 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?J=C3=A9r=C3=A9my=20Benoist?=
 <j0k3r@users.noreply.github.com>
Date: Thu, 2 Nov 2017 15:54:02 +0200
Subject: [PATCH 04/10] Create iphon.fr.txt (#356)

---
 iphon.fr.txt | 5 +++++
 1 file changed, 5 insertions(+)
 create mode 100644 iphon.fr.txt

diff --git a/iphon.fr.txt b/iphon.fr.txt
new file mode 100644
index 00000000..1683ffb8
--- /dev/null
+++ b/iphon.fr.txt
@@ -0,0 +1,5 @@
+date: //meta[@name="date"]/@content
+author: //meta[@name="author"]/@content
+strip_id_or_class: follow
+
+test_url: http://www.iphon.fr/post/iphone-8-x-recharge-sans-fil-cable-simultanee-891682

From 33f7780e2811b7b5f110b81cd8d04a787dcca151 Mon Sep 17 00:00:00 2001
From: Vladimir Shabanov <vshabanoff@gmail.com>
Date: Thu, 2 Nov 2017 16:54:33 +0300
Subject: [PATCH 05/10] Update lenta.ru.txt (#355)

Banner stripped
---
 lenta.ru.txt | 2 ++
 1 file changed, 2 insertions(+)

diff --git a/lenta.ru.txt b/lenta.ru.txt
index 00c3d935..c2814794 100644
--- a/lenta.ru.txt
+++ b/lenta.ru.txt
@@ -1,7 +1,9 @@
 body: //div[@itemprop='articleBody'] | //img[@itemprop='image']
 strip_id_or_class: b-inline-topics-box
+strip: //iframe[@name='banner']
 
 # do not prune embedded videos
 prune: no
 
 test_url: https://lenta.ru/news/2016/07/08/weakpound/
+test_url: https://lenta.ru/news/2017/11/01/bigdata/

From 754d6f80ff2a8cff8e6172cfa984794a7f48210a Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?J=C3=A9r=C3=A9my=20Benoist?=
 <j0k3r@users.noreply.github.com>
Date: Fri, 3 Nov 2017 15:00:51 +0200
Subject: [PATCH 06/10] Update slate.fr.txt (#357)

media_legend: it removes image within the article, bad!
col-right: most shared articles, we don't care
---
 slate.fr.txt | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/slate.fr.txt b/slate.fr.txt
index 21cb8589..5e25ccf6 100644
--- a/slate.fr.txt
+++ b/slate.fr.txt
@@ -4,10 +4,10 @@
 
 body: //article[contains(concat(' ',normalize-space(@class),' '),' full_article ')]
 
-strip_id_or_class: media_legend
 strip_id_or_class: article_author
 strip_id_or_class: tag_articles
 strip_id_or_class: article_insert
+strip_id_or_class: col-right
 
 strip_id_or_class: tag
 strip_id_or_class: category

From 9d21d9397adb972486e323dded8064a67a3acc7d Mon Sep 17 00:00:00 2001
From: Andy Forceno <t4exanadu@gmail.com>
Date: Fri, 3 Nov 2017 20:33:15 -0400
Subject: [PATCH 07/10] Nature branch (#358)

* Nature.com site_config

* Updated config due to site redesign
---
 nature.com.txt | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/nature.com.txt b/nature.com.txt
index c832fe4e..5a91504c 100644
--- a/nature.com.txt
+++ b/nature.com.txt
@@ -1,2 +1,2 @@
-body: //div[@id='content']
+body: //div[@id='article']
 test_url: https://www.nature.com/npp/journal/v42/n11/full/npp201786a.html
\ No newline at end of file

From 011de47582655ccd0d903133a198782edc3df245 Mon Sep 17 00:00:00 2001
From: "FiveFilters.org" <fivefilters@fivefilters.org>
Date: Wed, 8 Nov 2017 14:00:23 +0000
Subject: [PATCH 08/10] Create sowetanlive.co.za.txt

---
 sowetanlive.co.za.txt | 5 +++++
 1 file changed, 5 insertions(+)
 create mode 100644 sowetanlive.co.za.txt

diff --git a/sowetanlive.co.za.txt b/sowetanlive.co.za.txt
new file mode 100644
index 00000000..34f9ec69
--- /dev/null
+++ b/sowetanlive.co.za.txt
@@ -0,0 +1,5 @@
+body: //div[contains(concat(' ',normalize-space(@class),' '),' article-widgets ')]
+strip_id_or_class: related_articles
+
+test_url: https://www.sowetanlive.co.za/rss/?publication=sowetan-live
+test_url: https://www.sowetanlive.co.za/sundayworld/news/2017-11-08-manana-should-pay-medical-bills-for-assault-victims-court-told/

From f22471c37ba6e833a443fa89935150db188887d2 Mon Sep 17 00:00:00 2001
From: "FiveFilters.org" <fivefilters@fivefilters.org>
Date: Wed, 8 Nov 2017 14:04:58 +0000
Subject: [PATCH 09/10] Update sowetanlive.co.za.txt

---
 sowetanlive.co.za.txt | 2 ++
 1 file changed, 2 insertions(+)

diff --git a/sowetanlive.co.za.txt b/sowetanlive.co.za.txt
index 34f9ec69..2a1983cc 100644
--- a/sowetanlive.co.za.txt
+++ b/sowetanlive.co.za.txt
@@ -1,5 +1,7 @@
 body: //div[contains(concat(' ',normalize-space(@class),' '),' article-widgets ')]
 strip_id_or_class: related_articles
+find_string:=s180"
+replace_string:=s750"
 
 test_url: https://www.sowetanlive.co.za/rss/?publication=sowetan-live
 test_url: https://www.sowetanlive.co.za/sundayworld/news/2017-11-08-manana-should-pay-medical-bills-for-assault-victims-court-told/

From 12ceb4c057cd2c86cf1328db1a5ce12518e0a8e3 Mon Sep 17 00:00:00 2001
From: "FiveFilters.org" <fivefilters@fivefilters.org>
Date: Wed, 8 Nov 2017 14:14:32 +0000
Subject: [PATCH 10/10] Create apple.news.txt

---
 apple.news.txt | 2 ++
 1 file changed, 2 insertions(+)
 create mode 100644 apple.news.txt

diff --git a/apple.news.txt b/apple.news.txt
new file mode 100644
index 00000000..7cb13d81
--- /dev/null
+++ b/apple.news.txt
@@ -0,0 +1,2 @@
+single_page_link: //p//a[contains(., 'Click here')]
+test_url: https://apple.news/AHQREjzH0Ts6iikKhNe6o8w