From bc2b5e7c3d3538eb397dcc99f627092e4b703add Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Fr=C3=A9d=C3=A9ric=20Guillot?= <contact@fredericguillot.com>
Date: Mon, 30 Sep 2013 22:15:18 -0400
Subject: [PATCH] Add new content grabber rules

---
 README.markdown                         |  4 ++++
 vendor/PicoFeed/Grabber.php             | 18 ++++++++++++++++--
 vendor/PicoFeed/Rules/.phoronix.com.php |  9 +++++++++
 vendor/PicoFeed/Rules/lists.php         |  9 +++++++++
 vendor/PicoFeed/Rules/medium.com.php    |  9 +++++++++
 vendor/PicoFeed/Rules/pastebin.com.php  |  9 +++++++++
 6 files changed, 56 insertions(+), 2 deletions(-)
 create mode 100644 vendor/PicoFeed/Rules/.phoronix.com.php
 create mode 100644 vendor/PicoFeed/Rules/lists.php
 create mode 100644 vendor/PicoFeed/Rules/medium.com.php
 create mode 100644 vendor/PicoFeed/Rules/pastebin.com.php

diff --git a/README.markdown b/README.markdown
index cf1723d..8f5429a 100644
--- a/README.markdown
+++ b/README.markdown
@@ -344,6 +344,7 @@ Don't forget to send a pull request or a ticket to share your contribution with
 - *.blog.lemonde.fr
 - *.blog.nytimes.com
 - *.nytimes.com
+- *.phoronix.com
 - *.slate.com
 - *.theguardian.com
 - *.wikipedia.org
@@ -351,6 +352,9 @@ Don't forget to send a pull request or a ticket to share your contribution with
 - *.wsj.com
 - github.com
 - lifehacker.com
+- lists.*
+- medium.com
+- pastebin.com
 - plus.google.com
 - rue89.com
 - smallhousebliss.com
diff --git a/vendor/PicoFeed/Grabber.php b/vendor/PicoFeed/Grabber.php
index 5483d82..32a6a9a 100644
--- a/vendor/PicoFeed/Grabber.php
+++ b/vendor/PicoFeed/Grabber.php
@@ -25,6 +25,8 @@ class Grabber
         'post_content',
         'entry-content',
         'main-content',
+        'story_content',
+        'storycontent',
         'entryBox',
         'entrytext',
         'comic',
@@ -46,7 +48,12 @@ class Grabber
         'nav',
         'header',
         'social',
+        'tag',
+        'metadata',
         'entry-utility',
+        'related-posts',
+        'tweet',
+        'categories',
     );
 
     public $stripTags = array(
@@ -56,6 +63,7 @@ class Grabber
         'header',
         'footer',
         'aside',
+        'form',
     );
 
 
@@ -114,8 +122,14 @@ class Grabber
         $hostname = parse_url($this->url, PHP_URL_HOST);
         $files = array($hostname);
 
-        if (substr($hostname, 0, 4) == 'www.') $files[] = substr($hostname, 4);
-        if (($pos = strpos($hostname, '.')) !== false) $files[] = substr($hostname, $pos);
+        if (substr($hostname, 0, 4) == 'www.') {
+            $files[] = substr($hostname, 4);
+        }
+
+        if (($pos = strpos($hostname, '.')) !== false) {
+            $files[] = substr($hostname, $pos);
+            $files[] = substr($hostname, 0, $pos);
+        }
 
         foreach ($files as $file) {
 
diff --git a/vendor/PicoFeed/Rules/.phoronix.com.php b/vendor/PicoFeed/Rules/.phoronix.com.php
new file mode 100644
index 0000000..0d10eff
--- /dev/null
+++ b/vendor/PicoFeed/Rules/.phoronix.com.php
@@ -0,0 +1,9 @@
+<?php
+return array(
+    'test_url' => 'http://www.phoronix.com/scan.php?page=article&item=amazon_ec2_bare&num=1',
+    'body' => array(
+        '//article[@class="KonaBody"]',
+    ),
+    'strip' => array(
+    )
+);
diff --git a/vendor/PicoFeed/Rules/lists.php b/vendor/PicoFeed/Rules/lists.php
new file mode 100644
index 0000000..fb9c8d0
--- /dev/null
+++ b/vendor/PicoFeed/Rules/lists.php
@@ -0,0 +1,9 @@
+<?php
+return array(
+    'test_url' => 'http://lists.freebsd.org/pipermail/freebsd-announce/2013-September/001504.html',
+    'body' => array(
+        '//pre',
+    ),
+    'strip' => array(
+    )
+);
diff --git a/vendor/PicoFeed/Rules/medium.com.php b/vendor/PicoFeed/Rules/medium.com.php
new file mode 100644
index 0000000..79ed5bc
--- /dev/null
+++ b/vendor/PicoFeed/Rules/medium.com.php
@@ -0,0 +1,9 @@
+<?php
+return array(
+    'test_url' => 'https://medium.com/lessons-learned/917b8b63ae3e',
+    'body' => array(
+        '//div[contains(@class, "post-field body")]',
+    ),
+    'strip' => array(
+    )
+);
diff --git a/vendor/PicoFeed/Rules/pastebin.com.php b/vendor/PicoFeed/Rules/pastebin.com.php
new file mode 100644
index 0000000..9a576f7
--- /dev/null
+++ b/vendor/PicoFeed/Rules/pastebin.com.php
@@ -0,0 +1,9 @@
+<?php
+return array(
+    'test_url' => 'http://pastebin.com/ed1pP9Ak',
+    'body' => array(
+        '//div[@class="text"]',
+    ),
+    'strip' => array(
+    )
+);