update picofeed

39 files changed, 226 insertions, 163 deletions

diff --git a/vendor/autoload.php b/vendor/autoload.php
index 815d13376..e4ee3e08c 100644
--- a/vendor/autoload.php
+++ b/vendor/autoload.php
@@ -4,4 +4,4 @@
 
 require_once __DIR__ . '/composer' . '/autoload_real.php';
 
-return ComposerAutoloaderInitfee7a50d745041658cc0946610dc5951::getLoader();
+return ComposerAutoloaderInit878b2c9b24ed043c2ed72a45ff7ae283::getLoader();
diff --git a/vendor/composer/autoload_real.php b/vendor/composer/autoload_real.php
index 72636c6dd..3a2bd3c33 100644
--- a/vendor/composer/autoload_real.php
+++ b/vendor/composer/autoload_real.php
@@ -2,7 +2,7 @@
 
 // autoload_real.php @generated by Composer
 
-class ComposerAutoloaderInitfee7a50d745041658cc0946610dc5951
+class ComposerAutoloaderInit878b2c9b24ed043c2ed72a45ff7ae283
 {
     private static $loader;
 
@@ -19,9 +19,9 @@ class ComposerAutoloaderInitfee7a50d745041658cc0946610dc5951
             return self::$loader;
         }
 
-        spl_autoload_register(array('ComposerAutoloaderInitfee7a50d745041658cc0946610dc5951', 'loadClassLoader'), true, true);
+        spl_autoload_register(array('ComposerAutoloaderInit878b2c9b24ed043c2ed72a45ff7ae283', 'loadClassLoader'), true, true);
         self::$loader = $loader = new \Composer\Autoload\ClassLoader();
-        spl_autoload_unregister(array('ComposerAutoloaderInitfee7a50d745041658cc0946610dc5951', 'loadClassLoader'));
+        spl_autoload_unregister(array('ComposerAutoloaderInit878b2c9b24ed043c2ed72a45ff7ae283', 'loadClassLoader'));
 
         $includePaths = require __DIR__ . '/include_paths.php';
         array_push($includePaths, get_include_path());
@@ -46,14 +46,14 @@ class ComposerAutoloaderInitfee7a50d745041658cc0946610dc5951
 
         $includeFiles = require __DIR__ . '/autoload_files.php';
         foreach ($includeFiles as $file) {
-            composerRequirefee7a50d745041658cc0946610dc5951($file);
+            composerRequire878b2c9b24ed043c2ed72a45ff7ae283($file);
         }
 
         return $loader;
     }
 }
 
-function composerRequirefee7a50d745041658cc0946610dc5951($file)
+function composerRequire878b2c9b24ed043c2ed72a45ff7ae283($file)
 {
     require $file;
 }
diff --git a/vendor/composer/installed.json b/vendor/composer/installed.json
index 18cf0b2ed..1bfdaad8a 100644
--- a/vendor/composer/installed.json
+++ b/vendor/composer/installed.json
@@ -119,12 +119,12 @@
         "source": {
             "type": "git",
             "url": "https://github.com/fguillot/picoFeed.git",
-            "reference": "aa83e0c66525251cb6c6acab3babbc9e1879527b"
+            "reference": "035eed9d5d5b32441f44d59a9a75ea7e1c07a3aa"
         },
         "dist": {
             "type": "zip",
-            "url": "https://api.github.com/repos/fguillot/picoFeed/zipball/aa83e0c66525251cb6c6acab3babbc9e1879527b",
-            "reference": "aa83e0c66525251cb6c6acab3babbc9e1879527b",
+            "url": "https://api.github.com/repos/fguillot/picoFeed/zipball/035eed9d5d5b32441f44d59a9a75ea7e1c07a3aa",
+            "reference": "035eed9d5d5b32441f44d59a9a75ea7e1c07a3aa",
             "shasum": ""
         },
         "require": {
@@ -135,7 +135,7 @@
             "ext-xml": "*",
             "php": ">=5.3.0"
         },
-        "time": "2015-02-24 03:00:15",
+        "time": "2015-02-27 02:45:04",
         "type": "library",
         "installation-source": "dist",
         "autoload": {
diff --git a/vendor/fguillot/picofeed/docs/feed-parsing.markdown b/vendor/fguillot/picofeed/docs/feed-parsing.markdown
index d00e08364..6e7f2fdc2 100644
--- a/vendor/fguillot/picofeed/docs/feed-parsing.markdown
+++ b/vendor/fguillot/picofeed/docs/feed-parsing.markdown
@@ -45,23 +45,23 @@ Feed::id = tag:linuxfr.org,2005:/news
 Feed::title = LinuxFr.org : les dépêches
 Feed::feed_url = http://linuxfr.org/news.atom
 Feed::site_url = http://linuxfr.org/news
-Feed::date = 1415138079
 Feed::language = en-US
 Feed::description =
 Feed::logo =
-Feed::items = 15 items
+Feed::date = Thu, 26 Feb 15 09:33:08 +0100
 Feed::isRTL() = false
+Feed::items = 15 items
 ----
-Item::id = 38d8f48284fb03940cbb3aff9101089b81e44efb1281641bdd7c3e7e4bf3b0cd
-Item::title = openSUSE 13.2 : nouvelle version du caméléon disponible !
-Item::url = http://linuxfr.org/news/opensuse-13-2-nouvelle-version-du-cameleon-disponible
-Item::date = 1415122640
+Item::id = 56198c98ae852d21c369bfb5ffbc2ad13db2f3227236dde3e21ca1a9eb943faf
+Item::title = Les brevets logiciels : un frein à l'innovation et la recherche (un nouvel exemple aux États-Unis)
+Item::url = http://linuxfr.org/news/les-brevets-logiciels-un-frein-a-l-innovation-et-la-recherche-un-nouvel-exemple-aux-etats-unis
 Item::language = en-US
-Item::author = Syvolc
+Item::author = alenvers
 Item::enclosure_url =
 Item::enclosure_type =
+Item::date = Thu, 26 Feb 15 09:33:08 +0100
 Item::isRTL() = false
-Item::content = 18307 bytes
+Item::content = 6452 bytes
 ....
 ```
 
@@ -186,7 +186,7 @@ $feed->getId();              // Unique feed id
 $feed->getTitle();           // Feed title
 $feed->getFeedUrl();         // Feed url
 $feed->getSiteUrl();         // Website url
-$feed->getDate();            // Feed last updated date
+$feed->getDate();            // Feed last updated date (DateTime object)
 $feed->getLanguage();        // Feed language
 $feed->getDescription();     // Feed description
 $feed->getLogo();            // Feed logo (can be a large image, different from icon)
@@ -196,7 +196,7 @@ $feed->getItems();           // List of item objects
 $feed->items[0]->getId();                      // Item unique id (hash)
 $feed->items[0]->getTitle();                   // Item title
 $feed->items[0]->getUrl();                     // Item url
-$feed->items[0]->getDate();                    // Item published date (timestamp)
+$feed->items[0]->getDate();                    // Item published date (DateTime object)
 $feed->items[0]->getLanguage();                // Item language
 $feed->items[0]->getAuthor();                  // Item author
 $feed->items[0]->getEnclosureUrl();            // Enclosure url
diff --git a/vendor/fguillot/picofeed/docs/installation.markdown b/vendor/fguillot/picofeed/docs/installation.markdown
index f132b32cc..9bf1450d1 100644
--- a/vendor/fguillot/picofeed/docs/installation.markdown
+++ b/vendor/fguillot/picofeed/docs/installation.markdown
@@ -6,7 +6,8 @@ Versions
 
 - Development version: master
 - Available versions:
-    - v0.1.1 (stable)
+    - v0.1.2 (stable)
+    - v0.1.1
     - v0.1.0
     - v0.0.2
     - v0.0.1
@@ -21,7 +22,7 @@ Configure your `composer.json`:
 ```json
 {
     "require": {
-        "fguillot/picofeed": "0.1.1"
+        "fguillot/picofeed": "0.1.2"
     }
 }
 ```
@@ -29,7 +30,7 @@ Configure your `composer.json`:
 Or simply:
 
 ```bash
-composer require fguillot/picofeed:0.1.1
+composer require fguillot/picofeed:0.1.2
 ```
 
 And download the code:
diff --git a/vendor/fguillot/picofeed/lib/PicoFeed/Client/Grabber.php b/vendor/fguillot/picofeed/lib/PicoFeed/Client/Grabber.php
index 52f2f0bf1..fe4890400 100644
--- a/vendor/fguillot/picofeed/lib/PicoFeed/Client/Grabber.php
+++ b/vendor/fguillot/picofeed/lib/PicoFeed/Client/Grabber.php
@@ -123,8 +123,6 @@ class Grabber
      * @var array
      */
     private $stripTags = array(
-        'script',
-        'style',
         'nav',
         'header',
         'footer',
@@ -276,11 +274,11 @@ class Grabber
         }
 
         if ($this->html) {
+            $html_encoding = XmlParser::getEncodingFromMetaTag($this->html);
 
-            Logger::setMessage(get_called_class().': Fix encoding');
-            Logger::setMessage(get_called_class().': HTTP Encoding "'.$this->encoding.'"');
-
-            $this->html = Encoding::convert($this->html, $this->encoding);
+            // Encode everything in UTF-8
+            Logger::setMessage(get_called_class().': HTTP Encoding "'.$this->encoding.'" ; HTML Encoding "'.$html_encoding.'"');
+            $this->html = Encoding::convert($this->html, $html_encoding ?: $this->encoding);
             $this->html = Filter::stripHeadTags($this->html);
 
             Logger::setMessage(get_called_class().': Content length: '.strlen($this->html).' bytes');
diff --git a/vendor/fguillot/picofeed/lib/PicoFeed/Filter/Tag.php b/vendor/fguillot/picofeed/lib/PicoFeed/Filter/Tag.php
index 647b7352f..aa7efe4bd 100644
--- a/vendor/fguillot/picofeed/lib/PicoFeed/Filter/Tag.php
+++ b/vendor/fguillot/picofeed/lib/PicoFeed/Filter/Tag.php
@@ -158,6 +158,11 @@ class Tag
     public function removeBlacklistedTags($data)
     {
         $dom = XmlParser::getDomDocument($data);
+
+        if ($dom === false) {
+            return '';
+        }
+
         $xpath = new DOMXpath($dom);
 
         $nodes = $xpath->query(implode(' | ', $this->tag_blacklist));
diff --git a/vendor/fguillot/picofeed/lib/PicoFeed/Parser/Atom.php b/vendor/fguillot/picofeed/lib/PicoFeed/Parser/Atom.php
index 9f2f108d8..21566527e 100644
--- a/vendor/fguillot/picofeed/lib/PicoFeed/Parser/Atom.php
+++ b/vendor/fguillot/picofeed/lib/PicoFeed/Parser/Atom.php
@@ -131,7 +131,7 @@ class Atom extends Parser
      */
     public function findFeedDate(SimpleXMLElement $xml, Feed $feed)
     {
-        $feed->date = $this->date->getTimestamp((string) $xml->updated);
+        $feed->date = $this->date->getDateTime((string) $xml->updated);
     }
 
     /**
@@ -143,10 +143,15 @@ class Atom extends Parser
      */
     public function findItemDate(SimpleXMLElement $entry, Item $item)
     {
-        $published = isset($entry->published) ? $this->date->getTimestamp((string) $entry->published) : 0;
-        $updated = isset($entry->updated) ? $this->date->getTimestamp((string) $entry->updated) : 0;
+        $published = isset($entry->published) ? $this->date->getDateTime((string) $entry->published) : null;
+        $updated = isset($entry->updated) ? $this->date->getDateTime((string) $entry->updated) : null;
 
-        $item->date = max($published, $updated) ?: time();
+        if ($published !== null && $updated !== null) {
+            $item->date = max($published, $updated);
+        }
+        else {
+            $item->date = $updated ?: $published;
+        }
     }
 
     /**
diff --git a/vendor/fguillot/picofeed/lib/PicoFeed/Parser/DateParser.php b/vendor/fguillot/picofeed/lib/PicoFeed/Parser/DateParser.php
index 89f189e7a..92d0012ac 100644
--- a/vendor/fguillot/picofeed/lib/PicoFeed/Parser/DateParser.php
+++ b/vendor/fguillot/picofeed/lib/PicoFeed/Parser/DateParser.php
@@ -21,7 +21,7 @@ class DateParser
      */
     public $timezone = 'UTC';
 
-    /** 
+    /**
      * Supported formats [ 'format' => length ]
      *
      * @access public
@@ -60,9 +60,9 @@ class DateParser
      *
      * @access public
      * @param  string  $value  Original date format
-     * @return integer         Timestamp
+     * @return DateTime
      */
-    public function getTimestamp($value)
+    public function getDateTime($value)
     {
         $value = trim($value);
 
@@ -73,14 +73,13 @@ class DateParser
                 $truncated_value = substr($truncated_value, 0, $length);
             }
 
-            $timestamp = $this->getValidDate($format, $truncated_value);
-            if ($timestamp > 0) {
-                return $timestamp;
+            $date = $this->getValidDate($format, $truncated_value);
+            if ($date !== false) {
+                return $date;
             }
         }
 
-        $date = new DateTime('now', new DateTimeZone($this->timezone));
-        return $date->getTimestamp();
+        return $this->getCurrentDateTime();
     }
 
     /**
@@ -89,7 +88,7 @@ class DateParser
      * @access public
      * @param  string  $format   Date format
      * @param  string  $value    Original date value
-     * @return integer           Timestamp
+     * @return DateTime
      */
     public function getValidDate($format, $value)
     {
@@ -100,10 +99,21 @@ class DateParser
             $errors = DateTime::getLastErrors();
 
             if ($errors['error_count'] === 0 && $errors['warning_count'] === 0) {
-                return $date->getTimestamp();
+                return $date;
             }
         }
 
-        return 0;
+        return false;
+    }
+
+    /**
+     * Get the current datetime
+     *
+     * @access public
+     * @return DateTime
+     */
+    public function getCurrentDateTime()
+    {
+        return new DateTime('now', new DateTimeZone($this->timezone));
     }
 }
diff --git a/vendor/fguillot/picofeed/lib/PicoFeed/Parser/Feed.php b/vendor/fguillot/picofeed/lib/PicoFeed/Parser/Feed.php
index 74e9ab4da..93831e746 100644
--- a/vendor/fguillot/picofeed/lib/PicoFeed/Parser/Feed.php
+++ b/vendor/fguillot/picofeed/lib/PicoFeed/Parser/Feed.php
@@ -62,9 +62,9 @@ class Feed
      * Feed date
      *
      * @access public
-     * @var integer
+     * @var \DateTime
      */
-    public $date = 0;
+    public $date = null;
 
     /**
      * Feed language
@@ -100,10 +100,11 @@ class Feed
     {
         $output = '';
 
-        foreach (array('id', 'title', 'feed_url', 'site_url', 'date', 'language', 'description', 'logo') as $property) {
+        foreach (array('id', 'title', 'feed_url', 'site_url', 'language', 'description', 'logo') as $property) {
             $output .= 'Feed::'.$property.' = '.$this->$property.PHP_EOL;
         }
 
+        $output .= 'Feed::date = '.$this->date->format(DATE_RFC822).PHP_EOL;
         $output .= 'Feed::isRTL() = '.($this->isRTL() ? 'true' : 'false').PHP_EOL;
         $output .= 'Feed::items = '.count($this->items).' items'.PHP_EOL;
 
diff --git a/vendor/fguillot/picofeed/lib/PicoFeed/Parser/Item.php b/vendor/fguillot/picofeed/lib/PicoFeed/Parser/Item.php
index 3642cccea..1585131c7 100644
--- a/vendor/fguillot/picofeed/lib/PicoFeed/Parser/Item.php
+++ b/vendor/fguillot/picofeed/lib/PicoFeed/Parser/Item.php
@@ -63,9 +63,9 @@ class Item
      * Item date
      *
      * @access public
-     * @var integer
+     * @var \DateTime
      */
-    public $date = 0;
+    public $date = null;
 
     /**
      * Item content
@@ -109,10 +109,11 @@ class Item
     {
         $output = '';
 
-        foreach (array('id', 'title', 'url', 'date', 'language', 'author', 'enclosure_url', 'enclosure_type') as $property) {
+        foreach (array('id', 'title', 'url', 'language', 'author', 'enclosure_url', 'enclosure_type') as $property) {
             $output .= 'Item::'.$property.' = '.$this->$property.PHP_EOL;
         }
 
+        $output .= 'Item::date = '.$this->date->format(DATE_RFC822).PHP_EOL;
         $output .= 'Item::isRTL() = '.($this->isRTL() ? 'true' : 'false').PHP_EOL;
         $output .= 'Item::content = '.strlen($this->content).' bytes'.PHP_EOL;
 
diff --git a/vendor/fguillot/picofeed/lib/PicoFeed/Parser/Rss10.php b/vendor/fguillot/picofeed/lib/PicoFeed/Parser/Rss10.php
index da57327ed..16c66a016 100644
--- a/vendor/fguillot/picofeed/lib/PicoFeed/Parser/Rss10.php
+++ b/vendor/fguillot/picofeed/lib/PicoFeed/Parser/Rss10.php
@@ -33,7 +33,7 @@ class Rss10 extends Rss20
      */
     public function findFeedDate(SimpleXMLElement $xml, Feed $feed)
     {
-        $feed->date = $this->date->getTimestamp(XmlParser::getNamespaceValue($xml->channel, $this->namespaces, 'date'));
+        $feed->date = $this->date->getDateTime(XmlParser::getNamespaceValue($xml->channel, $this->namespaces, 'date'));
     }
 
     /**
diff --git a/vendor/fguillot/picofeed/lib/PicoFeed/Parser/Rss20.php b/vendor/fguillot/picofeed/lib/PicoFeed/Parser/Rss20.php
index 114308f51..b5fb09580 100644
--- a/vendor/fguillot/picofeed/lib/PicoFeed/Parser/Rss20.php
+++ b/vendor/fguillot/picofeed/lib/PicoFeed/Parser/Rss20.php
@@ -134,7 +134,7 @@ class Rss20 extends Parser
     public function findFeedDate(SimpleXMLElement $xml, Feed $feed)
     {
         $date = isset($xml->channel->pubDate) ? $xml->channel->pubDate : $xml->channel->lastBuildDate;
-        $feed->date = $this->date->getTimestamp((string) $date);
+        $feed->date = $this->date->getDateTime((string) $date);
     }
 
     /**
@@ -156,7 +156,7 @@ class Rss20 extends Parser
             $date = (string) $entry->pubDate;
         }
 
-        $item->date = $this->date->getTimestamp($date);
+        $item->date = $this->date->getDateTime($date);
     }
 
     /**
diff --git a/vendor/fguillot/picofeed/lib/PicoFeed/Parser/XmlParser.php b/vendor/fguillot/picofeed/lib/PicoFeed/Parser/XmlParser.php
index 73a4a43a6..2c68c50a5 100644
--- a/vendor/fguillot/picofeed/lib/PicoFeed/Parser/XmlParser.php
+++ b/vendor/fguillot/picofeed/lib/PicoFeed/Parser/XmlParser.php
@@ -90,10 +90,14 @@ class XmlParser
      * @static
      * @access public
      * @param  string   $input   XML content
-     * @return \DOMNode
+     * @return \DOMNDocument
      */
     public static function getDomDocument($input)
     {
+        if (empty($input)) {
+            return false;
+        }
+
         $dom = self::scanInput($input, function ($in) {
             $dom = new DomDocument;
             $dom->loadXml($in, LIBXML_NONET);
@@ -208,6 +212,49 @@ class XmlParser
     }
 
     /**
+     * Extract charset from meta tag
+     *
+     * @static
+     * @access public
+     * @param  string  $data  meta tag content
+     * @return string
+     */
+    public static function findCharset($data)
+    {
+        $result = explode('charset=', $data);
+        return isset($result[1]) ? $result[1] : $data;
+    }
+
+    /**
+     * Get the encoding from a xml tag
+     *
+     * @static
+     * @access public
+     * @param  string  $data  Input data
+     * @return string
+     */
+    public static function getEncodingFromMetaTag($data)
+    {
+        $encoding = '';
+
+        $dom = static::getHtmlDocument($data);
+        $xpath = new DOMXPath($dom);
+
+        $tags = array(
+            '/html/head/meta[translate(@http-equiv, "CENOPTY", "cenopty")="content-type"]/@content', //HTML4, convert upper to lower-case
+            '/html/head/meta/@charset', //HTML5
+        );
+
+        $nodes = $xpath->query(implode(' | ', $tags));
+
+        foreach ($nodes as $node) {
+            $encoding = static::findCharset($node->nodeValue);
+        }
+
+        return $encoding;
+    }
+
+    /**
      * Get xml:lang value
      *
      * @static
diff --git a/vendor/fguillot/picofeed/lib/PicoFeed/Rules/.wired.com.php b/vendor/fguillot/picofeed/lib/PicoFeed/Rules/.wired.com.php
index 6e25f2c2e..32c0475d3 100644
--- a/vendor/fguillot/picofeed/lib/PicoFeed/Rules/.wired.com.php
+++ b/vendor/fguillot/picofeed/lib/PicoFeed/Rules/.wired.com.php
@@ -5,8 +5,6 @@ return array(
          '//div[@class="entry"]',
     ),
     'strip' => array(
-        '//script',
-        '//style',
         '//*[@id="linker_widget"]',
         '//*[contains(@class, "bio")]',
         '//*[contains(@class, "entry-footer")]',
diff --git a/vendor/fguillot/picofeed/lib/PicoFeed/Rules/01net.com.php b/vendor/fguillot/picofeed/lib/PicoFeed/Rules/01net.com.php
index 85c0e7eab..615ad7702 100644
--- a/vendor/fguillot/picofeed/lib/PicoFeed/Rules/01net.com.php
+++ b/vendor/fguillot/picofeed/lib/PicoFeed/Rules/01net.com.php
@@ -5,7 +5,6 @@ return array(
         '//div[@class="article_ventre_box"]',
     ),
     'strip' => array(
-        '//script',
         '//link',
         '//*[contains(@class, "article_navigation")]',
         '//h1',
diff --git a/vendor/fguillot/picofeed/lib/PicoFeed/Rules/journaldugeek.com.php b/vendor/fguillot/picofeed/lib/PicoFeed/Rules/journaldugeek.com.php
index 72de69135..ad0d67a1c 100644
--- a/vendor/fguillot/picofeed/lib/PicoFeed/Rules/journaldugeek.com.php
+++ b/vendor/fguillot/picofeed/lib/PicoFeed/Rules/journaldugeek.com.php
@@ -3,8 +3,5 @@ return array(
     'test_url' => 'http://www./2014/05/20/le-playstation-now-arrive-en-beta-fermee-aux-etats-unis/',
     'body' => array(
         '//div[@class="post-content"]',
-    ),
-    'strip' => array(
-        '//style'
     )
 );
\ No newline at end of file
diff --git a/vendor/fguillot/picofeed/lib/PicoFeed/Rules/lejapon.fr.php b/vendor/fguillot/picofeed/lib/PicoFeed/Rules/lejapon.fr.php
index 8066c3d80..fbc2e53bc 100644
--- a/vendor/fguillot/picofeed/lib/PicoFeed/Rules/lejapon.fr.php
+++ b/vendor/fguillot/picofeed/lib/PicoFeed/Rules/lejapon.fr.php
@@ -5,7 +5,6 @@ return array(
         '//div[@class="entry"]'
     ),
     'strip' => array(
-        '//script',
         '//*[contains(@class, "addthis_toolbox")]',
         '//*[contains(@class, "addthis_default_style")]',
         '//*[@class="navigation small"]',
diff --git a/vendor/fguillot/picofeed/lib/PicoFeed/Rules/smallhousebliss.com.php b/vendor/fguillot/picofeed/lib/PicoFeed/Rules/smallhousebliss.com.php
index 395cadd8c..8bea4fbf9 100644
--- a/vendor/fguillot/picofeed/lib/PicoFeed/Rules/smallhousebliss.com.php
+++ b/vendor/fguillot/picofeed/lib/PicoFeed/Rules/smallhousebliss.com.php
@@ -5,8 +5,6 @@ return array(
         '//div[@class="post-content"]',
     ),
     'strip' => array(
-        '//style',
-        '//script',
         '//*[contains(@class, "gallery")]',
         '//*[contains(@class, "share")]',
         '//*[contains(@class, "wpcnt")]',
diff --git a/vendor/fguillot/picofeed/lib/PicoFeed/Rules/spiegel.de.php b/vendor/fguillot/picofeed/lib/PicoFeed/Rules/spiegel.de.ph