Improve content grabber

model.php

@@ -367,11 +367,9 @@ function update_feed_cache_infos($feed_id, $last_modified, $etag)
 function parse_content_with_readability($content, $url)
 {
     require_once 'vendor/Readability/Readability.php';
-    require_once 'vendor/PicoFeed/Encoding.php';
 
     if (! empty($content)) {
 
-        $content = \PicoFeed\Encoding::toUTF8($content);
         $readability = new \Readability($content, $url);
 
         if ($readability->init()) {
@@ -400,13 +398,14 @@ function download_content($url)
         // Try first with PicoFeed grabber and with Readability after
         $grabber = new \PicoFeed\Grabber($url);
         $grabber->html = $html;
+        $content = '';
 
         if ($grabber->parse()) {
             $content = $grabber->content;
         }
 
         if (empty($content)) {
-            $content = parse_content_with_readability($html, $url);
+            $content = parse_content_with_readability($grabber->html, $url);
         }
 
         // Filter content

vendor/PicoFeed/Filter.php

@@ -464,4 +464,10 @@ class Filter
 
         return $data;
     }
+
+
+    public static function stripMetaTags($data)
+    {
+        return preg_replace('/<meta\s.*?\/>/is', '', $data);
+    }
 }

vendor/PicoFeed/Grabber.php

@@ -5,6 +5,7 @@ namespace PicoFeed;
 require_once __DIR__.'/Client.php';
 require_once __DIR__.'/Encoding.php';
 require_once __DIR__.'/Logging.php';
+require_once __DIR__.'/Filter.php';
 
 class Grabber
 {
@@ -20,6 +21,7 @@ class Grabber
         'articlecontent',
         'articlePage',
         'post-content',
+        'entry-content',
         'content',
         'main',
     );
@@ -36,6 +38,7 @@ class Grabber
         'nav',
         'header',
         'social',
+        'entry-utility',
     );
 
     public $stripTags = array(
@@ -58,34 +61,23 @@ class Grabber
     {
         if ($this->html) {
 
-            Logging::log(\get_called_class().' HTML fetched');
+            Logging::log(\get_called_class().' Fix encoding');
+            $this->html = Filter::stripMetaTags($this->html);
+            $this->html = Encoding::toUtf8($this->html);
 
+            Logging::log(\get_called_class().' Try to find rules');
             $rules = $this->getRules();
 
-            \libxml_use_internal_errors(true);
-            $dom = new \DOMDocument;
-            $dom->loadHTML($this->html);
-
             if (is_array($rules)) {
                 Logging::log(\get_called_class().' Parse content with rules');
-                $this->parseContentWithRules($dom, $rules);
+                $this->parseContentWithRules($rules);
             }
             else {
-
                 Logging::log(\get_called_class().' Parse content with candidates');
-                $this->parseContentWithCandidates($dom);
-
-                if (strlen($this->content) < 50) {
-                    Logging::log(\get_called_class().' No enought content fetched, get the full body');
-                    $this->content = $dom->saveXML($dom->firstChild);
-                }
-
-                Logging::log(\get_called_class().' Strip garbage');
-                $this->stripGarbage();
+                $this->parseContentWithCandidates();
             }
         }
         else {
-
             Logging::log(\get_called_class().' No content fetched');
         }
 
@@ -129,8 +121,11 @@ class Grabber
     }
 
 
-    public function parseContentWithRules($dom, array $rules)
+    public function parseContentWithRules(array $rules)
     {
+        \libxml_use_internal_errors(true);
+        $dom = new \DOMDocument;
+        $dom->loadHTML('<?xml version="1.0" encoding="UTF-8">'.$this->html);
         $xpath = new \DOMXPath($dom);
 
         if (isset($rules['strip']) && is_array($rules['strip'])) {
@@ -147,21 +142,6 @@ class Grabber
             }
         }
 
-        if (isset($rules['strip_id_or_class']) && is_array($rules['strip_id_or_class'])) {
-
-            foreach ($rules['strip_id_or_class'] as $pattern) {
-
-                $pattern = strtr($pattern, array("'" => '', '"' => ''));
-                $nodes = $xpath->query("//*[contains(@class, '$pattern') or contains(@id, '$pattern')]");
-
-                if ($nodes !== false && $nodes->length > 0) {
-                    foreach ($nodes as $node) {
-                        $node->parentNode->removeChild($node);
-                    }
-                }
-            }
-        }
-
         if (isset($rules['body']) && is_array($rules['body'])) {
 
             foreach ($rules['body'] as $pattern) {
@@ -178,8 +158,11 @@ class Grabber
     }
 
 
-    public function parseContentWithCandidates($dom)
+    public function parseContentWithCandidates()
     {
+        \libxml_use_internal_errors(true);
+        $dom = new \DOMDocument;
+        $dom->loadHTML('<?xml version="1.0" encoding="UTF-8">'.$this->html);
         $xpath = new \DOMXPath($dom);
 
         // Try to fetch <article/>
@@ -187,27 +170,36 @@ class Grabber
 
         if ($nodes !== false && $nodes->length > 0) {
             $this->content = $dom->saveXML($nodes->item(0));
-            return;
         }
 
         // Try to lookup in each <div/>
+        if (! $this->content) {
+
             foreach ($this->candidatesAttributes as $candidate) {
 
                 $nodes = $xpath->query('//div[(contains(@class, "'.$candidate.'") or @id="'.$candidate.'") and not (contains(@class, "nav") or contains(@class, "page"))]');
 
                 if ($nodes !== false && $nodes->length > 0) {
                     $this->content = $dom->saveXML($nodes->item(0));
-                return;
                 }
             }
         }
 
+        if (strlen($this->content) < 50) {
+            Logging::log(\get_called_class().' No enought content fetched, get the full body');
+            $this->content = $dom->saveXML($dom->firstChild);
+        }
+
+        Logging::log(\get_called_class().' Strip garbage');
+        $this->stripGarbage();
+    }
+
 
     public function stripGarbage()
     {
         \libxml_use_internal_errors(true);
         $dom = new \DOMDocument;
-        $dom->loadXML($this->content);
+        $dom->loadXML('<?xml version="1.0" encoding="UTF-8">'.$this->content);
         $xpath = new \DOMXPath($dom);
 
         foreach ($this->stripTags as $tag) {

vendor/PicoFeed/Parser.php

@@ -18,8 +18,8 @@ abstract class Parser
     public $items = array();
     public $grabber = false;
     public $grabber_ignore_urls = array();
-    public $grabber_timeout = 5;
-    public $grabber_user_agent = 'PicoFeed (https://github.com/fguillot/picoFeed)';
+    public $grabber_timeout = null;
+    public $grabber_user_agent = null;
 
 
     abstract public function execute();
@@ -45,8 +45,7 @@ abstract class Parser
         if ($this->grabber && ! in_array($item_url, $this->grabber_ignore_urls)) {
             $grabber = new Grabber($item_url);
             $grabber->download($this->grabber_timeout, $this->grabber_user_agent);
-            $grabber->parse();
-            if ($grabber->content) $item_content = $grabber->content;
+            if ($grabber->parse()) $item_content = $grabber->content;
         }
 
         if ($item_content) {

vendor/PicoFeed/Rules/.blog.lemonde.fr.php

@@ -6,5 +6,5 @@ return array(
     ),
     'strip' => array(
         '//*[contains(@class, "fb-like") or contains(@class, "social")]'
-    )
+    ),
 );

vendor/PicoFeed/Rules/.blogs.nytimes.com.php

@@ -1,8 +1,6 @@
 <?php
 return array(
     'title' => '//header/h1',
-    'test_url' => 'http://opinionator.blogs.nytimes.com/2011/02/03/lost-and-gone-forever/',
-    'test_url' => 'http://krugman.blogs.nytimes.com/2012/09/12/a-vote-of-confidence/',
     'test_url' => 'http://bits.blogs.nytimes.com/2012/01/16/wikipedia-plans-to-go-dark-on-wednesday-to-protest-sopa/',
     'body' => array(
          '//div[@class="postContent"]',

vendor/PicoFeed/Rules/.wikipedia.org.php

@@ -0,0 +1,25 @@
+<?php
+return array(
+    'test_url' => 'https://en.wikipedia.org/wiki/Grace_Hopper',
+    'body' => array(
+        '//div[@id="bodyContent"]',
+    ),
+    'strip' => array(
+        "//div[@id='toc']",
+        "//div[@id='catlinks']",
+        "//div[@id='jump-to-nav']",
+        "//div[@class='thumbcaption']//div[@class='magnify']",
+        "//table[@class='navbox']",
+        "//table[contains(@class, 'infobox')]",
+        "//div[@class='dablink']",
+        "//div[@id='contentSub']",
+        "//div[@id='siteSub']",
+        "//table[@id='persondata']",
+        "//table[contains(@class, 'metadata')]",
+        "//*[contains(@class, 'noprint')]",
+        "//*[contains(@class, 'printfooter')]",
+        "//*[contains(@class, 'editsection')]",
+        "//*[contains(@class, 'error')]",
+        "//span[@title='pronunciation:']",
+    ),
+);

vendor/PicoFeed/Rules/techcrunch.com.php

@@ -0,0 +1,12 @@
+<?php
+return array(
+    'test_url' => 'http://techcrunch.com/2013/08/31/indias-visa-maze/',
+    'body' => array(
+        '//div[contains(@class, "media-container")]',
+        '//div[@class="body-copy"]',
+    ),
+    'strip' => array(
+        '//script',
+        '//style',
+    )
+);

vendor/PicoFeed/Rules/www.cnn.com.php

@@ -2,7 +2,21 @@
 return array(
     'test_url' => 'http://www.cnn.com/2013/08/31/world/meast/syria-civil-war/index.html?hpt=hp_t1',
     'body' => array(
-        '//*[contains(@class, "cnn_storypgraphtxt")]]',
-        '//*[contains(@class, "cnnvideo_wrapper")]]',
+        '//div[@class="cnn_strycntntlft"]',
     ),
+    'strip' => array(
+        '//script',
+        '//style',
+        '//div[@class="cnn_stryshrwdgtbtm"]',
+        '//div[@class="cnn_strybtmcntnt"]',
+        '//div[@class="cnn_strylftcntnt"]',
+        '//div[contains(@class, "cnnGalleryContainer")]',
+        '//div[contains(@class, "cnn_strylftcexpbx")]',
+        '//div[contains(@class, "articleGalleryNavContainer")]',
+        '//div[contains(@class, "cnnArticleGalleryCaptionControl")]',
+        '//div[contains(@class, "cnnArticleGalleryNavPrevNextDisabled")]',
+        '//div[contains(@class, "cnnArticleGalleryNavPrevNext")]',
+        '//div[contains(@class, "cnn_html_media_title_new")]',
+        '//div[contains(@id, "disqus")]',
+    )
 );

vendor/PicoFeed/Rules/www.theguardian.com.php

@@ -0,0 +1,9 @@
+<?php
+return array(
+    'test_url' => 'http://www.theguardian.com/law/2013/aug/31/microsoft-google-sue-us-fisa',
+    'body' => array(
+        '//div[@id="article-wrapper"]',
+    ),
+    'strip' => array(
+    ),
+);