Merge pull request #57 from j0k3r/fix/html5-parser

Use a new deps for HTML5 parser

.editorconfig

@@ -4,8 +4,6 @@ root = true
 ; Unix-style newlines
 [*]
 end_of_line = LF
-
-[*.php]
 indent_style = space
 indent_size = 4
 trim_trailing_whitespace = true

.gitattributes

@@ -4,6 +4,6 @@
 /.scrutinizer.yml export-ignore
 /.travis.yml export-ignore
 /.php_cs export-ignore
-/README.md export-ignore
 /phpunit.xml.dist export-ignore
+/phpstan.neon export-ignore
 /tests export-ignore

.gitignore

@@ -2,3 +2,4 @@ vendor/
 coverage/
 composer.lock
 .php_cs.cache
+.phpunit.result.cache

.scrutinizer.yml

@@ -1,2 +1,12 @@
 tools:
     external_code_coverage: false
+
+build:
+    nodes:
+        analysis:
+            tests:
+                override:
+                    - php-scrutinizer-run
+            environment:
+                php:
+                    version: 7.2

.travis.yml

@@ -1,40 +1,54 @@
+dist: xenial
+os: linux
 language: php
 
 php:
     - 5.6
     - 7.0
     - 7.1
-    - 7.2
-    - 7.3
-    - nightly
 
-matrix:
+jobs:
     include:
         - php: 7.2
-          env: CS_FIXER=run
+          env: SYMFONY_PHPUNIT_VERSION=7.5
+        - php: 7.2
+          env: SYMFONY_PHPUNIT_VERSION=7.5 COMPOSER_FLAGS="--prefer-lowest"
+        - php: 7.3
+          env: SYMFONY_PHPUNIT_VERSION=7.5 CS_FIXER=run COVERAGE=run
+        - php: 7.4
+          env: SYMFONY_PHPUNIT_VERSION=7.5
+        - php: nightly
+          env: SYMFONY_PHPUNIT_VERSION=7.5
     fast_finish: true
     allow_failures:
         - php: nightly
 
-# cache vendor dirs
 cache:
     directories:
-        - vendor
+        - $HOME/.composer/cache/files
-        - $HOME/.composer/cache
 
 before_install:
     - if [ -n "$GH_TOKEN" ]; then composer config github-oauth.github.com ${GH_TOKEN}; fi;
 
 install:
-    - composer self-update
+    # only install phpstan when we need it
+    - if [ "$CS_FIXER" = "run" ]; then composer require phpstan/phpstan phpstan/phpstan-phpunit --dev --no-progress --no-suggest ; fi;
+    - composer update --prefer-dist --no-progress --no-suggest -o $COMPOSER_FLAGS
+    - php vendor/bin/simple-phpunit install
 
 before_script:
-    - composer install -o --prefer-dist --no-interaction
+    - mkdir -p build/logs
+    - if [ "$COVERAGE" = "run" ]; then PHPUNIT_FLAGS="--coverage-clover build/logs/clover.xml"; fi;
 
 script:
-    - mkdir -p build/logs
+    - php vendor/bin/simple-phpunit -v $PHPUNIT_FLAGS
-    - php vendor/bin/simple-phpunit -v --coverage-clover build/logs/clover.xml
     - if [ "$CS_FIXER" = "run" ]; then php vendor/bin/php-cs-fixer fix --verbose --dry-run ; fi;
+    - if [ "$CS_FIXER" = "run" ]; then php vendor/bin/phpstan analyse src tests --no-progress --level 1 ; fi;
 
-after_script:
+after_success:
-    - php vendor/bin/php-coveralls -v -x build/logs/clover.xml
+    - |
+        if [ "$COVERAGE" = "run" ]; then
+            wget https://github.com/php-coveralls/php-coveralls/releases/download/v2.2.0/php-coveralls.phar
+            chmod +x php-coveralls.phar
+            php php-coveralls.phar -v -x build/logs/clover.xml
+        fi

composer.json

@@ -27,12 +27,11 @@
         "php": ">=5.6.0",
         "ext-mbstring": "*",
         "psr/log": "^1.0",
-        "electrolinux/php-html5lib": "^0.1.0"
+        "masterminds/html5": "^2.7"
     },
     "require-dev": {
-        "php-coveralls/php-coveralls": "^2.1",
         "friendsofphp/php-cs-fixer": "^2.14",
-        "monolog/monolog": "^1.24",
+        "monolog/monolog": "^1.24|^2.1",
         "symfony/phpunit-bridge": "^4.2.3"
     },
     "suggest": {
@@ -40,5 +39,8 @@
     },
     "autoload": {
         "psr-4": { "Readability\\": "src/" }
+    },
+    "autoload-dev": {
+        "psr-4": { "Tests\\Readability\\": "tests/" }
     }
 }

phpstan.neon

@@ -0,0 +1,8 @@
+includes:
+    - vendor/phpstan/phpstan-phpunit/extension.neon
+    - vendor/phpstan/phpstan-phpunit/rules.neon
+
+parameters:
+    # https://github.com/phpstan/phpstan/issues/694#issuecomment-350724288
+    autoload_files:
+        - vendor/bin/.phpunit/phpunit-7.5-0/vendor/autoload.php

phpunit.xml.dist

@@ -7,7 +7,6 @@
     convertWarningsToExceptions="true"
     processIsolation="false"
     stopOnFailure="false"
-    syntaxCheck="false"
     bootstrap="vendor/autoload.php"
     >
 
@@ -27,6 +26,6 @@
     </filter>
 
     <!-- <logging>
-        <log type="coverage-html" target="coverage" title="Readability" charset="UTF-8" yui="true" highlight="true" lowUpperBound="35" highLowerBound="70"/>
+        <log type="coverage-html" target="coverage" lowUpperBound="35" highLowerBound="70"/>
     </logging> -->
 </phpunit>

src/JSLikeHTMLElement.php

@@ -127,4 +127,14 @@ class JSLikeHTMLElement extends \DOMElement
     {
         return '[' . $this->tagName . ']';
     }
+
+    public function getInnerHtml()
+    {
+        return $this->__get('innerHTML');
+    }
+
+    public function setInnerHtml($value)
+    {
+        return $this->__set('innerHTML', $value);
+    }
 }

src/Readability.php

@@ -2,7 +2,7 @@
 
 namespace Readability;
 
-use HTML5Lib\Parser;
+use Masterminds\HTML5;
 use Psr\Log\LoggerAwareInterface;
 use Psr\Log\LoggerInterface;
 use Psr\Log\NullLogger;
@@ -61,7 +61,7 @@ class Readability implements LoggerAwareInterface
     // constants
     const SCORE_CHARS_IN_PARAGRAPH = 100;
     const SCORE_WORDS_IN_PARAGRAPH = 20;
-    const GRANDPARENT_SCORE_DIVISOR = 2.2;
+    const GRANDPARENT_SCORE_DIVISOR = 2;
     const MIN_PARAGRAPH_LENGTH = 20;
     const MIN_COMMAS_IN_PARAGRAPH = 6;
     const MIN_ARTICLE_LENGTH = 200;
@@ -72,6 +72,9 @@ class Readability implements LoggerAwareInterface
     public $articleTitle;
     public $articleContent;
     public $original_html;
+    /**
+     * @var \DOMDocument
+     */
     public $dom;
     // optional - URL where HTML was retrieved
     public $url = null;
@@ -89,10 +92,10 @@ class Readability implements LoggerAwareInterface
         'unlikelyCandidates' => '/display\s*:\s*none|ignore|\binfos?\b|annoy|clock|date|time|author|intro|hidd?e|about|archive|\bprint|bookmark|tags|tag-list|share|search|social|robot|published|combx|comment|mast(?:head)|subscri|community|category|disqus|extra|head|head(?:er|note)|floor|foot(?:er|note)|menu|tool\b|function|nav|remark|rss|shoutbox|widget|meta|banner|sponsor|adsense|inner-?ad|ad-|sponsor|\badv\b|\bads\b|agr?egate?|pager|sidebar|popup|tweet|twitter/i',
         'okMaybeItsACandidate' => '/article\b|contain|\bcontent|column|general|detail|shadow|lightbox|blog|body|entry|main|page|footnote/i',
         'positive' => '/read|full|article|body|\bcontent|contain|entry|main|markdown|media|page|attach|pagination|post|text|blog|story/i',
-        'negative' => '/bottom|stat|info|discuss|e[\-]?mail|comment|reply|log.{2}(n|ed)|sign|single|combx|com-|contact|_nav|link|media|\bout|promo|\bad-|related|scroll|shoutbox|sidebar|sponsor|shopping|teaser|recommend/i',
+        'negative' => '/bottom|stat|info|discuss|e[\-]?mail|comment|reply|log.{2}(n|ed)|sign|single|combx|com-|contact|_nav|link|media|promo|\bad-|related|scroll|shoutbox|sidebar|sponsor|shopping|teaser|recommend/i',
         'divToPElements' => '/<(?:blockquote|header|section|code|div|article|footer|aside|img|p|pre|dl|ol|ul)/mi',
         'killBreaks' => '/(<br\s*\/?>([ \r\n\s]|&nbsp;?)*)+/',
-        'media' => '!//(?:[^\.\?/]+\.)?(?:youtu(?:be)?|giphy|soundcloud|dailymotion|vimeo|pornhub|xvideos|twitvid|rutube|viddler)\.(?:com|be|org|net)/!i',
+        'media' => '!//(?:[^\.\?/]+\.)?(?:youtu(?:be)?|giphy|soundcloud|dailymotion|vimeo|pornhub|xvideos|twitvid|rutube|openload\.co|viddler)\.(?:com|be|org|net)/!i',
         'skipFootnoteLink' => '/^\s*(\[?[a-z0-9]{1,2}\]?|^|edit|citation needed)\s*$/i',
     ];
     public $tidy_config = [
@@ -169,10 +172,10 @@ class Readability implements LoggerAwareInterface
     /**
      * Create instance of Readability.
      *
-     * @param string UTF-8 encoded string
+     * @param string $html     UTF-8 encoded string
-     * @param string (optional) URL associated with HTML (for footnotes)
+     * @param string $url      URL associated with HTML (for footnotes)
-     * @param string (optional) Which parser to use for turning raw HTML into a DOMDocument
+     * @param string $parser   Which parser to use for turning raw HTML into a DOMDocument
-     * @param bool (optional) Use tidy
+     * @param bool   $use_tidy Use tidy
      */
     public function __construct($html, $url = null, $parser = 'libxml', $use_tidy = true)
     {
@@ -213,8 +216,8 @@ class Readability implements LoggerAwareInterface
     /**
      * Add pre filter for raw input HTML processing.
      *
-     * @param string RegExp for replace
+     * @param string $filter   RegExp for replace
-     * @param string (optional) Replacer
+     * @param string $replacer Replacer
      */
     public function addPreFilter($filter, $replacer = '')
     {
@@ -224,8 +227,8 @@ class Readability implements LoggerAwareInterface
     /**
      * Add post filter for raw output HTML processing.
      *
-     * @param string RegExp for replace
+     * @param string $filter   RegExp for replace
-     * @param string (optional) Replacer
+     * @param string $replacer Replacer
      */
     public function addPostFilter($filter, $replacer = '')
     {
@@ -258,7 +261,7 @@ class Readability implements LoggerAwareInterface
         if (null === $this->bodyCache) {
             $this->bodyCache = '';
             foreach ($bodyElems as $bodyNode) {
-                $this->bodyCache .= trim($bodyNode->innerHTML);
+                $this->bodyCache .= trim($bodyNode->getInnerHTML());
             }
         }
 
@@ -278,7 +281,7 @@ class Readability implements LoggerAwareInterface
             $this->success = false;
             $articleContent = $this->dom->createElement('div');
             $articleContent->setAttribute('class', 'readability-content');
-            $articleContent->innerHTML = '<p>Sorry, Readability was unable to parse this page for content.</p>';
+            $articleContent->setInnerHtml('<p>Sorry, Readability was unable to parse this page for content.</p>');
         }
 
         $overlay->setAttribute('class', 'readOverlay');
@@ -290,7 +293,7 @@ class Readability implements LoggerAwareInterface
         $overlay->appendChild($innerDiv);
 
         // Clear the old HTML, insert the new content.
-        $this->body->innerHTML = '';
+        $this->body->setInnerHtml('');
         $this->body->appendChild($overlay);
         $this->body->removeAttribute('style');
         $this->postProcessContent($articleContent);
@@ -304,10 +307,8 @@ class Readability implements LoggerAwareInterface
 
     /**
      * Run any post-process modifications to article content as necessary.
-     *
-     * @param \DOMElement $articleContent
      */
-    public function postProcessContent($articleContent)
+    public function postProcessContent(\DOMElement $articleContent)
     {
         if ($this->convertLinksToFootnotes && !preg_match('/\bwiki/', $this->url)) {
             $this->addFootnotes($articleContent);
@@ -318,14 +319,12 @@ class Readability implements LoggerAwareInterface
      * For easier reading, convert this document to have footnotes at the bottom rather than inline links.
      *
      * @see http://www.roughtype.com/archives/2010/05/experiments_in.php
-     *
-     * @param \DOMElement $articleContent
      */
-    public function addFootnotes($articleContent)
+    public function addFootnotes(\DOMElement $articleContent)
     {
         $footnotesWrapper = $this->dom->createElement('footer');
         $footnotesWrapper->setAttribute('class', 'readability-footnotes');
-        $footnotesWrapper->innerHTML = '<h3>References</h3>';
+        $footnotesWrapper->setInnerHtml('<h3>References</h3>');
         $articleFootnotes = $this->dom->createElement('ol');
         $articleFootnotes->setAttribute('class', 'readability-footnotes-list');
         $footnotesWrapper->appendChild($articleFootnotes);
@@ -351,7 +350,7 @@ class Readability implements LoggerAwareInterface
 
             // Add a superscript reference after the article link.
             $refLink->setAttribute('href', '#readabilityFootnoteLink-' . $linkCount);
-            $refLink->innerHTML = '<small><sup>[' . $linkCount . ']</sup></small>';
+            $refLink->setInnerHtml('<small><sup>[' . $linkCount . ']</sup></small>');
             $refLink->setAttribute('class', 'readability-DoNotFootnote');
             $refLink->setAttribute('style', 'color: inherit;');
 
@@ -363,13 +362,13 @@ class Readability implements LoggerAwareInterface
 
             $articleLink->setAttribute('style', 'color: inherit; text-decoration: none;');
             $articleLink->setAttribute('name', 'readabilityLink-' . $linkCount);
-            $footnote->innerHTML = '<small><sup><a href="#readabilityLink-' . $linkCount . '" title="Jump to Link in Article">^</a></sup></small> ';
+            $footnote->setInnerHtml('<small><sup><a href="#readabilityLink-' . $linkCount . '" title="Jump to Link in Article">^</a></sup></small> ');
-            $footnoteLink->innerHTML = ('' !== $footnoteLink->getAttribute('title') ? $footnoteLink->getAttribute('title') : $linkText);
+            $footnoteLink->setInnerHtml(('' !== $footnoteLink->getAttribute('title') ? $footnoteLink->getAttribute('title') : $linkText));
             $footnoteLink->setAttribute('name', 'readabilityFootnoteLink-' . $linkCount);
             $footnote->appendChild($footnoteLink);
 
             if ($linkDomain) {
-                $footnote->innerHTML = $footnote->innerHTML . '<small> (' . $linkDomain . ')</small>';
+                $footnote->setInnerHtml($footnote->getInnerHTML() . '<small> (' . $linkDomain . ')</small>');
             }
             $articleFootnotes->appendChild($footnote);
         }
@@ -382,11 +381,13 @@ class Readability implements LoggerAwareInterface
     /**
      * Prepare the article node for display. Clean out any inline styles,
      * iframes, forms, strip extraneous <p> tags, etc.
-     *
-     * @param \DOMElement $articleContent
      */
-    public function prepArticle($articleContent)
+    public function prepArticle(\DOMNode $articleContent)
     {
+        if (!$articleContent instanceof \DOMElement) {
+            return;
+        }
+
         $this->logger->debug($this->lightClean ? 'Light clean enabled.' : 'Standard clean enabled.');
 
         $this->cleanStyles($articleContent);
@@ -467,7 +468,7 @@ class Readability implements LoggerAwareInterface
         if (!$this->flagIsActive(self::FLAG_DISABLE_POSTFILTER)) {
             try {
                 foreach ($this->post_filters as $search => $replace) {
-                    $articleContent->innerHTML = preg_replace($search, $replace, $articleContent->innerHTML);
+                    $articleContent->setInnerHtml(preg_replace($search, $replace, $articleContent->getInnerHTML()));
                 }
                 unset($search, $replace);
             } catch (\Exception $e) {
@@ -551,12 +552,11 @@ class Readability implements LoggerAwareInterface
      * This is the amount of text that is inside a link divided by the total text in the node.
      * Can exclude external references to differentiate between simple text and menus/infoblocks.
      *
-     * @param \DOMElement $e
+     * @param bool $excludeExternal
-     * @param string      $excludeExternal
      *
      * @return int
      */
-    public function getLinkDensity($e, $excludeExternal = false)
+    public function getLinkDensity(\DOMElement $e, $excludeExternal = false)
     {
         $links = $e->getElementsByTagName('a');
         $textLength = mb_strlen($this->getInnerText($e, true, true));
@@ -579,11 +579,9 @@ class Readability implements LoggerAwareInterface
     /**
      * Get an element relative weight.
      *
-     * @param \DOMElement $e
-     *
      * @return int
      */
-    public function getWeight($e)
+    public function getWeight(\DOMElement $e)
     {
         if (!$this->flagIsActive(self::FLAG_WEIGHT_ATTRIBUTES)) {
             return 0;
@@ -600,14 +598,12 @@ class Readability implements LoggerAwareInterface
 
     /**
      * Remove extraneous break tags from a node.
-     *
-     * @param \DOMElement $node
      */
-    public function killBreaks($node)
+    public function killBreaks(\DOMElement $node)
     {
-        $html = $node->innerHTML;
+        $html = $node->getInnerHTML();
         $html = preg_replace($this->regexps['killBreaks'], '<br />', $html);
-        $node->innerHTML = $html;
+        $node->setInnerHtml($html);
     }
 
     /**
@@ -616,12 +612,10 @@ class Readability implements LoggerAwareInterface
      *
      * Updated 2012-09-18 to preserve youtube/vimeo iframes
      *
-     * @param \DOMElement $e
+     * @param string $tag
-     * @param string      $tag
      */
-    public function clean($e, $tag)
+    public function clean(\DOMElement $e, $tag)
     {
-        $currentItem = null;
         $targetList = $e->getElementsByTagName($tag);
         $isEmbed = ('audio' === $tag || 'video' === $tag || 'iframe' === $tag || 'object' === $tag || 'embed' === $tag);
 
@@ -638,7 +632,7 @@ class Readability implements LoggerAwareInterface
                 }
 
                 // Then check the elements inside this element for the same.
-                if (preg_match($this->regexps['media'], $targetList->item($y)->innerHTML)) {
+                if (preg_match($this->regexps['media'], $targetList->item($y)->getInnerHTML())) {
                     continue;
                 }
             }
@@ -652,10 +646,9 @@ class Readability implements LoggerAwareInterface
      * "Fishy" is an algorithm based on content length, classnames,
      * link density, number of images & embeds, etc.
      *
-     * @param \DOMElement $e
+     * @param string $tag
-     * @param string      $tag
      */
-    public function cleanConditionally($e, $tag)
+    public function cleanConditionally(\DOMElement $e, $tag)
     {
         if (!$this->flagIsActive(self::FLAG_CLEAN_CONDITIONALLY)) {
             return;
@@ -663,7 +656,6 @@ class Readability implements LoggerAwareInterface
 
         $tagsList = $e->getElementsByTagName($tag);
         $curTagsLength = $tagsList->length;
-        $node = null;
 
         /*
          * Gather counts for other typical elements embedded within.
@@ -765,10 +757,8 @@ class Readability implements LoggerAwareInterface
 
     /**
      * Clean out spurious headers from an Element. Checks things like classnames and link density.
-     *
-     * @param \DOMElement $e
      */
-    public function cleanHeaders($e)
+    public function cleanHeaders(\DOMElement $e)
     {
         for ($headerIndex = 1; $headerIndex < 3; ++$headerIndex) {
             $headers = $e->getElementsByTagName('h' . $headerIndex);
@@ -871,7 +861,7 @@ class Readability implements LoggerAwareInterface
         }
 
         $articleTitle = $this->dom->createElement('h1');
-        $articleTitle->innerHTML = $curTitle;
+        $articleTitle->setInnerHtml($curTitle);
 
         return $articleTitle;
     }
@@ -908,10 +898,8 @@ class Readability implements LoggerAwareInterface
     /**
      * Initialize a node with the readability object. Also checks the
      * className/id for special names to add to its score.
-     *
-     * @param \DOMElement $node
      */
-    protected function initializeNode($node)
+    protected function initializeNode(\DOMElement $node)
     {
         if (!isset($node->tagName)) {
             return;
@@ -981,9 +969,9 @@ class Readability implements LoggerAwareInterface
      *
      * @param \DOMElement $page
      *
-     * @return \DOMElement|bool
+     * @return \DOMElement|false
      */
-    protected function grabArticle($page = null)
+    protected function grabArticle(\DOMElement $page = null)
     {
         if (!$page) {
             $page = $this->dom;
@@ -1009,11 +997,11 @@ class Readability implements LoggerAwareInterface
             // Turn divs into P tags where they have been used inappropriately
             //  (as in, where they contain no other block level elements).
             if (0 === strcasecmp($tagName, 'div') || 0 === strcasecmp($tagName, 'article') || 0 === strcasecmp($tagName, 'section')) {
-                if (!preg_match($this->regexps['divToPElements'], $node->innerHTML)) {
+                if (!preg_match($this->regexps['divToPElements'], $node->getInnerHTML())) {
                     $newNode = $this->dom->createElement('p');
 
                     try {
-                        $newNode->innerHTML = $node->innerHTML;
+                        $newNode->setInnerHtml($node->getInnerHTML());
 
                         $node->parentNode->replaceChild($newNode, $node);
                         --$nodeIndex;
@@ -1040,7 +1028,7 @@ class Readability implements LoggerAwareInterface
 
                         if (XML_TEXT_NODE === $childNode->nodeType) {
                             $p = $this->dom->createElement('p');
-                            $p->innerHTML = $childNode->nodeValue;
+                            $p->setInnerHtml($childNode->nodeValue);
                             $p->setAttribute('data-readability-styled', 'true');
                             $childNode->parentNode->replaceChild($p, $childNode);
                         }
@@ -1108,7 +1096,7 @@ class Readability implements LoggerAwareInterface
             // Add the score to the parent. The grandparent gets half.
             $parentNode->getAttributeNode('readability')->value += $contentScore;
             if ($grandParentNode) {
-                $grandParentNode->getAttributeNode('readability')->value += $contentScore / self::GRANDPARENT_SCORE_DIVISOR;
+                $grandParentNode->getAttributeNode('readability')->value += round($contentScore / self::GRANDPARENT_SCORE_DIVISOR);
             }
         }
 
@@ -1118,7 +1106,6 @@ class Readability implements LoggerAwareInterface
          */
         if ($this->flagIsActive(self::FLAG_STRIP_UNLIKELYS) && $xpath) {
             $candidates = $xpath->query('.//*[(self::footer and count(//footer)<2) or (self::aside and count(//aside)<2)]', $page->documentElement);
-            $node = null;
 
             for ($c = $candidates->length - 1; $c >= 0; --$c) {
                 $node = $candidates->item($c);
@@ -1130,7 +1117,6 @@ class Readability implements LoggerAwareInterface
             }
 
             $candidates = $xpath->query('.//*[not(self::body) and (@class or @id or @style) and ((number(@readability) < 40) or not(@readability))]', $page->documentElement);
-            $node = null;
 
             for ($c = $candidates->length - 1; $c >= 0; --$c) {
                 $node = $candidates->item($c);
@@ -1190,14 +1176,14 @@ class Readability implements LoggerAwareInterface
                     $this->logger->debug('The page has no body!');
                 } else {
                     $this->logger->debug('Setting body to a raw HTML of original page!');
-                    $topCandidate->innerHTML = $page->documentElement->innerHTML;
+                    $topCandidate->setInnerHtml($page->documentElement->getInnerHTML());
-                    $page->documentElement->innerHTML = '';
+                    $page->documentElement->setInnerHtml('');
                     $this->reinitBody();
                     $page->documentElement->appendChild($topCandidate);
                 }
             } else {
-                $topCandidate->innerHTML = $page->innerHTML;
+                $topCandidate->setInnerHtml($page->getInnerHTML());
-                $page->innerHTML = '';
+                $page->setInnerHtml('');
                 $page->appendChild($topCandidate);
             }
 
@@ -1229,8 +1215,8 @@ class Readability implements LoggerAwareInterface
         $siblingScoreThreshold = max(10, ((int) $topCandidate->getAttribute('readability')) * 0.2);
         $siblingNodes = $topCandidate->parentNode->childNodes;
 
-        if (!isset($siblingNodes)) {
+        if (null === $siblingNodes) {
-            $siblingNodes = new stdClass();
+            $siblingNodes = new \stdClass();
             $siblingNodes->length = 0;
         }
 
@@ -1276,7 +1262,7 @@ class Readability implements LoggerAwareInterface
 
                     try {
                         $nodeToAppend->setAttribute('alt', $siblingNodeName);
-                        $nodeToAppend->innerHTML = $siblingNode->innerHTML;
+                        $nodeToAppend->setInnerHtml($siblingNode->getInnerHTML());
                     } catch (\Exception $e) {
                         $this->logger->debug('Could not alter siblingNode "' . $siblingNodeName . '" to "div", reverting to original.');
                         $nodeToAppend = $siblingNode;
@@ -1339,12 +1325,11 @@ class Readability implements LoggerAwareInterface
      * Get an element weight by attribute.
      * Uses regular expressions to tell if this element looks good or bad.
      *
-     * @param \DOMElement $element
+     * @param string $attribute
-     * @param string      $attribute
      *
      * @return int
      */
-    protected function weightAttribute($element, $attribute)
+    protected function weightAttribute(\DOMElement $element, $attribute)
     {
         if (!$element->hasAttribute($attribute)) {
             return 0;
@@ -1379,7 +1364,7 @@ class Readability implements LoggerAwareInterface
     {
         if (!isset($this->body->childNodes)) {
             $this->body = $this->dom->createElement('body');
-            $this->body->innerHTML = $this->bodyCache;
+            $this->body->setInnerHtml($this->bodyCache);
         }
     }
 
@@ -1435,17 +1420,16 @@ class Readability implements LoggerAwareInterface
 
         $this->html = mb_convert_encoding($this->html, 'HTML-ENTITIES', 'UTF-8');
 
-        if (!('html5lib' === $this->parser && ($this->dom = Parser::parse($this->html)))) {
+        if ('html5lib' === $this->parser || 'html5' === $this->parser) {
+            $this->dom = (new HTML5())->loadHTML($this->html);
+        }
+
+        if ('libxml' === $this->parser) {
             libxml_use_internal_errors(true);
 
             $this->dom = new \DOMDocument();
             $this->dom->preserveWhiteSpace = false;
-
+            $this->dom->loadHTML($this->html, LIBXML_NOBLANKS | LIBXML_COMPACT | LIBXML_NOERROR);
-            if (\PHP_VERSION_ID >= 50400) {
-                $this->dom->loadHTML($this->html, LIBXML_NOBLANKS | LIBXML_COMPACT | LIBXML_NOERROR);
-            } else {
-                $this->dom->loadHTML($this->html);
-            }
 
             libxml_use_internal_errors(false);
         }

tests/ReadabilityTest.php

@@ -73,8 +73,8 @@ class ReadabilityTest extends \PHPUnit\Framework\TestCase
         $this->assertFalse($res);
         $this->assertInstanceOf('Readability\JSLikeHTMLElement', $readability->getContent());
         $this->assertInstanceOf('Readability\JSLikeHTMLElement', $readability->getTitle());
-        $this->assertEmpty($readability->getTitle()->innerHTML);
+        $this->assertEmpty($readability->getTitle()->getInnerHtml());
-        $this->assertContains('Sorry, Readability was unable to parse this page for content.', $readability->getContent()->innerHTML);
+        $this->assertContains('Sorry, Readability was unable to parse this page for content.', $readability->getContent()->getInnerHtml());
     }
 
     public function testInitP()
@@ -85,9 +85,9 @@ class ReadabilityTest extends \PHPUnit\Framework\TestCase
         $this->assertTrue($res);
         $this->assertInstanceOf('Readability\JSLikeHTMLElement', $readability->getContent());
         $this->assertInstanceOf('Readability\JSLikeHTMLElement', $readability->getTitle());
-        $this->assertContains('<div readability=', $readability->getContent()->innerHTML);
+        $this->assertContains('<div readability=', $readability->getContent()->getInnerHtml());
-        $this->assertEmpty($readability->getTitle()->innerHTML);
+        $this->assertEmpty($readability->getTitle()->getInnerHtml());
-        $this->assertContains('This is the awesome content :)', $readability->getContent()->innerHTML);
+        $this->assertContains('This is the awesome content :)', $readability->getContent()->getInnerHtml());
     }
 
     public function testInitDivP()
@@ -98,9 +98,9 @@ class ReadabilityTest extends \PHPUnit\Framework\TestCase
         $this->assertTrue($res);
         $this->assertInstanceOf('Readability\JSLikeHTMLElement', $readability->getContent());
         $this->assertInstanceOf('Readability\JSLikeHTMLElement', $readability->getTitle());
-        $this->assertContains('<div readability=', $readability->getContent()->innerHTML);
+        $this->assertContains('<div readability=', $readability->getContent()->getInnerHtml());
-        $this->assertEmpty($readability->getTitle()->innerHTML);
+        $this->assertEmpty($readability->getTitle()->getInnerHtml());
-        $this->assertContains('This is the awesome content :)', $readability->getContent()->innerHTML);
+        $this->assertContains('This is the awesome content :)', $readability->getContent()->getInnerHtml());
     }
 
     public function testInitDiv()
@@ -112,9 +112,9 @@ class ReadabilityTest extends \PHPUnit\Framework\TestCase
         $this->assertTrue($res);
         $this->assertInstanceOf('Readability\JSLikeHTMLElement', $readability->getContent());
         $this->assertInstanceOf('Readability\JSLikeHTMLElement', $readability->getTitle());
-        $this->assertContains('<div readability=', $readability->getContent()->innerHTML);
+        $this->assertContains('<div readability=', $readability->getContent()->getInnerHtml());
-        $this->assertEmpty($readability->getTitle()->innerHTML);
+        $this->assertEmpty($readability->getTitle()->getInnerHtml());
-        $this->assertContains('This is the awesome content :)', $readability->getContent()->innerHTML);
+        $this->assertContains('This is the awesome content :)', $readability->getContent()->getInnerHtml());
     }
 
     public function testWithFootnotes()
@@ -127,11 +127,11 @@ class ReadabilityTest extends \PHPUnit\Framework\TestCase
         $this->assertTrue($res);
         $this->assertInstanceOf('Readability\JSLikeHTMLElement', $readability->getContent());
         $this->assertInstanceOf('Readability\JSLikeHTMLElement', $readability->getTitle());
-        $this->assertContains('<div readability=', $readability->getContent()->innerHTML);
+        $this->assertContains('<div readability=', $readability->getContent()->getInnerHtml());
-        $this->assertEmpty($readability->getTitle()->innerHTML);
+        $this->assertEmpty($readability->getTitle()->getInnerHtml());
-        $this->assertContains('This is an awesome text with some links, here there are', $readability->getContent()->innerHTML);
+        $this->assertContains('This is an awesome text with some links, here there are', $readability->getContent()->getInnerHtml());
-        $this->assertContains('readabilityFootnoteLink', $readability->getContent()->innerHTML);
+        $this->assertContains('readabilityFootnoteLink', $readability->getContent()->getInnerHtml());
-        $this->assertContains('readabilityLink-3', $readability->getContent()->innerHTML);
+        $this->assertContains('readabilityLink-3', $readability->getContent()->getInnerHtml());
     }
 
     public function testStandardClean()
@@ -144,11 +144,11 @@ class ReadabilityTest extends \PHPUnit\Framework\TestCase
         $this->assertTrue($res);
         $this->assertInstanceOf('Readability\JSLikeHTMLElement', $readability->getContent());
         $this->assertInstanceOf('Readability\JSLikeHTMLElement', $readability->getTitle());
-        $this->assertContains('<div readability=', $readability->getContent()->innerHTML);
+        $this->assertContains('<div readability=', $readability->getContent()->getInnerHtml());
-        $this->assertEmpty($readability->getTitle()->innerHTML);
+        $this->assertEmpty($readability->getTitle()->getInnerHtml());
-        $this->assertContains('This is an awesome text with some links, here there are', $readability->getContent()->innerHTML);
+        $this->assertContains('This is an awesome text with some links, here there are', $readability->getContent()->getInnerHtml());
-        $this->assertContains('will NOT be removed', $readability->getContent()->innerHTML);
+        $this->assertContains('will NOT be removed', $readability->getContent()->getInnerHtml());
-        $this->assertNotContains('<h2>', $readability->getContent()->innerHTML);
+        $this->assertNotContains('<h2>', $readability->getContent()->getInnerHtml());
     }
 
     public function testWithIframe()
@@ -160,10 +160,10 @@ class ReadabilityTest extends \PHPUnit\Framework\TestCase
         $this->assertTrue($res);
         $this->assertInstanceOf('Readability\JSLikeHTMLElement', $readability->getContent());
         $this->assertInstanceOf('Readability\JSLikeHTMLElement', $readability->getTitle());
-        $this->assertContains('<div readability=', $readability->getContent()->innerHTML);
+        $this->assertContains('<div readability=', $readability->getContent()->getInnerHtml());
-        $this->assertEmpty($readability->getTitle()->innerHTML);
+        $this->assertEmpty($readability->getTitle()->getInnerHtml());
-        $this->assertContains('This is an awesome text with some links, here there are', $readability->getContent()->innerHTML);
+        $this->assertContains('This is an awesome text with some links, here there are', $readability->getContent()->getInnerHtml());
-        $this->assertContains('nofollow', $readability->getContent()->innerHTML);
+        $this->assertContains('nofollow', $readability->getContent()->getInnerHtml());
     }
 
     public function testWithArticle()
@@ -175,10 +175,10 @@ class ReadabilityTest extends \PHPUnit\Framework\TestCase
         $this->assertTrue($res);
         $this->assertInstanceOf('Readability\JSLikeHTMLElement', $readability->getContent());
         $this->assertInstanceOf('Readability\JSLikeHTMLElement', $readability->getTitle());
-        $this->assertContains('alt="article"', $readability->getContent()->innerHTML);
+        $this->assertContains('alt="article"', $readability->getContent()->getInnerHtml());
-        $this->assertEmpty($readability->getTitle()->innerHTML);
+        $this->assertEmpty($readability->getTitle()->getInnerHtml());
-        $this->assertContains('This is an awesome text with some links, here there are', $readability->getContent()->innerHTML);
+        $this->assertContains('This is an awesome text with some links, here there are', $readability->getContent()->getInnerHtml());
-        $this->assertContains('nofollow', $readability->getContent()->innerHTML);
+        $this->assertContains('nofollow', $readability->getContent()->getInnerHtml());
     }
 
     public function testWithAside()
@@ -190,10 +190,10 @@ class ReadabilityTest extends \PHPUnit\Framework\TestCase
         $this->assertTrue($res);
         $this->assertInstanceOf('Readability\JSLikeHTMLElement', $readability->getContent());
         $this->assertInstanceOf('Readability\JSLikeHTMLElement', $readability->getTitle());
-        $this->assertEmpty($readability->getTitle()->innerHTML);
+        $this->assertEmpty($readability->getTitle()->getInnerHtml());
-        $this->assertContains('This is an awesome text with some links, here there are', $readability->getContent()->innerHTML);
+        $this->assertContains('This is an awesome text with some links, here there are', $readability->getContent()->getInnerHtml());
-        $this->assertNotContains('<aside>', $readability->getContent()->innerHTML);
+        $this->assertNotContains('<aside>', $readability->getContent()->getInnerHtml());
-        $this->assertContains('<footer readability="4"/>', $readability->getContent()->innerHTML);
+        $this->assertContains('<footer readability="5"/>', $readability->getContent()->getInnerHtml());
     }
 
     public function testWithClasses()
@@ -205,10 +205,10 @@ class ReadabilityTest extends \PHPUnit\Framework\TestCase
         $this->assertTrue($res);
         $this->assertInstanceOf('Readability\JSLikeHTMLElement', $readability->getContent());
         $this->assertInstanceOf('Readability\JSLikeHTMLElement', $readability->getTitle());
-        $this->assertContains('alt="article"', $readability->getContent()->innerHTML);
+        $this->assertContains('alt="article"', $readability->getContent()->getInnerHtml());
-        $this->assertEmpty($readability->getTitle()->innerHTML);
+        $this->assertEmpty($readability->getTitle()->getInnerHtml());
-        $this->assertContains('This is an awesome text with some links, here there are', $readability->getContent()->innerHTML);
+        $this->assertContains('This is an awesome text with some links, here there are', $readability->getContent()->getInnerHtml());
-        $this->assertNotContains('This text should be removed', $readability->getContent()->innerHTML);
+        $this->assertNotContains('This text should be removed', $readability->getContent()->getInnerHtml());
     }
 
     public function testWithClassesWithoutLightClean()
@@ -221,10 +221,10 @@ class ReadabilityTest extends \PHPUnit\Framework\TestCase
         $this->assertTrue($res);
         $this->assertInstanceOf('Readability\JSLikeHTMLElement', $readability->getContent());
         $this->assertInstanceOf('Readability\JSLikeHTMLElement', $readability->getTitle());
-        $this->assertContains('alt="article"', $readability->getContent()->innerHTML);
+        $this->assertContains('alt="article"', $readability->getContent()->getInnerHtml());
-        $this->assertEmpty($readability->getTitle()->innerHTML);
+        $this->assertEmpty($readability->getTitle()->getInnerHtml());
-        $this->assertContains('This is an awesome text with some links, here there are', $readability->getContent()->innerHTML);
+        $this->assertContains('This is an awesome text with some links, here there are', $readability->getContent()->getInnerHtml());
-        $this->assertNotContains('This text should be removed', $readability->getContent()->innerHTML);
+        $this->assertNotContains('This text should be removed', $readability->getContent()->getInnerHtml());
     }
 
     public function testWithTd()
@@ -236,8 +236,8 @@ class ReadabilityTest extends \PHPUnit\Framework\TestCase
         $this->assertTrue($res);
         $this->assertInstanceOf('Readability\JSLikeHTMLElement', $readability->getContent());
         $this->assertInstanceOf('Readability\JSLikeHTMLElement', $readability->getTitle());
-        $this->assertEmpty($readability->getTitle()->innerHTML);
+        $this->assertEmpty($readability->getTitle()->getInnerHtml());
-        $this->assertContains('This is an awesome text with some links, here there are', $readability->getContent()->innerHTML);
+        $this->assertContains('This is an awesome text with some links, here there are', $readability->getContent()->getInnerHtml());
     }
 
     public function testWithSameClasses()
@@ -249,9 +249,9 @@ class ReadabilityTest extends \PHPUnit\Framework\TestCase
         $this->assertTrue($res);
         $this->assertInstanceOf('Readability\JSLikeHTMLElement', $readability->getContent());
         $this->assertInstanceOf('Readability\JSLikeHTMLElement', $readability->getTitle());
-        $this->assertEmpty($readability->getTitle()->innerHTML);
+        $this->assertEmpty($readability->getTitle()->getInnerHtml());
-        $this->assertContains('This is an awesome text with some links, here there are', $readability->getContent()->innerHTML);
+        $this->assertContains('This is an awesome text with some links, here there are', $readability->getContent()->getInnerHtml());
-        $this->assertContains('This text is also an awesome text and you should know that', $readability->getContent()->innerHTML);
+        $this->assertContains('This text is also an awesome text and you should know that', $readability->getContent()->getInnerHtml());
     }
 
     public function testWithScript()
@@ -263,9 +263,9 @@ class ReadabilityTest extends \PHPUnit\Framework\TestCase
         $this->assertTrue($res);
         $this->assertInstanceOf('Readability\JSLikeHTMLElement', $readability->getContent());
         $this->assertInstanceOf('Readability\JSLikeHTMLElement', $readability->getTitle());
-        $this->assertEmpty($readability->getTitle()->innerHTML);
+        $this->assertEmpty($readability->getTitle()->getInnerHtml());
-        $this->assertContains('This is an awesome text with some links, here there are', $readability->getContent()->innerHTML);
+        $this->assertContains('This is an awesome text with some links, here there are', $readability->getContent()->getInnerHtml());
-        $this->assertNotContains('This text is also an awesome text and you should know that', $readability->getContent()->innerHTML);
+        $this->assertNotContains('This text is also an awesome text and you should know that', $readability->getContent()->getInnerHtml());
     }
 
     public function testTitle()
@@ -277,9 +277,9 @@ class ReadabilityTest extends \PHPUnit\Framework\TestCase
         $this->assertTrue($res);
         $this->assertInstanceOf('Readability\JSLikeHTMLElement', $readability->getContent());
         $this->assertInstanceOf('Readability\JSLikeHTMLElement', $readability->getTitle());
-        $this->assertEquals('this is my title', $readability->getTitle()->innerHTML);
+        $this->assertEquals('this is my title', $readability->getTitle()->getInnerHtml());
-        $this->assertContains('This is an awesome text with some links, here there are', $readability->getContent()->innerHTML);
+        $this->assertContains('This is an awesome text with some links, here there are', $readability->getContent()->getInnerHtml());
-        $this->assertNotContains('This text is also an awesome text and you should know that', $readability->getContent()->innerHTML);
+        $this->assertNotContains('This text is also an awesome text and you should know that', $readability->getContent()->getInnerHtml());
     }
 
     public function testTitleWithDash()
@@ -291,9 +291,9 @@ class ReadabilityTest extends \PHPUnit\Framework\TestCase
         $this->assertTrue($res);
         $this->assertInstanceOf('Readability\JSLikeHTMLElement', $readability->getContent());
         $this->assertInstanceOf('Readability\JSLikeHTMLElement', $readability->getTitle());
-        $this->assertEquals('title2 - title3', $readability->getTitle()->innerHTML);
+        $this->assertEquals('title2 - title3', $readability->getTitle()->getInnerHtml());
-        $this->assertContains('This is an awesome text with some links, here there are', $readability->getContent()->innerHTML);
+        $this->assertContains('This is an awesome text with some links, here there are', $readability->getContent()->getInnerHtml());
-        $this->assertNotContains('This text is also an awesome text and you should know that', $readability->getContent()->innerHTML);
+        $this->assertNotContains('This text is also an awesome text and you should know that', $readability->getContent()->getInnerHtml());
     }
 
     public function testTitleWithDoubleDot()
@@ -305,9 +305,9 @@ class ReadabilityTest extends \PHPUnit\Framework\TestCase
         $this->assertTrue($res);
         $this->assertInstanceOf('Readability\JSLikeHTMLElement', $readability->getContent());
         $this->assertInstanceOf('Readability\JSLikeHTMLElement', $readability->getTitle());
-        $this->assertEquals('title2 : title3', $readability->getTitle()->innerHTML);
+        $this->assertEquals('title2 : title3', $readability->getTitle()->getInnerHtml());
-        $this->assertContains('This is an awesome text with some links, here there are', $readability->getContent()->innerHTML);
+        $this->assertContains('This is an awesome text with some links, here there are', $readability->getContent()->getInnerHtml());
-        $this->assertNotContains('This text is also an awesome text and you should know that', $readability->getContent()->innerHTML);
+        $this->assertNotContains('This text is also an awesome text and you should know that', $readability->getContent()->getInnerHtml());
     }
 
     public function testTitleTooShortUseH1()
@@ -319,9 +319,9 @@ class ReadabilityTest extends \PHPUnit\Framework\TestCase
         $this->assertTrue($res);
         $this->assertInstanceOf('Readability\JSLikeHTMLElement', $readability->getContent());
         $this->assertInstanceOf('Readability\JSLikeHTMLElement', $readability->getTitle());
-        $this->assertEquals('this is my h1 title !', $readability->getTitle()->innerHTML);
+        $this->assertEquals('this is my h1 title !', $readability->getTitle()->getInnerHtml());
-        $this->assertContains('This is an awesome text with some links, here there are', $readability->getContent()->innerHTML);
+        $this->assertContains('This is an awesome text with some links, here there are', $readability->getContent()->getInnerHtml());
-        $this->assertNotContains('This text is also an awesome text and you should know that', $readability->getContent()->innerHTML);
+        $this->assertNotContains('This text is also an awesome text and you should know that', $readability->getContent()->getInnerHtml());
     }
 
     // dummy function to be used to the next test
@@ -369,8 +369,8 @@ class ReadabilityTest extends \PHPUnit\Framework\TestCase
         $this->assertTrue($res);
         $this->assertInstanceOf('Readability\JSLikeHTMLElement', $readability->getContent());
         $this->assertInstanceOf('Readability\JSLikeHTMLElement', $readability->getTitle());
-        $this->assertContains('<iframe src="https://www.youtube.com/embed/PUep6xNeKjA" width="560" height="315" frameborder="0" allowfullscreen="allowfullscreen"> </iframe>', $readability->getContent()->innerHTML);
+        $this->assertContains('<iframe src="https://www.youtube.com/embed/PUep6xNeKjA" width="560" height="315" frameborder="0" allowfullscreen="allowfullscreen"> </iframe>', $readability->getContent()->getInnerHtml());
-        $this->assertContains('3D Touch', $readability->getTitle()->innerHTML);
+        $this->assertContains('3D Touch', $readability->getTitle()->getInnerHtml());
     }
 
     /**
@@ -440,7 +440,7 @@ class ReadabilityTest extends \PHPUnit\Framework\TestCase
         $res = $readability->init();
 
         $this->assertTrue($res);
-        $this->assertContains('This  the awesome content :)', $readability->getContent()->innerHTML);
+        $this->assertContains('This  the awesome content :)', $readability->getContent()->getInnerHtml());
     }
 
     public function testPreFilters()
@@ -453,7 +453,7 @@ class ReadabilityTest extends \PHPUnit\Framework\TestCase
         $res = $readability->init();
 
         $this->assertTrue($res);
-        $this->assertContains('This the awesome and WONDERFUL content :)', $readability->getContent()->innerHTML);
+        $this->assertContains('This the awesome and WONDERFUL content :)', $readability->getContent()->getInnerHtml());
     }
 
     public function testChildNodeGoneNull()
@@ -479,8 +479,8 @@ class ReadabilityTest extends \PHPUnit\Framework\TestCase
         $res = $readability->init();
 
         $this->assertTrue($res);
-        $this->assertContains('<sup id="fnref1:fnfeed_2"><a href="#fn:fnfeed_2" class="footnote-ref">2</a></sup>', $readability->getContent()->innerHTML);
+        $this->assertContains('<sup id="fnref1:fnfeed_2"><a href="#fn:fnfeed_2" class="footnote-ref">2</a></sup>', $readability->getContent()->getInnerHtml());
-        $this->assertContains('<a href="#fnref1:fnfeed_2" rev="footnote"', $readability->getContent()->innerHTML);
+        $this->assertContains('<a href="#fnref1:fnfeed_2" rev="footnote"', $readability->getContent()->getInnerHtml());
     }
 
     private function getReadability($html, $url = null, $parser = 'libxml', $useTidy = true)