Merge pull request #30 from j0k3r/avoid-php-warning

Avoid php warning
Fix tests
2026-09-27 06:26:17 +00:00 · 2017-05-19 16:16:20 +02:00 · 2017-05-19 16:02:36 +02:00 · 2017-05-19 15:37:46 +02:00 · 2017-03-18 09:16:30 +01:00 · 2017-03-18 04:02:50 +01:00
6 changed files with 61 additions and 49 deletions
@@ -46,7 +46,7 @@ before_script:

 script:
    - mkdir -p build/logs
-    - phpunit -v --coverage-clover build/logs/clover.xml
+    - 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;

 after_script:
@@ -17,9 +17,9 @@ The default php-readability lib is really old and needs to be improved. I found

 ## Requirements

-By default, this lib will use the [Tidy extension](https://github.com/htacg/tidy-html5) if it's available. Tidy is only used to cleanup the given HTML and avoid problems with bad HTML structure, etc ..
+By default, this lib will use the [Tidy extension](https://github.com/htacg/tidy-html5) if it's available. Tidy is only used to cleanup the given HTML and avoid problems with bad HTML structure, etc .. It'll be suggested by Composer.

-Since Composer doesn't support suggestion on PHP extension, I write this suggestion here.
+Also, if you got problem from parsing a content without Tidy installed, please install it and try again.

 ## Usage

@@ -25,12 +25,17 @@
    }],
    "require": {
        "php": ">=5.3.3",
+        "ext-mbstring": "*",
        "psr/log": "^1.0"
    },
    "require-dev": {
        "satooshi/php-coveralls": "~0.6",
        "friendsofphp/php-cs-fixer": "<2",
-        "monolog/monolog": "^1.13"
+        "monolog/monolog": "^1.13",
+        "symfony/phpunit-bridge": "^3.2"
+    },
+    "suggest": {
+        "ext-tidy": "Used to clean up given HTML and to avoid problems with bad HTML structure."
    },
    "autoload": {
        "psr-4": { "Readability\\": "src/" }
@@ -45,52 +45,59 @@ class JSLikeHTMLElement extends \DOMElement
     */
    public function __set($name, $value)
    {
-        if ($name === 'innerHTML') {
-            // first, empty the element
-            for ($x = $this->childNodes->length - 1; $x >= 0; --$x) {
-                $this->removeChild($this->childNodes->item($x));
-            }
-
-            // $value holds our new inner HTML
-            if ($value !== '') {
-                $f = $this->ownerDocument->createDocumentFragment();
-
-                // appendXML() expects well-formed markup (XHTML)
-                // @ to suppress PHP warnings
-                $result = @$f->appendXML($value);
-                if ($result) {
-                    if ($f->hasChildNodes()) {
-                        $this->appendChild($f);
-                    }
-                } else {
-                    // $value is probably ill-formed
-                    $f = new \DOMDocument();
-                    $value = mb_convert_encoding($value, 'HTML-ENTITIES', 'UTF-8');
-
-                    // Using <htmlfragment> will generate a warning, but so will bad HTML
-                    // (and by this point, bad HTML is what we've got).
-                    // We use it (and suppress the warning) because an HTML fragment will
-                    // be wrapped around <html><body> tags which we don't really want to keep.
-                    // Note: despite the warning, if loadHTML succeeds it will return true.
-                    $result = @$f->loadHTML('<htmlfragment>' . $value . '</htmlfragment>');
-
-                    if ($result) {
-                        $import = $f->getElementsByTagName('htmlfragment')->item(0);
-
-                        foreach ($import->childNodes as $child) {
-                            $importedNode = $this->ownerDocument->importNode($child, true);
-                            $this->appendChild($importedNode);
-                        }
-                    } else {
-                        // oh well, we tried, we really did. :(
-                        // this element is now empty
-                    }
-                }
-            }
-        } else {
+        if ($name !== 'innerHTML') {
            $trace = debug_backtrace();
            trigger_error('Undefined property via __set(): ' . $name . ' in ' . $trace[0]['file'] . ' on line ' . $trace[0]['line'], E_USER_NOTICE);
+
+            return;
        }
+
+        // first, empty the element
+        for ($x = $this->childNodes->length - 1; $x >= 0; --$x) {
+            $this->removeChild($this->childNodes->item($x));
+        }
+
+        // $value holds our new inner HTML
+        $value = trim($value);
+        if (empty($value)) {
+            return;
+        }
+
+        // ensure bad entity won't generate warning
+        $previousError = libxml_use_internal_errors(true);
+
+        $f = $this->ownerDocument->createDocumentFragment();
+
+        // appendXML() expects well-formed markup (XHTML)
+        $result = $f->appendXML($value);
+        if ($result) {
+            if ($f->hasChildNodes()) {
+                $this->appendChild($f);
+            }
+        } else {
+            // $value is probably ill-formed
+            $f = new \DOMDocument();
+            $value = mb_convert_encoding($value, 'HTML-ENTITIES', 'UTF-8');
+
+            // Using <htmlfragment> will generate a warning, but so will bad HTML
+            // (and by this point, bad HTML is what we've got).
+            // We use it (and suppress the warning) because an HTML fragment will
+            // be wrapped around <html><body> tags which we don't really want to keep.
+            // Note: despite the warning, if loadHTML succeeds it will return true.
+            $result = $f->loadHTML('<htmlfragment>' . $value . '</htmlfragment>');
+
+            if ($result) {
+                $import = $f->getElementsByTagName('htmlfragment')->item(0);
+
+                foreach ($import->childNodes as $child) {
+                    $importedNode = $this->ownerDocument->importNode($child, true);
+                    $this->appendChild($importedNode);
+                }
+            }
+        }
+
+        libxml_clear_errors();
+        libxml_use_internal_errors($previousError);
    }

    /**
@@ -83,7 +83,7 @@ class Readability implements LoggerAwareInterface
     * Defined up here so we don't instantiate them repeatedly in loops.
     */
    public $regexps = array(
-        'unlikelyCandidates' => '/display\s*:\s*none|ignore|\binfos?\b|annoy|clock|date|time|author|intro|links|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',
+        '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/i',
        'positive' => '/read|full|article|body|\bcontent|contain|entry|main|markdown|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',
@@ -443,7 +443,7 @@ class ReadabilityTest extends \PHPUnit_Framework_TestCase

    public function testPostFilters()
    {
-        $readability = $this->getReadability('<div>' . str_repeat('<p>This <b>is</b> the awesome content :)</p>', 7) . '</div>', 'http://0.0.0.0');
+        $readability = $this->getReadability('<div>' . str_repeat('<p>This <b>is</b> the awesome content :)</p>', 10) . '</div>', 'http://0.0.0.0');
        $readability->addPostFilter('!<strong[^>]*>(.*?)</strong>!is', '');

        $res = $readability->init();
Author	SHA1	Message	Date
Jérémy BenoistandGitHub	5ac82f3c87	Merge pull request #30 from j0k3r/avoid-php-warning Avoid php warning	2017-05-19 16:16:20 +02:00
Jeremy Benoist	85fb92a042	Fix tests	2017-05-19 16:02:36 +02:00
Jeremy Benoist	f2a43b476c	Avoid PHP Warning This isn't the best solution but the previous one using `@` wasn't really better. Appending a string into a fragment might generate some warning if the string contains bad entity. For example `+`.	2017-05-19 15:37:46 +02:00
Jérémy BenoistandGitHub	6def743902	Merge pull request #27 from jtojnar/ext-mbstring Add mbstring as a dependency	2017-03-18 09:16:30 +01:00
Jan Tojnar	8a44926392	Add mbstring as a dependency	2017-03-18 04:02:50 +01:00
Jérémy BenoistandGitHub	8c7740f073	Merge pull request #26 from j0k3r/fix-phpunit-7.x Fix PHPUnit error on PHP 7+	2017-03-09 09:41:33 +01:00
Jeremy Benoist	6a4720c951	Fix PHPUnit error on PHP 7+	2017-03-09 09:21:13 +01:00
Jérémy BenoistandGitHub	7a9a82b543	Update README.md	2017-03-09 09:02:04 +01:00
Jérémy BenoistandGitHub	8a91d36294	Merge pull request #25 from jtojnar/patch-1 Add tidy to suggested dependenies	2017-03-09 09:01:05 +01:00
Jan TojnarandGitHub	669adfb20f	Add tidy to suggested dependenies	2017-02-16 15:12:45 +01:00
Jeremy BenoistandGitHub	445b889efb	Merge pull request #24 from j0k3r/links Don't be to hard on 'links' attribute	2017-02-02 16:12:48 +01:00
Jeremy Benoist	8b1c3f147d	Don't be to hard on 'links' attribute	2017-02-02 15:57:15 +01:00