Merge pull request #7 from blaaat/master

alcaeus · web-flow · commit 402f2e5fc9b3 · 2018-03-26T06:37:23.000+02:00
Ignore ; or space after charset in meta tag
diff --git a/src/PHPHtmlParser/Dom.php b/src/PHPHtmlParser/Dom.php
@@ -633,7 +633,7 @@ protected function detectCharset()
             return false;
         }
         $matches = [];
-        if (preg_match('/charset=(.+)/', $content, $matches)) {
+        if (preg_match('/charset=([^; ]+)/', $content, $matches)) {
             $encode->from(trim($matches[1]));
             $this->root->propagateEncoding($encode);
 
diff --git a/src/PHPHtmlParser/Dom/AbstractNode.php b/src/PHPHtmlParser/Dom/AbstractNode.php
@@ -70,8 +70,8 @@ public function __construct()
     public function __get($key)
     {
         // check attribute first
-        if ( ! is_null($this->getAttribute($key))) {
-            return $this->getAttribute($key);
+        if ( ! is_null($value = $this->getAttribute($key))) {
+            return $value;
         }
         switch (strtolower($key)) {
             case 'outerhtml':
diff --git a/src/PHPHtmlParser/Dom/Tag.php b/src/PHPHtmlParser/Dom/Tag.php
@@ -146,6 +146,7 @@ public function setAttribute($key, $value)
             $value = [
                 'value'       => $value,
                 'doubleQuote' => true,
+                'didConvertCharset' => true, 
             ];
         }
         $this->attr[$key] = $value;
@@ -217,9 +218,10 @@ public function getAttribute($key)
             return null;
         }
         $value = $this->attr[$key]['value'];
-        if (is_string($value) && ! is_null($this->encode)) {
+        if (is_string($value) && ! is_null($this->encode) && ! ($this->attr[$key]['didConvertCharset'] ?? false)) {
             // convert charset
             $this->attr[$key]['value'] = $this->encode->convert($value);
+            $this->attr[$key]['didConvertCharset'] = true;
         }
 
         return $this->attr[$key];
diff --git a/tests/DomTest.php b/tests/DomTest.php
@@ -33,6 +33,21 @@ public function testIncorrectAccess()
         $div = $dom->find('div', 0);
         $this->assertEquals(null, $div->foo);
     }
+    public function testIncorrectContentType()
+    {
+        $dom = new Dom;
+        $dom->load('<html><head><meta http-equiv="Content-Type" content="text/html; charset=UTF-8;"></head><body><div class="all"><p>Hey bro, <a href="google.com">click here</a><br /> :)</p></div></body></html>');
+        $a = $dom->find('a', 0);
+        $this->assertEquals($a->text, 'click here');
+    }
+
+    public function testCharsetConvertInAttribute()
+    {
+        $dom = new Dom;
+        $dom->loadFromFile(__DIR__ . '/files/ISO-8859-7.html', ['preserveLineBreaks' => true]);
+        $a = $dom->find('a', 0);
+        $this->assertEquals('/testη', $a->href);
+    }
 
     public function testLoadSelfclosingAttr()
     {
diff --git a/tests/files/ISO-8859-7.html b/tests/files/ISO-8859-7.html
@@ -0,0 +1,8 @@
+<html>
+	<head>
+		<meta http-equiv="Content-Type" content="text/html; charset=ISO-8859-7">
+	</head>
+	<body>
+		<a href="/test�">EE�<span>XX</span></a>
+	</body>
+</html>

Original file line number	Diff line number	Diff line change
`@@ -633,7 +633,7 @@ protected function detectCharset()`
`633`	`633`	`return false;`
`634`	`634`	`}`
`635`	`635`	`$matches = [];`
`636`		`- if (preg_match('/charset=(.+)/', $content, $matches)) {`
	`636`	`+ if (preg_match('/charset=([^; ]+)/', $content, $matches)) {`
`637`	`637`	`$encode->from(trim($matches[1]));`
`638`	`638`	`$this->root->propagateEncoding($encode);`
`639`	`639`
Original file line number	Diff line number	Diff line change
`@@ -70,8 +70,8 @@ public function __construct()`
`70`	`70`	`public function __get($key)`
`71`	`71`	`{`
`72`	`72`	`// check attribute first`
`73`		`- if ( ! is_null($this->getAttribute($key))) {`
`74`		`- return $this->getAttribute($key);`
	`73`	`+ if ( ! is_null($value = $this->getAttribute($key))) {`
	`74`	`+ return $value;`
`75`	`75`	`}`
`76`	`76`	`switch (strtolower($key)) {`
`77`	`77`	`case 'outerhtml':`
Original file line number	Diff line number	Diff line change
`@@ -146,6 +146,7 @@ public function setAttribute($key, $value)`
`146`	`146`	`$value = [`
`147`	`147`	`'value' => $value,`
`148`	`148`	`'doubleQuote' => true,`
	`149`	`+ 'didConvertCharset' => true,`
`149`	`150`	`];`
`150`	`151`	`}`
`151`	`152`	`$this->attr[$key] = $value;`
`@@ -217,9 +218,10 @@ public function getAttribute($key)`
`217`	`218`	`return null;`
`218`	`219`	`}`
`219`	`220`	`$value = $this->attr[$key]['value'];`
`220`		`- if (is_string($value) && ! is_null($this->encode)) {`
	`221`	`+ if (is_string($value) && ! is_null($this->encode) && ! ($this->attr[$key]['didConvertCharset'] ?? false)) {`
`221`	`222`	`// convert charset`
`222`	`223`	`$this->attr[$key]['value'] = $this->encode->convert($value);`
	`224`	`+ $this->attr[$key]['didConvertCharset'] = true;`
`223`	`225`	`}`
`224`	`226`
`225`	`227`	`return $this->attr[$key];`
-Original file line number
+Diff line change
@@ @@ -0,0 +1,8 @@ @@
 +<html>
 +	<head>
 +		<meta http-equiv="Content-Type" content="text/html; charset=ISO-8859-7">
 +	</head>
 +	<body>
 +		<a href="/testç">EEç<span>XX</span></a>
 +	</body>
 +</html>