Added charset and unicode support

Author: JanDC

src/Css/Processor.php

@@ -11,7 +11,7 @@ class Processor
      * Get the rules from a given CSS-string
      *
      * @param string $css
-     * @param array  $existingRules
+     * @param array $existingRules
      * @return Rule[]
      */
     public function getRules($css, $existingRules = [])
@@ -23,6 +23,17 @@ public function getRules($css, $existingRules = [])
         return $rulesProcessor->convertArrayToObjects($rulesByMediaQuery, $existingRules);
     }
 
+    public function getCharset($css)
+    {
+        preg_match_all('/@charset "[^"]++";/', $css, $matches);
+
+        if ($charset = reset($matches)) {
+            return reset($charset);
+        }
+
+        return null;
+    }
+
     /**
      * Get the CSS from the style-tags in the given HTML-string
      *

src/CssFromHTMLExtractor.php

@@ -59,6 +59,7 @@ public function getHtmlStore()
     public function addBaseRules($sourceCss)
     {
         $this->rules = $this->processor->getRules($sourceCss, $this->rules);
+        $this->getCssStore()->setCharset($this->processor->getCharset($sourceCss));
     }
 
     public function buildExtractedRuleSet()

src/CssStore.php

@@ -10,6 +10,9 @@ class CssStore
     /** @var array Property objects, grouped by selector */
     private $styles = [];
 
+    /** @var  string|null */
+    private $charset;
+
     public function addCssStyles($cssRules)
     {
         $this->styles = array_merge($this->styles, $cssRules);
@@ -38,19 +41,20 @@ public function compileStyles()
         $styles = $this->prepareStylesForProcessing();
 
 
+        $prefix = is_null($this->charset) ? '' : $this->charset;
 
-        return join(
-            '',
-            array_map(
-                function ($styleGroup) {
-                    $media = key($styleGroup);
-                    $rules = reset($styleGroup);
+        return $prefix . join(
+                '',
+                array_map(
+                    function ($styleGroup) {
+                        $media = key($styleGroup);
+                        $rules = reset($styleGroup);
 
-                    return $this->parseMediaToString($media, $rules);
-                },
-                $styles
-            )
-        );
+                        return $this->parseMediaToString($media, $rules);
+                    },
+                    $styles
+                )
+            );
     }
 
     /**
@@ -68,8 +72,7 @@ private function parseMediaToString($media, array $rules)
                 join(
                     '',
                     array_map(
-                        function ($rule) {
-                            /** @var Rule $rule */
+                        function (Rule $rule) {
                             return $this->parsePropertiesToString($rule->getSelector(), $rule->getProperties());
                         },
                         $rules
@@ -78,16 +81,15 @@ function ($rule) {
 
         }
 
-        return "$media { ".join(
+        return "$media { " . join(
                 '',
                 array_map(
-                    function ($rule) {
-                        /** @var Rule $rule */
+                    function (Rule $rule) {
                         return $this->parsePropertiesToString($rule->getSelector(), $rule->getProperties());
                     },
                     $rules
                 )
-            )."}";
+            ) . "}";
 
 
     }
@@ -101,16 +103,18 @@ function ($rule) {
      */
     private function parsePropertiesToString($selector, array $properties)
     {
-        return "$selector { ".
+        $selector = mb_convert_encoding($selector, 'unicode');
+
+        return "$selector { " .
             join(
                 '',
                 array_map(
                     function (Property $property) {
-                        return $property->getName().': '.$property->getValue().';';
+                        return mb_convert_encoding($property->getName(), 'unicode') . ': ' . mb_convert_encoding($property->getValue(), 'unicode') . ';';
                     },
                     $properties
                 )
-            ).
+            ) .
             "}";
     }
 
@@ -127,4 +131,10 @@ private function prepareStylesForProcessing()
 
         return $groupedStyles;
     }
+
+    public function setCharset($charset)
+    {
+        $this->charset = $charset;
+
+    }
 }