improved URL handling

Author: ju1ius

CSSParser.php

@@ -85,23 +85,21 @@ public function parseFile($sPath, $aLoadedFiles=array())
     if(!$this->aOptions['base_url']) {
       $this->aOptions['base_url'] = dirname($sPath);
     }
-    if($this->aOptions['absolute_urls']) {
+    if($this->aOptions['absolute_urls'] && !CSSUrlUtils::isAbsUrl($this->aOptions['base_url'])) {
       $this->aOptions['base_url'] = realpath($this->aOptions['base_url']);
     }
     $this->sImportMode = self::IMPORT_FILE;
     $sPath = realpath($sPath);
     $aLoadedFiles[] = $sPath;
     $this->aLoadedFiles = array_merge($this->aLoadedFiles, $aLoadedFiles);
     $sCss = file_get_contents($sPath);
-    var_dump($sPath);
-    
     return $this->parseString($sCss);
   }
 
   public function parseURL($sPath, $aLoadedFiles=array())
   {
     if(!$this->aOptions['base_url']) {
-      $this->aOptions['base_url'] = dirname($sPath);
+      $this->aOptions['base_url'] = CSSUrlUtils::dirname($sPath);
     }
     $this->sImportMode = self::IMPORT_URL;
     $aLoadedFiles[] =$sPath;
@@ -117,7 +115,7 @@ public function parseURL($sPath, $aLoadedFiles=array())
 
 
   public function parseString($sString, $sCharset=null) {
-    echo "First 4 bytes: " . CSSCharsetUtils::printBytes($sString, 4) . PHP_EOL;
+    //echo "First 4 bytes: " . CSSCharsetUtils::printBytes($sString, 4) . PHP_EOL;
     if(!$sCharset) {
       // detect charset from BOM and/or @charset rule
       $sCharset = CSSCharsetUtils::detectCharset($sString);

lib/CSSUrlUtils.php

@@ -4,7 +4,16 @@
 
 class CSSUrlUtils {
 
-  static function loadURL($sURL) {
+  /**
+   * Requests the contents of an URL
+   *
+   * @param   string $sURL the URL to fetch
+   * @returns array        an array in the form:
+   *   'charset'  => the charset of the response as specified by the
+   *                 HTTP Content-Type header, if specified
+   *   'response' => the response body
+   **/
+  static public function loadURL($sURL) {
     $rCurl = curl_init();
     curl_setopt($rCurl, CURLOPT_URL, $sURL);
     //curl_setopt($rCurl, CURLOPT_HEADER, true);
@@ -21,7 +30,7 @@ static function loadURL($sURL) {
       'response' => $mResponse  
     );
     if($aInfos['content_type']) {
-      if (preg_match('/charset=([a-zA-Z0-9-]*)/', $aInfos['content_type'], $aMatches)) {
+      if(preg_match('/charset=([a-zA-Z0-9-]*)/', $aInfos['content_type'], $aMatches)) {
         $aResult['charset'] = $aMatches[0];
       }
     }
@@ -31,12 +40,11 @@ static function loadURL($sURL) {
   /**
    * CSSUrlUtils::joinPaths( string $head, string $tail [, string $...] )
    *
-   * @param $head string the head component of the path
-   * @param $tail string at least one path component
-   * @returns string the resulting path
+   * @param   string $head the head component of the path
+   * @param   string $tail at least one path component
+   * @returns string       the resulting path
    **/
-  static function joinPaths()
-  {
+  static public function joinPaths() {
     $num_args = func_num_args();
     if($num_args < 1) return '';
     $args = func_get_args();
@@ -54,12 +62,10 @@ static function joinPaths()
   /**
    * Returns boolean based on whether given path is absolute or not.
    *
-   * @static
-   * @access  public
    * @param   string  $path Given path
-   * @return  boolean True if the path is absolute, false if it is not
+   * @return  boolean       True if the path is absolute, false if it is not
    */
-  static function isAbsPath($sPath) {
+  static public function isAbsPath($sPath) {
     if (preg_match('#(?:/|\\\)\.\.(?=/|$)#', $sPath)) {
       return false;
     }
@@ -69,8 +75,70 @@ static function isAbsPath($sPath) {
     return ($sPath[0] == '/') || ($sPath[0] == '~');
   }
 
-  static function isAbsURL($sPath)
+  /**
+   * Tests if an URL is absolute
+   *
+   * @param  string  $sURL
+   * @return boolean
+   **/
+  static public function isAbsURL($sURL) {
+    return preg_match('#^(http|https|ftp)://#', $sURL);
+  }
+
+  /**
+   * Returns the parent path of an URL or path
+   * 
+   * @param   string $sURL an URL
+   * @returns string       an URL
+   **/
+  static public function dirname($sURL) {
+    $aURL = parse_url($sURL);
+    if(isset($aURL['path'])) {
+      $sPath = dirname($aURL['path']);
+      if($sPath == '/') {
+        unset($aURL['path']);
+      } else {
+        $aURL['path'] = $sPath;
+      }
+    }
+    return self::buildURL($aURL);
+  }
+  
+  /**
+   * Builds an URL from an array of URL parts
+   *
+   * @param  array  $aURL   URL parts in the format returned by parse_url
+   * @return string         the builded URL
+   * @see http://php.net/manual/function.parse-url.php 
+   **/
+  static public function buildURL(array $aURL)
   {
-    return preg_match('#^(http|https|ftp)://#', $sPath);
+    if(isset($aURL['scheme'])) {
+      $sURL .= $aURL['scheme'] . '://';
+    }
+    if(isset($aURL['user'])) {
+      $sURL .= $aURL['user'];
+      if(isset($aURL['pass'])) {
+        $sURL .= ':' . $aURL['pass'];
+      }
+      $sURL .= '@';
+    }
+    if(isset($aURL['host'])) {
+      $sURL .= $aURL['host'];
+    }
+    if(isset($aURL['port'])) {
+      $sURL .= ':' . $aURL['port'];
+    }
+    if(isset($aURL['path'])) {
+      if(strpos($aURL['path'], '/') !== 0) $sURL .= '/';
+      $sURL .= $aURL['path'];
+    }
+    if(isset($aURL['query'])) {
+      $sURL .= '?' . $aURL['query'];
+    }
+    if(isset($aURL['fragment'])) {
+      $sURL .= '#' . $aURL['fragment'];
+    }
+    return $sURL;
   }
 }

tests/CSSImportTest.php

@@ -40,7 +40,8 @@ public function testAbsoluteUrlsProvider()
   public function testResolveImports()
   {
     $oParser = new CSSParser(array('resolve_imports' => true));
-    $oDoc = $oParser->parseFile("files/import.css");
+    $oDoc = $oParser->parseFile(dirname(__FILE__)."/files/import.css");
+    echo $oDoc . PHP_EOL;
   }
 
 }

tests/files/import.css

@@ -1,6 +1,5 @@
 @import "imported.css";
 @import "import/utf-16.css";
-@import "unicode.css" screen, print;
 
 body#icomelast{
   color: fuschia;