fixes leafo#395 - bin/pscss: add --iso8859-1 option; defaults to utf-8 (slower but more compatible with ruby scss)

Author: robocoder

bin/pscss

@@ -31,6 +31,7 @@ $changeDir = false;
 $debugInfo = false;
 $lineNumbers = false;
 $ignoreErrors = false;
+$encoding = false;
 
 /**
  * Parse argument
@@ -73,6 +74,7 @@ Options include:
     --debug-info        Annotate selectors with CSS referring to the source file and line number
     -f=format           Set the output format (compact, compressed, crunched, expanded, or nested)
     -i=path             Set import path
+    --iso8859-1         Use iso8859-1 encoding instead of utf-8 (default utf-8)
     --line-numbers      Annotate selectors with comments referring to the source file and line number
     -p=precision        Set decimal number precision (default 5)
     -T                  Dump formatted parse tree
@@ -96,6 +98,11 @@ EOT;
         continue;
     }
 
+    if ($argv[$i] === '--iso8859-1') {
+        $encoding = 'iso8859-1';
+        continue;
+    }
+
     if ($argv[$i] === '--line-numbers' || $argv[$i] === '--line-comments') {
         $lineNumbers = true;
         continue;
@@ -186,6 +193,10 @@ if ($style) {
     $scss->setFormatter('Leafo\\ScssPhp\\Formatter\\' . ucfirst($style));
 }
 
+if ($encoding) {
+    $scss->setEncoding($encoding);
+}
+
 echo $scss->compile($data, $inputFile);
 
 if ($changeDir) {

src/Compiler.php

@@ -117,6 +117,7 @@ class Compiler
         'global-variable-shadowing'   => false,
     );
 
+    protected $encoding = null;
     protected $lineNumberStyle = null;
 
     protected $formatter = 'Leafo\ScssPhp\Formatter\Nested';
@@ -208,7 +209,7 @@ public function compile($code, $path = null)
      */
     private function parserFactory($path)
     {
-        $parser = new Parser($path, count($this->sourceNames));
+        $parser = new Parser($path, count($this->sourceNames), $this->encoding);
 
         $this->sourceNames[] = $path;
         $this->addParsedFile($path);
@@ -3239,6 +3240,18 @@ public function findImport($url)
         return null;
     }
 
+    /**
+     * Set encoding
+     *
+     * @api
+     *
+     * @param string $encoding
+     */
+    public function setEncoding($encoding)
+    {
+        $this->encoding = $encoding;
+    }
+
     /**
      * Ignore errors?
      *

src/Parser.php

@@ -62,6 +62,7 @@ class Parser
     private $inParens;
     private $eatWhiteDefault;
     private $buffer;
+    private $utf8;
 
     /**
      * Constructor
@@ -70,12 +71,14 @@ class Parser
      *
      * @param string  $sourceName
      * @param integer $sourceIndex
+     * @param string  $encoding
      */
-    public function __construct($sourceName, $sourceIndex = 0)
+    public function __construct($sourceName, $sourceIndex = 0, $encoding = 'utf-8')
     {
         $this->sourceName  = $sourceName ?: '(stdin)';
         $this->sourceIndex = $sourceIndex;
         $this->charset     = null;
+        $this->utf8        = ! $encoding || strtolower($encoding) === 'utf-8';
 
         if (empty(self::$operatorPattern)) {
             self::$operatorPattern = '([*\/%+-]|[!=]\=|\>\=?|\<\=\>|\<\=?|and|or)';
@@ -85,7 +88,9 @@ public function __construct($sourceName, $sourceIndex = 0)
             $commentMultiRight  = '\*\/';
 
             self::$commentPattern = $commentMultiLeft . '.*?' . $commentMultiRight;
-            self::$whitePattern = '/' . $commentSingle . '[^\n]*\s*|(' . self::$commentPattern . ')\s*|\s+/AisuS';
+            self::$whitePattern = $this->utf8
+                ? '/' . $commentSingle . '[^\n]*\s*|(' . self::$commentPattern . ')\s*|\s+/AisuS'
+                : '/' . $commentSingle . '[^\n]*\s*|(' . self::$commentPattern . ')\s*|\s+/AisS';
         }
     }
 
@@ -762,7 +767,7 @@ protected function peek($regex, &$out, $from = null)
             $from = $this->count;
         }
 
-        $r = '/' . $regex . '/Aisu';
+        $r = $this->utf8 ? '/' . $regex . '/Aisu' : '/' . $regex . '/Ais';
         $result = preg_match($r, $this->buffer, $out, null, $from);
 
         return $result;
@@ -842,7 +847,7 @@ protected function match($regex, &$out, $eatWhitespace = null)
             $eatWhitespace = $this->eatWhiteDefault;
         }
 
-        $r = '/' . $regex . '/Aisu';
+        $r = $this->utf8 ? '/' . $regex . '/Aisu' : '/' . $regex . '/Ais';
 
         if (preg_match($r, $this->buffer, $out, null, $this->count)) {
             $this->count += strlen($out[0]);
@@ -2235,7 +2240,9 @@ protected function variable(&$out)
     protected function keyword(&$word, $eatWhitespace = null)
     {
         if ($this->match(
-            '(([\pL\w_\-\*!"\']|[\\\\].)([\pL\w\-_"\']|[\\\\].)*)',
+            $this->utf8
+                ? '(([\pL\w_\-\*!"\']|[\\\\].)([\pL\w\-_"\']|[\\\\].)*)'
+                : '(([\w_\-\*!"\']|[\\\\].)([\w\-_"\']|[\\\\].)*)',
             $m,
             $eatWhitespace
         )) {
@@ -2256,7 +2263,12 @@ protected function keyword(&$word, $eatWhitespace = null)
      */
     protected function placeholder(&$placeholder)
     {
-        if ($this->match('([\pL\w\-_]+|#[{][$][\pL\w\-_]+[}])', $m)) {
+        if ($this->match(
+            $this->utf8
+                ? '([\pL\w\-_]+|#[{][$][\pL\w\-_]+[}])'
+                : '([\w\-_]+|#[{][$][\w\-_]+[}])',
+            $m
+        )) {
             $placeholder = $m[1];
 
             return true;