fixed WordpressParser issue with offsets, fixed RegularParser bug whe…

…n nesting just-closed shortcodes, removed internal isRoot variable

src/Parser/RegularParser.php

@@ -47,7 +47,9 @@ public function parse($text)
             while($this->position < $this->tokensCount && !$this->lookahead(self::TOKEN_OPEN)) {
                 $this->position++;
             }
-            foreach($this->shortcode(true) ?: array() as $shortcode) {
+            $names = array();
+            $this->beginBacktrack();
+            foreach($this->shortcode($names) ?: array() as $shortcode) {
                 $shortcodes[] = $shortcode;
             }
         }
@@ -62,15 +64,14 @@ private function getObject($name, $parameters, $bbCode, $offset, $content, $text
 
     /* --- RULES ----------------------------------------------------------- */
 
-    private function shortcode($isRoot)
+    private function shortcode(array &$names)
     {
         $name = null;
         $offset = null;
 
         $setName = function(array $token) use(&$name) { $name = $token[1]; };
         $setOffset = function(array $token) use(&$offset) { $offset = $token[2]; };
 
-        $isRoot && $this->beginBacktrack();
         if(!$this->match(self::TOKEN_OPEN, $setOffset, true)) { return false; }
         if(!$this->match(self::TOKEN_STRING, $setName, false)) { return false; }
         if($this->lookahead(self::TOKEN_STRING, null)) { return false; }
@@ -89,22 +90,32 @@ private function shortcode($isRoot)
         // just-closed or with-content
         if(!$this->match(self::TOKEN_CLOSE)) { return false; }
         $this->beginBacktrack();
-        list($content, $shortcodes) = $this->content($name);
+        $names[] = $name;
+        list($content, $shortcodes, $closingName) = $this->content($names);
+        if(null !== $closingName && $closingName !== $name) {
+            array_pop($names);
+            array_pop($this->backtracks);
+            array_pop($this->backtracks);
+
+            return $closingName;
+        }
         if(false === $content) {
             $this->backtrack(false);
             $text = $this->backtrack(false);
+
             return array_merge(array($this->getObject($name, $parameters, $bbCode, $offset, null, $text)), $shortcodes);
         }
-        array_pop($this->backtracks);
-        if(!$this->close($name)) { return false; }
+        $content = $this->getBacktrack();
+        if(!$this->close($names)) { return false; }
 
         return array($this->getObject($name, $parameters, $bbCode, $offset, $content, $this->getBacktrack()));
     }
 
-    private function content($name)
+    private function content(array &$names)
     {
         $content = null;
         $shortcodes = array();
+        $closingName = null;
         $appendContent = function(array $token) use(&$content) { $content .= $token[1]; };
 
         while($this->position < $this->tokensCount) {
@@ -113,20 +124,25 @@ private function content($name)
             }
 
             $this->beginBacktrack();
-            $matchedShortcodes = $this->shortcode(false);
+            $matchedShortcodes = $this->shortcode($names);
+            if(is_string($matchedShortcodes)) {
+                $closingName = $matchedShortcodes;
+                break;
+            }
             if(false !== $matchedShortcodes) {
                 $shortcodes = array_merge($shortcodes, $matchedShortcodes);
                 continue;
             }
             $this->backtrack();
 
             $this->beginBacktrack();
-            if(false !== $this->close($name)) {
+            if(false !== ($closingName = $this->close($names))) {
                 if(null === $content) { $content = ''; }
                 $this->backtrack();
                 $shortcodes = array();
                 break;
             }
+            $closingName = null;
             $this->backtrack();
             if($this->position < $this->tokensCount) {
                 $shortcodes = array();
@@ -136,10 +152,10 @@ private function content($name)
             $this->match(null, $appendContent);
         }
 
-        return array($this->position < $this->tokensCount ? $content : false, $shortcodes);
+        return array($this->position < $this->tokensCount ? $content : false, $shortcodes, $closingName);
     }
 
-    private function close($openingName)
+    private function close(array &$names)
     {
         $closingName = null;
         $setName = function(array $token) use(&$closingName) { $closingName = $token[1]; };
@@ -149,7 +165,7 @@ private function close($openingName)
         if(!$this->match(self::TOKEN_STRING, $setName, true)) { return false; }
         if(!$this->match(self::TOKEN_CLOSE)) { return false; }
 
-        return $openingName === $closingName;
+        return in_array($closingName, $names) ? $closingName : false;
     }
 
     private function bbCode()
@@ -196,6 +212,18 @@ private function value()
 
     /* --- PARSER ---------------------------------------------------------- */
 
+    /**
+     * This method is used only for debugging purposes. DO NOT DELETE.
+     *
+     * @return array
+     */
+    private function listBacktracks()
+    {
+        return array_map(function(array $backtrack) {
+            return implode('', array_column($backtrack, 1));
+        }, $this->backtracks);
+    }
+
     private function beginBacktrack()
     {
         $this->backtracks[] = array();

src/Parser/WordpressParser.php

@@ -40,10 +40,7 @@ public function parse($text)
             $parameters = static::parseParameters($matches[3][$i][0]);
             $content = $matches[5][$i][0] ?: null;
             $match = $matches[0][$i][0];
-            // TODO check how WordPress deals with unicode offset problems
-            // which can be solved by changing the line below to
-            // $offset = mb_strlen(substr($text, 0, $matches[0][$i][1]), 'utf-8');
-            $offset = $matches[0][$i][1];
+            $offset = mb_strlen(substr($text, 0, $matches[0][$i][1]), 'utf-8');
 
             $shortcode = new Shortcode($name, $parameters, $content, null);
             $shortcodes[] = new ParsedShortcode($shortcode, $match, $offset);

tests/ParserTest.php

@@ -165,6 +165,14 @@ public function provideShortcodes()
             array($s, str_repeat('[a]', 20), array_map(function($offset) { // 20
                 return new ParsedShortcode(new Shortcode('a', array(), null), '[a]', $offset);
             }, range(0, 57, 3))),
+            array($s, '[b][a]x[a][/a][/a][/b] [b][a][a][/a]y[/a][/b]', array(
+                new ParsedShortcode(new Shortcode('b', array(), '[a]x[a][/a][/a]'), '[b][a]x[a][/a][/a][/b]', 0),
+                new ParsedShortcode(new Shortcode('b', array(), '[a][a][/a]y[/a]'), '[b][a][a][/a]y[/a][/b]', 23),
+            )),
+            array($s, '[b] [a][a][a] [/b] [b] [a][a][a] [/b]', array(
+                new ParsedShortcode(new Shortcode('b', array(), ' [a][a][a] '), '[b] [a][a][a] [/b]', 0),
+                new ParsedShortcode(new Shortcode('b', array(), ' [a][a][a] '), '[b] [a][a][a] [/b]', 19),
+            )),
         );
 
         /**
@@ -178,7 +186,7 @@ public function provideShortcodes()
          *
          * Tests cases from array above with identifiers in the array below must be skipped.
          */
-        $wordpressSkip = array(3, 6, 16, 21, 22, 23, 25, 32, 33, 34, 35);
+        $wordpressSkip = array(3, 6, 16, 21, 22, 23, 25, 32, 33, 34);
         $result = array();
         foreach($tests as $key => $test) {
             $syntax = array_shift($test);