From 637a920c28797cebc6ee53b494c39d102d39ae74 Mon Sep 17 00:00:00 2001
From: Alexander Ebert <ebert@woltlab.com>
Date: Tue, 19 Jul 2016 00:55:59 +0200
Subject: [PATCH] Fixed handling of source bbcodes

---
 .../system/bbcode/HtmlBBCodeParser.class.php  | 268 ++++++++++++------
 1 file changed, 175 insertions(+), 93 deletions(-)

diff --git a/wcfsetup/install/files/lib/system/bbcode/HtmlBBCodeParser.class.php b/wcfsetup/install/files/lib/system/bbcode/HtmlBBCodeParser.class.php
index c5b1ba57cc..f9a582daea 100644
--- a/wcfsetup/install/files/lib/system/bbcode/HtmlBBCodeParser.class.php
+++ b/wcfsetup/install/files/lib/system/bbcode/HtmlBBCodeParser.class.php
@@ -46,11 +46,170 @@ class HtmlBBCodeParser extends BBCodeParser {
 		// tags that will eventually be removed within the marker processor
 		$this->buildXMLStructure();
 		
+		$this->handleSourceBBCodes();
+		
 		$this->buildParsedString();
 		
 		return $this->parsedText;
 	}
 	
+	/**
+	 * @inheritDoc
+	 */
+	public function buildXMLStructure() {
+		// stack for open tags
+		$openTagStack = $openTagDataStack = [];
+		$newTagArray = [];
+		$newTextArray = [];
+		
+		$i = -1;
+		foreach ($this->tagArray as $i => $tag) {
+			if ($tag['closing']) {
+				// closing tag
+				if (in_array($tag['name'], $openTagStack) && $this->isAllowed($openTagStack, $tag['name'], true)) {
+					// close unclosed tags
+					while (($previousTag = end($openTagStack)) != $tag['name']) {
+						$nextIndex = count($newTagArray);
+						
+						// mark as invalid and do not flag as opened tag
+						$newTag = $this->buildTag('[/'.end($openTagStack).']');
+						$newTag['invalid'] = true;
+						
+						$newTagArray[$nextIndex] = $newTag;
+						if (!isset($newTextArray[$nextIndex])) $newTextArray[$nextIndex] = '';
+						$newTextArray[$nextIndex] .= $this->textArray[$i];
+						$this->textArray[$i] = '';
+						array_pop($openTagStack);
+						array_pop($openTagDataStack);
+					}
+					
+					$nextIndex = count($newTagArray);
+					$newTagArray[$nextIndex] = $tag;
+					array_pop($openTagStack);
+					array_pop($openTagDataStack);
+					if (!isset($newTextArray[$nextIndex])) $newTextArray[$nextIndex] = '';
+					$newTextArray[$nextIndex] .= $this->textArray[$i];
+				}
+				else {
+					// no such tag open
+					// handle as plain text
+					$this->textArray[$i] .= $tag['source'];
+					$last = count($newTagArray);
+					if (!isset($newTextArray[$last])) $newTextArray[$last] = '';
+					$newTextArray[$last] .= $this->textArray[$i];
+				}
+			}
+			else {
+				// opening tag
+				if ($this->isAllowed($openTagStack, $tag['name']) && $this->isValidTag($tag)) {
+					$openTagStack[] = $tag['name'];
+					$openTagDataStack[] = $tag;
+					$nextIndex = count($newTagArray);
+					$newTagArray[$nextIndex] = $tag;
+					if (!isset($newTextArray[$nextIndex])) $newTextArray[$nextIndex] = '';
+					$newTextArray[$nextIndex] .= $this->textArray[$i];
+				}
+				else {
+					// tag not allowed
+					$this->textArray[$i] .= $tag['source'];
+					$last = count($newTagArray);
+					if (!isset($newTextArray[$last])) $newTextArray[$last] = '';
+					$newTextArray[$last] .= $this->textArray[$i];
+				}
+			}
+		}
+		
+		$last = count($newTagArray);
+		if (!isset($newTextArray[$last])) $newTextArray[$last] = '';
+		$newTextArray[$last] .= $this->textArray[$i + 1];
+		
+		// close unclosed open tags
+		while (end($openTagStack)) {
+			$nextIndex = count($newTagArray);
+			
+			// mark as invalid
+			$newTag = $this->buildTag('[/'.end($openTagStack).']');
+			$newTag['invalid'] = true;
+			
+			$newTagArray[$nextIndex] = $newTag;
+			if (!isset($newTextArray[$nextIndex])) $newTextArray[$nextIndex] = '';
+			array_pop($openTagStack);
+			array_pop($openTagDataStack);
+		}
+		
+		$this->tagArray = $newTagArray;
+		$this->textArray = $newTextArray;
+	}
+	
+	/**
+	 * Flags bbcodes inside code bbcodes for reversal, turning them back
+	 * into their source state (= textual representation).
+	 */
+	protected function handleSourceBBCodes() {
+		$sourceBBCodes = $this->getSourceBBCodes();
+		
+		$inCode = '';
+		$openTagStack = [];
+		
+		for ($i = 0, $length = count($this->tagArray); $i < $length; $i++) {
+			$tag = $this->tagArray[$i];
+			
+			if (!empty($tag['invalid'])) {
+				continue;
+			}
+			
+			$name = $tag['name'];
+			
+			if ($tag['closing']) {
+				if ($inCode) {
+					// matches opening code tag
+					if ($inCode === $name) {
+						$inCode = '';
+						array_pop($openTagStack);
+					}
+					else {
+						// unrelated tag, flag as invalid
+						$this->tagArray[$i]['inCode'] = true;
+					}
+					
+					continue;
+				}
+				
+				array_pop($openTagStack);
+			}
+			else {
+				if ($inCode) {
+					// inside code block, flag as invalid
+					$this->tagArray[$i]['inCode'] = true;
+					continue;
+				}
+				
+				// starts a new code block
+				if (in_array($name, $sourceBBCodes)) {
+					// look ahead to see if there is a valid closing tag
+					$hasClosingTag = false;
+					for ($j = $i + 1; $j < $length; $j++) {
+						if ($this->tagArray[$j]['name'] === $name && empty($this->tagArray[$j]['invalid'])) {
+							$hasClosingTag = true;
+							break;
+						}
+					}
+					
+					if ($hasClosingTag) {
+						$inCode = $name;
+					}
+					else {
+						// no closing tag, flag as invalid to avoid the
+						// entire content afterwards being treated as code
+						$this->tagArray[$i]['inCode'] = true;
+					}
+				}
+				
+				$openTagStack[] = $name;
+			}
+		}
+	}
+	
 	/**
 	 * @inheritDoc
 	 */
@@ -74,6 +233,16 @@ class HtmlBBCodeParser extends BBCodeParser {
 			$buffer .= $this->textArray[$i];
 			
 			if ($tag['closing']) {
+				if (!empty($tag['invalid'])) {
+					// drop invalid closing tag
+					continue;
+				}
+				else if (!empty($tag['inCode'])) {
+					// revert bbcodes inside code
+					$buffer .= $tag['source'];
+					continue;
+				}
+				
 				// get buffered opening tag
 				$openingTag = end($bufferedTagStack);
 				
@@ -127,6 +296,12 @@ class HtmlBBCodeParser extends BBCodeParser {
 				}
 			}
 			else {
+				if (!empty($tag['inCode'])) {
+					// revert bbcodes inside code
+					$buffer .= $tag['source'];
+					continue;
+				}
+				
 				// opening tag
 				if ($this->needBuffering($tag)) {
 					// start buffering
@@ -143,94 +318,6 @@ class HtmlBBCodeParser extends BBCodeParser {
 		if (isset($this->textArray[$i + 1])) $this->parsedText .= $this->textArray[$i + 1];
 	}
 	
-	/**
-	 * @inheritDoc
-	 */
-	public function buildXMLStructure() {
-		// stack for open tags
-		$openTagStack = $openTagDataStack = [];
-		$newTagArray = [];
-		$newTextArray = [];
-		
-		$i = -1;
-		foreach ($this->tagArray as $i => $tag) {
-			if ($tag['closing']) {
-				// closing tag
-				if (in_array($tag['name'], $openTagStack) && $this->isAllowed($openTagStack, $tag['name'], true)) {
-					// close unclosed tags
-					while (($previousTag = end($openTagStack)) != $tag['name']) {
-						$nextIndex = count($newTagArray);
-						
-						// mark as invalid and do not flag as opened tag
-						$newTag = $this->buildTag('[/'.end($openTagStack).']');
-						$newTag['invalid'] = true;
-						
-						$newTagArray[$nextIndex] = $newTag;
-						if (!isset($newTextArray[$nextIndex])) $newTextArray[$nextIndex] = '';
-						$newTextArray[$nextIndex] .= $this->textArray[$i];
-						$this->textArray[$i] = '';
-						array_pop($openTagStack);
-						array_pop($openTagDataStack);
-					}
-					
-					$nextIndex = count($newTagArray);
-					$newTagArray[$nextIndex] = $tag;
-					array_pop($openTagStack);
-					array_pop($openTagDataStack);
-					if (!isset($newTextArray[$nextIndex])) $newTextArray[$nextIndex] = '';
-					$newTextArray[$nextIndex] .= $this->textArray[$i];
-				}
-				else {
-					// no such tag open
-					// handle as plain text
-					$this->textArray[$i] .= $tag['source'];
-					$last = count($newTagArray);
-					if (!isset($newTextArray[$last])) $newTextArray[$last] = '';
-					$newTextArray[$last] .= $this->textArray[$i];
-				}
-			}
-			else {
-				// opening tag
-				if ($this->isAllowed($openTagStack, $tag['name']) && $this->isValidTag($tag)) {
-					$openTagStack[] = $tag['name'];
-					$openTagDataStack[] = $tag;
-					$nextIndex = count($newTagArray);
-					$newTagArray[$nextIndex] = $tag;
-					if (!isset($newTextArray[$nextIndex])) $newTextArray[$nextIndex] = '';
-					$newTextArray[$nextIndex] .= $this->textArray[$i];
-				}
-				else {
-					// tag not allowed
-					$this->textArray[$i] .= $tag['source'];
-					$last = count($newTagArray);
-					if (!isset($newTextArray[$last])) $newTextArray[$last] = '';
-					$newTextArray[$last] .= $this->textArray[$i];
-				}
-			}
-		}
-		
-		$last = count($newTagArray);
-		if (!isset($newTextArray[$last])) $newTextArray[$last] = '';
-		$newTextArray[$last] .= $this->textArray[$i + 1];
-		
-		// close unclosed open tags
-		while (end($openTagStack)) {
-			$nextIndex = count($newTagArray);
-			
-			// mark as invalid
-			$newTag = $this->buildTag('[/'.end($openTagStack).']');
-			$newTag['invalid'] = true;
-			
-			$newTagArray[$nextIndex] = $newTag;
-			if (!isset($newTextArray[$nextIndex])) $newTextArray[$nextIndex] = '';
-			array_pop($openTagStack);
-			array_pop($openTagDataStack);
-		}
-		
-		$this->tagArray = $newTagArray;
-		$this->textArray = $newTextArray;
-	}
-	
 	/**
 	 * Builds the bbcode output.
 	 * 
@@ -376,11 +463,6 @@ class HtmlBBCodeParser extends BBCodeParser {
 			throw new SystemException("Tag mismatch, expected '".$name."', got '".$data['name']."'.");
 		}
 		
-		if (!empty($data['invalid'])) {
-			// drop invalid closing tags
-			return '';
-		}
-		
 		return '<woltlab-metacode-marker data-uuid="' . $data['uuid'] . '" data-source="' . base64_encode($tag['source']) . '" />';
 	}
 	
-- 
2.20.1