From: Gabriel Wicke <gwicke@users.mediawiki.org>
Date: Fri, 14 May 2004 00:18:43 +0000 (+0000)
Subject: Made strong/em handling more forgiving against unbalanced ticks
X-Git-Tag: 1.3.0beta1~70
X-Git-Url: http://git.cyclocoop.org/?a=commitdiff_plain;h=d181a35170a12f30b0ebea0658f9954622fff6ee;p=lhc%2Fweb%2Fwiklou.git

Made strong/em handling more forgiving against unbalanced ticks
* close any open em's or strong tags before each heading
* close any open em's or strong's at the end of the text
---

diff --git a/includes/Parser.php b/includes/Parser.php
index e65acb3683..05c6786e98 100644
--- a/includes/Parser.php
+++ b/includes/Parser.php
@@ -861,6 +861,19 @@ class Parser
 					# empty token
 					$txt="";
 					break;
+				case "h": 
+					#heading- used to close all unbalanced bold or em tags in this section
+					$txt = '';
+					if( $state['em'] !== false and 
+					( $state['strong'] === false or $state['em'] > $state['strong'] ) )
+					{ 
+						$s .= '</em>';
+						$state['em'] = false;
+					}
+					if ( $state['strong'] !== false ) $txt .= '</strong>';
+					if ( $state['em'] !== false ) $txt .= '</em>';
+					$state['strong'] = $state['em'] = false;
+					break;
 				case "RFC ":
 					if ( $tagIsOpen ) {
 						$txt = "RFC ";
@@ -904,6 +917,19 @@ class Parser
 				$s .= $txt;
 			}
 		} #end while
+
+		# make 100% sure all strong and em tags are closed
+		# doBlockLevels often messes the last bit up though, but invalid nesting is better than unclosed tags
+		# tidy solves this though
+		if( $state['em'] !== false and 
+		( $state['strong'] === false or $state['em'] > $state['strong'] ) )
+		{ 
+			$s .= '</em>';
+			$state['em'] = false;
+		}
+		if ( $state['strong'] !== false ) $s .= '</strong>';
+		if ( $state['em'] !== false ) $s .= '</em>';
+
 		if ( count( $tokenStack ) != 0 )
 		{
 			# still objects on stack. opened [[ tag without closing ]] tag.
diff --git a/includes/Tokenizer.php b/includes/Tokenizer.php
index 0b313d5e26..ab18b106db 100644
--- a/includes/Tokenizer.php
+++ b/includes/Tokenizer.php
@@ -147,7 +147,7 @@ class Tokenizer {
 						}
 						break;
 					case "\n": // for block levels, actually, only "----" is handled.
-					case "\r":
+					case "\r": // headings are detected to close any unbalanced em or strong tags in a section
 						if ( $this->continues( "----" ) )
 						{
 						     	$queueToken["type"] = "----";
@@ -160,6 +160,21 @@ class Tokenizer {
 								$this->mPos ++;
 							}
 							break 2;
+						} else if ( 
+							$this->continues( "<h" ) and (
+								$this->continues( "<h1" ) or
+								$this->continues( "<h2" ) or 
+								$this->continues( "<h3" ) or 
+								$this->continues( "<h4" ) or 
+								$this->continues( "<h5" ) or
+								$this->continues( "<h6" ) 
+							)
+						) { // heading
+							$queueToken["type"] = "h";
+							$queueToken["text"] = "";
+							$this->mQueuedToken[] = $queueToken;
+							$this->mPos ++;
+							break 2; // switch + while
 						}
 						break;
 					case "!": // French spacing rules have a space before exclamation