Merged TOC feature from jtopjian's fork

PHP Markdown Extra Readme.text

@@ -9,6 +9,8 @@ by Michel Fortin
 based on Markdown by John Gruber  
 <http://daringfireball.net/>
 
+table of contents support added by Joe Topjian
+<http://terrarum.net/>
 
 Introduction
 ------------
@@ -219,6 +221,20 @@ Extra 1.2.5 (8 Jan 2012):
 *	Fixed an issue where HTML tags inside fenced code blocks were sometime
 	not encoded with entities.
 
+Extra 1.2.4.1 (24 Sept 2010):
+*   Added Table of Contents support:
+
+    Create Table of Contents by specifying the following tag:
+
+        [TOC]
+
+    Table of Contents requirements:
+        * Only headings 2-6
+        * Headings must have an ID
+        * Builds TOC with headings _after_ the [TOC] tag
+
+    Table of Contents generation technnique was inspired by 
+    [ReMarkable](http://camendesign.com/code/remarkable)
 
 1.0.1n (10 Oct 2009):
 

markdown.php

@@ -1706,6 +1706,7 @@ function MarkdownExtra_Parser() {
 			"stripFootnotes"     => 15,
 			"stripAbbreviations" => 25,
 			"appendFootnotes"    => 50,
+            "doTOC"              => 55,
 			);
 		$this->block_gamut += array(
 			"doFencedCodeBlocks" => 5,
@@ -2252,23 +2253,23 @@ function doHeaders($text) {
 				(.+?)		# $2 = Header text
 				[ ]*
 				\#*			# optional closing #\'s (not counted)
-				(?:[ ]+\{\#([-_:a-zA-Z0-9]+)\})? # id attribute
+				(?:[ ]+\{\#([^\}#.]*)\})? # id attribute
 				[ ]*
 				\n+
 			}xm',
 			array(&$this, '_doHeaders_callback_atx'), $text);
 
 		return $text;
 	}
-	function _doHeaders_attr($attr) {
-		if (empty($attr))  return "";
-		return " id=\"$attr\"";
+	function _doHeaders_attr($attr, $text) {
+		$id = empty($attr) ? str_replace(' ', '-', $text) : $attr;
+		return " id=\"$id\"";
 	}
 	function _doHeaders_callback_setext($matches) {
 		if ($matches[3] == '-' && preg_match('{^- }', $matches[1]))
 			return $matches[0];
 		$level = $matches[3]{0} == '=' ? 1 : 2;
-		$attr  = $this->_doHeaders_attr($id =& $matches[2]);
+		$attr  = $this->_doHeaders_attr($id =& $matches[2], $matches[1]);
 		$block = "<h$level$attr>".$this->runSpanGamut($matches[1])."</h$level>";
 		return "\n" . $this->hashBlock($block) . "\n\n";
 	}
@@ -2840,6 +2841,27 @@ function _doAbbreviations_callback($matches) {
 		}
 	}
 
+    function doTOC($text) {
+    #    
+    # Adds TOC support by including the following on a single line:
+    #    
+    # [TOC]
+    #    
+    # TOC Requirements:
+    #     * Only headings 2-6
+    #     * Headings must have an ID
+    #     * Builds TOC with headings _after_ the [TOC] tag
+
+        if (preg_match ('/\[TOC\]/m', $text, $i, PREG_OFFSET_CAPTURE)) {
+			$toc = '';
+            preg_match_all ('/<h([2-6]) id="([^"]+)">(.*?)<\/h\1>/i', $text, $h, PREG_SET_ORDER, $i[0][1]);
+            foreach ($h as &$m) $toc .= str_repeat ("\t", (int) $m[1]-2)."*\t [${m[3]}](#${m[2]})\n";
+            $text = preg_replace ('/\[TOC\]/m', Markdown($toc), $text);
+        }
+        return trim ($text, "\n");
+    }
+
+
 }
 
 
@@ -2929,4 +2951,4 @@ function _doAbbreviations_callback($matches) {
 software, even if advised of the possibility of such damage.
 
 */
-?>
+?>