fixed handling of non-separated block elements

improved markdown handling of block elements that are not separated by a
blank line.

fixes #27

GithubMarkdown.php

@@ -41,7 +41,7 @@ protected function inlineMarkers()
 	 */
 	protected function identifyLine($lines, $current)
 	{
-		if (strncmp($lines[$current], '```', 3) === 0) {
+		if (isset($lines[$current]) && strncmp($lines[$current], '```', 3) === 0) {
 			return 'fencedCode';
 		}
 		return parent::identifyLine($lines, $current);

Markdown.php

@@ -103,7 +103,7 @@ protected function prepare()
 	 */
 	protected function identifyLine($lines, $current)
 	{
-		if (empty($lines[$current]) || ltrim($lines[$current]) === '') {
+		if (empty($lines[$current])) {
 			return 'empty';
 		}
 		$line = $lines[$current];
@@ -205,6 +205,28 @@ protected function identifyLine($lines, $current)
 		return 'paragraph';
 	}
 
+	/**
+	 * Consume lines for a paragraph
+	 */
+	public function consumeParagraph($lines, $current)
+	{
+		// consume until newline or intended line
+
+		$block = [
+			'type' => 'paragraph',
+			'content' => [],
+		];
+		for($i = $current, $count = count($lines); $i < $count; $i++) {
+			if (ltrim($lines[$i]) !== '' && $lines[$i][0] != "\t" && strncmp($lines[$i], '    ', 4) !== 0) {
+				$block['content'][] = $lines[$i];
+			} else {
+				break;
+			}
+		}
+
+		return [$block, --$i];
+	}
+
 	/**
 	 * Consume lines for a blockquote element
 	 */
@@ -247,7 +269,14 @@ protected function consumeCode($lines, $current)
 		];
 		for($i = $current, $count = count($lines); $i < $count; $i++) {
 			$line = $lines[$i];
-			if (ltrim($line) !== '' || isset($lines[$i + 1]) && $this->identifyLine($lines, $i + 1) === 'code') {
+
+			// a line is considered to belong to this code block as long as it is intended by 4 spaces or a tab
+			if (isset($line[0]) && ($line[0] === "\t" || strncmp($lines[$i], '    ', 4) === 0)) {
+				$line = $line[0] === "\t" ? substr($line, 1) : substr($line, 4);
+				$block['content'][] = $line;
+			// but also if it is empty and the next line is intended by 4 spaces or a tab
+			} elseif ((empty($line) || rtrim($line) === '') && isset($lines[$i + 1][0]) &&
+					  ($lines[$i + 1][0] === "\t" || strncmp($lines[$i + 1], '    ', 4) === 0)) {
 				if (!empty($line)) {
 					$line = $line[0] === "\t" ? substr($line, 1) : substr($line, 4);
 				}
@@ -257,7 +286,7 @@ protected function consumeCode($lines, $current)
 			}
 		}
 
-		return [$block, $i];
+		return [$block, --$i];
 	}
 
 	/**
@@ -311,7 +340,7 @@ private function consumeList($lines, $current, $block, $type)
 				$block['items'][++$item][] = $line;
 			} elseif (ltrim($line) === '') {
 				// next line after empty one is also a list or indented -> lazy list
-				if (isset($lines[$i + 1]) && (
+				if (isset($lines[$i + 1][0]) && (
 					$this->identifyLine($lines, $i + 1) === $type ||
 					(strncmp($lines[$i + 1], $indent, $len) === 0 || !empty($lines[$i + 1]) && $lines[$i + 1][0] == "\t"))) {
 					$block['items'][$item][] = $line;
@@ -355,15 +384,15 @@ protected function consumeHeadline($lines, $current)
 				'content' => trim($lines[$current], "# \t"),
 				'level' => $level,
 			];
-			return [$block, ++$current];
+			return [$block, $current];
 		}
 
 		$block = [
 			'type' => 'headline',
 			'content' => $lines[$current],
 			'level' => $lines[$current + 1][0] === '=' ? 1 : 2,
 		];
-		return [$block, $current + 2];
+		return [$block, $current + 1];
 	}
 
 	/**
@@ -409,7 +438,7 @@ protected function consumeHr($lines, $current)
 		$block = [
 			'type' => 'hr',
 		];
-		return [$block, $current + 1];
+		return [$block, $current];
 	}
 
 	/**
@@ -434,7 +463,7 @@ protected function consumeReference($lines, $current)
 			}
 			$current++;
 		}
-		return [false, $current];
+		return [false, --$current];
 	}
 
 
@@ -469,7 +498,7 @@ protected function renderList($block)
 			$output .= '<li>';
 			if (!isset($block['lazyItems'][$item])) {
 				$firstPar = [];
-				while(!empty($itemLines) && $this->identifyLine($itemLines, 0) === 'paragraph') {
+				while(!empty($itemLines) && rtrim($itemLines[0]) !== '' && $this->identifyLine($itemLines, 0) === 'paragraph') {
 					$firstPar[] = array_shift($itemLines);
 				}
 				$output .= $this->parseInline(implode("\n", $firstPar));

Parser.php

@@ -105,7 +105,7 @@ protected function parseBlocks($lines)
 		// convert lines to blocks
 
 		for($i = 0, $count = count($lines); $i < $count; $i++) {
-			if (!empty($lines[$i])) { // skip empty lines
+			if (!empty($lines[$i]) && rtrim($lines[$i]) !== '') { // skip empty lines
 				// identify a blocks beginning
 				$blockType = $this->identifyLine($lines, $i);
 

tests/markdown-data/dense-block-markers.html

@@ -0,0 +1,50 @@
+<h1>this is to test dense blocks (no newlines between them)</h1>
+<hr />
+<h2>what is Markdown?</h2>
+<p>see <a href="http://en.wikipedia.org/wiki/Markdown">Wikipedia</a></p>
+<h2>a h2</h2>
+<p>paragraph</p>
+<p>this is a paragraph, not a headline or list
+next line
+- whoo</p>
+<p>par</p>
+<pre><code>code
+code
+</code></pre>
+<p>par</p>
+<h3>Tasks list</h3>
+<ul>
+<li>list items</li>
+</ul>
+<h2>headline1</h2>
+<blockquote><p>quote
+quote</p>
+</blockquote>
+<h2>headline2</h2>
+<hr />
+<h1>h1</h1>
+<h2>h2</h2>
+<hr />
+<h3>h3</h3>
+<ol>
+<li>ol1</li>
+<li>ol2</li>
+</ol>
+<h4>h4</h4>
+<ul>
+<li>listA</li>
+<li>listB</li>
+</ul>
+<h5>h5</h5>
+<h6>h6</h6>
+<hr />
+<hr />
+<h2>changelog 1</h2>
+<ul>
+<li>17-Feb-2013 re-design</li>
+</ul>
+<hr />
+<h2>changelog 2</h2>
+<ul>
+<li>17-Feb-2013 re-design</li>
+</ul>

tests/markdown-data/dense-block-markers.md

@@ -0,0 +1,56 @@
+# this is to test dense blocks (no newlines between them)
+
+----
+## what is Markdown?
+see [Wikipedia][]
+
+a h2
+----
+paragraph
+
+this is a paragraph, not a headline or list
+next line
+- whoo
+
+par
+	code
+	code
+par
+
+### Tasks list
+- list items
+
+headline1
+---------
+> quote
+> quote
+
+[Wikipedia]: http://en.wikipedia.org/wiki/Markdown
+headline2
+---------
+
+----
+# h1
+## h2
+---
+### h3
+1. ol1
+2. ol2
+
+#### h4
+- listA
+- listB
+
+##### h5
+###### h6
+--------
+
+----
+
+## changelog 1
+
+* 17-Feb-2013 re-design
+
+----
+## changelog 2
+* 17-Feb-2013 re-design

tests/markdown-data/md1_inline_html_simple.html

@@ -32,6 +32,7 @@
 <!-- foo -->   
 <p>Code:</p>
 <pre><code>&lt;hr /&gt;
+
 </code></pre>
 <p>Hr's:</p>
 <hr>

tests/markdown-data/md1_markdown_documentation_basics.html

@@ -143,6 +143,7 @@ <h2>Lists</h2>
 &lt;p&gt;With multiple paragraphs.&lt;/p&gt;&lt;/li&gt;
 &lt;li&gt;&lt;p&gt;Another item in the list.&lt;/p&gt;&lt;/li&gt;
 &lt;/ul&gt;
+
 </code></pre>
 <h3>Links</h3>
 <p>Markdown supports two styles for creating links: <em>inline</em> and