From e4499350d8a94c4c693a7e784295eff7a717ae67 Mon Sep 17 00:00:00 2001
From: Carsten Brandt <mail@cebe.cc>
Date: Tue, 9 Feb 2016 23:09:46 +0100
Subject: [PATCH] better handling of continued inline HTML in paragraphs

fixes #114
---
 GithubMarkdown.php                  | 11 ++++++++++-
 Markdown.php                        | 19 +++++++++++++------
 block/HtmlTrait.php                 |  4 ++--
 tests/markdown-data/html-block.html | 18 ++++++++++++++++++
 tests/markdown-data/html-block.md   | 22 ++++++++++++++++++++++
 5 files changed, 65 insertions(+), 9 deletions(-)

diff --git a/GithubMarkdown.php b/GithubMarkdown.php
index 3a46b33..23c9d1a 100644
--- a/GithubMarkdown.php
+++ b/GithubMarkdown.php
@@ -68,7 +68,6 @@ protected function consumeParagraph($lines, $current)
 				|| ltrim($line) === ''
 				|| !ctype_alpha($line[0]) && (
 					$this->identifyQuote($line, $lines, $i) ||
-					$this->identifyCode($line, $lines, $i) ||
 					$this->identifyFencedCode($line, $lines, $i) ||
 					$this->identifyUl($line, $lines, $i) ||
 					$this->identifyOl($line, $lines, $i) ||
@@ -77,6 +76,16 @@ protected function consumeParagraph($lines, $current)
 				|| $this->identifyHeadline($line, $lines, $i))
 			{
 				break;
+			} elseif ($this->identifyCode($line, $lines, $i)) {
+				// possible beginning of a code block
+				// but check for continued inline HTML
+				// e.g. <img src="file.jpg"
+				//           alt="some alt aligned with src attribute" title="some text" />
+				if (preg_match('~<\w+([^>]+)$~s', implode("\n", $content))) {
+					$content[] = $line;
+				} else {
+					break;
+				}
 			} else {
 				$content[] = $line;
 			}
diff --git a/Markdown.php b/Markdown.php
index df29135..32c9b0f 100644
--- a/Markdown.php
+++ b/Markdown.php
@@ -92,13 +92,20 @@ protected function consumeParagraph($lines, $current)
 				break;
 			}
 
-			if ($line !== '' && ltrim($line) !== '' &&
-				!($line[0] === "\t" || $line[0] === " " && strncmp($line, '    ', 4) === 0) &&
-				!$this->identifyHeadline($line, $lines, $i))
-			{
-				$content[] = $line;
-			} else {
+			if ($line === '' || ltrim($line) === '' || $this->identifyHeadline($line, $lines, $i)) {
 				break;
+			} elseif ($line[0] === "\t" || $line[0] === " " && strncmp($line, '    ', 4) === 0) {
+				// possible beginning of a code block
+				// but check for continued inline HTML
+				// e.g. <img src="file.jpg"
+				//           alt="some alt aligned with src attribute" title="some text" />
+				if (preg_match('~<\w+([^>]+)$~s', implode("\n", $content))) {
+					$content[] = $line;
+				} else {
+					break;
+				}
+			} else {
+				$content[] = $line;
 			}
 		}
 		$block = [
diff --git a/block/HtmlTrait.php b/block/HtmlTrait.php
index 448518e..2f15135 100644
--- a/block/HtmlTrait.php
+++ b/block/HtmlTrait.php
@@ -146,10 +146,10 @@ protected function renderInlineHtml($block)
 	protected function parseInlineHtml($text)
 	{
 		if (strpos($text, '>') !== false) {
-			if (preg_match('~^</?(\w+\d?)( .*?)?>~', $text, $matches)) {
+			if (preg_match('~^</?(\w+\d?)( .*?)?>~s', $text, $matches)) {
 				// HTML tags
 				return [['inlineHtml', $matches[0]], strlen($matches[0])];
-			} elseif (preg_match('~^<!--.*?-->~', $text, $matches)) {
+			} elseif (preg_match('~^<!--.*?-->~s', $text, $matches)) {
 				// HTML comments
 				return [['inlineHtml', $matches[0]], strlen($matches[0])];
 			}
diff --git a/tests/markdown-data/html-block.html b/tests/markdown-data/html-block.html
index 49198f2..738b51b 100644
--- a/tests/markdown-data/html-block.html
+++ b/tests/markdown-data/html-block.html
@@ -12,6 +12,9 @@
 <p>more markdown here</p>
 <p>&lt; this is not an html tag</p>
 <p>&lt;thisisnotanhtmltag</p>
+<p>but this is:</p>
+<p><img src="file.jpg"
+     alt="some alt aligned with src attribute" title="some text" /></p>
 <p><span class="test">some inline <strong>md</strong></span></p>
 <p><span>some inline <strong>md</strong></span></p>
 <p>self-closing on block level:</p>
@@ -22,3 +25,18 @@
 <h1>h1</h1>
 <custom multi="line" something="hi" />
 <h2>h2</h2>
+<p>p <img src="file.jpg"
+       alt="some alt aligned with src attribute"
+       title="some text" />
+   something</p>
+<p>p <img src="file.jpg"
+       alt="some alt aligned with src attribute"
+       title="some text" /></p>
+<pre><code>something
+</code></pre>
+<p>p is &lt; than 5</p>
+<pre><code>this is code
+</code></pre>
+<p>this paragraph contains a <!-- multi
+line html comment -->
+newline</p>
diff --git a/tests/markdown-data/html-block.md b/tests/markdown-data/html-block.md
index 23ed1bc..ac8c9ae 100644
--- a/tests/markdown-data/html-block.md
+++ b/tests/markdown-data/html-block.md
@@ -17,6 +17,11 @@ more markdown here
 
 <thisisnotanhtmltag
 
+but this is:
+
+<img src="file.jpg"
+     alt="some alt aligned with src attribute" title="some text" />
+
 <span class="test">some inline **md**</span>
 
 <span>some inline **md**</span>
@@ -35,3 +40,20 @@ something **bold**.
 <custom multi="line" something="hi" />
 
 ## h2
+
+p <img src="file.jpg"
+       alt="some alt aligned with src attribute"
+       title="some text" />
+   something
+
+p <img src="file.jpg"
+       alt="some alt aligned with src attribute"
+       title="some text" />
+    something
+
+p is < than 5
+    this is code
+
+this paragraph contains a <!-- multi
+line html comment -->
+newline