Merge pull request #6845 from MrPetovan/bug/6633-add-markdown-autolinker

Add autolinker to Markdown parser

include/items.php

@@ -24,7 +24,8 @@ use Friendica\Util\ParseUrl;
 use Friendica\Util\Strings;
 use Friendica\Util\Temporal;
 
-require_once 'mod/share.php';
+require_once __DIR__ . '/../mod/share.php';
+
 function add_page_info_data(array $data, $no_photos = false)
 {
 	Hook::callAll('page_info_data', $data);

src/BaseObject.php

@@ -4,7 +4,7 @@
  */
 namespace Friendica;
 
-require_once 'boot.php';
+require_once __DIR__ . '/../boot.php';
 
 use Friendica\Network\HTTPException\InternalServerErrorException;
 

src/Content/Text/BBCode.php

@@ -1268,24 +1268,7 @@ class BBCode extends BaseObject
 
 		// if the HTML is used to generate plain text, then don't do this search, but replace all URL of that kind to text
 		if (!$for_plaintext) {
-			// Autolink feature (thanks to https://daringfireball.net/2010/07/improved_regex_for_matching_urls)
-			$autolink_regex = '@(?xi)
-(?<![=\'\]"/])          # Not preceded by =, \', ], ", /
-\b
-(                       # Capture 1: entire matched URL
-  https?://                 # http or https protocol
-  (?:
-    [^/.][^/]+[.][^/]+/?    # looks like domain name followed by a slash
-  )
-  (?:                       # One or more:
-    [^\s()<>]+                  # Run of non-space, non-()<>
-    |                           #   or
-    \(([^\s()<>]+|(\([^\s()<>]+\)))*\)  # balanced parens, up to 2 levels
-    |                               #   or
-    [^\s`!()\[\]{};:\'".,<>?«»“”‘’]        # not a space or one of these punct chars
-  )*
-)@';
-			$text = preg_replace($autolink_regex, '[url]$1[/url]', $text);
+			$text = preg_replace(Strings::autoLinkRegEx(), '[url]$1[/url]', $text);
 			if ($simple_html == 7) {
 				$text = preg_replace_callback("/\[url\]([$URLSearchString]*)\[\/url\]/ism", 'self::convertUrlForOStatusCallback', $text);
 				$text = preg_replace_callback("/\[url\=([$URLSearchString]*)\]([$URLSearchString]*)\[\/url\]/ism", 'self::convertUrlForOStatusCallback', $text);

src/Content/Text/Markdown.php

@@ -9,7 +9,6 @@ namespace Friendica\Content\Text;
 use Friendica\BaseObject;
 use Friendica\Core\System;
 use Friendica\Model\Contact;
-use Michelf\MarkdownExtra;
 
 /**
  * Friendica-specific usage of Markdown
@@ -31,11 +30,18 @@ class Markdown extends BaseObject
 	public static function convert($text, $hardwrap = true) {
 		$stamp1 = microtime(true);
 
-		$MarkdownParser = new MarkdownExtra();
-		$MarkdownParser->hard_wrap = $hardwrap;
-		$MarkdownParser->code_class_prefix = 'language-';
+		$MarkdownParser = new MarkdownParser();
+		$MarkdownParser->code_class_prefix  = 'language-';
+		$MarkdownParser->hard_wrap          = $hardwrap;
+		$MarkdownParser->hashtag_protection = true;
+		$MarkdownParser->url_filter_func    = function ($url) {
+			if (strpos($url, '#') === 0) {
+				$url = ltrim($_SERVER['REQUEST_URI'], '/') . $url;
+			}
+			return  $url;
+		};
+
 		$html = $MarkdownParser->transform($text);
-		$html = preg_replace('/<a(.*?)href="#/is', '<a$1href="' . ltrim($_SERVER['REQUEST_URI'], '/') . '#', $html);
 
 		self::getApp()->getProfiler()->saveTimestamp($stamp1, "parser", System::callstack());
 

src/Content/Text/MarkdownParser.php

@@ -0,0 +1,18 @@
+<?php
+namespace Friendica\Content\Text;
+
+use Friendica\Util\Strings;
+use Michelf\MarkdownExtra;
+
+class MarkdownParser extends MarkdownExtra
+{
+	protected function doAutoLinks($text)
+	{
+		$text = parent::doAutoLinks($text);
+
+		$text = preg_replace_callback(Strings::autoLinkRegEx(),
+			array($this, '_doAutoLinks_url_callback'), $text);
+
+		return $text;
+	}
+}

src/Render/FriendicaSmartyEngine.php

@@ -17,7 +17,7 @@ class FriendicaSmartyEngine implements ITemplateEngine
 
 	public function __construct()
 	{
-		if (!is_writable('view/smarty3/')) {
+		if (!is_writable(__DIR__ . '/../../view/smarty3/')) {
 			echo "<b>ERROR:</b> folder <tt>view/smarty3/</tt> must be writable by webserver.";
 			exit();
 		}

src/Util/Strings.php

@@ -346,4 +346,30 @@ class Strings
 
 		return $return;
 	}
+
+	/**
+	 * Returns the regular expression string to match URLs in a given text
+	 *
+	 * @return string
+	 * @see https://daringfireball.net/2010/07/improved_regex_for_matching_urls
+	 */
+	public static function autoLinkRegEx()
+	{
+		return '@(?xi)
+(?<![=\'\]"/])          # Not preceded by [, =, \', ], ", /
+\b
+(                       # Capture 1: entire matched URL
+  https?://                 # http or https protocol
+  (?:
+    [^/.][^/]+[.][^/]+/?    # looks like domain name followed by a slash
+  )
+  (?:                       # One or more:
+    [^\s()<>]+                  # Run of non-space, non-()<>
+    |                           #   or
+    \(([^\s()<>]+|(\([^\s()<>]+\)))*\)  # balanced parens, up to 2 levels
+    |                               #   or
+    [^\s`!()\[\]{};:\'".,<>?«»“”‘’]        # not a space or one of these punct chars
+  )*
+)@';
+	}
 }

tests/datasets/content/text/markdown/bug-6633.html

@@ -0,0 +1,9 @@
+<h1>Produção de cebola em sistema orgânico com uso de homeopatia</h1>
+
+<p><img src="https://mapadaagroecologia.org/system/midias/imagems/000/000/097/original/Cebola_em_sistema_org%C3%A2nico.jpg?1549640469" alt="Bulbos de cebola em sistema orgânico na fase de colheita
+" title="Bulbos de cebola em sistema orgânico na fase de colheita
+" /></p>
+
+<h2><a href="https://mapadaagroecologia.org/locais/epagri-estacao-experimental-de-ituporanga-sc?locale=pt-BR">https://mapadaagroecologia.org/locais/epagri-estacao-experimental-de-ituporanga-sc?locale=pt-BR</a></h2>
+
+<p>#agroecologia #ecologia #orgânico #agroecology #brazil</p>

tests/datasets/content/text/markdown/bug-6633.md

@@ -0,0 +1,9 @@
+# Produção de cebola em sistema orgânico com uso de homeopatia
+
+![Bulbos de cebola em sistema orgânico na fase de colheita
+](https://mapadaagroecologia.org/system/midias/imagems/000/000/097/original/Cebola_em_sistema_org%C3%A2nico.jpg?1549640469 "Bulbos de cebola em sistema orgânico na fase de colheita
+")
+
+## https://mapadaagroecologia.org/locais/epagri-estacao-experimental-de-ituporanga-sc?locale=pt-BR
+
+#agroecologia #ecologia #orgânico #agroecology #brazil

tests/src/Content/Text/MarkdownTest.php

@@ -0,0 +1,52 @@
+<?php
+
+namespace Friendica\Test\src\Content\Text;
+
+use Friendica\Content\Text\Markdown;
+use Friendica\Test\MockedTest;
+use Friendica\Test\Util\AppMockTrait;
+use Friendica\Test\Util\VFSTrait;
+
+class MarkdownTest extends MockedTest
+{
+	use VFSTrait;
+	use AppMockTrait;
+
+	protected function setUp()
+	{
+		parent::setUp();
+		$this->setUpVfsDir();
+		$this->mockApp($this->root);
+	}
+
+	public function dataMarkdown()
+	{
+		$inputFiles = glob(__DIR__ . '/../../../datasets/content/text/markdown/*.md');
+
+		$data = [];
+
+		foreach ($inputFiles as $file) {
+			$data[str_replace('.md', '', $file)] = [
+				'input'    => file_get_contents($file),
+				'expected' => file_get_contents(str_replace('.md', '.html', $file))
+			];
+		}
+
+		return $data;
+	}
+
+	/**
+	 * Test convert different input Markdown text into HTML
+	 * @dataProvider dataMarkdown
+	 *
+	 * @param string $input    The Markdown text to test
+	 * @param string $expected The expected HTML output
+	 * @throws \Exception
+	 */
+	public function testConvert($input, $expected)
+	{
+		$output = Markdown::convert($input);
+
+		$this->assertEquals($expected, $output);
+	}
+}