From b0accf4d4c674799ae210fce012dd0f31834717e Mon Sep 17 00:00:00 2001
From: Hypolite Petovan <mrpetovan@gmail.com>
Date: Thu, 6 Apr 2017 23:49:56 -0400
Subject: [PATCH] Fix code blocks to Diaspora

- Extracts code blocks before BBCode conversion to prevent code
highlighting and whitespace meddling
- Use the improved HTLM To Markdown library
- Use <code>  instead of <key> for Diaspora inline code blocks
---
 include/bb2diaspora.php | 40 +++++++++++++++++++++++++++++++---
 include/bbcode.php      | 48 ++++++++++++++++++++++++++++++-----------
 2 files changed, 73 insertions(+), 15 deletions(-)

diff --git a/include/bb2diaspora.php b/include/bb2diaspora.php
index 5a60cd945a..cef293c766 100644
--- a/include/bb2diaspora.php
+++ b/include/bb2diaspora.php
@@ -105,8 +105,18 @@ function diaspora_mentions($match) {
 	return $mention;
 }
 
-function bb2diaspora($Text,$preserve_nl = false, $fordiaspora = true) {
-
+/**
+ * @brief Converts a BBCode text into Markdown
+ *
+ * This function converts a BBCode item body to be sent to Markdown-enabled
+ * systems like Diaspora and Libertree
+ *
+ * @param string $Text
+ * @param bool $preserve_nl Effects unclear, unused in Friendica
+ * @param bool $fordiaspora Diaspora requires more changes than Libertree
+ * @return string
+ */
+function bb2diaspora($Text, $preserve_nl = false, $fordiaspora = true) {
 	$a = get_app();
 
 	$OriginalText = $Text;
@@ -129,6 +139,18 @@ function bb2diaspora($Text,$preserve_nl = false, $fordiaspora = true) {
 	// Converting images with size parameters to simple images. Markdown doesn't know it.
 	$Text = preg_replace("/\[img\=([0-9]*)x([0-9]*)\](.*?)\[\/img\]/ism", '[img]$3[/img]', $Text);
 
+	// Extracting multi-line code blocks before the whitespace processing/code highlighter in bbcode()
+	$codeblocks = [];
+	$Text = preg_replace_callback('#\[code(?:=([^\]]*))?\](?=\n)(.*?)\[\/code\]#is',
+		function ($matches) use (&$codeblocks) {
+			$return = '#codeblock-' . count($codeblocks) . '#';
+
+            $prefix = '````' . $matches[1] . PHP_EOL;
+			$codeblocks[] = $prefix . trim($matches[2]) . PHP_EOL . '````';
+			return $return;
+		}
+	, $Text);
+
 	// Convert it to HTML - don't try oembed
 	if ($fordiaspora) {
 		$Text = bbcode($Text, $preserve_nl, false, 3);
@@ -158,7 +180,8 @@ function bb2diaspora($Text,$preserve_nl = false, $fordiaspora = true) {
 	$stamp1 = microtime(true);
 
 	// Now convert HTML to Markdown
-	$Text = new HTML_To_Markdown($Text);
+	$converter = new HtmlConverter();
+	$Text = $converter->convert($Text);
 
 	// unmask the special chars back to HTML
 	$Text = str_replace(array('&_lt_;', '&_gt_;', '&_amp_;'), array('&lt;', '&gt;', '&amp;'), $Text);
@@ -177,6 +200,17 @@ function bb2diaspora($Text,$preserve_nl = false, $fordiaspora = true) {
 		$Text = preg_replace_callback("/([@]\[(.*?)\])\(([$URLSearchString]*?)\)/ism", 'diaspora_mentions', $Text);
 	}
 
+	// Restore code blocks
+	$Text = preg_replace_callback('/#codeblock-([0-9]+)#/iU',
+		function ($matches) use ($codeblocks) {
+            $return = '';
+            if (isset($codeblocks[intval($matches[1])])) {
+                $return = $codeblocks[$matches[1]];
+            }
+			return $return;
+		}
+	, $Text);
+
 	call_hooks('bb2diaspora',$Text);
 
 	return $Text;
diff --git a/include/bbcode.php b/include/bbcode.php
index f73f586d34..fd380edc99 100644
--- a/include/bbcode.php
+++ b/include/bbcode.php
@@ -159,13 +159,6 @@ function stripcode_br_cb($s) {
 	return '[code]' . str_replace('<br />', '', $s[1]) . '[/code]';
 }
 
-function bb_onelinecode_cb($match) {
-	if (strpos($match[1],"<br>")===false){
-		return "<key>".$match[1]."</key>";
-	}
-	return "<code>".$match[1]."</code>";
-}
-
 function tryoembed($match) {
 	$url = $match[1];
 
@@ -729,9 +722,31 @@ function bb_highlight($match) {
 	return $match[0];
 }
 
-	// BBcode 2 HTML was written by WAY2WEB.net
-	// extended to work with Mistpark/Friendica - Mike Macgirvin
-
+/**
+ * @brief Converts a BBCode message to HTML message
+ *
+ * BBcode 2 HTML was written by WAY2WEB.net
+ * extended to work with Mistpark/Friendica - Mike Macgirvin
+ *
+ * Simple HTML values meaning:
+ * - 0: Friendica display
+ * - 1: Unused
+ * - 2: Used for Facebook, Google+, Windows Phone push, Friendica API
+ * - 3: Used before converting to Markdown in bb2diaspora.php
+ * - 4: Used for WordPress, Libertree (before Markdown), pump.io and tumblr
+ * - 5: Unused
+ * - 6: Used for Appnet
+ * - 7: Used for dfrn, OStatus
+ * - 8: Used for WP backlink text setting
+ *
+ * @staticvar array $allowed_src_protocols
+ * @param string $Text
+ * @param bool $preserve_nl
+ * @param bool $tryoembed
+ * @param int $simplehtml
+ * @param bool $forplaintext
+ * @return string
+ */
 function bbcode($Text, $preserve_nl = false, $tryoembed = true, $simplehtml = false, $forplaintext = false) {
 
 	$a = get_app();
@@ -1158,8 +1173,17 @@ function bbcode($Text, $preserve_nl = false, $tryoembed = true, $simplehtml = fa
 	}
 
 
-	//replace oneliner <code> with <key>
-	$Text = preg_replace_callback("|(?!<br[^>]*>)<code>([^<]*)</code>(?!<br[^>]*>)|ism", 'bb_onelinecode_cb', $Text);
+	// Replace inline code blocks
+	$Text = preg_replace_callback("|(?!<br[^>]*>)<code>([^<]*)</code>(?!<br[^>]*>)|ism",
+		function ($match) use ($simplehtml) {
+			$return = '<key>' . $match[1] . '</key>';
+			// Use <code> for Diaspora inline code blocks
+			if ($simplehtml === 3) {
+				$return = '<code>' . $match[1] . '</code>';
+			}
+			return $return;
+		}
+	, $Text);
 
 	// Unhide all [noparse] contained bbtags unspacefying them
 	// and triming the [noparse] tag.