From 55dc5cedc338f56268d8e302cdc52e97206543de Mon Sep 17 00:00:00 2001
From: Michael Foster <savetheinternet@tinyboard.org>
Date: Sat, 7 Sep 2013 12:50:32 +1000
Subject: [PATCH] Steal make_comment_hex() from plainib

---
 inc/filters.php   |  6 +++---
 inc/functions.php | 22 +++++++++++++++++++++-
 2 files changed, 24 insertions(+), 4 deletions(-)

diff --git a/inc/filters.php b/inc/filters.php
index 212118bf..27b167ea 100644
--- a/inc/filters.php
+++ b/inc/filters.php
@@ -39,7 +39,7 @@ class Filter {
 									continue 3;
 								break;
 							case 'body':
-								if ($flood_post['posthash'] != md5($post['body_nomarkup']))
+								if ($flood_post['posthash'] != make_comment_hex($post['body_nomarkup']))
 									continue 3;
 								break;
 							case 'file':
@@ -216,12 +216,12 @@ function do_filters(array $post) {
 		if ($post['has_file']) {
 			$query = prepare("SELECT * FROM ``flood`` WHERE `ip` = :ip OR `posthash` = :posthash OR `filehash` = :filehash");
 			$query->bindValue(':ip', $_SERVER['REMOTE_ADDR']);
-			$query->bindValue(':posthash', md5($post['body_nomarkup']));
+			$query->bindValue(':posthash', make_comment_hex($post['body_nomarkup']));
 			$query->bindValue(':filehash', $post['filehash']);
 		} else {
 			$query = prepare("SELECT * FROM ``flood`` WHERE `ip` = :ip OR `posthash` = :posthash");
 			$query->bindValue(':ip', $_SERVER['REMOTE_ADDR']);
-			$query->bindValue(':posthash', md5($post['body_nomarkup']));
+			$query->bindValue(':posthash', make_comment_hex($post['body_nomarkup']));
 		}
 		$query->execute() or error(db_error($query));
 		$flood_check = $query->fetchAll(PDO::FETCH_ASSOC);
diff --git a/inc/functions.php b/inc/functions.php
index df3a4eeb..ba1b04c8 100644
--- a/inc/functions.php
+++ b/inc/functions.php
@@ -763,7 +763,7 @@ function insertFloodPost(array $post) {
 	$query->bindValue(':ip', $_SERVER['REMOTE_ADDR']);
 	$query->bindValue(':board', $board['uri']);
 	$query->bindValue(':time', time());
-	$query->bindValue(':posthash', md5($post['body_nomarkup']));
+	$query->bindValue(':posthash', make_comment_hex($post['body_nomarkup']));
 	if ($post['has_file'])
 		$query->bindValue(':filehash', $post['filehash']);
 	else
@@ -1191,6 +1191,26 @@ function getPages($mod=false) {
 	return $pages;
 }
 
+// Stolen with permission from PlainIB (by Frank Usrs)
+function make_comment_hex($str) {
+	// remove cross-board citations
+	// the numbers don't matter
+	$str = preg_replace('!>>>/[A-Za-z0-9]+/!', '', $str);
+
+	if (function_exists('iconv')) {
+		// remove diacritics and other noise
+		// FIXME: this removes cyrillic entirely
+		$str = iconv('UTF-8', 'ASCII//TRANSLIT//IGNORE', $str);
+	}
+
+	$str = strtolower($str);
+
+	// strip all non-alphabet characters
+	$str = preg_replace('/[^a-z]/', '', $str);
+
+	return md5($str);
+}
+
 function makerobot($body) {
 	global $config;
 	$body = strtolower($body);