From 4833c9a47521654f853d11641c87c7f3f3afc96a Mon Sep 17 00:00:00 2001
From: Bogdan Abaev <bogdan@zotero.org>
Date: Thu, 13 Jul 2023 01:15:36 +0000
Subject: [PATCH] strip titles of html tags + update regex

Strip all html tags from title in getSortTitle function
Update the regex in getSortTitle to not skip double brackets

Fixes: #159
---
 misc/db-updates/2023-07-12/sortTitleClean | 34 +++++++++++++++++++++++
 model/Items.inc.php                       |  3 +-
 2 files changed, 36 insertions(+), 1 deletion(-)
 create mode 100755 misc/db-updates/2023-07-12/sortTitleClean

diff --git a/misc/db-updates/2023-07-12/sortTitleClean b/misc/db-updates/2023-07-12/sortTitleClean
new file mode 100755
index 00000000..765ab26e
--- /dev/null
+++ b/misc/db-updates/2023-07-12/sortTitleClean
@@ -0,0 +1,34 @@
+#!/usr/bin/php -d mysqlnd.net_read_timeout=86400
+<?php
+set_include_path("../../../include");
+require("header.inc.php");
+
+Z_Core::$debug = true;
+
+$startShard = !empty($argv[1]) ? $argv[1] : 1;
+$shardIDs = Zotero_DB::columnQuery("SELECT shardID FROM shards WHERE shardID >= ? ORDER BY shardID", $startShard);
+
+foreach ($shardIDs as $shardID) {
+	echo "Shard: $shardID\n";
+
+	# Fetch all title fields with special characters: [, ], (, ), {, }, <, >, ", ', ”, ’, “, ‘ 
+	$rows = Zotero_DB::query("SELECT i.itemID, i.value, s.sortTitle FROM itemData as i INNER JOIN itemSortFields as s ON s.itemID = i.itemID WHERE fieldID = 110 AND value REGEXP '[{}><()\\\\[\\\\]\"\'”’“‘]' ", false, $shardID);
+	Zotero_DB::beginTransaction();
+	foreach($rows as $row) {
+		# Find the desired sort title for that title field
+		$sortTitle = Zotero_Items::getSortTitle($row['value']);
+		# Do nothing if first characters of existing sort title are the same as of desired sort title
+		if (isset($row['sortTitle']) && mb_substr($sortTitle ?? '', 0, 5) == mb_substr($row['sortTitle'] ?? '', 0, 5)) {
+			continue;
+		}
+		# Do nothing if first characters of desired sort title are the same as of title
+		if (mb_substr($sortTitle ?? '', 0, 5) == mb_substr($row['value'] ?? '', 0, 5)) {
+			continue;
+		}
+		# Update item sort title
+		Zotero_DB::query("UPDATE itemSortFields SET sortTitle = ? WHERE itemID = ?", [$sortTitle, $row['itemID']], $shardID);
+	}
+	Zotero_DB::commit();
+	
+}
+echo "Done";
\ No newline at end of file
diff --git a/model/Items.inc.php b/model/Items.inc.php
index df08d0c6..9652a4b2 100644
--- a/model/Items.inc.php
+++ b/model/Items.inc.php
@@ -2580,7 +2580,8 @@ public static function getSortTitle($title) {
 		if (!$title) {
 			return '';
 		}
-		return mb_strcut(preg_replace('/^[[({\-"\'“‘ ]+(.*)[\])}\-"\'”’ ]*?$/Uu', '$1', $title), 0, Zotero_Notes::$MAX_TITLE_LENGTH);
+		$cleaned_of_characters = preg_replace('/[\[({\\-"\'“‘\])}\-"\'”’]/u', '$1', $title);
+		return mb_strcut(strip_tags($cleaned_of_characters), 0, Zotero_Notes::$MAX_TITLE_LENGTH);
 	}
 }