From c92ec738559fe76a57be40ef67f53bf51f3947fe Mon Sep 17 00:00:00 2001
From: Eddie Kohler <ekohler@gmail.com>
Date: Fri, 10 Dec 2021 08:20:48 -0500
Subject: [PATCH] Add Dbl::utf8_charset and convert_utf8.

---
 lib/dbl.php            | 45 ++++++++++++++++++++++++++++++++----------
 src/confinvariants.php |  5 +++--
 2 files changed, 38 insertions(+), 12 deletions(-)

diff --git a/lib/dbl.php b/lib/dbl.php
index f077afa64..75b9fca44 100644
--- a/lib/dbl.php
+++ b/lib/dbl.php
@@ -835,23 +835,48 @@ static function shutdown() {
         self::$query_log = false;
     }
 
-    /** @return string */
-    static function utf8(/* [$dblink,] $qstr */) {
-        $args = func_get_args();
-        $dblink = count($args) > 1 ? $args[0] : self::$default_dblink;
+    /** @param \mysqli|string $dblink
+     * @param ?string $qstr
+     * @return string */
+    static function utf8($dblink, $qstr = null) {
+        if (is_string($dblink)) {
+            $qstr = $dblink;
+            $dblink = self::$default_dblink;
+        }
         $utf8 = $dblink->server_version >= 50503 ? "utf8mb4" : "utf8";
-        $qstr = count($args) > 1 ? $args[1] : $args[0];
         return "_{$utf8}{$qstr}";
     }
 
-    /** @return string */
-    static function utf8ci(/* [$dblink,] $qstr */) {
-        $args = func_get_args();
-        $dblink = count($args) > 1 ? $args[0] : self::$default_dblink;
+    /** @param \mysqli|string $dblink
+     * @param ?string $qstr
+     * @return string */
+    static function utf8ci($dblink, $qstr = null) {
+        if (is_string($dblink)) {
+            $qstr = $dblink;
+            $dblink = self::$default_dblink;
+        }
         $utf8 = $dblink->server_version >= 50503 ? "utf8mb4" : "utf8";
-        $qstr = count($args) > 1 ? $args[1] : $args[0];
         return "_{$utf8}{$qstr} collate {$utf8}_general_ci";
     }
+
+    /** @param \mysqli|string $dblink
+     * @param ?string $qstr
+     * @return string */
+    static function convert_utf8($dblink, $qstr = null) {
+        if (is_string($dblink)) {
+            $qstr = $dblink;
+            $dblink = self::$default_dblink;
+        }
+        $utf8 = $dblink->server_version >= 50503 ? "utf8mb4" : "utf8";
+        return "convert($qstr using $utf8)";
+    }
+
+    /** @param ?\mysqli $dblink
+     * @return string */
+    static function utf8_charset($dblink = null) {
+        $dblink = $dblink ?? self::$default_dblink;
+        return $dblink->server_version >= 50503 ? "utf8mb4" : "utf8";
+    }
 }
 
 // quoting for SQL
diff --git a/src/confinvariants.php b/src/confinvariants.php
index c95e68308..82b2b2165 100644
--- a/src/confinvariants.php
+++ b/src/confinvariants.php
@@ -162,8 +162,9 @@ function exec_main() {
         Dbl::free($result);
 
         // whitespace is simplified
-        $regex = Dbl::utf8($this->conf->dblink, "'^ | \$|  |[\\n\\r\\t]'");
-        $any = $this->invariantq("select email from ContactInfo where convert(firstName using utf8mb4) regexp $regex or convert(lastName using utf8mb4) regexp $regex or convert(affiliation using utf8mb4) regexp $regex limit 1");
+        $utf8cs = Dbl::utf8_charset($this->conf->dblink);
+        $regex = "_{$utf8cs}'^ | \$|  |[\\n\\r\\t]'";
+        $any = $this->invariantq("select email from ContactInfo where convert(firstName using $utf8cs) regexp $regex or convert(lastName using $utf8cs) regexp $regex or convert(affiliation using $utf8cs) regexp $regex limit 1");
         if ($any) {
             $this->invariant_error("user_whitespace", "user whitespace is not simplified");
         }