From b97db67d1d184046899a0046af547d40c486df7f Mon Sep 17 00:00:00 2001
From: Jakub Vrana <jakub.vrana@csfd.cz>
Date: Sat, 7 Feb 2026 19:44:18 +0100
Subject: [PATCH 1/2] Fix regular expression escaping

`/` was unnecessarily escaped even though the delimiter was `~`.
Using delimiter `~` required escaping it in `$boundaries` and elsewhere
even though these are `preg_quote`'d later.
Using `()` as the delimiter allows treating all characters the same.
---
 src/Tokenizer.php | 6 +++---
 1 file changed, 3 insertions(+), 3 deletions(-)

diff --git a/src/Tokenizer.php b/src/Tokenizer.php
index 5682c49..8b76db4 100644
--- a/src/Tokenizer.php
+++ b/src/Tokenizer.php
@@ -820,10 +820,10 @@ private function makeRegexFromList(array $values, bool $sorted = false): string
                 $prefix = (string) $prefix;
             }
 
-            $regex .= preg_quote($prefix, '/');
+            $regex .= preg_quote($prefix);
 
             $regex .= count($items) === 1
-                ? preg_quote(substr(reset($items), strlen($prefix)), '/')
+                ? preg_quote(substr(reset($items), strlen($prefix)))
                 : $this->makeRegexFromList(array_map(static fn ($v) => substr($v, strlen($prefix)), $items), true);
         }
 
@@ -887,7 +887,7 @@ private function makeTokenizeRegex(array $regexes): string
             $parts[] = '(?<t_' . $type . '>' . $regex . ')';
         }
 
-        return '~\G(?:' . implode('|', $parts) . ')~';
+        return '(\G(?:' . implode('|', $parts) . '))';
     }
 
     /**

From 6d638a78655cf83cf1b51047e222c9dd09c27f0a Mon Sep 17 00:00:00 2001
From: Jakub Vrana <jakub.vrana@csfd.cz>
Date: Sat, 7 Feb 2026 19:44:34 +0100
Subject: [PATCH 2/2] Treat ~* as operator

---
 src/Tokenizer.php           | 1 +
 tests/clihighlight.txt      | 3 +++
 tests/compress.txt          | 2 ++
 tests/format-highlight.html | 3 +++
 tests/format.txt            | 3 +++
 tests/highlight.html        | 2 ++
 tests/sql.sql               | 2 ++
 7 files changed, 16 insertions(+)

diff --git a/src/Tokenizer.php b/src/Tokenizer.php
index 8b76db4..a62bfe3 100644
--- a/src/Tokenizer.php
+++ b/src/Tokenizer.php
@@ -743,6 +743,7 @@ final class Tokenizer
         '>',
         '+',
         '-',
+        '~*', // https://www.postgresql.org/docs/current/functions-matching.html#FUNCTIONS-POSIX-REGEXP
         '*',
         '/',
         '!',
diff --git a/tests/clihighlight.txt b/tests/clihighlight.txt
index 75c8429..ceaf391 100644
--- a/tests/clihighlight.txt
+++ b/tests/clihighlight.txt
@@ -1211,3 +1211,6 @@ MY_NON_TOP_LEVEL_KEYWORD_FX_3();
 ---
 SELECT
   vector1 <#> vector2
+---
+SELECT
+  text ~* '\w+'
diff --git a/tests/compress.txt b/tests/compress.txt
index f7fcba4..6ec44ca 100644
--- a/tests/compress.txt
+++ b/tests/compress.txt
@@ -115,3 +115,5 @@ CREATE TABLE t (c VARCHAR(20)) DEFAULT CHARACTER SET utf8mb4 ENGINE = InnoDB
 SELECT '{}'::json #> '{}'
 ---
 SELECT vector1 <#> vector2
+---
+SELECT text ~* '\w+'
diff --git a/tests/format-highlight.html b/tests/format-highlight.html
index a059de3..bafb6f2 100644
--- a/tests/format-highlight.html
+++ b/tests/format-highlight.html
@@ -1211,3 +1211,6 @@
 ---
 <pre style="color: black; background-color: white;"><span style="font-weight:bold;">SELECT</span>
   <span style="color: #333;">vector1</span> <span >&lt;</span><span >#</span><span >&gt;</span> <span style="color: #333;">vector2</span></pre>
+---
+<pre style="color: black; background-color: white;"><span style="font-weight:bold;">SELECT</span>
+  <span style="color: #333;">text</span> <span >~*</span> <span style="color: blue;">'\w+'</span></pre>
diff --git a/tests/format.txt b/tests/format.txt
index 868db4b..d3bea98 100644
--- a/tests/format.txt
+++ b/tests/format.txt
@@ -1209,3 +1209,6 @@ SELECT
 ---
 SELECT
   vector1 <#> vector2
+---
+SELECT
+  text ~* '\w+'
diff --git a/tests/highlight.html b/tests/highlight.html
index e988db0..9d24e6c 100644
--- a/tests/highlight.html
+++ b/tests/highlight.html
@@ -429,3 +429,5 @@
 <pre style="color: black; background-color: white;"><span style="font-weight:bold;">SELECT</span> <span style="color: blue;">'{}'</span><span >::</span><span style="color: #333;">json</span> <span >#</span><span >&gt;</span> <span style="color: blue;">'{}'</span></pre>
 ---
 <pre style="color: black; background-color: white;"><span style="font-weight:bold;">SELECT</span> <span style="color: #333;">vector1</span> <span >&lt;</span><span >#</span><span >&gt;</span> <span style="color: #333;">vector2</span></pre>
+---
+<pre style="color: black; background-color: white;"><span style="font-weight:bold;">SELECT</span> <span style="color: #333;">text</span> <span >~*</span> <span style="color: blue;">'\w+'</span></pre>
diff --git a/tests/sql.sql b/tests/sql.sql
index c0811e5..2c72c6f 100644
--- a/tests/sql.sql
+++ b/tests/sql.sql
@@ -429,3 +429,5 @@ CREATE TABLE t (c VARCHAR(20)) DEFAULT CHARACTER SET utf8mb4 ENGINE = InnoDB
 SELECT '{}'::json #> '{}'
 ---
 SELECT vector1 <#> vector2
+---
+SELECT text ~* '\w+'