Merge pull request #173 from vrana/pgsql-regexp

Treat ~* as operator

Author: greg0ire

src/Tokenizer.php

@@ -743,6 +743,7 @@ final class Tokenizer
         '>',
         '+',
         '-',
+        '~*', // https://www.postgresql.org/docs/current/functions-matching.html#FUNCTIONS-POSIX-REGEXP
         '*',
         '/',
         '!',
@@ -820,10 +821,10 @@ 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 +888,7 @@ private function makeTokenizeRegex(array $regexes): string
             $parts[] = '(?<t_' . $type . '>' . $regex . ')';
         }
 
-        return '~\G(?:' . implode('|', $parts) . ')~';
+        return '(\G(?:' . implode('|', $parts) . '))';
     }
 
     /**

tests/clihighlight.txt

@@ -1211,3 +1211,6 @@ MY_NON_TOP_LEVEL_KEYWORD_FX_3();
 ---
 SELECT
   vector1 <#> vector2
+---
+SELECT
+  text ~* '\w+'

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+'

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>

tests/format.txt

@@ -1209,3 +1209,6 @@ SELECT
 ---
 SELECT
   vector1 <#> vector2
+---
+SELECT
+  text ~* '\w+'

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>

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+'