| 1 |
<?php |
| 2 |
/** |
| 3 |
* Glob_Matcher — translate user-friendly glob patterns to PCRE for the |
| 4 |
* cache engine's exclusion checks. |
| 5 |
* |
| 6 |
* Supported syntax (shell-style glob, NOT full PCRE): |
| 7 |
* * → match any run of characters (greedy), including '/' |
| 8 |
* ? → match exactly one character |
| 9 |
* [abc] → character class |
| 10 |
* \* → literal asterisk (escape) |
| 11 |
* |
| 12 |
* Compiles each user pattern once, caches the compiled regex in a |
| 13 |
* static array for the request, then matches with a single preg_match. |
| 14 |
* |
| 15 |
* A bare substring (no glob metacharacters) keeps the legacy "contains" |
| 16 |
* semantics: `/cart` matches `/cart`, `/cart/items`, `/foo/cart/bar`. |
| 17 |
* Adding ANY of `* ? [` switches the pattern to anchored glob mode: |
| 18 |
* `/cart/*` matches `/cart/items` but NOT `/foo/cart/bar`. |
| 19 |
* |
| 20 |
* @package XSpeed |
| 21 |
*/ |
| 22 |
|
| 23 |
declare(strict_types=1); |
| 24 |
|
| 25 |
namespace XSpeed; |
| 26 |
|
| 27 |
defined( 'ABSPATH' ) || exit; |
| 28 |
|
| 29 |
final class Glob_Matcher { |
| 30 |
|
| 31 |
/** |
| 32 |
* @var array<string,string> pattern => compiled regex |
| 33 |
*/ |
| 34 |
private static $compiled = array(); |
| 35 |
|
| 36 |
/** |
| 37 |
* Does any of `$patterns` match `$subject`? |
| 38 |
* |
| 39 |
* @param string[] $patterns |
| 40 |
*/ |
| 41 |
public static function any_match( array $patterns, string $subject ): bool { |
| 42 |
foreach ( $patterns as $p ) { |
| 43 |
$p = (string) $p; |
| 44 |
if ( '' === $p ) { |
| 45 |
continue; |
| 46 |
} |
| 47 |
if ( self::matches( $p, $subject ) ) { |
| 48 |
return true; |
| 49 |
} |
| 50 |
} |
| 51 |
return false; |
| 52 |
} |
| 53 |
|
| 54 |
public static function matches( string $pattern, string $subject ): bool { |
| 55 |
$regex = self::compile( $pattern ); |
| 56 |
// Anchored glob (regex returned starts with '#^') vs substring |
| 57 |
// (regex returned starts with '#'). Both use preg_match the |
| 58 |
// same way; the anchoring is baked into the pattern. |
| 59 |
return 1 === preg_match( $regex, $subject ); |
| 60 |
} |
| 61 |
|
| 62 |
/** |
| 63 |
* Compile a user pattern to a PCRE delimited with `#`. Cached |
| 64 |
* for the request lifetime. |
| 65 |
*/ |
| 66 |
public static function compile( string $pattern ): string { |
| 67 |
if ( isset( self::$compiled[ $pattern ] ) ) { |
| 68 |
return self::$compiled[ $pattern ]; |
| 69 |
} |
| 70 |
// Decide mode based on UNESCAPED glob metacharacters only. |
| 71 |
// `\*` alone keeps the pattern in substring mode (with escapes |
| 72 |
// resolved); `/cart/*` flips to anchored glob mode. |
| 73 |
$has_unescaped_glob = (bool) preg_match( '/(?<!\\\\)[*?\[]/', $pattern ); |
| 74 |
if ( $has_unescaped_glob ) { |
| 75 |
$regex = '#^' . self::glob_to_regex( $pattern ) . '$#'; |
| 76 |
} else { |
| 77 |
// Substring "contains" mode. Resolve `\X` → `X` first so |
| 78 |
// `\*` matches a literal asterisk anywhere in the subject. |
| 79 |
$resolved = preg_replace( '/\\\\(.)/', '$1', $pattern ); |
| 80 |
$regex = '#' . preg_quote( (string) $resolved, '#' ) . '#'; |
| 81 |
} |
| 82 |
self::$compiled[ $pattern ] = $regex; |
| 83 |
return $regex; |
| 84 |
} |
| 85 |
|
| 86 |
/** |
| 87 |
* Translate glob syntax → regex body (no delimiters, no anchors). |
| 88 |
* Mirrors fnmatch's FNM_PATHNAME-disabled semantics: `*` matches |
| 89 |
* across `/` so `/cart/*` correctly catches `/cart/items/sub`. |
| 90 |
*/ |
| 91 |
private static function glob_to_regex( string $glob ): string { |
| 92 |
$out = ''; |
| 93 |
$in_class = false; |
| 94 |
$len = strlen( $glob ); |
| 95 |
$escape = false; |
| 96 |
|
| 97 |
for ( $i = 0; $i < $len; $i++ ) { |
| 98 |
$ch = $glob[ $i ]; |
| 99 |
|
| 100 |
if ( $escape ) { |
| 101 |
$out .= preg_quote( $ch, '#' ); |
| 102 |
$escape = false; |
| 103 |
continue; |
| 104 |
} |
| 105 |
|
| 106 |
if ( '\\' === $ch ) { |
| 107 |
$escape = true; |
| 108 |
continue; |
| 109 |
} |
| 110 |
|
| 111 |
if ( $in_class ) { |
| 112 |
if ( ']' === $ch ) { |
| 113 |
$out .= ']'; |
| 114 |
$in_class = false; |
| 115 |
} else { |
| 116 |
// Inside a character class, dash + letters are passed |
| 117 |
// through; we still preg_quote dangerous chars. |
| 118 |
$out .= preg_quote( $ch, '#' ); |
| 119 |
} |
| 120 |
continue; |
| 121 |
} |
| 122 |
|
| 123 |
switch ( $ch ) { |
| 124 |
case '*': |
| 125 |
$out .= '.*'; |
| 126 |
break; |
| 127 |
case '?': |
| 128 |
$out .= '.'; |
| 129 |
break; |
| 130 |
case '[': |
| 131 |
$out .= '['; |
| 132 |
$in_class = true; |
| 133 |
break; |
| 134 |
default: |
| 135 |
$out .= preg_quote( $ch, '#' ); |
| 136 |
} |
| 137 |
} |
| 138 |
|
| 139 |
return $out; |
| 140 |
} |
| 141 |
|
| 142 |
/** |
| 143 |
* Test-only: clear the compile cache. Production code never needs |
| 144 |
* this (PHP request lifetime handles it). |
| 145 |
*/ |
| 146 |
public static function reset_cache(): void { |
| 147 |
self::$compiled = array(); |
| 148 |
} |
| 149 |
} |
| 150 |
|
