| 1 |
<?php |
| 2 |
/** |
| 3 |
* MCP Name Sanitizer utility for normalizing component names. |
| 4 |
* |
| 5 |
* @package McpAdapter |
| 6 |
*/ |
| 7 |
|
| 8 |
declare( strict_types=1 ); |
| 9 |
|
| 10 |
namespace WP\MCP\Domain\Utils; |
| 11 |
|
| 12 |
use WP_Error; |
| 13 |
|
| 14 |
/** |
| 15 |
* Utility class for sanitizing names to MCP-valid format. |
| 16 |
* |
| 17 |
* Implements best-effort sanitization per MCP 2025-11-25 spec: |
| 18 |
* - Length: 1-128 characters |
| 19 |
* - Charset: A-Za-z0-9_.- |
| 20 |
* |
| 21 |
* Used for both tools and prompts (same naming rules). |
| 22 |
* NOT used for resources (which use URIs as identifiers). |
| 23 |
* |
| 24 |
* @since 0.5.0 |
| 25 |
*/ |
| 26 |
class McpNameSanitizer { |
| 27 |
|
| 28 |
/** |
| 29 |
* Maximum length for MCP tool/prompt names per spec. |
| 30 |
* |
| 31 |
* @var int |
| 32 |
*/ |
| 33 |
public const MAX_LENGTH = 128; |
| 34 |
|
| 35 |
/** |
| 36 |
* Length of the hash suffix used for truncation uniqueness. |
| 37 |
* |
| 38 |
* @var int |
| 39 |
*/ |
| 40 |
public const HASH_LENGTH = 12; |
| 41 |
|
| 42 |
/** |
| 43 |
* Maximum characters to keep when truncating (MAX_LENGTH - 1 separator - HASH_LENGTH). |
| 44 |
* |
| 45 |
* @var int |
| 46 |
*/ |
| 47 |
public const TRUNCATE_LENGTH = 115; |
| 48 |
|
| 49 |
/** |
| 50 |
* Sanitize a name to be MCP-valid for tools and prompts. |
| 51 |
* |
| 52 |
* Normalization steps: |
| 53 |
* 1. Trim whitespace |
| 54 |
* 2. Replace `/` with `-` (forward slash not allowed in MCP) |
| 55 |
* 3. If valid, return as-is |
| 56 |
* 4. Otherwise: transliterate accents, replace invalid chars, collapse hyphens, trim edges |
| 57 |
* 5. If too long: truncate + add hash suffix for uniqueness |
| 58 |
* 6. If still invalid/empty: return WP_Error |
| 59 |
* |
| 60 |
* @param string $name Original name. |
| 61 |
* |
| 62 |
* @return string|\WP_Error Sanitized name or WP_Error if unsalvageable. |
| 63 |
* @since 0.5.0 |
| 64 |
* |
| 65 |
*/ |
| 66 |
public static function sanitize_name( string $name ) { |
| 67 |
$original = $name; |
| 68 |
|
| 69 |
// Step 1: Trim whitespace. |
| 70 |
$name = trim( $name ); |
| 71 |
|
| 72 |
// Step 2: Replace / with - (forward slash not allowed in MCP). |
| 73 |
$name = str_replace( '/', '-', $name ); |
| 74 |
|
| 75 |
// Step 3: Early validation - if already valid, return as-is. |
| 76 |
if ( McpValidator::validate_name( $name ) ) { |
| 77 |
return $name; |
| 78 |
} |
| 79 |
|
| 80 |
// Step 4a: Transliterate accented characters to ASCII equivalents. |
| 81 |
// Uses WordPress core function: é→e, ü→u, ñ→n, etc. |
| 82 |
$name = remove_accents( $name ); |
| 83 |
|
| 84 |
// Step 4b: Replace any remaining non-allowed chars with hyphen. |
| 85 |
$name = (string) preg_replace( '/[^a-zA-Z0-9_.-]/', '-', $name ); |
| 86 |
|
| 87 |
// Step 4c: Collapse consecutive hyphens. |
| 88 |
$name = (string) preg_replace( '/-+/', '-', $name ); |
| 89 |
|
| 90 |
// Step 4d: Trim leading/trailing hyphens and underscores. |
| 91 |
$name = trim( $name, '-_' ); |
| 92 |
|
| 93 |
// Step 5: Handle length > 128. |
| 94 |
if ( strlen( $name ) > self::MAX_LENGTH ) { |
| 95 |
$hash = substr( md5( $original ), 0, self::HASH_LENGTH ); |
| 96 |
$name = substr( $name, 0, self::TRUNCATE_LENGTH ) . '-' . $hash; |
| 97 |
} |
| 98 |
|
| 99 |
// Step 6: Final check - only empty is possible failure after sanitization. |
| 100 |
// Characters are guaranteed valid (replaced), length is handled (truncated). |
| 101 |
if ( empty( $name ) ) { |
| 102 |
return new WP_Error( |
| 103 |
'mcp_name_invalid', |
| 104 |
sprintf( |
| 105 |
/* translators: %s: original ability name */ |
| 106 |
__( 'Unable to derive valid MCP name from: %s', 'mcp-adapter' ), |
| 107 |
$original |
| 108 |
) |
| 109 |
); |
| 110 |
} |
| 111 |
|
| 112 |
return $name; |
| 113 |
} |
| 114 |
} |
| 115 |
|
