| 1 |
<?php |
| 2 |
|
| 3 |
namespace Code_Snippets\Utils; |
| 4 |
|
| 5 |
/** |
| 6 |
* Validates code prior to execution. |
| 7 |
* |
| 8 |
* @package Code_Snippets |
| 9 |
*/ |
| 10 |
class Validator { |
| 11 |
|
| 12 |
/** |
| 13 |
* Code to validate. |
| 14 |
* |
| 15 |
* @var string |
| 16 |
*/ |
| 17 |
private string $code; |
| 18 |
|
| 19 |
/** |
| 20 |
* List of tokens. |
| 21 |
* |
| 22 |
* @var array<string> |
| 23 |
*/ |
| 24 |
private array $tokens; |
| 25 |
|
| 26 |
/** |
| 27 |
* The index of the token currently being examined. |
| 28 |
* |
| 29 |
* @var int |
| 30 |
*/ |
| 31 |
private int $current; |
| 32 |
|
| 33 |
/** |
| 34 |
* The total number of tokens. |
| 35 |
* |
| 36 |
* @var int |
| 37 |
*/ |
| 38 |
private int $length; |
| 39 |
|
| 40 |
/** |
| 41 |
* Array to keep track of the various function, class and interface identifiers which have been defined. |
| 42 |
* |
| 43 |
* @var array<string, string[]> |
| 44 |
*/ |
| 45 |
private array $defined_identifiers = []; |
| 46 |
|
| 47 |
/** |
| 48 |
* Exclude certain tokens from being checked. |
| 49 |
* |
| 50 |
* @var array<string, string[]> |
| 51 |
*/ |
| 52 |
private array $exceptions = []; |
| 53 |
|
| 54 |
/** |
| 55 |
* Identifiers already claimed by other snippets being validated alongside |
| 56 |
* this one. |
| 57 |
* |
| 58 |
* A snippet is validated against everything PHP has declared so far, which |
| 59 |
* does not include a snippet that is about to be activated in the same |
| 60 |
* batch. Two snippets declaring the same function therefore both passed and |
| 61 |
* both activated, and the site crashed on the next request. |
| 62 |
* |
| 63 |
* @var array<string, string[]> |
| 64 |
*/ |
| 65 |
private array $claimed_identifiers; |
| 66 |
|
| 67 |
/** |
| 68 |
* Namespace the code being read currently declares, lower-cased, or empty for the global namespace. |
| 69 |
* |
| 70 |
* @var string |
| 71 |
*/ |
| 72 |
private string $namespace = ''; |
| 73 |
|
| 74 |
/** |
| 75 |
* Class constructor. |
| 76 |
* |
| 77 |
* @param string $code Snippet code for parsing. |
| 78 |
* @param array<string, string[]> $claimed_identifiers Identifiers already claimed by |
| 79 |
* snippets validated alongside this one. |
| 80 |
*/ |
| 81 |
public function __construct( string $code, array $claimed_identifiers = [] ) { |
| 82 |
$this->claimed_identifiers = $claimed_identifiers; |
| 83 |
$this->code = $code; |
| 84 |
$this->tokens = token_get_all( "<?php\n" . $this->code ); |
| 85 |
$this->length = count( $this->tokens ); |
| 86 |
$this->current = 0; |
| 87 |
} |
| 88 |
|
| 89 |
/** |
| 90 |
* Retrieve the identifiers claimed so far, including this snippet's own. |
| 91 |
* |
| 92 |
* Pass the result to the next Validator in a batch so that two snippets |
| 93 |
* cannot both claim the same name. |
| 94 |
* |
| 95 |
* @return array<string, string[]> |
| 96 |
*/ |
| 97 |
public function get_claimed_identifiers(): array { |
| 98 |
return $this->claimed_identifiers; |
| 99 |
} |
| 100 |
|
| 101 |
/** |
| 102 |
* Determine whether the parser has reached the end of the list of tokens. |
| 103 |
* |
| 104 |
* @return bool |
| 105 |
*/ |
| 106 |
private function end(): bool { |
| 107 |
return $this->current === $this->length; |
| 108 |
} |
| 109 |
|
| 110 |
/** |
| 111 |
* Retrieve the next token without moving the pointer |
| 112 |
* |
| 113 |
* @return string|array<string|int>|null The current token if the list has not been expended, null otherwise. |
| 114 |
*/ |
| 115 |
private function peek() { |
| 116 |
return $this->end() ? null : $this->tokens[ $this->current ]; |
| 117 |
} |
| 118 |
|
| 119 |
/** |
| 120 |
* Move the pointer to the next token, if there is one. |
| 121 |
* |
| 122 |
* If the first argument is provided, only move the pointer if the tokens match. |
| 123 |
*/ |
| 124 |
private function next() { |
| 125 |
if ( ! $this->end() ) { |
| 126 |
++$this->current; |
| 127 |
} |
| 128 |
} |
| 129 |
|
| 130 |
/** |
| 131 |
* Check whether a particular identifier has been used previously. |
| 132 |
* |
| 133 |
* @param string $type Which type of identifier this is. Supports T_FUNCTION, T_CLASS and T_INTERFACE. |
| 134 |
* @param string $identifier The name of the identifier itself. |
| 135 |
* |
| 136 |
* @return bool true if the identifier is not unique. |
| 137 |
*/ |
| 138 |
private function check_duplicate_identifier( string $type, string $identifier ): bool { |
| 139 |
$identifier = strtolower( ltrim( $identifier, '\\' ) ); |
| 140 |
|
| 141 |
// PHP keeps declared names fully qualified, so that is the form compared |
| 142 |
// and claimed: the same short name in two namespaces is two names. |
| 143 |
$qualified = '' === $this->namespace ? $identifier : $this->namespace . '\\' . $identifier; |
| 144 |
$namespaced_identifier = 'code_snippets\\' . $identifier; |
| 145 |
|
| 146 |
if ( ! isset( $this->defined_identifiers[ $type ] ) ) { |
| 147 |
switch ( $type ) { |
| 148 |
case T_FUNCTION: |
| 149 |
$this->defined_identifiers[ T_FUNCTION ] = array_map( |
| 150 |
'strtolower', |
| 151 |
array_merge( get_defined_functions()['internal'], get_defined_functions()['user'] ) |
| 152 |
); |
| 153 |
break; |
| 154 |
|
| 155 |
case T_CLASS: |
| 156 |
$this->defined_identifiers[ T_CLASS ] = array_map( 'strtolower', get_declared_classes() ); |
| 157 |
break; |
| 158 |
|
| 159 |
case T_INTERFACE: |
| 160 |
$this->defined_identifiers[ T_INTERFACE ] = array_map( 'strtolower', get_declared_interfaces() ); |
| 161 |
break; |
| 162 |
|
| 163 |
default: |
| 164 |
return false; |
| 165 |
} |
| 166 |
} |
| 167 |
|
| 168 |
$known = array_merge( |
| 169 |
$this->defined_identifiers[ $type ], |
| 170 |
$this->claimed_identifiers[ $type ] ?? [] |
| 171 |
); |
| 172 |
|
| 173 |
$duplicate_identifier = in_array( $qualified, $known, true ); |
| 174 |
$duplicate_namespaced = '' === $this->namespace && in_array( $namespaced_identifier, $known, true ); |
| 175 |
$exceptions = $this->exceptions[ $type ] ?? []; |
| 176 |
$exception_identifier = in_array( $identifier, $exceptions, true ) || in_array( $qualified, $exceptions, true ); |
| 177 |
$exception_namespaced = in_array( $identifier, $exceptions, true ) || in_array( $namespaced_identifier, $exceptions, true ); |
| 178 |
|
| 179 |
array_unshift( $this->defined_identifiers[ $type ], $qualified ); |
| 180 |
$this->claimed_identifiers[ $type ][] = $qualified; |
| 181 |
|
| 182 |
return ( $duplicate_identifier && ! $exception_identifier ) || ( $duplicate_namespaced && ! $exception_namespaced ); |
| 183 |
} |
| 184 |
|
| 185 |
/** |
| 186 |
* Read the name a namespace declaration introduces, leaving the cursor after it. |
| 187 |
* |
| 188 |
* A bare "namespace {" opens the global namespace; "namespace\\foo()" is a |
| 189 |
* relative name rather than a declaration and is left alone. |
| 190 |
* |
| 191 |
* @return string Lower-cased namespace, or empty for the global namespace. |
| 192 |
*/ |
| 193 |
private function read_namespace_declaration(): string { |
| 194 |
$name = ''; |
| 195 |
|
| 196 |
while ( ! $this->end() ) { |
| 197 |
$token = $this->peek(); |
| 198 |
|
| 199 |
if ( is_array( $token ) ) { |
| 200 |
if ( T_WHITESPACE === $token[0] || T_COMMENT === $token[0] || T_DOC_COMMENT === $token[0] ) { |
| 201 |
$this->next(); |
| 202 |
continue; |
| 203 |
} |
| 204 |
|
| 205 |
if ( T_NS_SEPARATOR === $token[0] && '' === $name ) { |
| 206 |
return $this->namespace; |
| 207 |
} |
| 208 |
|
| 209 |
if ( defined( 'T_NAME_RELATIVE' ) && T_NAME_RELATIVE === $token[0] ) { |
| 210 |
return $this->namespace; |
| 211 |
} |
| 212 |
|
| 213 |
if ( T_STRING === $token[0] || T_NS_SEPARATOR === $token[0] |
| 214 |
|| ( defined( 'T_NAME_QUALIFIED' ) && T_NAME_QUALIFIED === $token[0] ) ) { |
| 215 |
$name .= $token[1]; |
| 216 |
$this->next(); |
| 217 |
continue; |
| 218 |
} |
| 219 |
|
| 220 |
return $this->namespace; |
| 221 |
} |
| 222 |
|
| 223 |
if ( ';' === $token || '{' === $token ) { |
| 224 |
$this->next(); |
| 225 |
return strtolower( trim( $name, '\\' ) ); |
| 226 |
} |
| 227 |
|
| 228 |
return $this->namespace; |
| 229 |
} |
| 230 |
|
| 231 |
return strtolower( trim( $name, '\\' ) ); |
| 232 |
} |
| 233 |
|
| 234 |
/** |
| 235 |
* Validate the given PHP code and return the result. |
| 236 |
* |
| 237 |
* @return array<string, mixed>|false Array containing message if an error was encountered, false if validation was successful. |
| 238 |
*/ |
| 239 |
public function validate() { |
| 240 |
|
| 241 |
while ( ! $this->end() ) { |
| 242 |
$token = $this->peek(); |
| 243 |
$this->next(); |
| 244 |
|
| 245 |
if ( ! is_array( $token ) ) { |
| 246 |
continue; |
| 247 |
} |
| 248 |
|
| 249 |
if ( T_NAMESPACE === $token[0] ) { |
| 250 |
$this->namespace = $this->read_namespace_declaration(); |
| 251 |
continue; |
| 252 |
} |
| 253 |
|
| 254 |
// If this is a function or class exists check, then allow this function or class to be defined. |
| 255 |
if ( T_STRING === $token[0] && ( 'function_exists' === $token[1] || 'class_exists' === $token[1] ) ) { |
| 256 |
$type = 'function_exists' === $token[1] ? T_FUNCTION : T_CLASS; |
| 257 |
|
| 258 |
// Eat tokens until we find the function or class name. |
| 259 |
while ( ! $this->end() && T_CONSTANT_ENCAPSED_STRING !== $token[0] ) { |
| 260 |
$token = $this->peek(); |
| 261 |
$this->next(); |
| 262 |
} |
| 263 |
|
| 264 |
// Add the identifier to the list of exceptions. |
| 265 |
$identifier = strtolower( ltrim( trim( $token[1], '\'"' ), '\\' ) ); |
| 266 |
|
| 267 |
if ( '' !== $identifier ) { |
| 268 |
$this->exceptions[ $type ] = $this->exceptions[ $type ] ?? []; |
| 269 |
$this->exceptions[ $type ][] = $identifier; |
| 270 |
} |
| 271 |
continue; |
| 272 |
} |
| 273 |
|
| 274 |
// If we have a double colon, followed by a class, then consume it before the next section. |
| 275 |
if ( T_DOUBLE_COLON === $token[0] ) { |
| 276 |
$token = $this->peek(); |
| 277 |
$this->next(); |
| 278 |
|
| 279 |
if ( T_CLASS === $token[0] ) { |
| 280 |
$this->next(); |
| 281 |
$token = $this->peek(); |
| 282 |
} |
| 283 |
} |
| 284 |
|
| 285 |
// Only look for class and function declaration tokens. |
| 286 |
if ( T_CLASS !== $token[0] && T_FUNCTION !== $token[0] ) { |
| 287 |
continue; |
| 288 |
} |
| 289 |
|
| 290 |
/** |
| 291 |
* Ensure the type of $token is inferred correctly. |
| 292 |
* |
| 293 |
* @var string|array<string|int> $token |
| 294 |
*/ |
| 295 |
$structure_type = $token[0]; |
| 296 |
|
| 297 |
// Continue eating tokens until we find the name of the class or function. |
| 298 |
while ( ! $this->end() && T_STRING !== $token[0] && |
| 299 |
( T_FUNCTION !== $structure_type || '(' !== $token ) && ( T_CLASS !== $structure_type || '{' !== $token ) ) { |
| 300 |
$token = $this->peek(); |
| 301 |
$this->next(); |
| 302 |
} |
| 303 |
|
| 304 |
// If we've eaten all the tokens without discovering a name, then there must be a syntax error, so return appropriately. |
| 305 |
if ( $this->end() ) { |
| 306 |
return array( |
| 307 |
'message' => __( 'Parse error: syntax error, unexpected end of snippet.', 'code-snippets' ), |
| 308 |
'line' => $token[2], |
| 309 |
); |
| 310 |
} |
| 311 |
|
| 312 |
// If the function or class is anonymous, with no name, then no need to check. |
| 313 |
if ( ! ( T_FUNCTION === $structure_type && '(' === $token ) && ! ( T_CLASS === $structure_type && '{' === $token ) ) { |
| 314 |
|
| 315 |
// Check whether the name has already been defined. |
| 316 |
if ( $this->check_duplicate_identifier( $structure_type, $token[1] ) ) { |
| 317 |
switch ( $structure_type ) { |
| 318 |
case T_FUNCTION: |
| 319 |
/* translators: %s: PHP function name */ |
| 320 |
$message = __( 'Cannot redeclare function %s.', 'code-snippets' ); |
| 321 |
break; |
| 322 |
case T_CLASS: |
| 323 |
/* translators: %s: PHP class name */ |
| 324 |
$message = __( 'Cannot redeclare class %s.', 'code-snippets' ); |
| 325 |
break; |
| 326 |
case T_INTERFACE: |
| 327 |
/* translators: %s: PHP interface name */ |
| 328 |
$message = __( 'Cannot redeclare interface %s.', 'code-snippets' ); |
| 329 |
break; |
| 330 |
default: |
| 331 |
/* translators: %s: PHP identifier name*/ |
| 332 |
$message = __( 'Cannot redeclare %s.', 'code-snippets' ); |
| 333 |
} |
| 334 |
|
| 335 |
return array( |
| 336 |
'message' => sprintf( $message, $token[1] ), |
| 337 |
'line' => $token[2], |
| 338 |
); |
| 339 |
} |
| 340 |
} |
| 341 |
|
| 342 |
// If we have entered into a class, eat tokens until we find the closing brace. |
| 343 |
if ( T_CLASS !== $structure_type ) { |
| 344 |
continue; |
| 345 |
} |
| 346 |
|
| 347 |
// Find the opening brace for the class. |
| 348 |
while ( ! $this->end() && '{' !== $token ) { |
| 349 |
$token = $this->peek(); |
| 350 |
$this->next(); |
| 351 |
} |
| 352 |
|
| 353 |
// Continue traversing the class tokens until we have found the class closing brace. |
| 354 |
$depth = 1; |
| 355 |
while ( ! $this->end() && $depth > 0 ) { |
| 356 |
$token = $this->peek(); |
| 357 |
|
| 358 |
if ( '{' === $token ) { |
| 359 |
++$depth; |
| 360 |
} elseif ( '}' === $token ) { |
| 361 |
--$depth; |
| 362 |
} |
| 363 |
|
| 364 |
$this->next(); |
| 365 |
} |
| 366 |
|
| 367 |
// If we did not make it out of the class, then there's a problem. |
| 368 |
if ( $depth > 0 ) { |
| 369 |
return array( |
| 370 |
'message' => __( 'Parse error: syntax error, unexpected end of snippet.', 'code-snippets' ), |
| 371 |
'line' => $token[2], |
| 372 |
); |
| 373 |
} |
| 374 |
} |
| 375 |
|
| 376 |
return false; |
| 377 |
} |
| 378 |
} |
| 379 |
|
