| 1 |
<?php |
| 2 |
/** |
| 3 |
* This file is a port of the Token class from the PHPMyAdmin/sql-parser library. |
| 4 |
* |
| 5 |
* @package wp-sqlite-integration |
| 6 |
* @see https://github.com/phpmyadmin/sql-parser |
| 7 |
*/ |
| 8 |
|
| 9 |
/** |
| 10 |
* Defines a token along with a set of types and flags and utility functions. |
| 11 |
* |
| 12 |
* An array of tokens will result after parsing the query. |
| 13 |
* |
| 14 |
* A structure representing a lexeme that explicitly indicates its categorization for the purpose of parsing. |
| 15 |
*/ |
| 16 |
class WP_SQLite_Token { |
| 17 |
|
| 18 |
/** |
| 19 |
* This type is used when the token is invalid or its type cannot be |
| 20 |
* determined because of the ambiguous context. Further analysis might be |
| 21 |
* required to detect its type. |
| 22 |
*/ |
| 23 |
const TYPE_NONE = 0; |
| 24 |
|
| 25 |
/** |
| 26 |
* SQL specific keywords: SELECT, UPDATE, INSERT, etc. |
| 27 |
*/ |
| 28 |
const TYPE_KEYWORD = 1; |
| 29 |
|
| 30 |
/** |
| 31 |
* Any type of legal operator. |
| 32 |
* |
| 33 |
* Arithmetic operators: +, -, *, /, etc. |
| 34 |
* Logical operators: ===, <>, !==, etc. |
| 35 |
* Bitwise operators: &, |, ^, etc. |
| 36 |
* Assignment operators: =, +=, -=, etc. |
| 37 |
* SQL specific operators: . (e.g. .. WHERE database.table ..), |
| 38 |
* * (e.g. SELECT * FROM ..) |
| 39 |
*/ |
| 40 |
const TYPE_OPERATOR = 2; |
| 41 |
|
| 42 |
/** |
| 43 |
* Spaces, tabs, new lines, etc. |
| 44 |
*/ |
| 45 |
const TYPE_WHITESPACE = 3; |
| 46 |
|
| 47 |
/** |
| 48 |
* Any type of legal comment. |
| 49 |
* |
| 50 |
* Bash (#), C (/* *\/) or SQL (--) comments: |
| 51 |
* |
| 52 |
* -- SQL-comment |
| 53 |
* |
| 54 |
* #Bash-like comment |
| 55 |
* |
| 56 |
* /*C-like comment*\/ |
| 57 |
* |
| 58 |
* or: |
| 59 |
* |
| 60 |
* /*C-like |
| 61 |
* comment*\/ |
| 62 |
* |
| 63 |
* Backslashes were added to respect PHP's comments syntax. |
| 64 |
*/ |
| 65 |
const TYPE_COMMENT = 4; |
| 66 |
|
| 67 |
/** |
| 68 |
* Boolean values: true or false. |
| 69 |
*/ |
| 70 |
const TYPE_BOOL = 5; |
| 71 |
|
| 72 |
/** |
| 73 |
* Numbers: 4, 0x8, 15.16, 23e42, etc. |
| 74 |
*/ |
| 75 |
const TYPE_NUMBER = 6; |
| 76 |
|
| 77 |
/** |
| 78 |
* Literal strings: 'string', "test". |
| 79 |
* Some of these strings are actually symbols. |
| 80 |
*/ |
| 81 |
const TYPE_STRING = 7; |
| 82 |
|
| 83 |
/** |
| 84 |
* Database, table names, variables, etc. |
| 85 |
* For example: ```SELECT `foo`, `bar` FROM `database`.`table`;```. |
| 86 |
*/ |
| 87 |
const TYPE_SYMBOL = 8; |
| 88 |
|
| 89 |
/** |
| 90 |
* Delimits an unknown string. |
| 91 |
* For example: ```SELECT * FROM test;```, `test` is a delimiter. |
| 92 |
*/ |
| 93 |
const TYPE_DELIMITER = 9; |
| 94 |
|
| 95 |
/** |
| 96 |
* Labels in LOOP statement, ITERATE statement etc. |
| 97 |
* For example (only for begin label): |
| 98 |
* begin_label: BEGIN [statement_list] END [end_label] |
| 99 |
* begin_label: LOOP [statement_list] END LOOP [end_label] |
| 100 |
* begin_label: REPEAT [statement_list] ... END REPEAT [end_label] |
| 101 |
* begin_label: WHILE ... DO [statement_list] END WHILE [end_label]. |
| 102 |
*/ |
| 103 |
const TYPE_LABEL = 10; |
| 104 |
|
| 105 |
// Flags that describe the tokens in more detail. |
| 106 |
// All keywords must have flag 1 so `Context::isKeyword` method doesn't |
| 107 |
// require strict comparison. |
| 108 |
const FLAG_KEYWORD_RESERVED = 2; |
| 109 |
const FLAG_KEYWORD_COMPOSED = 4; |
| 110 |
const FLAG_KEYWORD_DATA_TYPE = 8; |
| 111 |
const FLAG_KEYWORD_KEY = 16; |
| 112 |
const FLAG_KEYWORD_FUNCTION = 32; |
| 113 |
|
| 114 |
// Numbers related flags. |
| 115 |
const FLAG_NUMBER_HEX = 1; |
| 116 |
const FLAG_NUMBER_FLOAT = 2; |
| 117 |
const FLAG_NUMBER_APPROXIMATE = 4; |
| 118 |
const FLAG_NUMBER_NEGATIVE = 8; |
| 119 |
const FLAG_NUMBER_BINARY = 16; |
| 120 |
|
| 121 |
// Strings related flags. |
| 122 |
const FLAG_STRING_SINGLE_QUOTES = 1; |
| 123 |
const FLAG_STRING_DOUBLE_QUOTES = 2; |
| 124 |
|
| 125 |
// Comments related flags. |
| 126 |
const FLAG_COMMENT_BASH = 1; |
| 127 |
const FLAG_COMMENT_C = 2; |
| 128 |
const FLAG_COMMENT_SQL = 4; |
| 129 |
const FLAG_COMMENT_MYSQL_CMD = 8; |
| 130 |
|
| 131 |
// Operators related flags. |
| 132 |
const FLAG_OPERATOR_ARITHMETIC = 1; |
| 133 |
const FLAG_OPERATOR_LOGICAL = 2; |
| 134 |
const FLAG_OPERATOR_BITWISE = 4; |
| 135 |
const FLAG_OPERATOR_ASSIGNMENT = 8; |
| 136 |
const FLAG_OPERATOR_SQL = 16; |
| 137 |
|
| 138 |
// Symbols related flags. |
| 139 |
const FLAG_SYMBOL_VARIABLE = 1; |
| 140 |
const FLAG_SYMBOL_BACKTICK = 2; |
| 141 |
const FLAG_SYMBOL_USER = 4; |
| 142 |
const FLAG_SYMBOL_SYSTEM = 8; |
| 143 |
const FLAG_SYMBOL_PARAMETER = 16; |
| 144 |
|
| 145 |
/** |
| 146 |
* The token it its raw string representation. |
| 147 |
* |
| 148 |
* @var string |
| 149 |
*/ |
| 150 |
public $token; |
| 151 |
|
| 152 |
/** |
| 153 |
* The value this token contains (i.e. token after some evaluation). |
| 154 |
* |
| 155 |
* @var mixed |
| 156 |
*/ |
| 157 |
public $value; |
| 158 |
|
| 159 |
/** |
| 160 |
* The keyword value this token contains, always uppercase. |
| 161 |
* |
| 162 |
* @var mixed|string|null |
| 163 |
*/ |
| 164 |
public $keyword = null; |
| 165 |
|
| 166 |
/** |
| 167 |
* The type of this token. |
| 168 |
* |
| 169 |
* @var int |
| 170 |
*/ |
| 171 |
public $type; |
| 172 |
|
| 173 |
/** |
| 174 |
* The flags of this token. |
| 175 |
* |
| 176 |
* @var int |
| 177 |
*/ |
| 178 |
public $flags; |
| 179 |
|
| 180 |
/** |
| 181 |
* The position in the initial string where this token started. |
| 182 |
* |
| 183 |
* The position is counted in chars, not bytes, so you should |
| 184 |
* use mb_* functions to properly handle utf-8 multibyte chars. |
| 185 |
* |
| 186 |
* @var int|null |
| 187 |
*/ |
| 188 |
public $position; |
| 189 |
|
| 190 |
/** |
| 191 |
* Constructor. |
| 192 |
* |
| 193 |
* @param string $token The value of the token. |
| 194 |
* @param int $type The type of the token. |
| 195 |
* @param int $flags The flags of the token. |
| 196 |
*/ |
| 197 |
public function __construct( $token, $type = 0, $flags = 0 ) { |
| 198 |
$this->token = $token; |
| 199 |
$this->type = $type; |
| 200 |
$this->flags = $flags; |
| 201 |
$this->value = $this->extract(); |
| 202 |
} |
| 203 |
|
| 204 |
/** |
| 205 |
* Check if the token matches the given parameters. |
| 206 |
* |
| 207 |
* @param int|null $type The type of the token. |
| 208 |
* @param int|null $flags The flags of the token. |
| 209 |
* @param array|null $values The values of the token. |
| 210 |
* |
| 211 |
* @return bool |
| 212 |
*/ |
| 213 |
public function matches( $type = null, $flags = null, $values = null ) { |
| 214 |
if ( null === $type && null === $flags && ( null === $values || array() === $values ) ) { |
| 215 |
return ! $this->is_semantically_void(); |
| 216 |
} |
| 217 |
|
| 218 |
return ( |
| 219 |
( null === $type || $this->type === $type ) |
| 220 |
&& ( null === $flags || ( $this->flags & $flags ) ) |
| 221 |
&& ( null === $values || in_array( strtoupper( $this->value ?? '' ), $values, true ) ) |
| 222 |
); |
| 223 |
} |
| 224 |
|
| 225 |
/** |
| 226 |
* Check if the token is semantically void (i.e. whitespace or comment). |
| 227 |
* |
| 228 |
* @return bool |
| 229 |
*/ |
| 230 |
public function is_semantically_void() { |
| 231 |
return $this->matches( self::TYPE_WHITESPACE ) || $this->matches( self::TYPE_COMMENT ); |
| 232 |
} |
| 233 |
|
| 234 |
/** |
| 235 |
* Does little processing to the token to extract a value. |
| 236 |
* |
| 237 |
* If no processing can be done it will return the initial string. |
| 238 |
* |
| 239 |
* @return mixed |
| 240 |
*/ |
| 241 |
private function extract() { |
| 242 |
switch ( $this->type ) { |
| 243 |
case self::TYPE_KEYWORD: |
| 244 |
$this->keyword = strtoupper( $this->token ?? '' ); |
| 245 |
if ( ! ( $this->flags & self::FLAG_KEYWORD_RESERVED ) ) { |
| 246 |
/* |
| 247 |
* Unreserved keywords should stay the way they are |
| 248 |
* because they might represent field names. |
| 249 |
*/ |
| 250 |
return $this->token; |
| 251 |
} |
| 252 |
|
| 253 |
return $this->keyword; |
| 254 |
|
| 255 |
case self::TYPE_WHITESPACE: |
| 256 |
return ' '; |
| 257 |
|
| 258 |
case self::TYPE_BOOL: |
| 259 |
return strtoupper( $this->token ?? '' ) === 'TRUE'; |
| 260 |
|
| 261 |
case self::TYPE_NUMBER: |
| 262 |
$ret = str_replace( '--', '', $this->token ); // e.g. ---42 === -42. |
| 263 |
if ( $this->flags & self::FLAG_NUMBER_HEX ) { |
| 264 |
$ret = str_replace( array( '-', '+' ), '', $this->token ); |
| 265 |
if ( $this->flags & self::FLAG_NUMBER_NEGATIVE ) { |
| 266 |
$ret = -hexdec( $ret ); |
| 267 |
} else { |
| 268 |
$ret = hexdec( $ret ); |
| 269 |
} |
| 270 |
} elseif ( ( $this->flags & self::FLAG_NUMBER_APPROXIMATE ) || ( $this->flags & self::FLAG_NUMBER_FLOAT ) ) { |
| 271 |
$ret = (float) $ret; |
| 272 |
} elseif ( ! ( $this->flags & self::FLAG_NUMBER_BINARY ) ) { |
| 273 |
$ret = (int) $ret; |
| 274 |
} |
| 275 |
|
| 276 |
return $ret; |
| 277 |
|
| 278 |
case self::TYPE_STRING: |
| 279 |
// Trims quotes. |
| 280 |
$str = $this->token; |
| 281 |
$str = mb_substr( $str, 1, -1, 'UTF-8' ); |
| 282 |
|
| 283 |
// Removes surrounding quotes. |
| 284 |
$quote = $this->token[0]; |
| 285 |
$str = str_replace( $quote . $quote, $quote, $str ); |
| 286 |
|
| 287 |
/* |
| 288 |
* Finally unescapes the string. |
| 289 |
* |
| 290 |
* `stripcslashes` replaces escape sequences with their |
| 291 |
* representation. |
| 292 |
*/ |
| 293 |
$str = stripcslashes( $str ); |
| 294 |
|
| 295 |
return $str; |
| 296 |
|
| 297 |
case self::TYPE_SYMBOL: |
| 298 |
$str = $this->token; |
| 299 |
if ( isset( $str[0] ) && ( '@' === $str[0] ) ) { |
| 300 |
/* |
| 301 |
* `mb_strlen($str)` must be used instead of `null` because |
| 302 |
* in PHP 5.3- the `null` parameter isn't handled correctly. |
| 303 |
*/ |
| 304 |
$str = mb_substr( |
| 305 |
$str, |
| 306 |
! empty( $str[1] ) && ( '@' === $str[1] ) ? 2 : 1, |
| 307 |
mb_strlen( $str ), |
| 308 |
'UTF-8' |
| 309 |
); |
| 310 |
} |
| 311 |
|
| 312 |
if ( isset( $str[0] ) && ( ':' === $str[0] ) ) { |
| 313 |
$str = mb_substr( $str, 1, mb_strlen( $str ), 'UTF-8' ); |
| 314 |
} |
| 315 |
|
| 316 |
if ( isset( $str[0] ) && ( ( '`' === $str[0] ) || ( '"' === $str[0] ) || ( '\'' === $str[0] ) ) ) { |
| 317 |
$quote = $str[0]; |
| 318 |
$str = str_replace( $quote . $quote, $quote, $str ); |
| 319 |
$str = mb_substr( $str, 1, -1, 'UTF-8' ); |
| 320 |
} |
| 321 |
|
| 322 |
return $str; |
| 323 |
} |
| 324 |
|
| 325 |
return $this->token; |
| 326 |
} |
| 327 |
} |
| 328 |
|
