Code Coverage |
||||||||||
Lines |
Functions and Methods |
Classes and Traits |
||||||||
| Total | |
0.00% |
0 / 92 |
|
0.00% |
0 / 8 |
CRAP | |
0.00% |
0 / 1 |
| Converter | |
0.00% |
0 / 92 |
|
0.00% |
0 / 8 |
1482 | |
0.00% |
0 / 1 |
| convert | |
0.00% |
0 / 2 |
|
0.00% |
0 / 1 |
2 | |||
| __construct | |
0.00% |
0 / 3 |
|
0.00% |
0 / 1 |
2 | |||
| doConvert | |
0.00% |
0 / 28 |
|
0.00% |
0 / 1 |
132 | |||
| nextToken | |
0.00% |
0 / 50 |
|
0.00% |
0 / 1 |
420 | |||
| doOperation | |
0.00% |
0 / 6 |
|
0.00% |
0 / 1 |
6 | |||
| newNumber | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
2 | |||
| newOperator | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
2 | |||
| error | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
2 | |||
| 1 | <?php |
| 2 | /** |
| 3 | * @author Tim Starling |
| 4 | * @author Niklas Laxström |
| 5 | * @license GPL-2.0-or-later |
| 6 | * @file |
| 7 | */ |
| 8 | |
| 9 | namespace CLDRPluralRuleParser; |
| 10 | |
| 11 | use CLDRPluralRuleParser\Converter\Expression; |
| 12 | use CLDRPluralRuleParser\Converter\Fragment; |
| 13 | use CLDRPluralRuleParser\Converter\Operator; |
| 14 | |
| 15 | /** |
| 16 | * Helper class for converting rules to reverse polish notation (RPN). |
| 17 | */ |
| 18 | class Converter { |
| 19 | /** |
| 20 | * The input string |
| 21 | * |
| 22 | * @var string |
| 23 | */ |
| 24 | public $rule; |
| 25 | |
| 26 | /** |
| 27 | * The current position |
| 28 | * |
| 29 | * @var int |
| 30 | */ |
| 31 | public $pos; |
| 32 | |
| 33 | /** |
| 34 | * The past-the-end position |
| 35 | * |
| 36 | * @var int |
| 37 | */ |
| 38 | public $end; |
| 39 | |
| 40 | /** |
| 41 | * The operator stack |
| 42 | * |
| 43 | * @var array |
| 44 | */ |
| 45 | public $operators = []; |
| 46 | |
| 47 | /** |
| 48 | * The operand stack |
| 49 | * |
| 50 | * @var array |
| 51 | */ |
| 52 | public $operands = []; |
| 53 | |
| 54 | /** |
| 55 | * Precedence levels. Note that there's no need to worry about associativity |
| 56 | * for the level 4 operators, since they return boolean and don't accept |
| 57 | * boolean inputs. |
| 58 | * |
| 59 | * @var array |
| 60 | */ |
| 61 | private const PRECEDENCE_LEVELS = [ |
| 62 | 'or' => 2, |
| 63 | 'and' => 3, |
| 64 | 'is' => 4, |
| 65 | 'is-not' => 4, |
| 66 | 'in' => 4, |
| 67 | 'not-in' => 4, |
| 68 | 'within' => 4, |
| 69 | 'not-within' => 4, |
| 70 | 'mod' => 5, |
| 71 | ',' => 6, |
| 72 | '..' => 7, |
| 73 | ]; |
| 74 | |
| 75 | /** |
| 76 | * A character list defining whitespace, for use in strspn() etc. |
| 77 | */ |
| 78 | private const WHITESPACE_CLASS = " \t\r\n"; |
| 79 | |
| 80 | /** |
| 81 | * Same for digits. Note that the grammar given in UTS #35 doesn't allow |
| 82 | * negative numbers or decimal separators. |
| 83 | */ |
| 84 | private const NUMBER_CLASS = '0123456789'; |
| 85 | |
| 86 | /** |
| 87 | * A character list of symbolic operands. |
| 88 | */ |
| 89 | private const OPERAND_SYMBOLS = 'nivwft'; |
| 90 | |
| 91 | /** |
| 92 | * An anchored regular expression which matches a word at the current offset. |
| 93 | */ |
| 94 | private const WORD_REGEX = '/[a-zA-Z@]+/A'; |
| 95 | |
| 96 | /** |
| 97 | * Convert a rule to RPN. This is the only public entry point. |
| 98 | * |
| 99 | * @param string $rule The rule to convert |
| 100 | * @return string The RPN representation of the rule |
| 101 | */ |
| 102 | public static function convert( $rule ): string { |
| 103 | $parser = new self( $rule ); |
| 104 | |
| 105 | return $parser->doConvert(); |
| 106 | } |
| 107 | |
| 108 | /** |
| 109 | * Private constructor. |
| 110 | * @param string $rule |
| 111 | */ |
| 112 | protected function __construct( string $rule ) { |
| 113 | $this->rule = $rule; |
| 114 | $this->pos = 0; |
| 115 | $this->end = strlen( $rule ); |
| 116 | } |
| 117 | |
| 118 | /** |
| 119 | * Do the operation. |
| 120 | * |
| 121 | * @return string The RPN representation of the rule (e.g. "5 3 mod n is") |
| 122 | */ |
| 123 | protected function doConvert(): string { |
| 124 | $expectOperator = true; |
| 125 | |
| 126 | // Iterate through all tokens, saving the operators and operands to a |
| 127 | // stack per Dijkstra's shunting yard algorithm. |
| 128 | /** @var Operator $token */ |
| 129 | $token = $this->nextToken(); |
| 130 | while ( $token !== false ) { |
| 131 | // In this grammar, there are only binary operators, so every valid |
| 132 | // rule string will alternate between operator and operand tokens. |
| 133 | $expectOperator = !$expectOperator; |
| 134 | |
| 135 | if ( $token instanceof Expression ) { |
| 136 | // Operand |
| 137 | if ( $expectOperator ) { |
| 138 | $token->error( 'unexpected operand' ); |
| 139 | } |
| 140 | $this->operands[] = $token; |
| 141 | } else { |
| 142 | // Operator |
| 143 | if ( !$expectOperator ) { |
| 144 | $token->error( 'unexpected operator' ); |
| 145 | } |
| 146 | // Resolve higher precedence levels |
| 147 | /** @var Operator $lastOp */ |
| 148 | $lastOp = end( $this->operators ); |
| 149 | // @phan-suppress-next-next-line PhanUndeclaredProperty |
| 150 | while ( $lastOp && |
| 151 | self::PRECEDENCE_LEVELS[$token->name] <= self::PRECEDENCE_LEVELS[$lastOp->name] |
| 152 | ) { |
| 153 | $this->doOperation( $lastOp ); |
| 154 | array_pop( $this->operators ); |
| 155 | $lastOp = end( $this->operators ); |
| 156 | } |
| 157 | $this->operators[] = $token; |
| 158 | } |
| 159 | |
| 160 | $token = $this->nextToken(); |
| 161 | } |
| 162 | |
| 163 | // Finish off the stack |
| 164 | while ( $this->operators ) { |
| 165 | $this->doOperation( array_pop( $this->operators ) ); |
| 166 | } |
| 167 | |
| 168 | // Make sure the result is sensible. The first case is possible for an empty |
| 169 | // string input, the second should be unreachable. |
| 170 | if ( !count( $this->operands ) ) { |
| 171 | $this->error( 'condition expected' ); |
| 172 | } elseif ( count( $this->operands ) > 1 ) { |
| 173 | $this->error( 'missing operator or too many operands' ); |
| 174 | } |
| 175 | |
| 176 | $value = $this->operands[0]; |
| 177 | if ( $value->type !== 'boolean' ) { |
| 178 | $this->error( 'the result must have a boolean type' ); |
| 179 | } |
| 180 | |
| 181 | return $this->operands[0]->rpn; |
| 182 | } |
| 183 | |
| 184 | /** |
| 185 | * Fetch the next token from the input string. |
| 186 | * |
| 187 | * @return Fragment|false The next token |
| 188 | */ |
| 189 | protected function nextToken() { |
| 190 | if ( $this->pos >= $this->end ) { |
| 191 | return false; |
| 192 | } |
| 193 | |
| 194 | // Whitespace |
| 195 | $length = strspn( $this->rule, self::WHITESPACE_CLASS, $this->pos ); |
| 196 | $this->pos += $length; |
| 197 | |
| 198 | if ( $this->pos >= $this->end ) { |
| 199 | return false; |
| 200 | } |
| 201 | |
| 202 | // Number |
| 203 | $length = strspn( $this->rule, self::NUMBER_CLASS, $this->pos ); |
| 204 | if ( $length !== 0 ) { |
| 205 | $token = $this->newNumber( substr( $this->rule, $this->pos, $length ), $this->pos ); |
| 206 | $this->pos += $length; |
| 207 | |
| 208 | return $token; |
| 209 | } |
| 210 | |
| 211 | // Two-character operators |
| 212 | $op2 = substr( $this->rule, $this->pos, 2 ); |
| 213 | if ( $op2 === '..' || $op2 === '!=' ) { |
| 214 | $token = $this->newOperator( $op2, $this->pos, 2 ); |
| 215 | $this->pos += 2; |
| 216 | |
| 217 | return $token; |
| 218 | } |
| 219 | |
| 220 | // Single-character operators |
| 221 | $op1 = $this->rule[$this->pos]; |
| 222 | if ( $op1 === ',' || $op1 === '=' || $op1 === '%' ) { |
| 223 | $token = $this->newOperator( $op1, $this->pos, 1 ); |
| 224 | $this->pos++; |
| 225 | |
| 226 | return $token; |
| 227 | } |
| 228 | |
| 229 | // Word |
| 230 | if ( !preg_match( self::WORD_REGEX, $this->rule, $m, 0, $this->pos ) ) { |
| 231 | $this->error( 'unexpected character "' . $this->rule[$this->pos] . '"' ); |
| 232 | } |
| 233 | $word1 = strtolower( $m[0] ); |
| 234 | $word2 = ''; |
| 235 | $nextTokenPos = $this->pos + strlen( $word1 ); |
| 236 | if ( $word1 === 'not' || $word1 === 'is' ) { |
| 237 | // Look ahead one word |
| 238 | $nextTokenPos += strspn( $this->rule, self::WHITESPACE_CLASS, $nextTokenPos ); |
| 239 | if ( $nextTokenPos < $this->end |
| 240 | && preg_match( self::WORD_REGEX, $this->rule, $m, 0, $nextTokenPos ) |
| 241 | ) { |
| 242 | $word2 = strtolower( $m[0] ); |
| 243 | $nextTokenPos += strlen( $word2 ); |
| 244 | } |
| 245 | } |
| 246 | |
| 247 | // Two-word operators like "is not" take precedence over single-word operators like "is" |
| 248 | if ( $word2 !== '' ) { |
| 249 | $bothWords = "{$word1}-{$word2}"; |
| 250 | if ( isset( self::PRECEDENCE_LEVELS[$bothWords] ) ) { |
| 251 | $token = $this->newOperator( $bothWords, $this->pos, $nextTokenPos - $this->pos ); |
| 252 | $this->pos = $nextTokenPos; |
| 253 | |
| 254 | return $token; |
| 255 | } |
| 256 | } |
| 257 | |
| 258 | // Single-word operators |
| 259 | if ( isset( self::PRECEDENCE_LEVELS[$word1] ) ) { |
| 260 | $token = $this->newOperator( $word1, $this->pos, strlen( $word1 ) ); |
| 261 | $this->pos += strlen( $word1 ); |
| 262 | |
| 263 | return $token; |
| 264 | } |
| 265 | |
| 266 | // The single-character operand symbols |
| 267 | // @phan-suppress-next-line PhanParamSuspiciousOrder |
| 268 | if ( strpos( self::OPERAND_SYMBOLS, $word1 ) !== false ) { |
| 269 | $token = $this->newNumber( $word1, $this->pos ); |
| 270 | $this->pos++; |
| 271 | |
| 272 | return $token; |
| 273 | } |
| 274 | |
| 275 | // Samples |
| 276 | if ( $word1 === '@integer' || $word1 === '@decimal' ) { |
| 277 | // Samples are like comments, they have no effect on rule evaluation. |
| 278 | // They run from the first sample indicator to the end of the string. |
| 279 | $this->pos = $this->end; |
| 280 | |
| 281 | return false; |
| 282 | } |
| 283 | |
| 284 | $this->error( 'unrecognised word' ); |
| 285 | } |
| 286 | |
| 287 | /** |
| 288 | * For the binary operator $op, pop its operands off the stack and push |
| 289 | * a fragment with rpn and type members describing the result of that |
| 290 | * operation. |
| 291 | * |
| 292 | * @param Operator $op |
| 293 | */ |
| 294 | protected function doOperation( Operator $op ) { |
| 295 | if ( count( $this->operands ) < 2 ) { |
| 296 | $op->error( 'missing operand' ); |
| 297 | } |
| 298 | $right = array_pop( $this->operands ); |
| 299 | $left = array_pop( $this->operands ); |
| 300 | $result = $op->operate( $left, $right ); |
| 301 | $this->operands[] = $result; |
| 302 | } |
| 303 | |
| 304 | /** |
| 305 | * Create a numerical expression object |
| 306 | * |
| 307 | * @param string $text |
| 308 | * @param int $pos |
| 309 | * @return Expression The numerical expression |
| 310 | */ |
| 311 | protected function newNumber( string $text, int $pos ): Expression { |
| 312 | return new Expression( $this, 'number', $text, $pos, strlen( $text ) ); |
| 313 | } |
| 314 | |
| 315 | /** |
| 316 | * Create a binary operator |
| 317 | * |
| 318 | * @param string $type |
| 319 | * @param int $pos |
| 320 | * @param int $length |
| 321 | * @return Operator The operator |
| 322 | */ |
| 323 | protected function newOperator( string $type, int $pos, int $length ): Operator { |
| 324 | return new Operator( $this, $type, $pos, $length ); |
| 325 | } |
| 326 | |
| 327 | /** |
| 328 | * Throw an error |
| 329 | * @param string $message |
| 330 | * @throws Error |
| 331 | * @return never |
| 332 | */ |
| 333 | protected function error( string $message ) { |
| 334 | throw new Error( $message ); |
| 335 | } |
| 336 | } |