3f46453d7b201908951fa8c043d290b5f4b161fc
6 * @author Paul Copperman <paul.copperman@gmail.com>
7 * @license Choose any of Apache, MIT, GPL, LGPL
11 * This class is meant to safely minify javascript code, while leaving syntactically correct
12 * programs intact. Other libraries, such as JSMin require a certain coding style to work
13 * correctly. OTOH, libraries like jsminplus, that do parse the code correctly are rather
14 * slow, because they construct a complete parse tree before outputting the code minified.
15 * So this class is meant to allow arbitrary (but syntactically correct) input, while being
16 * fast enough to be used for on-the-fly minifying.
18 class JavaScriptMinifier
{
21 * The state machine is only necessary to decide whether to parse a slash as division
22 * operator or as regexp literal.
23 * States are named after the next expected item. We only distinguish states when the
24 * distinction is relevant for our purpose.
28 const PROPERTY_ASSIGNMENT
= 2;
30 const EXPRESSION_NO_NL
= 4; // only relevant for semicolon insertion
31 const EXPRESSION_OP
= 5;
32 const EXPRESSION_FUNC
= 6;
33 const EXPRESSION_TERNARY
= 7; // used to determine the role of a colon
34 const EXPRESSION_TERNARY_OP
= 8;
35 const EXPRESSION_TERNARY_FUNC
= 9;
36 const PAREN_EXPRESSION
= 10; // expression which is not on the top level
37 const PAREN_EXPRESSION_OP
= 11;
38 const PAREN_EXPRESSION_FUNC
= 12;
39 const PROPERTY_EXPRESSION
= 13; // expression which is within an object literal
40 const PROPERTY_EXPRESSION_OP
= 14;
41 const PROPERTY_EXPRESSION_FUNC
= 15;
44 const TYPE_UN_OP
= 1; // unary operators
45 const TYPE_INCR_OP
= 2; // ++ and --
46 const TYPE_BIN_OP
= 3; // binary operators
47 const TYPE_ADD_OP
= 4; // + and - which can be either unary or binary ops
48 const TYPE_HOOK
= 5; // ?
49 const TYPE_COLON
= 6; // :
50 const TYPE_COMMA
= 7; // ,
51 const TYPE_SEMICOLON
= 8; // ;
52 const TYPE_BRACE_OPEN
= 9; // {
53 const TYPE_BRACE_CLOSE
= 10; // }
54 const TYPE_PAREN_OPEN
= 11; // ( and [
55 const TYPE_PAREN_CLOSE
= 12; // ) and ]
56 const TYPE_RETURN
= 13; // keywords: break, continue, return, throw
57 const TYPE_IF
= 14; // keywords: catch, for, with, switch, while, if
58 const TYPE_DO
= 15; // keywords: case, var, finally, else, do, try
59 const TYPE_FUNC
= 16; // keywords: function
60 const TYPE_LITERAL
= 17; // all literals, identifiers and unrecognised tokens
62 // Sanity limit to avoid excessive memory usage
63 const STACK_LIMIT
= 1000;
66 * Returns minified JavaScript code.
68 * NOTE: $maxLineLength isn't a strict maximum. Longer lines will be produced when
69 * literals (e.g. quoted strings) longer than $maxLineLength are encountered
70 * or when required to guard against semicolon insertion.
72 * @param string $s JavaScript code to minify
73 * @param int $maxLineLength Maximum length of a single line, or -1 for no maximum.
74 * @return String Minified code
76 public static function minify( $s, $maxLineLength = 1000 ) {
77 // First we declare a few tables that contain our parsing rules
79 // $opChars : characters, which can be combined without whitespace in between them
109 // $tokenTypes : maps keywords and operators to their corresponding token type
111 '!' => self
::TYPE_UN_OP
,
112 '~' => self
::TYPE_UN_OP
,
113 'delete' => self
::TYPE_UN_OP
,
114 'new' => self
::TYPE_UN_OP
,
115 'typeof' => self
::TYPE_UN_OP
,
116 'void' => self
::TYPE_UN_OP
,
117 '++' => self
::TYPE_INCR_OP
,
118 '--' => self
::TYPE_INCR_OP
,
119 '!=' => self
::TYPE_BIN_OP
,
120 '!==' => self
::TYPE_BIN_OP
,
121 '%' => self
::TYPE_BIN_OP
,
122 '%=' => self
::TYPE_BIN_OP
,
123 '&' => self
::TYPE_BIN_OP
,
124 '&&' => self
::TYPE_BIN_OP
,
125 '&=' => self
::TYPE_BIN_OP
,
126 '*' => self
::TYPE_BIN_OP
,
127 '*=' => self
::TYPE_BIN_OP
,
128 '+=' => self
::TYPE_BIN_OP
,
129 '-=' => self
::TYPE_BIN_OP
,
130 '.' => self
::TYPE_BIN_OP
,
131 '/' => self
::TYPE_BIN_OP
,
132 '/=' => self
::TYPE_BIN_OP
,
133 '<' => self
::TYPE_BIN_OP
,
134 '<<' => self
::TYPE_BIN_OP
,
135 '<<=' => self
::TYPE_BIN_OP
,
136 '<=' => self
::TYPE_BIN_OP
,
137 '=' => self
::TYPE_BIN_OP
,
138 '==' => self
::TYPE_BIN_OP
,
139 '===' => self
::TYPE_BIN_OP
,
140 '>' => self
::TYPE_BIN_OP
,
141 '>=' => self
::TYPE_BIN_OP
,
142 '>>' => self
::TYPE_BIN_OP
,
143 '>>=' => self
::TYPE_BIN_OP
,
144 '>>>' => self
::TYPE_BIN_OP
,
145 '>>>=' => self
::TYPE_BIN_OP
,
146 '^' => self
::TYPE_BIN_OP
,
147 '^=' => self
::TYPE_BIN_OP
,
148 '|' => self
::TYPE_BIN_OP
,
149 '|=' => self
::TYPE_BIN_OP
,
150 '||' => self
::TYPE_BIN_OP
,
151 'in' => self
::TYPE_BIN_OP
,
152 'instanceof' => self
::TYPE_BIN_OP
,
153 '+' => self
::TYPE_ADD_OP
,
154 '-' => self
::TYPE_ADD_OP
,
155 '?' => self
::TYPE_HOOK
,
156 ':' => self
::TYPE_COLON
,
157 ',' => self
::TYPE_COMMA
,
158 ';' => self
::TYPE_SEMICOLON
,
159 '{' => self
::TYPE_BRACE_OPEN
,
160 '}' => self
::TYPE_BRACE_CLOSE
,
161 '(' => self
::TYPE_PAREN_OPEN
,
162 '[' => self
::TYPE_PAREN_OPEN
,
163 ')' => self
::TYPE_PAREN_CLOSE
,
164 ']' => self
::TYPE_PAREN_CLOSE
,
165 'break' => self
::TYPE_RETURN
,
166 'continue' => self
::TYPE_RETURN
,
167 'return' => self
::TYPE_RETURN
,
168 'throw' => self
::TYPE_RETURN
,
169 'catch' => self
::TYPE_IF
,
170 'for' => self
::TYPE_IF
,
171 'if' => self
::TYPE_IF
,
172 'switch' => self
::TYPE_IF
,
173 'while' => self
::TYPE_IF
,
174 'with' => self
::TYPE_IF
,
175 'case' => self
::TYPE_DO
,
176 'do' => self
::TYPE_DO
,
177 'else' => self
::TYPE_DO
,
178 'finally' => self
::TYPE_DO
,
179 'try' => self
::TYPE_DO
,
180 'var' => self
::TYPE_DO
,
181 'function' => self
::TYPE_FUNC
184 // $goto : This is the main table for our state machine. For every state/token pair
185 // the following state is defined. When no rule exists for a given pair,
186 // the state is left unchanged.
189 self
::TYPE_UN_OP
=> self
::EXPRESSION
,
190 self
::TYPE_INCR_OP
=> self
::EXPRESSION
,
191 self
::TYPE_ADD_OP
=> self
::EXPRESSION
,
192 self
::TYPE_PAREN_OPEN
=> self
::PAREN_EXPRESSION
,
193 self
::TYPE_RETURN
=> self
::EXPRESSION_NO_NL
,
194 self
::TYPE_IF
=> self
::CONDITION
,
195 self
::TYPE_FUNC
=> self
::CONDITION
,
196 self
::TYPE_LITERAL
=> self
::EXPRESSION_OP
199 self
::TYPE_PAREN_OPEN
=> self
::PAREN_EXPRESSION
201 self
::PROPERTY_ASSIGNMENT
=> [
202 self
::TYPE_COLON
=> self
::PROPERTY_EXPRESSION
,
203 self
::TYPE_BRACE_OPEN
=> self
::STATEMENT
205 self
::EXPRESSION
=> [
206 self
::TYPE_SEMICOLON
=> self
::STATEMENT
,
207 self
::TYPE_BRACE_OPEN
=> self
::PROPERTY_ASSIGNMENT
,
208 self
::TYPE_PAREN_OPEN
=> self
::PAREN_EXPRESSION
,
209 self
::TYPE_FUNC
=> self
::EXPRESSION_FUNC
,
210 self
::TYPE_LITERAL
=> self
::EXPRESSION_OP
212 self
::EXPRESSION_NO_NL
=> [
213 self
::TYPE_SEMICOLON
=> self
::STATEMENT
,
214 self
::TYPE_BRACE_OPEN
=> self
::PROPERTY_ASSIGNMENT
,
215 self
::TYPE_PAREN_OPEN
=> self
::PAREN_EXPRESSION
,
216 self
::TYPE_FUNC
=> self
::EXPRESSION_FUNC
,
217 self
::TYPE_LITERAL
=> self
::EXPRESSION_OP
219 self
::EXPRESSION_OP
=> [
220 self
::TYPE_BIN_OP
=> self
::EXPRESSION
,
221 self
::TYPE_ADD_OP
=> self
::EXPRESSION
,
222 self
::TYPE_HOOK
=> self
::EXPRESSION_TERNARY
,
223 self
::TYPE_COLON
=> self
::STATEMENT
,
224 self
::TYPE_COMMA
=> self
::EXPRESSION
,
225 self
::TYPE_SEMICOLON
=> self
::STATEMENT
,
226 self
::TYPE_PAREN_OPEN
=> self
::PAREN_EXPRESSION
228 self
::EXPRESSION_FUNC
=> [
229 self
::TYPE_BRACE_OPEN
=> self
::STATEMENT
231 self
::EXPRESSION_TERNARY
=> [
232 self
::TYPE_BRACE_OPEN
=> self
::PROPERTY_ASSIGNMENT
,
233 self
::TYPE_PAREN_OPEN
=> self
::PAREN_EXPRESSION
,
234 self
::TYPE_FUNC
=> self
::EXPRESSION_TERNARY_FUNC
,
235 self
::TYPE_LITERAL
=> self
::EXPRESSION_TERNARY_OP
237 self
::EXPRESSION_TERNARY_OP
=> [
238 self
::TYPE_BIN_OP
=> self
::EXPRESSION_TERNARY
,
239 self
::TYPE_ADD_OP
=> self
::EXPRESSION_TERNARY
,
240 self
::TYPE_HOOK
=> self
::EXPRESSION_TERNARY
,
241 self
::TYPE_COMMA
=> self
::EXPRESSION_TERNARY
,
242 self
::TYPE_PAREN_OPEN
=> self
::PAREN_EXPRESSION
244 self
::EXPRESSION_TERNARY_FUNC
=> [
245 self
::TYPE_BRACE_OPEN
=> self
::STATEMENT
247 self
::PAREN_EXPRESSION
=> [
248 self
::TYPE_BRACE_OPEN
=> self
::PROPERTY_ASSIGNMENT
,
249 self
::TYPE_PAREN_OPEN
=> self
::PAREN_EXPRESSION
,
250 self
::TYPE_FUNC
=> self
::PAREN_EXPRESSION_FUNC
,
251 self
::TYPE_LITERAL
=> self
::PAREN_EXPRESSION_OP
253 self
::PAREN_EXPRESSION_OP
=> [
254 self
::TYPE_BIN_OP
=> self
::PAREN_EXPRESSION
,
255 self
::TYPE_ADD_OP
=> self
::PAREN_EXPRESSION
,
256 self
::TYPE_HOOK
=> self
::PAREN_EXPRESSION
,
257 self
::TYPE_COLON
=> self
::PAREN_EXPRESSION
,
258 self
::TYPE_COMMA
=> self
::PAREN_EXPRESSION
,
259 self
::TYPE_SEMICOLON
=> self
::PAREN_EXPRESSION
,
260 self
::TYPE_PAREN_OPEN
=> self
::PAREN_EXPRESSION
262 self
::PAREN_EXPRESSION_FUNC
=> [
263 self
::TYPE_BRACE_OPEN
=> self
::STATEMENT
265 self
::PROPERTY_EXPRESSION
=> [
266 self
::TYPE_BRACE_OPEN
=> self
::PROPERTY_ASSIGNMENT
,
267 self
::TYPE_PAREN_OPEN
=> self
::PAREN_EXPRESSION
,
268 self
::TYPE_FUNC
=> self
::PROPERTY_EXPRESSION_FUNC
,
269 self
::TYPE_LITERAL
=> self
::PROPERTY_EXPRESSION_OP
271 self
::PROPERTY_EXPRESSION_OP
=> [
272 self
::TYPE_BIN_OP
=> self
::PROPERTY_EXPRESSION
,
273 self
::TYPE_ADD_OP
=> self
::PROPERTY_EXPRESSION
,
274 self
::TYPE_HOOK
=> self
::PROPERTY_EXPRESSION
,
275 self
::TYPE_COMMA
=> self
::PROPERTY_ASSIGNMENT
,
276 self
::TYPE_PAREN_OPEN
=> self
::PAREN_EXPRESSION
278 self
::PROPERTY_EXPRESSION_FUNC
=> [
279 self
::TYPE_BRACE_OPEN
=> self
::STATEMENT
283 // $push : This table contains the rules for when to push a state onto the stack.
284 // The pushed state is the state to return to when the corresponding
285 // closing token is found
288 self
::TYPE_BRACE_OPEN
=> self
::STATEMENT
,
289 self
::TYPE_PAREN_OPEN
=> self
::EXPRESSION_OP
292 self
::TYPE_PAREN_OPEN
=> self
::STATEMENT
294 self
::PROPERTY_ASSIGNMENT
=> [
295 self
::TYPE_BRACE_OPEN
=> self
::PROPERTY_ASSIGNMENT
297 self
::EXPRESSION
=> [
298 self
::TYPE_BRACE_OPEN
=> self
::EXPRESSION_OP
,
299 self
::TYPE_PAREN_OPEN
=> self
::EXPRESSION_OP
301 self
::EXPRESSION_NO_NL
=> [
302 self
::TYPE_BRACE_OPEN
=> self
::EXPRESSION_OP
,
303 self
::TYPE_PAREN_OPEN
=> self
::EXPRESSION_OP
305 self
::EXPRESSION_OP
=> [
306 self
::TYPE_HOOK
=> self
::EXPRESSION
,
307 self
::TYPE_PAREN_OPEN
=> self
::EXPRESSION_OP
309 self
::EXPRESSION_FUNC
=> [
310 self
::TYPE_BRACE_OPEN
=> self
::EXPRESSION_OP
312 self
::EXPRESSION_TERNARY
=> [
313 self
::TYPE_BRACE_OPEN
=> self
::EXPRESSION_TERNARY_OP
,
314 self
::TYPE_PAREN_OPEN
=> self
::EXPRESSION_TERNARY_OP
316 self
::EXPRESSION_TERNARY_OP
=> [
317 self
::TYPE_HOOK
=> self
::EXPRESSION_TERNARY
,
318 self
::TYPE_PAREN_OPEN
=> self
::EXPRESSION_TERNARY_OP
320 self
::EXPRESSION_TERNARY_FUNC
=> [
321 self
::TYPE_BRACE_OPEN
=> self
::EXPRESSION_TERNARY_OP
323 self
::PAREN_EXPRESSION
=> [
324 self
::TYPE_BRACE_OPEN
=> self
::PAREN_EXPRESSION_OP
,
325 self
::TYPE_PAREN_OPEN
=> self
::PAREN_EXPRESSION_OP
327 self
::PAREN_EXPRESSION_OP
=> [
328 self
::TYPE_PAREN_OPEN
=> self
::PAREN_EXPRESSION_OP
330 self
::PAREN_EXPRESSION_FUNC
=> [
331 self
::TYPE_BRACE_OPEN
=> self
::PAREN_EXPRESSION_OP
333 self
::PROPERTY_EXPRESSION
=> [
334 self
::TYPE_BRACE_OPEN
=> self
::PROPERTY_EXPRESSION_OP
,
335 self
::TYPE_PAREN_OPEN
=> self
::PROPERTY_EXPRESSION_OP
337 self
::PROPERTY_EXPRESSION_OP
=> [
338 self
::TYPE_PAREN_OPEN
=> self
::PROPERTY_EXPRESSION_OP
340 self
::PROPERTY_EXPRESSION_FUNC
=> [
341 self
::TYPE_BRACE_OPEN
=> self
::PROPERTY_EXPRESSION_OP
345 // $pop : Rules for when to pop a state from the stack
347 self
::STATEMENT
=> [ self
::TYPE_BRACE_CLOSE
=> true ],
348 self
::PROPERTY_ASSIGNMENT
=> [ self
::TYPE_BRACE_CLOSE
=> true ],
349 self
::EXPRESSION
=> [ self
::TYPE_BRACE_CLOSE
=> true ],
350 self
::EXPRESSION_NO_NL
=> [ self
::TYPE_BRACE_CLOSE
=> true ],
351 self
::EXPRESSION_OP
=> [ self
::TYPE_BRACE_CLOSE
=> true ],
352 self
::EXPRESSION_TERNARY_OP
=> [ self
::TYPE_COLON
=> true ],
353 self
::PAREN_EXPRESSION
=> [ self
::TYPE_PAREN_CLOSE
=> true ],
354 self
::PAREN_EXPRESSION_OP
=> [ self
::TYPE_PAREN_CLOSE
=> true ],
355 self
::PROPERTY_EXPRESSION
=> [ self
::TYPE_BRACE_CLOSE
=> true ],
356 self
::PROPERTY_EXPRESSION_OP
=> [ self
::TYPE_BRACE_CLOSE
=> true ]
359 // $semicolon : Rules for when a semicolon insertion is appropriate
361 self
::EXPRESSION_NO_NL
=> [
362 self
::TYPE_UN_OP
=> true,
363 self
::TYPE_INCR_OP
=> true,
364 self
::TYPE_ADD_OP
=> true,
365 self
::TYPE_BRACE_OPEN
=> true,
366 self
::TYPE_PAREN_OPEN
=> true,
367 self
::TYPE_RETURN
=> true,
368 self
::TYPE_IF
=> true,
369 self
::TYPE_DO
=> true,
370 self
::TYPE_FUNC
=> true,
371 self
::TYPE_LITERAL
=> true
373 self
::EXPRESSION_OP
=> [
374 self
::TYPE_UN_OP
=> true,
375 self
::TYPE_INCR_OP
=> true,
376 self
::TYPE_BRACE_OPEN
=> true,
377 self
::TYPE_RETURN
=> true,
378 self
::TYPE_IF
=> true,
379 self
::TYPE_DO
=> true,
380 self
::TYPE_FUNC
=> true,
381 self
::TYPE_LITERAL
=> true
385 // $divStates : Contains all states that can be followed by a division operator
387 self
::EXPRESSION_OP
=> true,
388 self
::EXPRESSION_TERNARY_OP
=> true,
389 self
::PAREN_EXPRESSION_OP
=> true,
390 self
::PROPERTY_EXPRESSION_OP
=> true
393 // Here's where the minifying takes place: Loop through the input, looking for tokens
394 // and output them to $out, taking actions to the above defined rules when appropriate.
397 $length = strlen( $s );
399 $newlineFound = true;
400 $state = self
::STATEMENT
;
402 $last = ';'; // Pretend that we have seen a semicolon yet
403 while ( $pos < $length ) {
404 // First, skip over any whitespace and multiline comments, recording whether we
405 // found any newline character
406 $skip = strspn( $s, " \t\n\r\xb\xc", $pos );
409 if ( $ch === '/' && substr( $s, $pos, 2 ) === '/*' ) {
410 // Multiline comment. Search for the end token or EOT.
411 $end = strpos( $s, '*/', $pos +
2 );
412 $skip = $end === false ?
$length - $pos : $end - $pos +
2;
416 // The semicolon insertion mechanism needs to know whether there was a newline
417 // between two tokens, so record it now.
418 if ( !$newlineFound && strcspn( $s, "\r\n", $pos, $skip ) !== $skip ) {
419 $newlineFound = true;
424 // Handle C++-style comments and html comments, which are treated as single line
425 // comments by the browser, regardless of whether the end tag is on the same line.
426 // Handle --> the same way, but only if it's at the beginning of the line
427 if ( ( $ch === '/' && substr( $s, $pos, 2 ) === '//' )
428 ||
( $ch === '<' && substr( $s, $pos, 4 ) === '<!--' )
429 ||
( $ch === '-' && $newlineFound && substr( $s, $pos, 3 ) === '-->' )
431 $pos +
= strcspn( $s, "\r\n", $pos );
435 // Find out which kind of token we're handling.
436 // Note: $end must point past the end of the current token
437 // so that `substr($s, $pos, $end - $pos)` would be the entire token.
438 // In order words, $end will be the offset of the last relevant character
439 // in the stream + 1, or simply put: The offset of the first character
440 // of any next token in the stream.
442 // Handle string literals
443 if ( $ch === "'" ||
$ch === '"' ) {
444 // Search to the end of the string literal, skipping over backslash escapes
445 $search = $ch . '\\';
447 // Speculatively add 2 to the end so that if we see a backslash,
448 // the next iteration will start 2 characters further (one for the
449 // backslash, one for the escaped character).
450 // We'll correct this outside the loop.
451 $end +
= strcspn( $s, $search, $end ) +
2;
452 // If the last character in our search for a quote or a backlash
453 // matched a backslash and we haven't reached the end, keep searching..
454 } while ( $end - 2 < $length && $s[$end - 2] === '\\' );
455 // Correction (1): Undo speculative add, keep only one (end of string literal)
457 if ( $end > $length ) {
458 // Correction (2): Loop wrongly assumed an end quote ended the search,
459 // but search ended because we've reached the end. Correct $end.
460 // TODO: This is invalid and should throw.
463 // We have to distinguish between regexp literals and division operators
464 // A division operator is only possible in certain states
465 } elseif ( $ch === '/' && !isset( $divStates[$state] ) ) {
468 // Search until we find "/" (end of regexp), "\" (backslash escapes),
469 // or "[" (start of character classes).
471 // Speculatively add 2 to ensure next iteration skips
472 // over backslash and escaped character.
473 // We'll correct this outside the loop.
474 $end +
= strcspn( $s, '/[\\', $end ) +
2;
475 // If backslash escape, keep searching...
476 } while ( $end - 2 < $length && $s[$end - 2] === '\\' );
477 // Correction (1): Undo speculative add, keep only one (end of regexp)
479 if ( $end > $length ) {
480 // Correction (2): Loop wrongly assumed end slash was seen
481 // String ended without end of regexp. Correct $end.
482 // TODO: This is invalid and should throw.
486 if ( $s[$end - 1] === '/' ) {
489 // (Implicit else), we must've found the start of a char class,
490 // skip until we find "]" (end of char class), or "\" (backslash escape)
492 // Speculatively add 2 for backslash escape.
493 // We'll substract one outside the loop.
494 $end +
= strcspn( $s, ']\\', $end ) +
2;
495 // If backslash escape, keep searching...
496 } while ( $end - 2 < $length && $s[$end - 2] === '\\' );
497 // Correction (1): Undo speculative add, keep only one (end of regexp)
500 // Search past the regexp modifiers (gi)
501 while ( $end < $length && ctype_alpha( $s[$end] ) ) {
506 && ( $pos +
1 < $length ) && ( $s[$pos +
1] === 'x' ||
$s[$pos +
1] === 'X' )
508 // Hex numeric literal
510 $len = strspn( $s, '0123456789ABCDEFabcdef', $end );
512 return self
::parseError(
515 'Expected a hexadecimal number but found ' . substr( $s, $pos, 5 ) . '...'
521 ||
( $ch === '.' && $pos +
1 < $length && ctype_digit( $s[$pos +
1] ) )
523 $end +
= strspn( $s, '0123456789', $end );
524 $decimal = strspn( $s, '.', $end );
526 if ( $decimal > 2 ) {
527 return self
::parseError( $s, $end, 'The number has too many decimal points' );
529 $end +
= strspn( $s, '0123456789', $end +
1 ) +
$decimal;
531 $exponent = strspn( $s, 'eE', $end );
533 if ( $exponent > 1 ) {
534 return self
::parseError( $s, $end, 'Number with several E' );
538 // + sign is optional; - sign is required.
539 $end +
= strspn( $s, '-+', $end );
540 $len = strspn( $s, '0123456789', $end );
542 return self
::parseError(
545 'No decimal digits after e, how many zeroes should be added?'
550 } elseif ( isset( $opChars[$ch] ) ) {
551 // Punctuation character. Search for the longest matching operator.
554 && isset( $tokenTypes[substr( $s, $pos, $end - $pos +
1 )] )
559 // Identifier or reserved word. Search for the end by excluding whitespace and
561 $end +
= strcspn( $s, " \t\n.;,=<>+-{}()[]?:*/%'\"!&|^~\xb\xc\r", $end );
564 // Now get the token type from our type array
565 $token = substr( $s, $pos, $end - $pos ); // so $end - $pos == strlen( $token )
566 $type = isset( $tokenTypes[$token] ) ?
$tokenTypes[$token] : self
::TYPE_LITERAL
;
568 if ( $newlineFound && isset( $semicolon[$state][$type] ) ) {
569 // This token triggers the semicolon insertion mechanism of javascript. While we
570 // could add the ; token here ourselves, keeping the newline has a few advantages.
572 $state = self
::STATEMENT
;
574 } elseif ( $maxLineLength > 0 && $lineLength +
$end - $pos > $maxLineLength &&
575 !isset( $semicolon[$state][$type] ) && $type !== self
::TYPE_INCR_OP
) {
576 // This line would get too long if we added $token, so add a newline first.
577 // Only do this if it won't trigger semicolon insertion and if it won't
578 // put a postfix increment operator on its own line, which is illegal in js.
581 // Check, whether we have to separate the token from the last one with whitespace
582 } elseif ( !isset( $opChars[$last] ) && !isset( $opChars[$ch] ) ) {
585 // Don't accidentally create ++, -- or // tokens
586 } elseif ( $last === $ch && ( $ch === '+' ||
$ch === '-' ||
$ch === '/' ) ) {
591 $type === self
::TYPE_LITERAL
592 && ( $token === 'true' ||
$token === 'false' )
593 && ( $state === self
::EXPRESSION ||
$state === self
::PROPERTY_EXPRESSION
)
596 $token = ( $token === 'true' ) ?
'!0' : '!1';
600 $lineLength +
= $end - $pos; // += strlen( $token )
601 $last = $s[$end - 1];
603 $newlineFound = false;
605 // Now that we have output our token, transition into the new state.
606 if ( isset( $push[$state][$type] ) && count( $stack ) < self
::STACK_LIMIT
) {
607 $stack[] = $push[$state][$type];
609 if ( $stack && isset( $pop[$state][$type] ) ) {
610 $state = array_pop( $stack );
611 } elseif ( isset( $goto[$state][$type] ) ) {
612 $state = $goto[$state][$type];
618 static function parseError( $fullJavascript, $position, $errorMsg ) {
619 // TODO: Handle the error: trigger_error, throw exception, return false...