diff options
author | Li Jin <dragon-fly@qq.com> | 2020-01-10 16:30:34 +0800 |
---|---|---|
committer | Li Jin <dragon-fly@qq.com> | 2020-01-10 16:30:34 +0800 |
commit | 52a6536103f46c26a3ba9b149b0fe7b40d524d8c (patch) | |
tree | 67e4759f8e1ea922079d0e162d84ecba5e558261 /MoonParser/parser.cpp | |
parent | 975167856ed0b11c2ede03c6eb750ca4e4a6a7fc (diff) | |
download | yuescript-52a6536103f46c26a3ba9b149b0fe7b40d524d8c.tar.gz yuescript-52a6536103f46c26a3ba9b149b0fe7b40d524d8c.tar.bz2 yuescript-52a6536103f46c26a3ba9b149b0fe7b40d524d8c.zip |
update.
Diffstat (limited to 'MoonParser/parser.cpp')
-rw-r--r-- | MoonParser/parser.cpp | 1430 |
1 files changed, 0 insertions, 1430 deletions
diff --git a/MoonParser/parser.cpp b/MoonParser/parser.cpp deleted file mode 100644 index 03857c6..0000000 --- a/MoonParser/parser.cpp +++ /dev/null | |||
@@ -1,1430 +0,0 @@ | |||
1 | #include <cstdlib> | ||
2 | #include <cstring> | ||
3 | #include <cassert> | ||
4 | #include <stdexcept> | ||
5 | #include <unordered_map> | ||
6 | #include <unordered_set> | ||
7 | |||
8 | #include "parser.hpp" | ||
9 | |||
10 | |||
11 | namespace parserlib { | ||
12 | |||
13 | |||
14 | //internal map from rules to parse procs | ||
15 | typedef std::unordered_map<rule *, parse_proc> _parse_proc_map_t; | ||
16 | |||
17 | //on exit, it deletes the parse proc map | ||
18 | static _parse_proc_map_t& _get_parse_proc_map() { | ||
19 | static _parse_proc_map_t _parse_proc_map; | ||
20 | return _parse_proc_map; | ||
21 | } | ||
22 | |||
23 | |||
24 | //get the parse proc from the map | ||
25 | static parse_proc _get_parse_proc(rule *r) { | ||
26 | _parse_proc_map_t& _parse_proc_map = _get_parse_proc_map(); | ||
27 | _parse_proc_map_t::iterator it = _parse_proc_map.find(r); | ||
28 | if (it == _parse_proc_map.end()) return 0; | ||
29 | return it->second; | ||
30 | } | ||
31 | |||
32 | |||
33 | //internal private class that manages access to the public classes' internals. | ||
34 | class _private { | ||
35 | public: | ||
36 | //get the internal expression object from the expression. | ||
37 | static _expr *get_expr(const expr &e) { | ||
38 | return e.m_expr; | ||
39 | } | ||
40 | |||
41 | //create new expression from given expression | ||
42 | static expr construct_expr(_expr *e) { | ||
43 | return e; | ||
44 | } | ||
45 | |||
46 | //get the internal expression object from the rule. | ||
47 | static _expr *get_expr(rule &r) { | ||
48 | return r.m_expr; | ||
49 | } | ||
50 | |||
51 | //get the internal parse proc from the rule. | ||
52 | static parse_proc get_parse_proc(rule &r) { | ||
53 | return r.m_parse_proc; | ||
54 | } | ||
55 | }; | ||
56 | |||
57 | |||
58 | class _context; | ||
59 | |||
60 | |||
61 | //parser state | ||
62 | class _state { | ||
63 | public: | ||
64 | //position | ||
65 | pos m_pos; | ||
66 | |||
67 | //size of match vector | ||
68 | size_t m_matches; | ||
69 | |||
70 | //constructor | ||
71 | _state(_context &con); | ||
72 | }; | ||
73 | |||
74 | |||
75 | //match | ||
76 | class _match { | ||
77 | public: | ||
78 | //rule matched | ||
79 | rule *m_rule; | ||
80 | |||
81 | //begin position | ||
82 | pos m_begin; | ||
83 | |||
84 | //end position | ||
85 | pos m_end; | ||
86 | |||
87 | //null constructor | ||
88 | _match() {} | ||
89 | |||
90 | //constructor from parameters | ||
91 | _match(rule *r, const pos &b, const pos &e) : | ||
92 | m_rule(r), | ||
93 | m_begin(b), | ||
94 | m_end(e) | ||
95 | { | ||
96 | } | ||
97 | }; | ||
98 | |||
99 | |||
100 | //match vector | ||
101 | typedef std::vector<_match> _match_vector; | ||
102 | |||
103 | |||
104 | //parsing context | ||
105 | class _context { | ||
106 | public: | ||
107 | //user data | ||
108 | void* m_user_data; | ||
109 | |||
110 | //current position | ||
111 | pos m_pos; | ||
112 | |||
113 | //error position | ||
114 | pos m_error_pos; | ||
115 | |||
116 | //input begin | ||
117 | input::iterator m_begin; | ||
118 | |||
119 | //input end | ||
120 | input::iterator m_end; | ||
121 | |||
122 | //matches | ||
123 | _match_vector m_matches; | ||
124 | |||
125 | //constructor | ||
126 | _context(input &i, void* ud) : | ||
127 | m_user_data(ud), | ||
128 | m_pos(i), | ||
129 | m_error_pos(i), | ||
130 | m_begin(i.begin()), | ||
131 | m_end(i.end()) | ||
132 | { | ||
133 | } | ||
134 | |||
135 | //check if the end is reached | ||
136 | bool end() const { | ||
137 | return m_pos.m_it == m_end; | ||
138 | } | ||
139 | |||
140 | //get the current symbol | ||
141 | input::value_type symbol() const { | ||
142 | assert(!end()); | ||
143 | return *m_pos.m_it; | ||
144 | } | ||
145 | |||
146 | //set the longest possible error | ||
147 | void set_error_pos() { | ||
148 | if (m_pos.m_it > m_error_pos.m_it) { | ||
149 | m_error_pos = m_pos; | ||
150 | } | ||
151 | } | ||
152 | |||
153 | //next column | ||
154 | void next_col() { | ||
155 | ++m_pos.m_it; | ||
156 | ++m_pos.m_col; | ||
157 | } | ||
158 | |||
159 | //next line | ||
160 | void next_line() { | ||
161 | ++m_pos.m_line; | ||
162 | m_pos.m_col = 1; | ||
163 | } | ||
164 | |||
165 | //restore the state | ||
166 | void restore(const _state &st) { | ||
167 | m_pos = st.m_pos; | ||
168 | m_matches.resize(st.m_matches); | ||
169 | } | ||
170 | |||
171 | //parse non-term rule. | ||
172 | bool parse_non_term(rule &r); | ||
173 | |||
174 | //parse term rule. | ||
175 | bool parse_term(rule &r); | ||
176 | |||
177 | //execute all the parse procs | ||
178 | void do_parse_procs(void *d) const { | ||
179 | for(_match_vector::const_iterator it = m_matches.begin(); | ||
180 | it != m_matches.end(); | ||
181 | ++it) | ||
182 | { | ||
183 | const _match &m = *it; | ||
184 | parse_proc p = _private::get_parse_proc(*m.m_rule); | ||
185 | p(m.m_begin, m.m_end, d); | ||
186 | } | ||
187 | } | ||
188 | |||
189 | private: | ||
190 | //parse non-term rule. | ||
191 | bool _parse_non_term(rule &r); | ||
192 | |||
193 | //parse term rule. | ||
194 | bool _parse_term(rule &r); | ||
195 | }; | ||
196 | |||
197 | |||
198 | //base class for expressions | ||
199 | class _expr { | ||
200 | public: | ||
201 | //destructor. | ||
202 | virtual ~_expr() { | ||
203 | } | ||
204 | |||
205 | //parse with whitespace | ||
206 | virtual bool parse_non_term(_context &con) const = 0; | ||
207 | |||
208 | //parse terminal | ||
209 | virtual bool parse_term(_context &con) const = 0; | ||
210 | }; | ||
211 | |||
212 | |||
213 | //single character expression. | ||
214 | class _char : public _expr { | ||
215 | public: | ||
216 | //constructor. | ||
217 | _char(char c) : | ||
218 | m_char(c) | ||
219 | { | ||
220 | } | ||
221 | |||
222 | //parse with whitespace | ||
223 | virtual bool parse_non_term(_context &con) const { | ||
224 | return _parse(con); | ||
225 | } | ||
226 | |||
227 | //parse terminal | ||
228 | virtual bool parse_term(_context &con) const { | ||
229 | return _parse(con); | ||
230 | } | ||
231 | |||
232 | private: | ||
233 | //character | ||
234 | input::value_type m_char; | ||
235 | |||
236 | //internal parse | ||
237 | bool _parse(_context &con) const { | ||
238 | if (!con.end()) { | ||
239 | input::value_type ch = con.symbol(); | ||
240 | if (ch == m_char) { | ||
241 | con.next_col(); | ||
242 | return true; | ||
243 | } | ||
244 | } | ||
245 | con.set_error_pos(); | ||
246 | return false; | ||
247 | } | ||
248 | }; | ||
249 | |||
250 | |||
251 | //string expression. | ||
252 | class _string : public _expr { | ||
253 | public: | ||
254 | //constructor from ansi string. | ||
255 | _string(const char *s) : | ||
256 | m_string(Converter{}.from_bytes(s)) | ||
257 | { | ||
258 | } | ||
259 | |||
260 | //parse with whitespace | ||
261 | virtual bool parse_non_term(_context &con) const { | ||
262 | return _parse(con); | ||
263 | } | ||
264 | |||
265 | //parse terminal | ||
266 | virtual bool parse_term(_context &con) const { | ||
267 | return _parse(con); | ||
268 | } | ||
269 | |||
270 | private: | ||
271 | //string | ||
272 | input m_string; | ||
273 | |||
274 | //parse the string | ||
275 | bool _parse(_context &con) const { | ||
276 | for(input::const_iterator it = m_string.begin(), | ||
277 | end = m_string.end();;) | ||
278 | { | ||
279 | if (it == end) return true; | ||
280 | if (con.end()) break; | ||
281 | if (con.symbol() != *it) break; | ||
282 | ++it; | ||
283 | con.next_col(); | ||
284 | } | ||
285 | con.set_error_pos(); | ||
286 | return false; | ||
287 | } | ||
288 | }; | ||
289 | |||
290 | |||
291 | //set expression. | ||
292 | class _set : public _expr { | ||
293 | public: | ||
294 | //constructor from ansi string. | ||
295 | _set(const char *s) { | ||
296 | auto str = Converter{}.from_bytes(s); | ||
297 | for (auto ch : str) { | ||
298 | _add(ch); | ||
299 | } | ||
300 | } | ||
301 | |||
302 | //constructor from range. | ||
303 | _set(int min, int max) { | ||
304 | assert(min >= 0); | ||
305 | assert(min <= max); | ||
306 | m_quick_set.resize((size_t)max + 1U); | ||
307 | for(; min <= max; ++min) { | ||
308 | m_quick_set[(size_t)min] = true; | ||
309 | } | ||
310 | } | ||
311 | |||
312 | //parse with whitespace | ||
313 | virtual bool parse_non_term(_context &con) const { | ||
314 | return _parse(con); | ||
315 | } | ||
316 | |||
317 | //parse terminal | ||
318 | virtual bool parse_term(_context &con) const { | ||
319 | return _parse(con); | ||
320 | } | ||
321 | |||
322 | private: | ||
323 | //set is kept as an array of flags, for quick access | ||
324 | std::vector<bool> m_quick_set; | ||
325 | std::unordered_set<size_t> m_large_set; | ||
326 | |||
327 | //add character | ||
328 | void _add(size_t i) { | ||
329 | if (i <= m_quick_set.size() || i <= 255) { | ||
330 | if (i >= m_quick_set.size()) { | ||
331 | m_quick_set.resize(i + 1); | ||
332 | } | ||
333 | m_quick_set[i] = true; | ||
334 | } else { | ||
335 | m_large_set.insert(i); | ||
336 | } | ||
337 | } | ||
338 | |||
339 | //internal parse | ||
340 | bool _parse(_context &con) const { | ||
341 | if (!con.end()) { | ||
342 | size_t ch = con.symbol(); | ||
343 | if (ch < m_quick_set.size()) { | ||
344 | if (m_quick_set[ch]) { | ||
345 | con.next_col(); | ||
346 | return true; | ||
347 | } | ||
348 | } else if (m_large_set.find(ch) != m_large_set.end()) { | ||
349 | con.next_col(); | ||
350 | return true; | ||
351 | } | ||
352 | } | ||
353 | con.set_error_pos(); | ||
354 | return false; | ||
355 | } | ||
356 | }; | ||
357 | |||
358 | |||
359 | //base class for unary expressions | ||
360 | class _unary : public _expr { | ||
361 | public: | ||
362 | //constructor. | ||
363 | _unary(_expr *e) : | ||
364 | m_expr(e) | ||
365 | { | ||
366 | } | ||
367 | |||
368 | //destructor. | ||
369 | virtual ~_unary() { | ||
370 | delete m_expr; | ||
371 | } | ||
372 | |||
373 | protected: | ||
374 | //expression | ||
375 | _expr *m_expr; | ||
376 | }; | ||
377 | |||
378 | |||
379 | //terminal | ||
380 | class _term : public _unary { | ||
381 | public: | ||
382 | //constructor. | ||
383 | _term(_expr *e) : | ||
384 | _unary(e) | ||
385 | { | ||
386 | } | ||
387 | |||
388 | //parse with whitespace | ||
389 | virtual bool parse_non_term(_context &con) const { | ||
390 | return m_expr->parse_term(con); | ||
391 | } | ||
392 | |||
393 | //parse terminal | ||
394 | virtual bool parse_term(_context &con) const { | ||
395 | return m_expr->parse_term(con); | ||
396 | } | ||
397 | }; | ||
398 | |||
399 | |||
400 | //user | ||
401 | class _user : public _unary { | ||
402 | public: | ||
403 | //constructor. | ||
404 | _user(_expr *e, const user_handler &callback) : | ||
405 | _unary(e), | ||
406 | m_handler(callback) | ||
407 | { | ||
408 | } | ||
409 | |||
410 | //parse with whitespace | ||
411 | virtual bool parse_non_term(_context &con) const { | ||
412 | pos pos = con.m_pos; | ||
413 | if (m_expr->parse_non_term(con)) { | ||
414 | item_t item = {pos.m_it, con.m_pos.m_it, con.m_user_data}; | ||
415 | return m_handler(item); | ||
416 | } | ||
417 | return false; | ||
418 | } | ||
419 | |||
420 | //parse terminal | ||
421 | virtual bool parse_term(_context &con) const { | ||
422 | pos pos = con.m_pos; | ||
423 | if (m_expr->parse_term(con)) { | ||
424 | item_t item = {pos.m_it, con.m_pos.m_it, con.m_user_data}; | ||
425 | return m_handler(item); | ||
426 | } | ||
427 | return false; | ||
428 | } | ||
429 | private: | ||
430 | user_handler m_handler; | ||
431 | }; | ||
432 | |||
433 | |||
434 | //loop 0 | ||
435 | class _loop0 : public _unary { | ||
436 | public: | ||
437 | //constructor. | ||
438 | _loop0(_expr *e) : | ||
439 | _unary(e) | ||
440 | { | ||
441 | } | ||
442 | |||
443 | //parse with whitespace | ||
444 | virtual bool parse_non_term(_context &con) const { | ||
445 | //if parsing of the first fails, restore the context and stop | ||
446 | _state st(con); | ||
447 | if (!m_expr->parse_non_term(con)) { | ||
448 | con.restore(st); | ||
449 | return true; | ||
450 | } | ||
451 | |||
452 | //parse the rest | ||
453 | for(;;) { | ||
454 | _state st(con); | ||
455 | if (!m_expr->parse_non_term(con)) { | ||
456 | con.restore(st); | ||
457 | break; | ||
458 | } | ||
459 | } | ||
460 | |||
461 | return true; | ||
462 | } | ||
463 | |||
464 | //parse terminal | ||
465 | virtual bool parse_term(_context &con) const { | ||
466 | //if parsing of the first fails, restore the context and stop | ||
467 | _state st(con); | ||
468 | if (!m_expr->parse_term(con)) { | ||
469 | con.restore(st); | ||
470 | return true; | ||
471 | } | ||
472 | |||
473 | //parse the rest until no more parsing is possible | ||
474 | for(;;) { | ||
475 | _state st(con); | ||
476 | if (!m_expr->parse_term(con)) { | ||
477 | con.restore(st); | ||
478 | break; | ||
479 | } | ||
480 | } | ||
481 | |||
482 | return true; | ||
483 | } | ||
484 | }; | ||
485 | |||
486 | |||
487 | //loop 1 | ||
488 | class _loop1 : public _unary { | ||
489 | public: | ||
490 | //constructor. | ||
491 | _loop1(_expr *e) : | ||
492 | _unary(e) | ||
493 | { | ||
494 | } | ||
495 | |||
496 | //parse with whitespace | ||
497 | virtual bool parse_non_term(_context &con) const { | ||
498 | //parse the first; if the first fails, stop | ||
499 | if (!m_expr->parse_non_term(con)) return false; | ||
500 | |||
501 | //parse the rest until no more parsing is possible | ||
502 | for(;;) { | ||
503 | _state st(con); | ||
504 | if (!m_expr->parse_non_term(con)) { | ||
505 | con.restore(st); | ||
506 | break; | ||
507 | } | ||
508 | } | ||
509 | |||
510 | return true; | ||
511 | } | ||
512 | |||
513 | //parse terminal | ||
514 | virtual bool parse_term(_context &con) const { | ||
515 | //parse the first; if the first fails, stop | ||
516 | if (!m_expr->parse_term(con)) return false; | ||
517 | |||
518 | //parse the rest until no more parsing is possible | ||
519 | for(;;) { | ||
520 | _state st(con); | ||
521 | if (!m_expr->parse_term(con)) { | ||
522 | con.restore(st); | ||
523 | break; | ||
524 | } | ||
525 | } | ||
526 | |||
527 | return true; | ||
528 | } | ||
529 | }; | ||
530 | |||
531 | |||
532 | //optional | ||
533 | class _optional : public _unary { | ||
534 | public: | ||
535 | //constructor. | ||
536 | _optional(_expr *e) : | ||
537 | _unary(e) | ||
538 | { | ||
539 | } | ||
540 | |||
541 | //parse with whitespace | ||
542 | virtual bool parse_non_term(_context &con) const { | ||
543 | _state st(con); | ||
544 | if (!m_expr->parse_non_term(con)) con.restore(st); | ||
545 | return true; | ||
546 | } | ||
547 | |||
548 | //parse terminal | ||
549 | virtual bool parse_term(_context &con) const { | ||
550 | _state st(con); | ||
551 | if (!m_expr->parse_term(con)) con.restore(st); | ||
552 | return true; | ||
553 | } | ||
554 | }; | ||
555 | |||
556 | |||
557 | //and | ||
558 | class _and : public _unary { | ||
559 | public: | ||
560 | //constructor. | ||
561 | _and(_expr *e) : | ||
562 | _unary(e) | ||
563 | { | ||
564 | } | ||
565 | |||
566 | //parse with whitespace | ||
567 | virtual bool parse_non_term(_context &con) const { | ||
568 | _state st(con); | ||
569 | bool ok = m_expr->parse_non_term(con); | ||
570 | con.restore(st); | ||
571 | return ok; | ||
572 | } | ||
573 | |||
574 | //parse terminal | ||
575 | virtual bool parse_term(_context &con) const { | ||
576 | _state st(con); | ||
577 | bool ok = m_expr->parse_term(con); | ||
578 | con.restore(st); | ||
579 | return ok; | ||
580 | } | ||
581 | }; | ||
582 | |||
583 | |||
584 | //not | ||
585 | class _not : public _unary { | ||
586 | public: | ||
587 | //constructor. | ||
588 | _not(_expr *e) : | ||
589 | _unary(e) | ||
590 | { | ||
591 | } | ||
592 | |||
593 | //parse with whitespace | ||
594 | virtual bool parse_non_term(_context &con) const { | ||
595 | _state st(con); | ||
596 | bool ok = !m_expr->parse_non_term(con); | ||
597 | con.restore(st); | ||
598 | return ok; | ||
599 | } | ||
600 | |||
601 | //parse terminal | ||
602 | virtual bool parse_term(_context &con) const { | ||
603 | _state st(con); | ||
604 | bool ok = !m_expr->parse_term(con); | ||
605 | con.restore(st); | ||
606 | return ok; | ||
607 | } | ||
608 | }; | ||
609 | |||
610 | |||
611 | //newline | ||
612 | class _nl : public _unary { | ||
613 | public: | ||
614 | //constructor. | ||
615 | _nl(_expr *e) : | ||
616 | _unary(e) | ||
617 | { | ||
618 | } | ||
619 | |||
620 | //parse with whitespace | ||
621 | virtual bool parse_non_term(_context &con) const { | ||
622 | if (!m_expr->parse_non_term(con)) return false; | ||
623 | con.next_line(); | ||
624 | return true; | ||
625 | } | ||
626 | |||
627 | //parse terminal | ||
628 | virtual bool parse_term(_context &con) const { | ||
629 | if (!m_expr->parse_term(con)) return false; | ||
630 | con.next_line(); | ||
631 | return true; | ||
632 | } | ||
633 | }; | ||
634 | |||
635 | |||
636 | //base class for binary expressions | ||
637 | class _binary : public _expr { | ||
638 | public: | ||
639 | //constructor. | ||
640 | _binary(_expr *left, _expr *right) : | ||
641 | m_left(left), m_right(right) | ||
642 | { | ||
643 | } | ||
644 | |||
645 | //destructor. | ||
646 | virtual ~_binary() { | ||
647 | delete m_left; | ||
648 | delete m_right; | ||
649 | } | ||
650 | |||
651 | protected: | ||
652 | //left and right expressions | ||
653 | _expr *m_left, *m_right; | ||
654 | }; | ||
655 | |||
656 | |||
657 | //sequence | ||
658 | class _seq : public _binary { | ||
659 | public: | ||
660 | //constructor. | ||
661 | _seq(_expr *left, _expr *right) : | ||
662 | _binary(left, right) | ||
663 | { | ||
664 | } | ||
665 | |||
666 | //parse with whitespace | ||
667 | virtual bool parse_non_term(_context &con) const { | ||
668 | if (!m_left->parse_non_term(con)) return false; | ||
669 | return m_right->parse_non_term(con); | ||
670 | } | ||
671 | |||
672 | //parse terminal | ||
673 | virtual bool parse_term(_context &con) const { | ||
674 | if (!m_left->parse_term(con)) return false; | ||
675 | return m_right->parse_term(con); | ||
676 | } | ||
677 | }; | ||
678 | |||
679 | |||
680 | //choice | ||
681 | class _choice : public _binary { | ||
682 | public: | ||
683 | //constructor. | ||
684 | _choice(_expr *left, _expr *right) : | ||
685 | _binary(left, right) | ||
686 | { | ||
687 | } | ||
688 | |||
689 | //parse with whitespace | ||
690 | virtual bool parse_non_term(_context &con) const { | ||
691 | _state st(con); | ||
692 | if (m_left->parse_non_term(con)) return true; | ||
693 | con.restore(st); | ||
694 | return m_right->parse_non_term(con); | ||
695 | } | ||
696 | |||
697 | //parse terminal | ||
698 | virtual bool parse_term(_context &con) const { | ||
699 | _state st(con); | ||
700 | if (m_left->parse_term(con)) return true; | ||
701 | con.restore(st); | ||
702 | return m_right->parse_term(con); | ||
703 | } | ||
704 | }; | ||
705 | |||
706 | |||
707 | //reference to rule | ||
708 | class _ref : public _expr { | ||
709 | public: | ||
710 | //constructor. | ||
711 | _ref(rule &r) : | ||
712 | m_rule(r) | ||
713 | { | ||
714 | } | ||
715 | |||
716 | //parse with whitespace | ||
717 | virtual bool parse_non_term(_context &con) const { | ||
718 | return con.parse_non_term(m_rule); | ||
719 | } | ||
720 | |||
721 | //parse terminal | ||
722 | virtual bool parse_term(_context &con) const { | ||
723 | return con.parse_term(m_rule); | ||
724 | } | ||
725 | |||
726 | private: | ||
727 | //reference | ||
728 | rule &m_rule; | ||
729 | }; | ||
730 | |||
731 | |||
732 | //eof | ||
733 | class _eof : public _expr { | ||
734 | public: | ||
735 | //parse with whitespace | ||
736 | virtual bool parse_non_term(_context &con) const { | ||
737 | return parse_term(con); | ||
738 | } | ||
739 | |||
740 | //parse terminal | ||
741 | virtual bool parse_term(_context &con) const { | ||
742 | return con.end(); | ||
743 | } | ||
744 | }; | ||
745 | |||
746 | |||
747 | //any | ||
748 | class _any : public _expr { | ||
749 | public: | ||
750 | //parse with whitespace | ||
751 | virtual bool parse_non_term(_context &con) const { | ||
752 | return parse_term(con); | ||
753 | } | ||
754 | |||
755 | //parse terminal | ||
756 | virtual bool parse_term(_context &con) const { | ||
757 | if (!con.end()) { | ||
758 | con.next_col(); | ||
759 | return true; | ||
760 | } | ||
761 | con.set_error_pos(); | ||
762 | return false; | ||
763 | } | ||
764 | }; | ||
765 | |||
766 | |||
767 | //true | ||
768 | class _true : public _expr { | ||
769 | public: | ||
770 | //parse with whitespace | ||
771 | virtual bool parse_non_term(_context &con) const { | ||
772 | return true; | ||
773 | } | ||
774 | |||
775 | //parse terminal | ||
776 | virtual bool parse_term(_context &con) const { | ||
777 | return true; | ||
778 | } | ||
779 | }; | ||
780 | |||
781 | |||
782 | //false | ||
783 | class _false: public _expr { | ||
784 | public: | ||
785 | //parse with whitespace | ||
786 | virtual bool parse_non_term(_context &con) const { | ||
787 | return false; | ||
788 | } | ||
789 | |||
790 | //parse terminal | ||
791 | virtual bool parse_term(_context &con) const { | ||
792 | return false; | ||
793 | } | ||
794 | }; | ||
795 | |||
796 | //exception thrown when left recursion terminates successfully | ||
797 | struct _lr_ok { | ||
798 | rule *m_rule; | ||
799 | _lr_ok(rule *r) : m_rule(r) {} | ||
800 | }; | ||
801 | |||
802 | |||
803 | //constructor | ||
804 | _state::_state(_context &con) : | ||
805 | m_pos(con.m_pos), | ||
806 | m_matches(con.m_matches.size()) | ||
807 | { | ||
808 | } | ||
809 | |||
810 | |||
811 | //parse non-term rule. | ||
812 | bool _context::parse_non_term(rule &r) { | ||
813 | //save the state of the rule | ||
814 | rule::_state old_state = r.m_state; | ||
815 | |||
816 | //success/failure result | ||
817 | bool ok; | ||
818 | |||
819 | //compute the new position | ||
820 | size_t new_pos = m_pos.m_it - m_begin; | ||
821 | |||
822 | //check if we have left recursion | ||
823 | bool lr = new_pos == r.m_state.m_pos; | ||
824 | |||
825 | //update the rule's state | ||
826 | r.m_state.m_pos = new_pos; | ||
827 | |||
828 | //handle the mode of the rule | ||
829 | switch (r.m_state.m_mode) { | ||
830 | //normal parse | ||
831 | case rule::_PARSE: | ||
832 | if (lr) { | ||
833 | //first try to parse the rule by rejecting it, so alternative branches are examined | ||
834 | r.m_state.m_mode = rule::_REJECT; | ||
835 | ok = _parse_non_term(r); | ||
836 | |||
837 | //if the first try is successful, try accepting the rule, | ||
838 | //so other elements of the sequence are parsed | ||
839 | if (ok) { | ||
840 | r.m_state.m_mode = rule::_ACCEPT; | ||
841 | |||
842 | //loop until no more parsing can be done | ||
843 | for(;;) { | ||
844 | //store the correct state, in order to backtrack if the call fails | ||
845 | _state st(*this); | ||
846 | |||
847 | //update the rule position to the current position, | ||
848 | //because at this state the rule is resolving the left recursion | ||
849 | r.m_state.m_pos = m_pos.m_it - m_begin; | ||
850 | |||
851 | //if parsing fails, restore the last good state and stop | ||
852 | if (!_parse_non_term(r)) { | ||
853 | restore(st); | ||
854 | break; | ||
855 | } | ||
856 | } | ||
857 | |||
858 | //since the left recursion was resolved successfully, | ||
859 | //return via a non-local exit | ||
860 | r.m_state = old_state; | ||
861 | throw _lr_ok(r.this_ptr()); | ||
862 | } | ||
863 | } | ||
864 | else { | ||
865 | try { | ||
866 | ok = _parse_non_term(r); | ||
867 | } | ||
868 | catch (const _lr_ok &ex) { | ||
869 | //since left recursions may be mutual, we must test which rule's left recursion | ||
870 | //was ended successfully | ||
871 | if (ex.m_rule == r.this_ptr()) { | ||
872 | ok = true; | ||
873 | } | ||
874 | else { | ||
875 | r.m_state = old_state; | ||
876 | throw; | ||
877 | } | ||
878 | } | ||
879 | } | ||
880 | break; | ||
881 | |||
882 | //reject the left recursive rule | ||
883 | case rule::_REJECT: | ||
884 | if (lr) { | ||
885 | ok = false; | ||
886 | } | ||
887 | else { | ||
888 | r.m_state.m_mode = rule::_PARSE; | ||
889 | ok = _parse_non_term(r); | ||
890 | r.m_state.m_mode = rule::_REJECT; | ||
891 | } | ||
892 | break; | ||
893 | |||
894 | //accept the left recursive rule | ||
895 | case rule::_ACCEPT: | ||
896 | if (lr) { | ||
897 | ok = true; | ||
898 | } | ||
899 | else { | ||
900 | r.m_state.m_mode = rule::_PARSE; | ||
901 | ok = _parse_non_term(r); | ||
902 | r.m_state.m_mode = rule::_ACCEPT; | ||
903 | } | ||
904 | break; | ||
905 | } | ||
906 | |||
907 | //restore the rule's state | ||
908 | r.m_state = old_state; | ||
909 | |||
910 | return ok; | ||
911 | } | ||
912 | |||
913 | |||
914 | //parse term rule. | ||
915 | bool _context::parse_term(rule &r) { | ||
916 | //save the state of the rule | ||
917 | rule::_state old_state = r.m_state; | ||
918 | |||
919 | //success/failure result | ||
920 | bool ok; | ||
921 | |||
922 | //compute the new position | ||
923 | size_t new_pos = m_pos.m_it - m_begin; | ||
924 | |||
925 | //check if we have left recursion | ||
926 | bool lr = new_pos == r.m_state.m_pos; | ||
927 | |||
928 | //update the rule's state | ||
929 | r.m_state.m_pos = new_pos; | ||
930 | |||
931 | //handle the mode of the rule | ||
932 | switch (r.m_state.m_mode) { | ||
933 | //normal parse | ||
934 | case rule::_PARSE: | ||
935 | if (lr) { | ||
936 | //first try to parse the rule by rejecting it, so alternative branches are examined | ||
937 | r.m_state.m_mode = rule::_REJECT; | ||
938 | ok = _parse_term(r); | ||
939 | |||
940 | //if the first try is successful, try accepting the rule, | ||
941 | //so other elements of the sequence are parsed | ||
942 | if (ok) { | ||
943 | r.m_state.m_mode = rule::_ACCEPT; | ||
944 | |||
945 | //loop until no more parsing can be done | ||
946 | for(;;) { | ||
947 | //store the correct state, in order to backtrack if the call fails | ||
948 | _state st(*this); | ||
949 | |||
950 | //update the rule position to the current position, | ||
951 | //because at this state the rule is resolving the left recursion | ||
952 | r.m_state.m_pos = m_pos.m_it - m_begin; | ||
953 | |||
954 | //if parsing fails, restore the last good state and stop | ||
955 | if (!_parse_term(r)) { | ||
956 | restore(st); | ||
957 | break; | ||
958 | } | ||
959 | } | ||
960 | |||
961 | //since the left recursion was resolved successfully, | ||
962 | //return via a non-local exit | ||
963 | r.m_state = old_state; | ||
964 | throw _lr_ok(r.this_ptr()); | ||
965 | } | ||
966 | } | ||
967 | else { | ||
968 | try { | ||
969 | ok = _parse_term(r); | ||
970 | } | ||
971 | catch (const _lr_ok &ex) { | ||
972 | //since left recursions may be mutual, we must test which rule's left recursion | ||
973 | //was ended successfully | ||
974 | if (ex.m_rule == r.this_ptr()) { | ||
975 | ok = true; | ||
976 | } | ||
977 | else { | ||
978 | r.m_state = old_state; | ||
979 | throw; | ||
980 | } | ||
981 | } | ||
982 | } | ||
983 | break; | ||
984 | |||
985 | //reject the left recursive rule | ||
986 | case rule::_REJECT: | ||
987 | if (lr) { | ||
988 | ok = false; | ||
989 | } | ||
990 | else { | ||
991 | r.m_state.m_mode = rule::_PARSE; | ||
992 | ok = _parse_term(r); | ||
993 | r.m_state.m_mode = rule::_REJECT; | ||
994 | } | ||
995 | break; | ||
996 | |||
997 | //accept the left recursive rule | ||
998 | case rule::_ACCEPT: | ||
999 | if (lr) { | ||
1000 | ok = true; | ||
1001 | } | ||
1002 | else { | ||
1003 | r.m_state.m_mode = rule::_PARSE; | ||
1004 | ok = _parse_term(r); | ||
1005 | r.m_state.m_mode = rule::_ACCEPT; | ||
1006 | } | ||
1007 | break; | ||
1008 | } | ||
1009 | |||
1010 | //restore the rule's state | ||
1011 | r.m_state = old_state; | ||
1012 | |||
1013 | return ok; | ||
1014 | } | ||
1015 | |||
1016 | |||
1017 | //parse non-term rule internal. | ||
1018 | bool _context::_parse_non_term(rule &r) { | ||
1019 | bool ok; | ||
1020 | if (_private::get_parse_proc(r)) { | ||
1021 | pos b = m_pos; | ||
1022 | ok = _private::get_expr(r)->parse_non_term(*this); | ||
1023 | if (ok) { | ||
1024 | m_matches.push_back(_match(r.this_ptr(), b, m_pos)); | ||
1025 | } | ||
1026 | } | ||
1027 | else { | ||
1028 | ok = _private::get_expr(r)->parse_non_term(*this); | ||
1029 | } | ||
1030 | return ok; | ||
1031 | } | ||
1032 | |||
1033 | |||
1034 | //parse term rule internal. | ||
1035 | bool _context::_parse_term(rule &r) { | ||
1036 | bool ok; | ||
1037 | if (_private::get_parse_proc(r)) { | ||
1038 | pos b = m_pos; | ||
1039 | ok = _private::get_expr(r)->parse_term(*this); | ||
1040 | if (ok) { | ||
1041 | m_matches.push_back(_match(r.this_ptr(), b, m_pos)); | ||
1042 | } | ||
1043 | } | ||
1044 | else { | ||
1045 | ok = _private::get_expr(r)->parse_term(*this); | ||
1046 | } | ||
1047 | return ok; | ||
1048 | } | ||
1049 | |||
1050 | |||
1051 | //get syntax error | ||
1052 | static error _syntax_error(_context &con) { | ||
1053 | return error(con.m_error_pos, con.m_error_pos, ERROR_SYNTAX_ERROR); | ||
1054 | } | ||
1055 | |||
1056 | |||
1057 | //get eof error | ||
1058 | static error _eof_error(_context &con) { | ||
1059 | return error(con.m_error_pos, con.m_error_pos, ERROR_INVALID_EOF); | ||
1060 | } | ||
1061 | |||
1062 | |||
1063 | /** constructor from input. | ||
1064 | @param i input. | ||
1065 | */ | ||
1066 | pos::pos(input &i) : | ||
1067 | m_it(i.begin()), | ||
1068 | m_line(1), | ||
1069 | m_col(1) | ||
1070 | { | ||
1071 | } | ||
1072 | |||
1073 | |||
1074 | /** character terminal constructor. | ||
1075 | @param c character. | ||
1076 | */ | ||
1077 | expr::expr(char c) : | ||
1078 | m_expr(new _char(c)) | ||
1079 | { | ||
1080 | } | ||
1081 | |||
1082 | |||
1083 | /** null-terminated string terminal constructor. | ||
1084 | @param s null-terminated string. | ||
1085 | */ | ||
1086 | expr::expr(const char *s) : | ||
1087 | m_expr(new _string(s)) | ||
1088 | { | ||
1089 | } | ||
1090 | |||
1091 | |||
1092 | /** rule reference constructor. | ||
1093 | @param r rule. | ||
1094 | */ | ||
1095 | expr::expr(rule &r) : | ||
1096 | m_expr(new _ref(r)) | ||
1097 | { | ||
1098 | } | ||
1099 | |||
1100 | |||
1101 | /** creates a zero-or-more loop out of this expression. | ||
1102 | @return a zero-or-more loop expression. | ||
1103 | */ | ||
1104 | expr expr::operator *() const { | ||
1105 | return _private::construct_expr(new _loop0(m_expr)); | ||
1106 | } | ||
1107 | |||
1108 | |||
1109 | /** creates a one-or-more loop out of this expression. | ||
1110 | @return a one-or-more loop expression. | ||
1111 | */ | ||
1112 | expr expr::operator +() const { | ||
1113 | return _private::construct_expr(new _loop1(m_expr)); | ||
1114 | } | ||
1115 | |||
1116 | |||
1117 | /** creates an optional out of this expression. | ||
1118 | @return an optional expression. | ||
1119 | */ | ||
1120 | expr expr::operator -() const { | ||
1121 | return _private::construct_expr(new _optional(m_expr)); | ||
1122 | } | ||
1123 | |||
1124 | |||
1125 | /** creates an AND-expression. | ||
1126 | @return an AND-expression. | ||
1127 | */ | ||
1128 | expr expr::operator &() const { | ||
1129 | return _private::construct_expr((new _and(m_expr))); | ||
1130 | } | ||
1131 | |||
1132 | |||
1133 | /** creates a NOT-expression. | ||
1134 | @return a NOT-expression. | ||
1135 | */ | ||
1136 | expr expr::operator !() const { | ||
1137 | return _private::construct_expr(new _not(m_expr)); | ||
1138 | } | ||
1139 | |||
1140 | |||
1141 | /** constructor. | ||
1142 | @param b begin position. | ||
1143 | @param e end position. | ||
1144 | */ | ||
1145 | input_range::input_range(const pos &b, const pos &e) : | ||
1146 | m_begin(b), | ||
1147 | m_end(e) {} | ||
1148 | |||
1149 | |||
1150 | /** constructor. | ||
1151 | @param b begin position. | ||
1152 | @param e end position. | ||
1153 | @param t error type. | ||
1154 | */ | ||
1155 | error::error(const pos &b, const pos &e, int t) : | ||
1156 | input_range(b, e), | ||
1157 | m_type(t) {} | ||
1158 | |||
1159 | |||
1160 | /** compare on begin position. | ||
1161 | @param e the other error to compare this with. | ||
1162 | @return true if this comes before the previous error, false otherwise. | ||
1163 | */ | ||
1164 | bool error::operator < (const error &e) const { | ||
1165 | return m_begin.m_it < e.m_begin.m_it; | ||
1166 | } | ||
1167 | |||
1168 | |||
1169 | /** character terminal constructor. | ||
1170 | @param c character. | ||
1171 | */ | ||
1172 | rule::rule(char c) : | ||
1173 | m_expr(new _char(c)) | ||
1174 | { | ||
1175 | m_parse_proc = _get_parse_proc(this); | ||
1176 | } | ||
1177 | |||
1178 | |||
1179 | /** null-terminated string terminal constructor. | ||
1180 | @param s null-terminated string. | ||
1181 | */ | ||
1182 | rule::rule(const char *s) : | ||
1183 | m_expr(new _string(s)) | ||
1184 | { | ||
1185 | m_parse_proc = _get_parse_proc(this); | ||
1186 | } | ||
1187 | |||
1188 | |||
1189 | /** constructor from expression. | ||
1190 | @param e expression. | ||
1191 | */ | ||
1192 | rule::rule(const expr &e) : | ||
1193 | m_expr(_private::get_expr(e)) | ||
1194 | { | ||
1195 | m_parse_proc = _get_parse_proc(this); | ||
1196 | } | ||
1197 | |||
1198 | |||
1199 | /** constructor from rule. | ||
1200 | @param r rule. | ||
1201 | */ | ||
1202 | rule::rule(rule &r) : | ||
1203 | m_expr(new _ref(r)), | ||
1204 | m_parse_proc(0) | ||
1205 | { | ||
1206 | m_parse_proc = _get_parse_proc(this); | ||
1207 | } | ||
1208 | |||
1209 | |||
1210 | /** invalid constructor from rule (required by gcc). | ||
1211 | @param r rule. | ||
1212 | @exception std::logic_error always thrown. | ||
1213 | */ | ||
1214 | rule::rule(const rule &r) { | ||
1215 | throw std::logic_error("invalid operation"); | ||
1216 | } | ||
1217 | |||
1218 | |||
1219 | /** deletes the internal object that represents the expression. | ||
1220 | */ | ||
1221 | rule::~rule() { | ||
1222 | delete m_expr; | ||
1223 | } | ||
1224 | |||
1225 | |||
1226 | /** creates a zero-or-more loop out of this rule. | ||
1227 | @return a zero-or-more loop rule. | ||
1228 | */ | ||
1229 | expr rule::operator *() { | ||
1230 | return _private::construct_expr(new _loop0(new _ref(*this))); | ||
1231 | } | ||
1232 | |||
1233 | |||
1234 | /** creates a one-or-more loop out of this rule. | ||
1235 | @return a one-or-more loop rule. | ||
1236 | */ | ||
1237 | expr rule::operator +() { | ||
1238 | return _private::construct_expr(new _loop1(new _ref(*this))); | ||
1239 | } | ||
1240 | |||
1241 | |||
1242 | /** creates an optional out of this rule. | ||
1243 | @return an optional rule. | ||
1244 | */ | ||
1245 | expr rule::operator -() { | ||
1246 | return _private::construct_expr(new _optional(new _ref(*this))); | ||
1247 | } | ||
1248 | |||
1249 | |||
1250 | /** creates an AND-expression out of this rule. | ||
1251 | @return an AND-expression out of this rule. | ||
1252 | */ | ||
1253 | expr rule::operator &() { | ||
1254 | return _private::construct_expr(new _and(new _ref(*this))); | ||
1255 | } | ||
1256 | |||
1257 | |||
1258 | /** creates a NOT-expression out of this rule. | ||
1259 | @return a NOT-expression out of this rule. | ||
1260 | */ | ||
1261 | expr rule::operator !() { | ||
1262 | return _private::construct_expr(new _not(new _ref(*this))); | ||
1263 | } | ||
1264 | |||
1265 | |||
1266 | /** sets the parse procedure. | ||
1267 | @param p procedure. | ||
1268 | */ | ||
1269 | void rule::set_parse_proc(parse_proc p) { | ||
1270 | assert(p); | ||
1271 | m_parse_proc = p; | ||
1272 | _parse_proc_map_t& _parse_proc_map = _get_parse_proc_map(); | ||
1273 | _parse_proc_map[this] = p; | ||
1274 | } | ||
1275 | |||
1276 | |||
1277 | /** creates a sequence of expressions. | ||
1278 | @param left left operand. | ||
1279 | @param right right operand. | ||
1280 | @return an expression which parses a sequence. | ||
1281 | */ | ||
1282 | expr operator >> (const expr &left, const expr &right) { | ||
1283 | return _private::construct_expr( | ||
1284 | new _seq(_private::get_expr(left), _private::get_expr(right))); | ||
1285 | } | ||
1286 | |||
1287 | |||
1288 | /** creates a choice of expressions. | ||
1289 | @param left left operand. | ||
1290 | @param right right operand. | ||
1291 | @return an expression which parses a choice. | ||
1292 | */ | ||
1293 | expr operator | (const expr &left, const expr &right) { | ||
1294 | return _private::construct_expr( | ||
1295 | new _choice(_private::get_expr(left), _private::get_expr(right))); | ||
1296 | } | ||
1297 | |||
1298 | |||
1299 | /** converts a parser expression into a terminal. | ||
1300 | @param e expression. | ||
1301 | @return an expression which parses a terminal. | ||
1302 | */ | ||
1303 | expr term(const expr &e) { | ||
1304 | return _private::construct_expr( | ||
1305 | new _term(_private::get_expr(e))); | ||
1306 | } | ||
1307 | |||
1308 | |||
1309 | /** creates a set expression from a null-terminated string. | ||
1310 | @param s null-terminated string with characters of the set. | ||
1311 | @return an expression which parses a single character out of a set. | ||
1312 | */ | ||
1313 | expr set(const char *s) { | ||
1314 | return _private::construct_expr(new _set(s)); | ||
1315 | } | ||
1316 | |||
1317 | |||
1318 | /** creates a range expression. | ||
1319 | @param min min character. | ||
1320 | @param max max character. | ||
1321 | @return an expression which parses a single character out of range. | ||
1322 | */ | ||
1323 | expr range(int min, int max) { | ||
1324 | return _private::construct_expr(new _set(min, max)); | ||
1325 | } | ||
1326 | |||
1327 | |||
1328 | /** creates an expression which increments the line counter | ||
1329 | and resets the column counter when the given expression | ||
1330 | is parsed successfully; used for newline characters. | ||
1331 | @param e expression to wrap into a newline parser. | ||
1332 | @return an expression that handles newlines. | ||
1333 | */ | ||
1334 | expr nl(const expr &e) { | ||
1335 | return _private::construct_expr(new _nl(_private::get_expr(e))); | ||
1336 | } | ||
1337 | |||
1338 | |||
1339 | /** creates an expression which tests for the end of input. | ||
1340 | @return an expression that handles the end of input. | ||
1341 | */ | ||
1342 | expr eof() { | ||
1343 | return _private::construct_expr(new _eof()); | ||
1344 | } | ||
1345 | |||
1346 | |||
1347 | /** creates a not expression. | ||
1348 | @param e expression. | ||
1349 | @return the appropriate expression. | ||
1350 | */ | ||
1351 | expr not_(const expr &e) { | ||
1352 | return !e; | ||
1353 | } | ||
1354 | |||
1355 | |||
1356 | /** creates an and expression. | ||
1357 | @param e expression. | ||
1358 | @return the appropriate expression. | ||
1359 | */ | ||
1360 | expr and_(const expr &e) { | ||
1361 | return &e; | ||
1362 | } | ||
1363 | |||
1364 | |||
1365 | /** creates an expression that parses any character. | ||
1366 | @return the appropriate expression. | ||
1367 | */ | ||
1368 | expr any() { | ||
1369 | return _private::construct_expr(new _any()); | ||
1370 | } | ||
1371 | |||
1372 | |||
1373 | /** parsing succeeds without consuming any input. | ||
1374 | */ | ||
1375 | expr true_() { | ||
1376 | return _private::construct_expr(new _true()); | ||
1377 | } | ||
1378 | |||
1379 | |||
1380 | /** parsing fails without consuming any input. | ||
1381 | */ | ||
1382 | expr false_() { | ||
1383 | return _private::construct_expr(new _false()); | ||
1384 | } | ||
1385 | |||
1386 | |||
1387 | /** parse with target expression and let user handle result. | ||
1388 | */ | ||
1389 | expr user(const expr &e, const user_handler& handler) { | ||
1390 | return _private::construct_expr(new _user(_private::get_expr(e), handler)); | ||
1391 | } | ||
1392 | |||
1393 | |||
1394 | /** parses the given input. | ||
1395 | The parse procedures of each rule parsed are executed | ||
1396 | before this function returns, if parsing succeeds. | ||
1397 | @param i input. | ||
1398 | @param g root rule of grammar. | ||
1399 | @param el list of errors. | ||
1400 | @param d user data, passed to the parse procedures. | ||
1401 | @return true on parsing success, false on failure. | ||
1402 | */ | ||
1403 | bool parse(input &i, rule &g, error_list &el, void *d, void* ud) { | ||
1404 | //prepare context | ||
1405 | _context con(i, ud); | ||
1406 | |||
1407 | //parse grammar | ||
1408 | if (!con.parse_non_term(g)) { | ||
1409 | el.push_back(_syntax_error(con)); | ||
1410 | return false; | ||
1411 | } | ||
1412 | |||
1413 | //if end is not reached, there was an error | ||
1414 | if (!con.end()) { | ||
1415 | if (con.m_error_pos.m_it < con.m_end) { | ||
1416 | el.push_back(_syntax_error(con)); | ||
1417 | } | ||
1418 | else { | ||
1419 | el.push_back(_eof_error(con)); | ||
1420 | } | ||
1421 | return false; | ||
1422 | } | ||
1423 | |||
1424 | //success; execute the parse procedures | ||
1425 | con.do_parse_procs(d); | ||
1426 | return true; | ||
1427 | } | ||
1428 | |||
1429 | |||
1430 | } //namespace parserlib | ||