?  search:
summary | log | graphiclog1 | graphiclog2 | commit | commitdiff | tree | refs | edit | fork
blame | history | raw | HEAD
compiler: fix parsing issue with non-ASCII first package char
[official-gcc.git] / gcc / go / gofrontend / lex.cc
blobe641b7eaf39f33a1593905fd2b34b9f572cb50e9
1 // lex.cc -- Go frontend lexer.
2
3 // Copyright 2009 The Go Authors. All rights reserved.
4 // Use of this source code is governed by a BSD-style
5 // license that can be found in the LICENSE file.
6
7 #include "go-system.h"
8 #include "go-diagnostics.h"
9
10 #include "lex.h"
11
12 // Manage mapping from keywords to the Keyword codes.
13
14 class Keywords
15 {
16 public:
17 // The structure which maps keywords to codes.
18 struct Mapping
19 {
20 // Keyword string.
21 const char* keystring;
22 // Keyword code.
23 Keyword keycode;
24 };
25
26 // Return the parsecode corresponding to KEYSTRING, or
27 // KEYWORD_INVALID if it is not a keyword.
28 Keyword
29 keyword_to_code(const char* keyword, size_t len) const;
30
31 // Return the string for a keyword.
32 const char*
33 keyword_to_string(Keyword) const;
34
35 private:
36 static const Mapping mapping_[];
37 static const int count_;
38 };
39
40 // Mapping from keyword string to keyword code. This array must be
41 // kept in sorted order, and the order must match the Keyword enum.
42 // Strings are looked up using bsearch.
43
44 const Keywords::Mapping
45 Keywords::mapping_[] =
46 {
47 { NULL, KEYWORD_INVALID },
48 { "__asm__", KEYWORD_ASM },
49 { "break", KEYWORD_BREAK },
50 { "case", KEYWORD_CASE },
51 { "chan", KEYWORD_CHAN },
52 { "const", KEYWORD_CONST },
53 { "continue", KEYWORD_CONTINUE },
54 { "default", KEYWORD_DEFAULT },
55 { "defer", KEYWORD_DEFER },
56 { "else", KEYWORD_ELSE },
57 { "fallthrough", KEYWORD_FALLTHROUGH },
58 { "for", KEYWORD_FOR },
59 { "func", KEYWORD_FUNC },
60 { "go", KEYWORD_GO },
61 { "goto", KEYWORD_GOTO },
62 { "if", KEYWORD_IF },
63 { "import", KEYWORD_IMPORT },
64 { "interface", KEYWORD_INTERFACE },
65 { "map", KEYWORD_MAP },
66 { "package", KEYWORD_PACKAGE },
67 { "range", KEYWORD_RANGE },
68 { "return", KEYWORD_RETURN },
69 { "select", KEYWORD_SELECT },
70 { "struct", KEYWORD_STRUCT },
71 { "switch", KEYWORD_SWITCH },
72 { "type", KEYWORD_TYPE },
73 { "var", KEYWORD_VAR }
74 };
75
76 // Number of entries in the map.
77
78 const int Keywords::count_ =
79 sizeof(Keywords::mapping_) / sizeof(Keywords::mapping_[0]);
80
81 // Comparison function passed to bsearch.
82
83 extern "C"
84 {
85
86 struct Keywords_search_key
87 {
88 const char* str;
89 size_t len;
90 };
91
92 static int
93 keyword_compare(const void* keyv, const void* mapv)
94 {
95 const Keywords_search_key* key =
96 static_cast<const Keywords_search_key*>(keyv);
97 const Keywords::Mapping* map =
98 static_cast<const Keywords::Mapping*>(mapv);
99 if (map->keystring == NULL)
100 return 1;
101 int i = strncmp(key->str, map->keystring, key->len);
102 if (i != 0)
103 return i;
104 if (map->keystring[key->len] != '\0')
105 return -1;
106 return 0;
107 }
108
109 } // End extern "C".
110
111 // Convert a string to a keyword code. Return KEYWORD_INVALID if the
112 // string is not a keyword.
113
114 Keyword
115 Keywords::keyword_to_code(const char* keyword, size_t len) const
116 {
117 Keywords_search_key key;
118 key.str = keyword;
119 key.len = len;
120 void* mapv = bsearch(&key,
121 this->mapping_,
122 this->count_,
123 sizeof(this->mapping_[0]),
124 keyword_compare);
125 if (mapv == NULL)
126 return KEYWORD_INVALID;
127 Mapping* map = static_cast<Mapping*>(mapv);
128 return map->keycode;
129 }
130
131 // Convert a keyword code to a string.
132
133 const char*
134 Keywords::keyword_to_string(Keyword code) const
135 {
136 go_assert(code > KEYWORD_INVALID && code < this->count_);
137 const Mapping* map = &this->mapping_[code];
138 go_assert(map->keycode == code);
139 return map->keystring;
140 }
141
142 // There is one instance of the Keywords class.
143
144 static Keywords keywords;
145
146 // Class Token.
147
148 // Make a general token.
149
150 Token::Token(Classification classification, Location location)
151 : classification_(classification), location_(location)
152 {
153 }
154
155 // Destroy a token.
156
157 Token::~Token()
158 {
159 this->clear();
160 }
161
162 // Clear a token--release memory.
163
164 void
165 Token::clear()
166 {
167 if (this->classification_ == TOKEN_INTEGER
168 || this->classification_ == TOKEN_CHARACTER)
169 mpz_clear(this->u_.integer_value);
170 else if (this->classification_ == TOKEN_FLOAT
171 || this->classification_ == TOKEN_IMAGINARY)
172 mpfr_clear(this->u_.float_value);
173 }
174
175 // Construct a token.
176
177 Token::Token(const Token& tok)
178 : classification_(tok.classification_), location_(tok.location_)
179 {
180 switch (this->classification_)
181 {
182 case TOKEN_INVALID:
183 case TOKEN_EOF:
184 break;
185 case TOKEN_KEYWORD:
186 this->u_.keyword = tok.u_.keyword;
187 break;
188 case TOKEN_IDENTIFIER:
189 case TOKEN_STRING:
190 this->u_.string_value = tok.u_.string_value;
191 break;
192 case TOKEN_OPERATOR:
193 this->u_.op = tok.u_.op;
194 break;
195 case TOKEN_CHARACTER:
196 case TOKEN_INTEGER:
197 mpz_init_set(this->u_.integer_value, tok.u_.integer_value);
198 break;
199 case TOKEN_FLOAT:
200 case TOKEN_IMAGINARY:
201 mpfr_init_set(this->u_.float_value, tok.u_.float_value, GMP_RNDN);
202 break;
203 default:
204 go_unreachable();
205 }
206 }
207
208 // Assign to a token.
209
210 Token&
211 Token::operator=(const Token& tok)
212 {
213 this->clear();
214 this->classification_ = tok.classification_;
215 this->location_ = tok.location_;
216 switch (tok.classification_)
217 {
218 case TOKEN_INVALID:
219 case TOKEN_EOF:
220 break;
221 case TOKEN_KEYWORD:
222 this->u_.keyword = tok.u_.keyword;
223 break;
224 case TOKEN_IDENTIFIER:
225 this->u_.identifier_value.name = tok.u_.identifier_value.name;
226 this->u_.identifier_value.is_exported =
227 tok.u_.identifier_value.is_exported;
228 break;
229 case TOKEN_STRING:
230 this->u_.string_value = tok.u_.string_value;
231 break;
232 case TOKEN_OPERATOR:
233 this->u_.op = tok.u_.op;
234 break;
235 case TOKEN_CHARACTER:
236 case TOKEN_INTEGER:
237 mpz_init_set(this->u_.integer_value, tok.u_.integer_value);
238 break;
239 case TOKEN_FLOAT:
240 case TOKEN_IMAGINARY:
241 mpfr_init_set(this->u_.float_value, tok.u_.float_value, GMP_RNDN);
242 break;
243 default:
244 go_unreachable();
245 }
246 return *this;
247 }
248
249 // Print the token for debugging.
250
251 void
252 Token::print(FILE* file) const
253 {
254 switch (this->classification_)
255 {
256 case TOKEN_INVALID:
257 fprintf(file, "invalid");
258 break;
259 case TOKEN_EOF:
260 fprintf(file, "EOF");
261 break;
262 case TOKEN_KEYWORD:
263 fprintf(file, "keyword %s", keywords.keyword_to_string(this->u_.keyword));
264 break;
265 case TOKEN_IDENTIFIER:
266 fprintf(file, "identifier \"%s\"", this->u_.string_value->c_str());
267 break;
268 case TOKEN_STRING:
269 fprintf(file, "quoted string \"%s\"", this->u_.string_value->c_str());
270 break;
271 case TOKEN_CHARACTER:
272 fprintf(file, "character ");
273 mpz_out_str(file, 10, this->u_.integer_value);
274 break;
275 case TOKEN_INTEGER:
276 fprintf(file, "integer ");
277 mpz_out_str(file, 10, this->u_.integer_value);
278 break;
279 case TOKEN_FLOAT:
280 fprintf(file, "float ");
281 mpfr_out_str(file, 10, 0, this->u_.float_value, GMP_RNDN);
282 break;
283 case TOKEN_IMAGINARY:
284 fprintf(file, "imaginary ");
285 mpfr_out_str(file, 10, 0, this->u_.float_value, GMP_RNDN);
286 break;
287 case TOKEN_OPERATOR:
288 fprintf(file, "operator ");
289 switch (this->u_.op)
290 {
291 case OPERATOR_INVALID:
292 fprintf(file, "invalid");
293 break;
294 case OPERATOR_OROR:
295 fprintf(file, "||");
296 break;
297 case OPERATOR_ANDAND:
298 fprintf(file, "&&");
299 break;
300 case OPERATOR_EQEQ:
301 fprintf(file, "==");
302 break;
303 case OPERATOR_NOTEQ:
304 fprintf(file, "!=");
305 break;
306 case OPERATOR_LT:
307 fprintf(file, "<");
308 break;
309 case OPERATOR_LE:
310 fprintf(file, "<=");
311 break;
312 case OPERATOR_GT:
313 fprintf(file, ">");
314 break;
315 case OPERATOR_GE:
316 fprintf(file, ">=");
317 break;
318 case OPERATOR_PLUS:
319 fprintf(file, "+");
320 break;
321 case OPERATOR_MINUS:
322 fprintf(file, "-");
323 break;
324 case OPERATOR_OR:
325 fprintf(file, "|");
326 break;
327 case OPERATOR_XOR:
328 fprintf(file, "^");
329 break;
330 case OPERATOR_MULT:
331 fprintf(file, "*");
332 break;
333 case OPERATOR_DIV:
334 fprintf(file, "/");
335 break;
336 case OPERATOR_MOD:
337 fprintf(file, "%%");
338 break;
339 case OPERATOR_LSHIFT:
340 fprintf(file, "<<");
341 break;
342 case OPERATOR_RSHIFT:
343 fprintf(file, ">>");
344 break;
345 case OPERATOR_AND:
346 fprintf(file, "&");
347 break;
348 case OPERATOR_BITCLEAR:
349 fprintf(file, "&^");
350 break;
351 case OPERATOR_NOT:
352 fprintf(file, "!");
353 break;
354 case OPERATOR_CHANOP:
355 fprintf(file, "<-");
356 break;
357 case OPERATOR_EQ:
358 fprintf(file, "=");
359 break;
360 case OPERATOR_PLUSEQ:
361 fprintf(file, "+=");
362 break;
363 case OPERATOR_MINUSEQ:
364 fprintf(file, "-=");
365 break;
366 case OPERATOR_OREQ:
367 fprintf(file, "|=");
368 break;
369 case OPERATOR_XOREQ:
370 fprintf(file, "^=");
371 break;
372 case OPERATOR_MULTEQ:
373 fprintf(file, "*=");
374 break;
375 case OPERATOR_DIVEQ:
376 fprintf(file, "/=");
377 break;
378 case OPERATOR_MODEQ:
379 fprintf(file, "%%=");
380 break;
381 case OPERATOR_LSHIFTEQ:
382 fprintf(file, "<<=");
383 break;
384 case OPERATOR_RSHIFTEQ:
385 fprintf(file, ">>=");
386 break;
387 case OPERATOR_ANDEQ:
388 fprintf(file, "&=");
389 break;
390 case OPERATOR_BITCLEAREQ:
391 fprintf(file, "&^=");
392 break;
393 case OPERATOR_PLUSPLUS:
394 fprintf(file, "++");
395 break;
396 case OPERATOR_MINUSMINUS:
397 fprintf(file, "--");
398 break;
399 case OPERATOR_COLON:
400 fprintf(file, ":");
401 break;
402 case OPERATOR_COLONEQ:
403 fprintf(file, ":=");
404 break;
405 case OPERATOR_SEMICOLON:
406 fprintf(file, ";");
407 break;
408 case OPERATOR_DOT:
409 fprintf(file, ".");
410 break;
411 case OPERATOR_COMMA:
412 fprintf(file, ",");
413 break;
414 case OPERATOR_LPAREN:
415 fprintf(file, "(");
416 break;
417 case OPERATOR_RPAREN:
418 fprintf(file, ")");
419 break;
420 case OPERATOR_LCURLY:
421 fprintf(file, "{");
422 break;
423 case OPERATOR_RCURLY:
424 fprintf(file, "}");
425 break;
426 case OPERATOR_LSQUARE:
427 fprintf(file, "[");
428 break;
429 case OPERATOR_RSQUARE:
430 fprintf(file, "]");
431 break;
432 default:
433 go_unreachable();
434 }
435 break;
436 default:
437 go_unreachable();
438 }
439 }
440
441 // Class Lex.
442
443 Lex::Lex(const char* input_file_name, FILE* input_file, Linemap* linemap)
444 : input_file_name_(input_file_name), input_file_(input_file),
445 linemap_(linemap), linebuf_(NULL), linebufsize_(120), linesize_(0),
446 lineoff_(0), lineno_(0), add_semi_at_eol_(false), pragmas_(0),
447 extern_(), linknames_(NULL)
448 {
449 this->linebuf_ = new char[this->linebufsize_];
450 this->linemap_->start_file(input_file_name, 0);
451 }
452
453 Lex::~Lex()
454 {
455 delete[] this->linebuf_;
456 }
457
458 // Read a new line from the file.
459
460 ssize_t
461 Lex::get_line()
462 {
463 char* buf = this->linebuf_;
464 size_t size = this->linebufsize_;
465
466 FILE* file = this->input_file_;
467 size_t cur = 0;
468 while (true)
469 {
470 int c = getc(file);
471 if (c == EOF)
472 {
473 if (cur == 0)
474 return -1;
475 break;
476 }
477 if (cur + 1 >= size)
478 {
479 size_t ns = 2 * size + 1;
480 if (ns < size || static_cast<ssize_t>(ns) < 0)
481 go_error_at(this->location(), "out of memory");
482 char* nb = new char[ns];
483 memcpy(nb, buf, cur);
484 delete[] buf;
485 buf = nb;
486 size = ns;
487 }
488 buf[cur] = c;
489 ++cur;
490
491 if (c == '\n')
492 break;
493 }
494
495 buf[cur] = '\0';
496
497 this->linebuf_ = buf;
498 this->linebufsize_ = size;
499
500 return cur;
501 }
502
503 // See if we need to read a new line. Return true if there is a new
504 // line, false if we are at EOF.
505
506 bool
507 Lex::require_line()
508 {
509 if (this->lineoff_ < this->linesize_)
510 return true;
511
512 ssize_t got = this->get_line();
513 if (got < 0)
514 return false;
515 ++this->lineno_;
516 this->linesize_= got;
517 this->lineoff_ = 0;
518
519 this->linemap_->start_line(this->lineno_, this->linesize_);
520
521 return true;
522 }
523
524 // Get the current location.
525
526 Location
527 Lex::location() const
528 {
529 return this->linemap_->get_location(this->lineoff_ + 1);
530 }
531
532 // Get a location slightly before the current one. This is used for
533 // slightly more efficient handling of operator tokens.
534
535 Location
536 Lex::earlier_location(int chars) const
537 {
538 return this->linemap_->get_location(this->lineoff_ + 1 - chars);
539 }
540
541 // Get the next token.
542
543 Token
544 Lex::next_token()
545 {
546 bool saw_cpp_comment = false;
547 while (true)
548 {
549 if (!this->require_line())
550 {
551 bool add_semi_at_eol = this->add_semi_at_eol_;
552 this->add_semi_at_eol_ = false;
553 if (add_semi_at_eol)
554 return this->make_operator(OPERATOR_SEMICOLON, 1);
555 return this->make_eof_token();
556 }
557
558 if (!saw_cpp_comment)
559 this->extern_.clear();
560 saw_cpp_comment = false;
561
562 const char* p = this->linebuf_ + this->lineoff_;
563 const char* pend = this->linebuf_ + this->linesize_;
564
565 while (p < pend)
566 {
567 unsigned char cc = *p;
568 switch (cc)
569 {
570 case ' ': case '\t': case '\r':
571 ++p;
572 // Skip whitespace quickly.
573 while (*p == ' ' || *p == '\t' || *p == '\r')
574 ++p;
575 break;
576
577 case '\n':
578 {
579 ++p;
580 bool add_semi_at_eol = this->add_semi_at_eol_;
581 this->add_semi_at_eol_ = false;
582 if (add_semi_at_eol)
583 {
584 this->lineoff_ = p - this->linebuf_;
585 return this->make_operator(OPERATOR_SEMICOLON, 1);
586 }
587 }
588 break;
589
590 case '/':
591 if (p[1] == '/')
592 {
593 this->lineoff_ = p + 2 - this->linebuf_;
594 this->skip_cpp_comment();
595 p = pend;
596 if (p[-1] == '\n' && this->add_semi_at_eol_)
597 --p;
598 saw_cpp_comment = true;
599 }
600 else if (p[1] == '*')
601 {
602 this->lineoff_ = p + 2 - this->linebuf_;
603 Location location = this->location();
604 bool found_newline = false;
605 if (!this->skip_c_comment(&found_newline))
606 return Token::make_invalid_token(location);
607 if (found_newline && this->add_semi_at_eol_)
608 {
609 this->add_semi_at_eol_ = false;
610 return this->make_operator(OPERATOR_SEMICOLON, 1);
611 }
612 p = this->linebuf_ + this->lineoff_;
613 pend = this->linebuf_ + this->linesize_;
614 }
615 else if (p[1] == '=')
616 {
617 this->add_semi_at_eol_ = false;
618 this->lineoff_ = p + 2 - this->linebuf_;
619 return this->make_operator(OPERATOR_DIVEQ, 2);
620 }
621 else
622 {
623 this->add_semi_at_eol_ = false;
624 this->lineoff_ = p + 1 - this->linebuf_;
625 return this->make_operator(OPERATOR_DIV, 1);
626 }
627 break;
628
629 case 'A': case 'B': case 'C': case 'D': case 'E': case 'F':
630 case 'G': case 'H': case 'I': case 'J': case 'K': case 'L':
631 case 'M': case 'N': case 'O': case 'P': case 'Q': case 'R':
632 case 'S': case 'T': case 'U': case 'V': case 'W': case 'X':
633 case 'Y': case 'Z':
634 case 'a': case 'b': case 'c': case 'd': case 'e': case 'f':
635 case 'g': case 'h': case 'i': case 'j': case 'k': case 'l':
636 case 'm': case 'n': case 'o': case 'p': case 'q': case 'r':
637 case 's': case 't': case 'u': case 'v': case 'w': case 'x':
638 case 'y': case 'z':
639 case '_':
640 this->lineoff_ = p - this->linebuf_;
641 return this->gather_identifier();
642
643 case '0': case '1': case '2': case '3': case '4':
644 case '5': case '6': case '7': case '8': case '9':
645 this->add_semi_at_eol_ = true;
646 this->lineoff_ = p - this->linebuf_;
647 return this->gather_number();
648
649 case '\'':
650 this->add_semi_at_eol_ = true;
651 this->lineoff_ = p - this->linebuf_;
652 return this->gather_character();
653
654 case '"':
655 this->add_semi_at_eol_ = true;
656 this->lineoff_ = p - this->linebuf_;
657 return this->gather_string();
658
659 case '`':
660 this->add_semi_at_eol_ = true;
661 this->lineoff_ = p - this->linebuf_;
662 return this->gather_raw_string();
663
664 case '<':
665 case '>':
666 case '&':
667 if (p + 2 < pend)
668 {
669 this->add_semi_at_eol_ = false;
670 Operator op = this->three_character_operator(cc, p[1], p[2]);
671 if (op != OPERATOR_INVALID)
672 {
673 this->lineoff_ = p + 3 - this->linebuf_;
674 return this->make_operator(op, 3);
675 }
676 }
677 // Fall through.
678 case '|':
679 case '=':
680 case '!':
681 case '+':
682 case '-':
683 case '^':
684 case '*':
685 // '/' handled above.
686 case '%':
687 case ':':
688 case ';':
689 case ',':
690 case '(': case ')':
691 case '{': case '}':
692 case '[': case ']':
693 {
694 this->add_semi_at_eol_ = false;
695 Operator op = this->two_character_operator(cc, p[1]);
696 int chars;
697 if (op != OPERATOR_INVALID)
698 {
699 ++p;
700 chars = 2;
701 }
702 else
703 {
704 op = this->one_character_operator(cc);
705 chars = 1;
706 }
707 this->lineoff_ = p + 1 - this->linebuf_;
708 return this->make_operator(op, chars);
709 }
710
711 case '.':
712 if (p[1] >= '0' && p[1] <= '9')
713 {
714 this->add_semi_at_eol_ = true;
715 this->lineoff_ = p - this->linebuf_;
716 return this->gather_number();
717 }
718 if (p[1] == '.' && p[2] == '.')
719 {
720 this->add_semi_at_eol_ = false;
721 this->lineoff_ = p + 3 - this->linebuf_;
722 return this->make_operator(OPERATOR_ELLIPSIS, 3);
723 }
724 this->add_semi_at_eol_ = false;
725 this->lineoff_ = p + 1 - this->linebuf_;
726 return this->make_operator(OPERATOR_DOT, 1);
727
728 default:
729 {
730 unsigned int ci;
731 bool issued_error;
732 this->lineoff_ = p - this->linebuf_;
733 const char *pnext = this->advance_one_utf8_char(p, &ci,
734 &issued_error);
735
736 // Ignore byte order mark at start of file.
737 if (ci == 0xfeff)
738 {
739 p = pnext;
740 break;
741 }
742
743 if (Lex::is_unicode_letter(ci))
744 return this->gather_identifier();
745
746 if (!issued_error)
747 go_error_at(this->location(),
748 "invalid character 0x%x in input file",
749 ci);
750
751 p = pend;
752
753 break;
754 }
755 }
756 }
757
758 this->lineoff_ = p - this->linebuf_;
759 }
760 }
761
762 // Fetch one UTF-8 character from a string. Set *VALUE to the value.
763 // Return the number of bytes read from the string. Returns 0 if the
764 // string does not point to a valid UTF-8 character.
765
766 int
767 Lex::fetch_char(const char* p, unsigned int* value)
768 {
769 unsigned char c = *p;
770 if (c <= 0x7f)
771 {
772 *value = c;
773 return 1;
774 }
775 else if ((c & 0xe0) == 0xc0
776 && (p[1] & 0xc0) == 0x80)
777 {
778 *value = (((c & 0x1f) << 6)
779 + (p[1] & 0x3f));
780 if (*value <= 0x7f)
781 {
782 *value = 0xfffd;
783 return 0;
784 }
785 return 2;
786 }
787 else if ((c & 0xf0) == 0xe0
788 && (p[1] & 0xc0) == 0x80
789 && (p[2] & 0xc0) == 0x80)
790 {
791 *value = (((c & 0xf) << 12)
792 + ((p[1] & 0x3f) << 6)
793 + (p[2] & 0x3f));
794 if (*value <= 0x7ff)
795 {
796 *value = 0xfffd;
797 return 0;
798 }
799 return 3;
800 }
801 else if ((c & 0xf8) == 0xf0
802 && (p[1] & 0xc0) == 0x80
803 && (p[2] & 0xc0) == 0x80
804 && (p[3] & 0xc0) == 0x80)
805 {
806 *value = (((c & 0x7) << 18)
807 + ((p[1] & 0x3f) << 12)
808 + ((p[2] & 0x3f) << 6)
809 + (p[3] & 0x3f));
810 if (*value <= 0xffff)
811 {
812 *value = 0xfffd;
813 return 0;
814 }
815 return 4;
816 }
817 else
818 {
819 /* Invalid encoding. Return the Unicode replacement
820 character. */
821 *value = 0xfffd;
822 return 0;
823 }
824 }
825
826 // Advance one UTF-8 character. Return the pointer beyond the
827 // character. Set *VALUE to the value. Set *ISSUED_ERROR if an error
828 // was issued.
829
830 const char*
831 Lex::advance_one_utf8_char(const char* p, unsigned int* value,
832 bool* issued_error)
833 {
834 *issued_error = false;
835
836 if (*p == '\0')
837 {
838 go_error_at(this->location(), "invalid NUL byte");
839 *issued_error = true;
840 *value = 0;
841 return p + 1;
842 }
843
844 int adv = Lex::fetch_char(p, value);
845 if (adv == 0)
846 {
847 go_error_at(this->location(), "invalid UTF-8 encoding");
848 *issued_error = true;
849 return p + 1;
850 }
851
852 // Warn about byte order mark, except at start of file.
853 if (*value == 0xfeff && (this->lineno_ != 1 || this->lineoff_ != 0))
854 {
855 go_error_at(this->location(), "Unicode (UTF-8) BOM in middle of file");
856 *issued_error = true;
857 }
858
859 return p + adv;
860 }
861
862 // Pick up an identifier.
863
864 Token
865 Lex::gather_identifier()
866 {
867 const char* pstart = this->linebuf_ + this->lineoff_;
868 const char* p = pstart;
869 const char* pend = this->linebuf_ + this->linesize_;
870 bool is_first = true;
871 bool is_exported = false;
872 bool has_non_ascii_char = false;
873 std::string buf;
874 while (p < pend)
875 {
876 unsigned char cc = *p;
877 if (cc <= 0x7f)
878 {
879 if ((cc < 'A' || cc > 'Z')
880 && (cc < 'a' || cc > 'z')
881 && cc != '_'
882 && (cc < '0' || cc > '9'))
883 {
884 // Check for an invalid character here, as we get better
885 // error behaviour if we swallow them as part of the
886 // identifier we are building.
887 if ((cc >= ' ' && cc < 0x7f)
888 || cc == '\t'
889 || cc == '\r'
890 || cc == '\n')
891 break;
892
893 this->lineoff_ = p - this->linebuf_;
894 go_error_at(this->location(),
895 "invalid character 0x%x in identifier",
896 cc);
897 if (!has_non_ascii_char)
898 {
899 buf.assign(pstart, p - pstart);
900 has_non_ascii_char = true;
901 }
902 if (!Lex::is_invalid_identifier(buf))
903 buf.append("$INVALID$");
904 }
905 ++p;
906 if (is_first)
907 {
908 is_exported = cc >= 'A' && cc <= 'Z';
909 is_first = false;
910 }
911 if (has_non_ascii_char)
912 buf.push_back(cc);
913 }
914 else
915 {
916 unsigned int ci;
917 bool issued_error;
918 this->lineoff_ = p - this->linebuf_;
919 const char* pnext = this->advance_one_utf8_char(p, &ci,
920 &issued_error);
921 bool is_invalid = false;
922 if (!Lex::is_unicode_letter(ci) && !Lex::is_unicode_digit(ci))
923 {
924 // There is no valid place for a non-ASCII character
925 // other than an identifier, so we get better error
926 // handling behaviour if we swallow this character after
927 // giving an error.
928 if (!issued_error)
929 go_error_at(this->location(),
930 "invalid character 0x%x in identifier",
931 ci);
932 is_invalid = true;
933 }
934 if (is_first)
935 {
936 is_exported = Lex::is_unicode_uppercase(ci);
937 is_first = false;
938 }
939 if (!has_non_ascii_char)
940 {
941 buf.assign(pstart, p - pstart);
942 has_non_ascii_char = true;
943 }
944 if (is_invalid && !Lex::is_invalid_identifier(buf))
945 buf.append("$INVALID$");
946 buf.append(p, pnext - p);
947 p = pnext;
948 }
949 }
950 Location location = this->location();
951 this->add_semi_at_eol_ = true;
952 this->lineoff_ = p - this->linebuf_;
953 if (has_non_ascii_char)
954 return Token::make_identifier_token(buf, is_exported, location);
955 else
956 {
957 Keyword code = keywords.keyword_to_code(pstart, p - pstart);
958 if (code == KEYWORD_INVALID)
959 return Token::make_identifier_token(std::string(pstart, p - pstart),
960 is_exported, location);
961 else
962 {
963 switch (code)
964 {
965 case KEYWORD_BREAK:
966 case KEYWORD_CONTINUE:
967 case KEYWORD_FALLTHROUGH:
968 case KEYWORD_RETURN:
969 break;
970 default:
971 this->add_semi_at_eol_ = false;
972 break;
973 }
974 return Token::make_keyword_token(code, location);
975 }
976 }
977 }
978
979 // Return whether C is a hex digit.
980
981 bool
982 Lex::is_hex_digit(char c)
983 {
984 return ((c >= '0' && c <= '9')
985 || (c >= 'A' && c <= 'F')
986 || (c >= 'a' && c <= 'f'));
987 }
988
989 // not a hex value
990 #define NHV 100
991
992 // for use by Lex::hex_val
993 static const unsigned char hex_value_lookup_table[256] =
994 {
995 NHV, NHV, NHV, NHV, NHV, NHV, NHV, NHV, // NUL SOH STX ETX EOT ENQ ACK BEL
996 NHV, NHV, NHV, NHV, NHV, NHV, NHV, NHV, // BS HT LF VT FF CR SO SI
997 NHV, NHV, NHV, NHV, NHV, NHV, NHV, NHV, // DLE DC1 DC2 DC3 DC4 NAK SYN ETB
998 NHV, NHV, NHV, NHV, NHV, NHV, NHV, NHV, // CAN EM SUB ESC FS GS RS US
999 NHV, NHV, NHV, NHV, NHV, NHV, NHV, NHV, // SP ! " # $ % & '
1000 NHV, NHV, NHV, NHV, NHV, NHV, NHV, NHV, // ( ) * + , - . /
1001 0, 1, 2, 3, 4, 5, 6, 7, // 0 1 2 3 4 5 6 7
1002 8, 9, NHV, NHV, NHV, NHV, NHV, NHV, // 8 9 : ; < = > ?
1003 NHV, 10, 11, 12, 13, 14, 15, NHV, // @ A B C D E F G
1004 NHV, NHV, NHV, NHV, NHV, NHV, NHV, NHV, // H I J K L M N O
1005 NHV, NHV, NHV, NHV, NHV, NHV, NHV, NHV, // P Q R S T U V W
1006 NHV, NHV, NHV, NHV, NHV, NHV, NHV, NHV, // X Y Z [ \ ] ^ _
1007 NHV, 10, 11, 12, 13, 14, 15, NHV, // ` a b c d e f g
1008 NHV, NHV, NHV, NHV, NHV, NHV, NHV, NHV, // h i j k l m n o
1009 NHV, NHV, NHV, NHV, NHV, NHV, NHV, NHV, // p q r s t u v w
1010 NHV, NHV, NHV, NHV, NHV, NHV, NHV, NHV, // x y z { | } ~
1011 NHV, NHV, NHV, NHV, NHV, NHV, NHV, NHV, //
1012 NHV, NHV, NHV, NHV, NHV, NHV, NHV, NHV, //
1013 NHV, NHV, NHV, NHV, NHV, NHV, NHV, NHV, //
1014 NHV, NHV, NHV, NHV, NHV, NHV, NHV, NHV, //
1015 NHV, NHV, NHV, NHV, NHV, NHV, NHV, NHV, //
1016 NHV, NHV, NHV, NHV, NHV, NHV, NHV, NHV, //
1017 NHV, NHV, NHV, NHV, NHV, NHV, NHV, NHV, //
1018 NHV, NHV, NHV, NHV, NHV, NHV, NHV, NHV, //
1019 NHV, NHV, NHV, NHV, NHV, NHV, NHV, NHV, //
1020 NHV, NHV, NHV, NHV, NHV, NHV, NHV, NHV, //
1021 NHV, NHV, NHV, NHV, NHV, NHV, NHV, NHV, //
1022 NHV, NHV, NHV, NHV, NHV, NHV, NHV, NHV, //
1023 NHV, NHV, NHV, NHV, NHV, NHV, NHV, NHV, //
1024 NHV, NHV, NHV, NHV, NHV, NHV, NHV, NHV, //
1025 NHV, NHV, NHV, NHV, NHV, NHV, NHV, NHV, //
1026 NHV, NHV, NHV, NHV, NHV, NHV, NHV, NHV //
1027 };
1028
1029 unsigned
1030 Lex::hex_val(char c)
1031 {
1032 return hex_value_lookup_table[static_cast<unsigned char>(c)];
1033 }
1034
1035 // Return whether an exponent could start at P.
1036
1037 bool
1038 Lex::could_be_exponent(const char* p, const char* pend)
1039 {
1040 if (*p != 'e' && *p != 'E')
1041 return false;
1042 ++p;
1043 if (p >= pend)
1044 return false;
1045 if (*p == '+' || *p == '-')
1046 {
1047 ++p;
1048 if (p >= pend)
1049 return false;
1050 }
1051 return *p >= '0' && *p <= '9';
1052 }
1053
1054 // Pick up a number.
1055
1056 Token
1057 Lex::gather_number()
1058 {
1059 const char* pstart = this->linebuf_ + this->lineoff_;
1060 const char* p = pstart;
1061 const char* pend = this->linebuf_ + this->linesize_;
1062
1063 Location location = this->location();
1064
1065 bool neg = false;
1066 if (*p == '+')
1067 ++p;
1068 else if (*p == '-')
1069 {
1070 ++p;
1071 neg = true;
1072 }
1073
1074 const char* pnum = p;
1075 if (*p == '0')
1076 {
1077 int base;
1078 if ((p[1] == 'x' || p[1] == 'X')
1079 && Lex::is_hex_digit(p[2]))
1080 {
1081 base = 16;
1082 p += 2;
1083 pnum = p;
1084 while (p < pend)
1085 {
1086 if (!Lex::is_hex_digit(*p))
1087 break;
1088 ++p;
1089 }
1090 }
1091 else
1092 {
1093 base = 8;
1094 pnum = p;
1095 while (p < pend)
1096 {
1097 if (*p < '0' || *p > '9')
1098 break;
1099 ++p;
1100 }
1101 }
1102
1103 // A partial token that looks like an octal literal might actually be the
1104 // beginning of a floating-point or imaginary literal.
1105 if (base == 16 || (*p != '.' && *p != 'i' && !Lex::could_be_exponent(p, pend)))
1106 {
1107 std::string s(pnum, p - pnum);
1108 mpz_t val;
1109 int r = mpz_init_set_str(val, s.c_str(), base);
1110 if (r != 0)
1111 {
1112 if (base == 8)
1113 go_error_at(this->location(), "invalid octal literal");
1114 else
1115 go_error_at(this->location(), "invalid hex literal");
1116 }
1117
1118 if (neg)
1119 mpz_neg(val, val);
1120
1121 this->lineoff_ = p - this->linebuf_;
1122 Token ret = Token::make_integer_token(val, location);
1123 mpz_clear(val);
1124 return ret;
1125 }
1126 }
1127
1128 while (p < pend)
1129 {
1130 if (*p < '0' || *p > '9')
1131 break;
1132 ++p;
1133 }
1134
1135 if (*p != '.' && *p != 'i' && !Lex::could_be_exponent(p, pend))
1136 {
1137 std::string s(pnum, p - pnum);
1138 mpz_t val;
1139 int r = mpz_init_set_str(val, s.c_str(), 10);
1140 go_assert(r == 0);
1141
1142 if (neg)
1143 mpz_neg(val, val);
1144
1145 this->lineoff_ = p - this->linebuf_;
1146 Token ret = Token::make_integer_token(val, location);
1147 mpz_clear(val);
1148 return ret;
1149 }
1150
1151 if (*p != 'i')
1152 {
1153 bool dot = *p == '.';
1154
1155 ++p;
1156
1157 if (!dot)
1158 {
1159 if (*p == '+' || *p == '-')
1160 ++p;
1161 }
1162
1163 while (p < pend)
1164 {
1165 if (*p < '0' || *p > '9')
1166 break;
1167 ++p;
1168 }
1169
1170 if (dot && Lex::could_be_exponent(p, pend))
1171 {
1172 ++p;
1173 if (*p == '+' || *p == '-')
1174 ++p;
1175 while (p < pend)
1176 {
1177 if (*p < '0' || *p > '9')
1178 break;
1179 ++p;
1180 }
1181 }
1182 }
1183
1184 std::string s(pnum, p - pnum);
1185 mpfr_t val;
1186 int r = mpfr_init_set_str(val, s.c_str(), 10, GMP_RNDN);
1187 go_assert(r == 0);
1188
1189 if (neg)
1190 mpfr_neg(val, val, GMP_RNDN);
1191
1192 bool is_imaginary = *p == 'i';
1193 if (is_imaginary)
1194 ++p;
1195
1196 this->lineoff_ = p - this->linebuf_;
1197 if (is_imaginary)
1198 {
1199 Token ret = Token::make_imaginary_token(val, location);
1200 mpfr_clear(val);
1201 return ret;
1202 }
1203 else
1204 {
1205 Token ret = Token::make_float_token(val, location);
1206 mpfr_clear(val);
1207 return ret;
1208 }
1209 }
1210
1211 // Advance one character, possibly escaped. Return the pointer beyond
1212 // the character. Set *VALUE to the character. Set *IS_CHARACTER if
1213 // this is a character (e.g., 'a' or '\u1234') rather than a byte
1214 // value (e.g., '\001').
1215
1216 const char*
1217 Lex::advance_one_char(const char* p, bool is_single_quote, unsigned int* value,
1218 bool* is_character)
1219 {
1220 *value = 0;
1221 *is_character = true;
1222 if (*p != '\\')
1223 {
1224 bool issued_error;
1225 const char* ret = this->advance_one_utf8_char(p, value, &issued_error);
1226 if (is_single_quote
1227 && (*value == '\'' || *value == '\n')
1228 && !issued_error)
1229 go_error_at(this->location(), "invalid character literal");
1230 return ret;
1231 }
1232 else
1233 {
1234 ++p;
1235 switch (*p)
1236 {
1237 case '0': case '1': case '2': case '3':
1238 case '4': case '5': case '6': case '7':
1239 *is_character = false;
1240 if (p[1] >= '0' && p[1] <= '7'
1241 && p[2] >= '0' && p[2] <= '7')
1242 {
1243 *value = ((Lex::octal_value(p[0]) << 6)
1244 + (Lex::octal_value(p[1]) << 3)
1245 + Lex::octal_value(p[2]));
1246 if (*value > 255)
1247 {
1248 go_error_at(this->location(), "invalid octal constant");
1249 *value = 255;
1250 }
1251 return p + 3;
1252 }
1253 go_error_at(this->location(), "invalid octal character");
1254 return (p[1] >= '0' && p[1] <= '7'
1255 ? p + 2
1256 : p + 1);
1257
1258 case 'x':
1259 *is_character = false;
1260 if (Lex::is_hex_digit(p[1]) && Lex::is_hex_digit(p[2]))
1261 {
1262 *value = (Lex::hex_val(p[1]) << 4) + Lex::hex_val(p[2]);
1263 return p + 3;
1264 }
1265 go_error_at(this->location(), "invalid hex character");
1266 return (Lex::is_hex_digit(p[1])
1267 ? p + 2
1268 : p + 1);
1269
1270 case 'a':
1271 *value = '\a';
1272 return p + 1;
1273 case 'b':
1274 *value = '\b';
1275 return p + 1;
1276 case 'f':
1277 *value = '\f';
1278 return p + 1;
1279 case 'n':
1280 *value = '\n';
1281 return p + 1;
1282 case 'r':
1283 *value = '\r';
1284 return p + 1;
1285 case 't':
1286 *value = '\t';
1287 return p + 1;
1288 case 'v':
1289 *value = '\v';
1290 return p + 1;
1291 case '\\':
1292 *value = '\\';
1293 return p + 1;
1294 case '\'':
1295 if (!is_single_quote)
1296 go_error_at(this->location(), "invalid quoted character");
1297 *value = '\'';
1298 return p + 1;
1299 case '"':
1300 if (is_single_quote)
1301 go_error_at(this->location(), "invalid quoted character");
1302 *value = '"';
1303 return p + 1;
1304
1305 case 'u':
1306 if (Lex::is_hex_digit(p[1]) && Lex::is_hex_digit(p[2])
1307 && Lex::is_hex_digit(p[3]) && Lex::is_hex_digit(p[4]))
1308 {
1309 *value = ((Lex::hex_val(p[1]) << 12)
1310 + (Lex::hex_val(p[2]) << 8)
1311 + (Lex::hex_val(p[3]) << 4)
1312 + Lex::hex_val(p[4]));
1313 if (*value >= 0xd800 && *value < 0xe000)
1314 {
1315 go_error_at(this->location(),
1316 "invalid unicode code point 0x%x",
1317 *value);
1318 // Use the replacement character.
1319 *value = 0xfffd;
1320 }
1321 return p + 5;
1322 }
1323 go_error_at(this->location(), "invalid little unicode code point");
1324 return p + 1;
1325
1326 case 'U':
1327 if (Lex::is_hex_digit(p[1]) && Lex::is_hex_digit(p[2])
1328 && Lex::is_hex_digit(p[3]) && Lex::is_hex_digit(p[4])
1329 && Lex::is_hex_digit(p[5]) && Lex::is_hex_digit(p[6])
1330 && Lex::is_hex_digit(p[7]) && Lex::is_hex_digit(p[8]))
1331 {
1332 *value = ((Lex::hex_val(p[1]) << 28)
1333 + (Lex::hex_val(p[2]) << 24)
1334 + (Lex::hex_val(p[3]) << 20)
1335 + (Lex::hex_val(p[4]) << 16)
1336 + (Lex::hex_val(p[5]) << 12)
1337 + (Lex::hex_val(p[6]) << 8)
1338 + (Lex::hex_val(p[7]) << 4)
1339 + Lex::hex_val(p[8]));
1340 if (*value > 0x10ffff
1341 || (*value >= 0xd800 && *value < 0xe000))
1342 {
1343 go_error_at(this->location(),
1344 "invalid unicode code point 0x%x",
1345 *value);
1346 // Use the replacement character.
1347 *value = 0xfffd;
1348 }
1349 return p + 9;
1350 }
1351 go_error_at(this->location(), "invalid big unicode code point");
1352 return p + 1;
1353
1354 default:
1355 go_error_at(this->location(), "invalid character after %<\\%>");
1356 *value = *p;
1357 return p + 1;
1358 }
1359 }
1360 }
1361
1362 // Append V to STR. IS_CHARACTER is true for a character which should
1363 // be stored in UTF-8, false for a general byte value which should be
1364 // stored directly.
1365
1366 void
1367 Lex::append_char(unsigned int v, bool is_character, std::string* str,
1368 Location location)
1369 {
1370 char buf[4];
1371 size_t len;
1372 if (v <= 0x7f || !is_character)
1373 {
1374 buf[0] = v;
1375 len = 1;
1376 }
1377 else if (v <= 0x7ff)
1378 {
1379 buf[0] = 0xc0 + (v >> 6);
1380 buf[1] = 0x80 + (v & 0x3f);
1381 len = 2;
1382 }
1383 else
1384 {
1385 if (v > 0x10ffff)
1386 {
1387 go_warning_at(location, 0,
1388 "unicode code point 0x%x out of range in string", v);
1389 // Turn it into the "replacement character".
1390 v = 0xfffd;
1391 }
1392 if (v >= 0xd800 && v < 0xe000)
1393 {
1394 go_warning_at(location, 0,
1395 "unicode code point 0x%x is invalid surrogate pair", v);
1396 v = 0xfffd;
1397 }
1398 if (v <= 0xffff)
1399 {
1400 buf[0] = 0xe0 + (v >> 12);
1401 buf[1] = 0x80 + ((v >> 6) & 0x3f);
1402 buf[2] = 0x80 + (v & 0x3f);
1403 len = 3;
1404 }
1405 else
1406 {
1407 buf[0] = 0xf0 + (v >> 18);
1408 buf[1] = 0x80 + ((v >> 12) & 0x3f);
1409 buf[2] = 0x80 + ((v >> 6) & 0x3f);
1410 buf[3] = 0x80 + (v & 0x3f);
1411 len = 4;
1412 }
1413 }
1414 str->append(buf, len);
1415 }
1416
1417 // Pick up a character literal.
1418
1419 Token
1420 Lex::gather_character()
1421 {
1422 ++this->lineoff_;
1423 const char* pstart = this->linebuf_ + this->lineoff_;
1424 const char* p = pstart;
1425
1426 unsigned int value;
1427 bool is_character;
1428 p = this->advance_one_char(p, true, &value, &is_character);
1429
1430 if (*p != '\'')
1431 {
1432 go_error_at(this->location(), "unterminated character constant");
1433 this->lineoff_ = p - this->linebuf_;
1434 return this->make_invalid_token();
1435 }
1436
1437 mpz_t val;
1438 mpz_init_set_ui(val, value);
1439
1440 Location location = this->location();
1441 this->lineoff_ = p + 1 - this->linebuf_;
1442 Token ret = Token::make_character_token(val, location);
1443 mpz_clear(val);
1444 return ret;
1445 }
1446
1447 // Pick up a quoted string.
1448
1449 Token
1450 Lex::gather_string()
1451 {
1452 const char* pstart = this->linebuf_ + this->lineoff_ + 1;
1453 const char* p = pstart;
1454 const char* pend = this->linebuf_ + this->linesize_;
1455
1456 std::string value;
1457 while (*p != '"')
1458 {
1459 Location loc = this->location();
1460 unsigned int c;
1461 bool is_character;
1462 this->lineoff_ = p - this->linebuf_;
1463 p = this->advance_one_char(p, false, &c, &is_character);
1464 if (p >= pend)
1465 {
1466 go_error_at(this->location(), "unterminated string");
1467 --p;
1468 break;
1469 }
1470 Lex::append_char(c, is_character, &value, loc);
1471 }
1472
1473 Location location = this->location();
1474 this->lineoff_ = p + 1 - this->linebuf_;
1475 return Token::make_string_token(value, location);
1476 }
1477
1478 // Pick up a raw string.
1479
1480 Token
1481 Lex::gather_raw_string()
1482 {
1483 const char* p = this->linebuf_ + this->lineoff_ + 1;
1484 const char* pend = this->linebuf_ + this->linesize_;
1485 Location location = this->location();
1486
1487 std::string value;
1488 while (true)
1489 {
1490 while (p < pend)
1491 {
1492 if (*p == '`')
1493 {
1494 this->lineoff_ = p + 1 - this->linebuf_;
1495 return Token::make_string_token(value, location);
1496 }
1497 Location loc = this->location();
1498 unsigned int c;
1499 bool issued_error;
1500 this->lineoff_ = p - this->linebuf_;
1501 p = this->advance_one_utf8_char(p, &c, &issued_error);
1502 // "Carriage return characters ('\r') inside raw string literals
1503 // are discarded from the raw string value."
1504 if (c != '\r')
1505 Lex::append_char(c, true, &value, loc);
1506 }
1507 this->lineoff_ = p - this->linebuf_;
1508 if (!this->require_line())
1509 {
1510 go_error_at(location, "unterminated raw string");
1511 return Token::make_string_token(value, location);
1512 }
1513 p = this->linebuf_ + this->lineoff_;
1514 pend = this->linebuf_ + this->linesize_;
1515 }
1516 }
1517
1518 // If C1 C2 C3 are a three character operator, return the code.
1519
1520 Operator
1521 Lex::three_character_operator(char c1, char c2, char c3)
1522 {
1523 if (c3 == '=')
1524 {
1525 if (c1 == '<' && c2 == '<')
1526 return OPERATOR_LSHIFTEQ;
1527 else if (c1 == '>' && c2 == '>')
1528 return OPERATOR_RSHIFTEQ;
1529 else if (c1 == '&' && c2 == '^')
1530 return OPERATOR_BITCLEAREQ;
1531 }
1532 return OPERATOR_INVALID;
1533 }
1534
1535 // If C1 C2 are a two character operator, return the code.
1536
1537 Operator
1538 Lex::two_character_operator(char c1, char c2)
1539 {
1540 switch (c1)
1541 {
1542 case '|':
1543 if (c2 == '|')
1544 return OPERATOR_OROR;
1545 else if (c2 == '=')
1546 return OPERATOR_OREQ;
1547 break;
1548 case '&':
1549 if (c2 == '&')
1550 return OPERATOR_ANDAND;
1551 else if (c2 == '^')
1552 return OPERATOR_BITCLEAR;
1553 else if (c2 == '=')
1554 return OPERATOR_ANDEQ;
1555 break;
1556 case '^':
1557 if (c2 == '=')
1558 return OPERATOR_XOREQ;
1559 break;
1560 case '=':
1561 if (c2 == '=')
1562 return OPERATOR_EQEQ;
1563 break;
1564 case '!':
1565 if (c2 == '=')
1566 return OPERATOR_NOTEQ;
1567 break;
1568 case '<':
1569 if (c2 == '=')
1570 return OPERATOR_LE;
1571 else if (c2 == '<')
1572 return OPERATOR_LSHIFT;
1573 else if (c2 == '-')
1574 return OPERATOR_CHANOP;
1575 break;
1576 case '>':
1577 if (c2 == '=')
1578 return OPERATOR_GE;
1579 else if (c2 == '>')
1580 return OPERATOR_RSHIFT;
1581 break;
1582 case '*':
1583 if (c2 == '=')
1584 return OPERATOR_MULTEQ;
1585 break;
1586 case '/':
1587 if (c2 == '=')
1588 return OPERATOR_DIVEQ;
1589 break;
1590 case '%':
1591 if (c2 == '=')
1592 return OPERATOR_MODEQ;
1593 break;
1594 case '+':
1595 if (c2 == '+')
1596 {
1597 this->add_semi_at_eol_ = true;
1598 return OPERATOR_PLUSPLUS;
1599 }
1600 else if (c2 == '=')
1601 return OPERATOR_PLUSEQ;
1602 break;
1603 case '-':
1604 if (c2 == '-')
1605 {
1606 this->add_semi_at_eol_ = true;
1607 return OPERATOR_MINUSMINUS;
1608 }
1609 else if (c2 == '=')
1610 return OPERATOR_MINUSEQ;
1611 break;
1612 case ':':
1613 if (c2 == '=')
1614 return OPERATOR_COLONEQ;
1615 break;
1616 default:
1617 break;
1618 }
1619 return OPERATOR_INVALID;
1620 }
1621
1622 // If character C is an operator, return the code.
1623
1624 Operator
1625 Lex::one_character_operator(char c)
1626 {
1627 switch (c)
1628 {
1629 case '<':
1630 return OPERATOR_LT;
1631 case '>':
1632 return OPERATOR_GT;
1633 case '+':
1634 return OPERATOR_PLUS;
1635 case '-':
1636 return OPERATOR_MINUS;
1637 case '|':
1638 return OPERATOR_OR;
1639 case '^':
1640 return OPERATOR_XOR;
1641 case '*':
1642 return OPERATOR_MULT;
1643 case '/':
1644 return OPERATOR_DIV;
1645 case '%':
1646 return OPERATOR_MOD;
1647 case '&':
1648 return OPERATOR_AND;
1649 case '!':
1650 return OPERATOR_NOT;
1651 case '=':
1652 return OPERATOR_EQ;
1653 case ':':
1654 return OPERATOR_COLON;
1655 case ';':
1656 return OPERATOR_SEMICOLON;
1657 case '.':
1658 return OPERATOR_DOT;
1659 case ',':
1660 return OPERATOR_COMMA;
1661 case '(':
1662 return OPERATOR_LPAREN;
1663 case ')':
1664 this->add_semi_at_eol_ = true;
1665 return OPERATOR_RPAREN;
1666 case '{':
1667 return OPERATOR_LCURLY;
1668 case '}':
1669 this->add_semi_at_eol_ = true;
1670 return OPERATOR_RCURLY;
1671 case '[':
1672 return OPERATOR_LSQUARE;
1673 case ']':
1674 this->add_semi_at_eol_ = true;
1675 return OPERATOR_RSQUARE;
1676 default:
1677 return OPERATOR_INVALID;
1678 }
1679 }
1680
1681 // Skip a C-style comment.
1682
1683 bool
1684 Lex::skip_c_comment(bool* found_newline)
1685 {
1686 while (true)
1687 {
1688 if (!this->require_line())
1689 {
1690 go_error_at(this->location(), "unterminated comment");
1691 return false;
1692 }
1693
1694 const char* p = this->linebuf_ + this->lineoff_;
1695 const char* pend = this->linebuf_ + this->linesize_;
1696
1697 while (p < pend)
1698 {
1699 if (p[0] == '*' && p + 1 < pend && p[1] == '/')
1700 {
1701 this->lineoff_ = p + 2 - this->linebuf_;
1702 return true;
1703 }
1704
1705 if (p[0] == '\n')
1706 *found_newline = true;
1707
1708 this->lineoff_ = p - this->linebuf_;
1709 unsigned int c;
1710 bool issued_error;
1711 p = this->advance_one_utf8_char(p, &c, &issued_error);
1712 }
1713
1714 this->lineoff_ = p - this->linebuf_;
1715 }
1716 }
1717
1718 // Skip a C++-style comment.
1719
1720 void
1721 Lex::skip_cpp_comment()
1722 {
1723 // Ensure that if EXTERN_ is set, it means that we just saw a
1724 // //extern comment.
1725 this->extern_.clear();
1726
1727 Location loc = this->location();
1728 size_t lineoff = this->lineoff_;
1729
1730 const char* p = this->linebuf_ + lineoff;
1731 const char* pend = this->linebuf_ + this->linesize_;
1732
1733 const char* pcheck = p;
1734 bool saw_error = false;
1735 while (pcheck < pend)
1736 {
1737 this->lineoff_ = pcheck - this->linebuf_;
1738 unsigned int c;
1739 bool issued_error;
1740 pcheck = this->advance_one_utf8_char(pcheck, &c, &issued_error);
1741 if (issued_error)
1742 saw_error = true;
1743 }
1744
1745 if (saw_error)
1746 return;
1747
1748 // Recognize various magic comments at the start of a line.
1749
1750 if (lineoff != 2)
1751 {
1752 // Not at the start of the line. (lineoff == 2 because of the
1753 // two characters in "//").
1754 return;
1755 }
1756
1757 while (pend > p
1758 && (pend[-1] == ' ' || pend[-1] == '\t'
1759 || pend[-1] == '\r' || pend[-1] == '\n'))
1760 --pend;
1761
1762 // A C++ comment at the start of the line of the form
1763 // //line FILE:LINENO
1764 // is interpreted as setting the file name and line number of the
1765 // next source line.
1766 if (pend - p > 5 && memcmp(p, "line ", 5) == 0)
1767 {
1768 p += 5;
1769 while (p < pend && *p == ' ')
1770 ++p;
1771 const char* pcolon = static_cast<const char*>(memchr(p, ':', pend - p));
1772 if (pcolon != NULL
1773 && pcolon[1] >= '0'
1774 && pcolon[1] <= '9')
1775 {
1776 char* plend;
1777 long lineno = strtol(pcolon + 1, &plend, 10);
1778 if (plend > pcolon + 1
1779 && (plend == pend
1780 || *plend < '0'
1781 || *plend > '9')
1782 && lineno > 0
1783 && lineno < 0x7fffffff)
1784 {
1785 unsigned int filelen = pcolon - p;
1786 char* file = new char[filelen + 1];
1787 memcpy(file, p, filelen);
1788 file[filelen] = '\0';
1789
1790 this->linemap_->start_file(file, lineno);
1791 this->lineno_ = lineno - 1;
1792
1793 p = plend;
1794 }
1795 }
1796 return;
1797 }
1798
1799 // As a special gccgo extension, a C++ comment at the start of the
1800 // line of the form
1801 // //extern NAME
1802 // which immediately precedes a function declaration means that the
1803 // external name of the function declaration is NAME. This is
1804 // normally used to permit Go code to call a C function.
1805 if (pend - p > 7 && memcmp(p, "extern ", 7) == 0)
1806 {
1807 p += 7;
1808 while (p < pend && (*p == ' ' || *p == '\t'))
1809 ++p;
1810 if (pend > p)
1811 this->extern_ = std::string(p, pend - p);
1812 return;
1813 }
1814
1815 // All other special comments start with "go:".
1816
1817 if (pend - p < 4 || memcmp(p, "go:", 3) != 0)
1818 return;
1819
1820 const char *ps = p + 3;
1821 while (ps < pend && *ps != ' ' && *ps != '\t')
1822 ++ps;
1823 std::string verb = std::string(p, ps - p);
1824
1825 if (verb == "go:linkname")
1826 {
1827 // As in the gc compiler, set the external link name for a Go symbol.
1828 std::string go_name;
1829 std::string ext_name;
1830 bool is_exported = false;
1831 if (ps < pend)
1832 {
1833 while (ps < pend && (*ps == ' ' || *ps == '\t'))
1834 ++ps;
1835 if (ps < pend)
1836 {
1837 const char* pg = ps;
1838
1839 unsigned int c;
1840 bool issued_error;
1841 ps = this->advance_one_utf8_char(ps, &c, &issued_error);
1842 is_exported = Lex::is_unicode_uppercase(c);
1843
1844 while (ps < pend && *ps != ' ' && *ps != '\t')
1845 ++ps;
1846 if (ps < pend)
1847 go_name = std::string(pg, ps - pg);
1848 while (ps < pend && (*ps == ' ' || *ps == '\t'))
1849 ++ps;
1850 }
1851 if (ps < pend)
1852 {
1853 const char* pc = ps;
1854 while (ps < pend && *ps != ' ' && *ps != '\t')
1855 ++ps;
1856 if (ps <= pend)
1857 ext_name = std::string(pc, ps - pc);
1858 }
1859 if (ps != pend)
1860 {
1861 go_name.clear();
1862 ext_name.clear();
1863 }
1864 }
1865 if (go_name.empty() || ext_name.empty())
1866 go_error_at(loc, "usage: //go:linkname localname linkname");
1867 else
1868 {
1869 if (this->linknames_ == NULL)
1870 this->linknames_ = new Linknames();
1871 (*this->linknames_)[go_name] = Linkname(ext_name, is_exported, loc);
1872 }
1873 }
1874 else if (verb == "go:nointerface")
1875 {
1876 // For field tracking analysis: a //go:nointerface comment means
1877 // that the next interface method should not be stored in the
1878 // type descriptor. This permits it to be discarded if it is
1879 // not needed.
1880 this->pragmas_ |= GOPRAGMA_NOINTERFACE;
1881 }
1882 else if (verb == "go:noescape")
1883 {
1884 // Applies to the next function declaration. Any arguments do
1885 // not escape.
1886 // FIXME: Not implemented.
1887 this->pragmas_ |= GOPRAGMA_NOESCAPE;
1888 }
1889 else if (verb == "go:nosplit")
1890 {
1891 // Applies to the next function. Do not split the stack when
1892 // entering the function.
1893 this->pragmas_ |= GOPRAGMA_NOSPLIT;
1894 }
1895 else if (verb == "go:noinline")
1896 {
1897 // Applies to the next function. Do not inline the function.
1898 this->pragmas_ |= GOPRAGMA_NOINLINE;
1899 }
1900 else if (verb == "go:notinheap")
1901 {
1902 // Applies to the next type. The type does not live in the heap.
1903 this->pragmas_ |= GOPRAGMA_NOTINHEAP;
1904 }
1905 else if (verb == "go:systemstack")
1906 {
1907 // Applies to the next function. It must run on the system stack.
1908 // FIXME: Should only work when compiling the runtime package.
1909 // FIXME: Not implemented.
1910 this->pragmas_ |= GOPRAGMA_SYSTEMSTACK;
1911 }
1912 else if (verb == "go:nowritebarrier")
1913 {
1914 // Applies to the next function. If the function needs to use
1915 // any write barriers, it should emit an error instead.
1916 // FIXME: Should only work when compiling the runtime package.
1917 this->pragmas_ |= GOPRAGMA_NOWRITEBARRIER;
1918 }
1919 else if (verb == "go:nowritebarrierrec")
1920 {
1921 // Applies to the next function. If the function, or any
1922 // function that it calls, needs to use any write barriers, it
1923 // should emit an error instead.
1924 // FIXME: Should only work when compiling the runtime package.
1925 this->pragmas_ |= GOPRAGMA_NOWRITEBARRIERREC;
1926 }
1927 else if (verb == "go:yeswritebarrierrec")
1928 {
1929 // Applies to the next function. Disables go:nowritebarrierrec
1930 // when looking at callees; write barriers are permitted here.
1931 // FIXME: Should only work when compiling the runtime package.
1932 this->pragmas_ |= GOPRAGMA_YESWRITEBARRIERREC;
1933 }
1934 else if (verb == "go:cgo_unsafe_args")
1935 {
1936 // Applies to the next function. Taking the address of any
1937 // argument implies taking the address of all arguments.
1938 // FIXME: Not implemented.
1939 this->pragmas_ |= GOPRAGMA_CGOUNSAFEARGS;
1940 }
1941 else if (verb == "go:uintptrescapes")
1942 {
1943 // Applies to the next function. If an argument is a pointer
1944 // converted to uintptr, then the pointer escapes.
1945 // FIXME: Not implemented.
1946 this->pragmas_ |= GOPRAGMA_UINTPTRESCAPES;
1947 }
1948 }
1949
1950 // The Unicode tables use this struct.
1951
1952 struct Unicode_range
1953 {
1954 // The low end of the range.
1955 unsigned int low;
1956 // The high end of the range.
1957 unsigned int high;
1958 // The stride. This entries represents low, low + stride, low + 2 *
1959 // stride, etc., up to high.
1960 unsigned int stride;
1961 };
1962
1963 // A table of whitespace characters--Unicode code points classified as
1964 // "Space", "C" locale whitespace characters, the "next line" control
1965 // character (0085), the line separator (2028), the paragraph
1966 // separator (2029), and the "zero-width non-break space" (feff).
1967
1968 static const Unicode_range unicode_space[] =
1969 {
1970 { 0x0009, 0x000d, 1 },
1971 { 0x0020, 0x0020, 1 },
1972 { 0x0085, 0x0085, 1 },
1973 { 0x00a0, 0x00a0, 1 },
1974 { 0x1680, 0x1680, 1 },
1975 { 0x180e, 0x180e, 1 },
1976 { 0x2000, 0x200a, 1 },
1977 { 0x2028, 0x2029, 1 },
1978 { 0x202f, 0x202f, 1 },
1979 { 0x205f, 0x205f, 1 },
1980 { 0x3000, 0x3000, 1 },
1981 { 0xfeff, 0xfeff, 1 },
1982 };
1983
1984 // A table of Unicode digits--Unicode code points classified as
1985 // "Digit".
1986
1987 static const Unicode_range unicode_digits[] =
1988 {
1989 { 0x0030, 0x0039, 1},
1990 { 0x0660, 0x0669, 1},
1991 { 0x06f0, 0x06f9, 1},
1992 { 0x07c0, 0x07c9, 1},
1993 { 0x0966, 0x096f, 1},
1994 { 0x09e6, 0x09ef, 1},
1995 { 0x0a66, 0x0a6f, 1},
1996 { 0x0ae6, 0x0aef, 1},
1997 { 0x0b66, 0x0b6f, 1},
1998 { 0x0be6, 0x0bef, 1},
1999 { 0x0c66, 0x0c6f, 1},
2000 { 0x0ce6, 0x0cef, 1},
2001 { 0x0d66, 0x0d6f, 1},
2002 { 0x0e50, 0x0e59, 1},
2003 { 0x0ed0, 0x0ed9, 1},
2004 { 0x0f20, 0x0f29, 1},
2005 { 0x1040, 0x1049, 1},
2006 { 0x17e0, 0x17e9, 1},
2007 { 0x1810, 0x1819, 1},
2008 { 0x1946, 0x194f, 1},
2009 { 0x19d0, 0x19d9, 1},
2010 { 0x1b50, 0x1b59, 1},
2011 { 0xff10, 0xff19, 1},
2012 { 0x104a0, 0x104a9, 1},
2013 { 0x1d7ce, 0x1d7ff, 1},
2014 };
2015
2016 // A table of Unicode letters--Unicode code points classified as
2017 // "Letter".
2018
2019 static const Unicode_range unicode_letters[] =
2020 {
2021 { 0x0041, 0x005a, 1},
2022 { 0x0061, 0x007a, 1},
2023 { 0x00aa, 0x00b5, 11},
2024 { 0x00ba, 0x00c0, 6},
2025 { 0x00c1, 0x00d6, 1},
2026 { 0x00d8, 0x00f6, 1},
2027 { 0x00f8, 0x02c1, 1},
2028 { 0x02c6, 0x02d1, 1},
2029 { 0x02e0, 0x02e4, 1},
2030 { 0x02ec, 0x02ee, 2},
2031 { 0x0370, 0x0374, 1},
2032 { 0x0376, 0x0377, 1},
2033 { 0x037a, 0x037d, 1},
2034 { 0x037f, 0x0386, 7},
2035 { 0x0388, 0x038a, 1},
2036 { 0x038c, 0x038e, 2},
2037 { 0x038f, 0x03a1, 1},
2038 { 0x03a3, 0x03f5, 1},
2039 { 0x03f7, 0x0481, 1},
2040 { 0x048a, 0x052f, 1},
2041 { 0x0531, 0x0556, 1},
2042 { 0x0559, 0x0561, 8},
2043 { 0x0562, 0x0587, 1},
2044 { 0x05d0, 0x05ea, 1},
2045 { 0x05f0, 0x05f2, 1},
2046 { 0x0620, 0x064a, 1},
2047 { 0x066e, 0x066f, 1},
2048 { 0x0671, 0x06d3, 1},
2049 { 0x06d5, 0x06e5, 16},
2050 { 0x06e6, 0x06ee, 8},
2051 { 0x06ef, 0x06fa, 11},
2052 { 0x06fb, 0x06fc, 1},
2053 { 0x06ff, 0x0710, 17},
2054 { 0x0712, 0x072f, 1},
2055 { 0x074d, 0x07a5, 1},
2056 { 0x07b1, 0x07ca, 25},
2057 { 0x07cb, 0x07ea, 1},
2058 { 0x07f4, 0x07f5, 1},
2059 { 0x07fa, 0x0800, 6},
2060 { 0x0801, 0x0815, 1},
2061 { 0x081a, 0x0824, 10},
2062 { 0x0828, 0x0840, 24},
2063 { 0x0841, 0x0858, 1},
2064 { 0x08a0, 0x08b4, 1},
2065 { 0x0904, 0x0939, 1},
2066 { 0x093d, 0x0950, 19},
2067 { 0x0958, 0x0961, 1},
2068 { 0x0971, 0x0980, 1},
2069 { 0x0985, 0x098c, 1},
2070 { 0x098f, 0x0990, 1},
2071 { 0x0993, 0x09a8, 1},
2072 { 0x09aa, 0x09b0, 1},
2073 { 0x09b2, 0x09b6, 4},
2074 { 0x09b7, 0x09b9, 1},
2075 { 0x09bd, 0x09ce, 17},
2076 { 0x09dc, 0x09dd, 1},
2077 { 0x09df, 0x09e1, 1},
2078 { 0x09f0, 0x09f1, 1},
2079 { 0x0a05, 0x0a0a, 1},
2080 { 0x0a0f, 0x0a10, 1},
2081 { 0x0a13, 0x0a28, 1},
2082 { 0x0a2a, 0x0a30, 1},
2083 { 0x0a32, 0x0a33, 1},
2084 { 0x0a35, 0x0a36, 1},
2085 { 0x0a38, 0x0a39, 1},
2086 { 0x0a59, 0x0a5c, 1},
2087 { 0x0a5e, 0x0a72, 20},
2088 { 0x0a73, 0x0a74, 1},
2089 { 0x0a85, 0x0a8d, 1},
2090 { 0x0a8f, 0x0a91, 1},
2091 { 0x0a93, 0x0aa8, 1},
2092 { 0x0aaa, 0x0ab0, 1},
2093 { 0x0ab2, 0x0ab3, 1},
2094 { 0x0ab5, 0x0ab9, 1},
2095 { 0x0abd, 0x0ad0, 19},
2096 { 0x0ae0, 0x0ae1, 1},
2097 { 0x0af9, 0x0b05, 12},
2098 { 0x0b06, 0x0b0c, 1},
2099 { 0x0b0f, 0x0b10, 1},
2100 { 0x0b13, 0x0b28, 1},
2101 { 0x0b2a, 0x0b30, 1},
2102 { 0x0b32, 0x0b33, 1},
2103 { 0x0b35, 0x0b39, 1},
2104 { 0x0b3d, 0x0b5c, 31},
2105 { 0x0b5d, 0x0b5f, 2},
2106 { 0x0b60, 0x0b61, 1},
2107 { 0x0b71, 0x0b83, 18},
2108 { 0x0b85, 0x0b8a, 1},
2109 { 0x0b8e, 0x0b90, 1},
2110 { 0x0b92, 0x0b95, 1},
2111 { 0x0b99, 0x0b9a, 1},
2112 { 0x0b9c, 0x0b9e, 2},
2113 { 0x0b9f, 0x0ba3, 4},
2114 { 0x0ba4, 0x0ba8, 4},
2115 { 0x0ba9, 0x0baa, 1},
2116 { 0x0bae, 0x0bb9, 1},
2117 { 0x0bd0, 0x0c05, 53},
2118 { 0x0c06, 0x0c0c, 1},
2119 { 0x0c0e, 0x0c10, 1},
2120 { 0x0c12, 0x0c28, 1},
2121 { 0x0c2a, 0x0c39, 1},
2122 { 0x0c3d, 0x0c58, 27},
2123 { 0x0c59, 0x0c5a, 1},
2124 { 0x0c60, 0x0c61, 1},
2125 { 0x0c85, 0x0c8c, 1},
2126 { 0x0c8e, 0x0c90, 1},
2127 { 0x0c92, 0x0ca8, 1},
2128 { 0x0caa, 0x0cb3, 1},
2129 { 0x0cb5, 0x0cb9, 1},
2130 { 0x0cbd, 0x0cde, 33},
2131 { 0x0ce0, 0x0ce1, 1},
2132 { 0x0cf1, 0x0cf2, 1},
2133 { 0x0d05, 0x0d0c, 1},
2134 { 0x0d0e, 0x0d10, 1},
2135 { 0x0d12, 0x0d3a, 1},
2136 { 0x0d3d, 0x0d5f, 17},
2137 { 0x0d60, 0x0d61, 1},
2138 { 0x0d7a, 0x0d7f, 1},
2139 { 0x0d85, 0x0d96, 1},
2140 { 0x0d9a, 0x0db1, 1},
2141 { 0x0db3, 0x0dbb, 1},
2142 { 0x0dbd, 0x0dc0, 3},
2143 { 0x0dc1, 0x0dc6, 1},
2144 { 0x0e01, 0x0e30, 1},
2145 { 0x0e32, 0x0e33, 1},
2146 { 0x0e40, 0x0e46, 1},
2147 { 0x0e81, 0x0e82, 1},
2148 { 0x0e84, 0x0e87, 3},
2149 { 0x0e88, 0x0e8a, 2},
2150 { 0x0e8d, 0x0e94, 7},
2151 { 0x0e95, 0x0e97, 1},
2152 { 0x0e99, 0x0e9f, 1},
2153 { 0x0ea1, 0x0ea3, 1},
2154 { 0x0ea5, 0x0ea7, 2},
2155 { 0x0eaa, 0x0eab, 1},
2156 { 0x0ead, 0x0eb0, 1},
2157 { 0x0eb2, 0x0eb3, 1},
2158 { 0x0ebd, 0x0ec0, 3},
2159 { 0x0ec1, 0x0ec4, 1},
2160 { 0x0ec6, 0x0edc, 22},
2161 { 0x0edd, 0x0edf, 1},
2162 { 0x0f00, 0x0f40, 64},
2163 { 0x0f41, 0x0f47, 1},
2164 { 0x0f49, 0x0f6c, 1},
2165 { 0x0f88, 0x0f8c, 1},
2166 { 0x1000, 0x102a, 1},
2167 { 0x103f, 0x1050, 17},
2168 { 0x1051, 0x1055, 1},
2169 { 0x105a, 0x105d, 1},
2170 { 0x1061, 0x1065, 4},
2171 { 0x1066, 0x106e, 8},
2172 { 0x106f, 0x1070, 1},
2173 { 0x1075, 0x1081, 1},
2174 { 0x108e, 0x10a0, 18},
2175 { 0x10a1, 0x10c5, 1},
2176 { 0x10c7, 0x10cd, 6},
2177 { 0x10d0, 0x10fa, 1},
2178 { 0x10fc, 0x1248, 1},
2179 { 0x124a, 0x124d, 1},
2180 { 0x1250, 0x1256, 1},
2181 { 0x1258, 0x125a, 2},
2182 { 0x125b, 0x125d, 1},
2183 { 0x1260, 0x1288, 1},
2184 { 0x128a, 0x128d, 1},
2185 { 0x1290, 0x12b0, 1},
2186 { 0x12b2, 0x12b5, 1},
2187 { 0x12b8, 0x12be, 1},
2188 { 0x12c0, 0x12c2, 2},
2189 { 0x12c3, 0x12c5, 1},
2190 { 0x12c8, 0x12d6, 1},
2191 { 0x12d8, 0x1310, 1},
2192 { 0x1312, 0x1315, 1},
2193 { 0x1318, 0x135a, 1},
2194 { 0x1380, 0x138f, 1},
2195 { 0x13a0, 0x13f5, 1},
2196 { 0x13f8, 0x13fd, 1},
2197 { 0x1401, 0x166c, 1},
2198 { 0x166f, 0x167f, 1},
2199 { 0x1681, 0x169a, 1},
2200 { 0x16a0, 0x16ea, 1},
2201 { 0x16f1, 0x16f8, 1},
2202 { 0x1700, 0x170c, 1},
2203 { 0x170e, 0x1711, 1},
2204 { 0x1720, 0x1731, 1},
2205 { 0x1740, 0x1751, 1},
2206 { 0x1760, 0x176c, 1},
2207 { 0x176e, 0x1770, 1},
2208 { 0x1780, 0x17b3, 1},
2209 { 0x17d7, 0x17dc, 5},
2210 { 0x1820, 0x1877, 1},
2211 { 0x1880, 0x18a8, 1},
2212 { 0x18aa, 0x18b0, 6},
2213 { 0x18b1, 0x18f5, 1},
2214 { 0x1900, 0x191e, 1},
2215 { 0x1950, 0x196d, 1},
2216 { 0x1970, 0x1974, 1},
2217 { 0x1980, 0x19ab, 1},
2218 { 0x19b0, 0x19c9, 1},
2219 { 0x1a00, 0x1a16, 1},
2220 { 0x1a20, 0x1a54, 1},
2221 { 0x1aa7, 0x1b05, 94},
2222 { 0x1b06, 0x1b33, 1},
2223 { 0x1b45, 0x1b4b, 1},
2224 { 0x1b83, 0x1ba0, 1},
2225 { 0x1bae, 0x1baf, 1},
2226 { 0x1bba, 0x1be5, 1},
2227 { 0x1c00, 0x1c23, 1},
2228 { 0x1c4d, 0x1c4f, 1},
2229 { 0x1c5a, 0x1c7d, 1},
2230 { 0x1ce9, 0x1cec, 1},
2231 { 0x1cee, 0x1cf1, 1},
2232 { 0x1cf5, 0x1cf6, 1},
2233 { 0x1d00, 0x1dbf, 1},
2234 { 0x1e00, 0x1f15, 1},
2235 { 0x1f18, 0x1f1d, 1},
2236 { 0x1f20, 0x1f45, 1},
2237 { 0x1f48, 0x1f4d, 1},
2238 { 0x1f50, 0x1f57, 1},
2239 { 0x1f59, 0x1f5f, 2},
2240 { 0x1f60, 0x1f7d, 1},
2241 { 0x1f80, 0x1fb4, 1},
2242 { 0x1fb6, 0x1fbc, 1},
2243 { 0x1fbe, 0x1fc2, 4},
2244 { 0x1fc3, 0x1fc4, 1},
2245 { 0x1fc6, 0x1fcc, 1},
2246 { 0x1fd0, 0x1fd3, 1},
2247 { 0x1fd6, 0x1fdb, 1},
2248 { 0x1fe0, 0x1fec, 1},
2249 { 0x1ff2, 0x1ff4, 1},
2250 { 0x1ff6, 0x1ffc, 1},
2251 { 0x2071, 0x207f, 14},
2252 { 0x2090, 0x209c, 1},
2253 { 0x2102, 0x2107, 5},
2254 { 0x210a, 0x2113, 1},
2255 { 0x2115, 0x2119, 4},
2256 { 0x211a, 0x211d, 1},
2257 { 0x2124, 0x212a, 2},
2258 { 0x212b, 0x212d, 1},
2259 { 0x212f, 0x2139, 1},
2260 { 0x213c, 0x213f, 1},
2261 { 0x2145, 0x2149, 1},
2262 { 0x214e, 0x2183, 53},
2263 { 0x2184, 0x2c00, 2684},
2264 { 0x2c01, 0x2c2e, 1},
2265 { 0x2c30, 0x2c5e, 1},
2266 { 0x2c60, 0x2ce4, 1},
2267 { 0x2ceb, 0x2cee, 1},
2268 { 0x2cf2, 0x2cf3, 1},
2269 { 0x2d00, 0x2d25, 1},
2270 { 0x2d27, 0x2d2d, 6},
2271 { 0x2d30, 0x2d67, 1},
2272 { 0x2d6f, 0x2d80, 17},
2273 { 0x2d81, 0x2d96, 1},
2274 { 0x2da0, 0x2da6, 1},
2275 { 0x2da8, 0x2dae, 1},
2276 { 0x2db0, 0x2db6, 1},
2277 { 0x2db8, 0x2dbe, 1},
2278 { 0x2dc0, 0x2dc6, 1},
2279 { 0x2dc8, 0x2dce, 1},
2280 { 0x2dd0, 0x2dd6, 1},
2281 { 0x2dd8, 0x2dde, 1},
2282 { 0x2e2f, 0x3005, 470},
2283 { 0x3006, 0x3031, 43},
2284 { 0x3032, 0x3035, 1},
2285 { 0x303b, 0x303c, 1},
2286 { 0x3041, 0x3096, 1},
2287 { 0x309d, 0x309f, 1},
2288 { 0x30a1, 0x30fa, 1},
2289 { 0x30fc, 0x30ff, 1},
2290 { 0x3105, 0x312d, 1},
2291 { 0x3131, 0x318e, 1},
2292 { 0x31a0, 0x31ba, 1},
2293 { 0x31f0, 0x31ff, 1},
2294 { 0x3400, 0x4db5, 1},
2295 { 0x4e00, 0x9fd5, 1},
2296 { 0xa000, 0xa48c, 1},
2297 { 0xa4d0, 0xa4fd, 1},
2298 { 0xa500, 0xa60c, 1},
2299 { 0xa610, 0xa61f, 1},
2300 { 0xa62a, 0xa62b, 1},
2301 { 0xa640, 0xa66e, 1},
2302 { 0xa67f, 0xa69d, 1},
2303 { 0xa6a0, 0xa6e5, 1},
2304 { 0xa717, 0xa71f, 1},
2305 { 0xa722, 0xa788, 1},
2306 { 0xa78b, 0xa7ad, 1},
2307 { 0xa7b0, 0xa7b7, 1},
2308 { 0xa7f7, 0xa801, 1},
2309 { 0xa803, 0xa805, 1},
2310 { 0xa807, 0xa80a, 1},
2311 { 0xa80c, 0xa822, 1},
2312 { 0xa840, 0xa873, 1},
2313 { 0xa882, 0xa8b3, 1},
2314 { 0xa8f2, 0xa8f7, 1},
2315 { 0xa8fb, 0xa8fd, 2},
2316 { 0xa90a, 0xa925, 1},
2317 { 0xa930, 0xa946, 1},
2318 { 0xa960, 0xa97c, 1},
2319 { 0xa984, 0xa9b2, 1},
2320 { 0xa9cf, 0xa9e0, 17},
2321 { 0xa9e1, 0xa9e4, 1},
2322 { 0xa9e6, 0xa9ef, 1},
2323 { 0xa9fa, 0xa9fe, 1},
2324 { 0xaa00, 0xaa28, 1},
2325 { 0xaa40, 0xaa42, 1},
2326 { 0xaa44, 0xaa4b, 1},
2327 { 0xaa60, 0xaa76, 1},
2328 { 0xaa7a, 0xaa7e, 4},
2329 { 0xaa7f, 0xaaaf, 1},
2330 { 0xaab1, 0xaab5, 4},
2331 { 0xaab6, 0xaab9, 3},
2332 { 0xaaba, 0xaabd, 1},
2333 { 0xaac0, 0xaac2, 2},
2334 { 0xaadb, 0xaadd, 1},
2335 { 0xaae0, 0xaaea, 1},
2336 { 0xaaf2, 0xaaf4, 1},
2337 { 0xab01, 0xab06, 1},
2338 { 0xab09, 0xab0e, 1},
2339 { 0xab11, 0xab16, 1},
2340 { 0xab20, 0xab26, 1},
2341 { 0xab28, 0xab2e, 1},
2342 { 0xab30, 0xab5a, 1},
2343 { 0xab5c, 0xab65, 1},
2344 { 0xab70, 0xabe2, 1},
2345 { 0xac00, 0xd7a3, 1},
2346 { 0xd7b0, 0xd7c6, 1},
2347 { 0xd7cb, 0xd7fb, 1},
2348 { 0xf900, 0xfa6d, 1},
2349 { 0xfa70, 0xfad9, 1},
2350 { 0xfb00, 0xfb06, 1},
2351 { 0xfb13, 0xfb17, 1},
2352 { 0xfb1d, 0xfb1f, 2},
2353 { 0xfb20, 0xfb28, 1},
2354 { 0xfb2a, 0xfb36, 1},
2355 { 0xfb38, 0xfb3c, 1},
2356 { 0xfb3e, 0xfb40, 2},
2357 { 0xfb41, 0xfb43, 2},
2358 { 0xfb44, 0xfb46, 2},
2359 { 0xfb47, 0xfbb1, 1},
2360 { 0xfbd3, 0xfd3d, 1},
2361 { 0xfd50, 0xfd8f, 1},
2362 { 0xfd92, 0xfdc7, 1},
2363 { 0xfdf0, 0xfdfb, 1},
2364 { 0xfe70, 0xfe74, 1},
2365 { 0xfe76, 0xfefc, 1},
2366 { 0xff21, 0xff3a, 1},
2367 { 0xff41, 0xff5a, 1},
2368 { 0xff66, 0xffbe, 1},
2369 { 0xffc2, 0xffc7, 1},
2370 { 0xffca, 0xffcf, 1},
2371 { 0xffd2, 0xffd7, 1},
2372 { 0xffda, 0xffdc, 1},
2373 { 0x10000, 0x1000b, 1},
2374 { 0x1000d, 0x10026, 1},
2375 { 0x10028, 0x1003a, 1},
2376 { 0x1003c, 0x1003d, 1},
2377 { 0x1003f, 0x1004d, 1},
2378 { 0x10050, 0x1005d, 1},
2379 { 0x10080, 0x100fa, 1},
2380 { 0x10280, 0x1029c, 1},
2381 { 0x102a0, 0x102d0, 1},
2382 { 0x10300, 0x1031f, 1},
2383 { 0x10330, 0x10340, 1},
2384 { 0x10342, 0x10349, 1},
2385 { 0x10350, 0x10375, 1},
2386 { 0x10380, 0x1039d, 1},
2387 { 0x103a0, 0x103c3, 1},
2388 { 0x103c8, 0x103cf, 1},
2389 { 0x10400, 0x1049d, 1},
2390 { 0x10500, 0x10527, 1},
2391 { 0x10530, 0x10563, 1},
2392 { 0x10600, 0x10736, 1},
2393 { 0x10740, 0x10755, 1},
2394 { 0x10760, 0x10767, 1},
2395 { 0x10800, 0x10805, 1},
2396 { 0x10808, 0x1080a, 2},
2397 { 0x1080b, 0x10835, 1},
2398 { 0x10837, 0x10838, 1},
2399 { 0x1083c, 0x1083f, 3},
2400 { 0x10840, 0x10855, 1},
2401 { 0x10860, 0x10876, 1},
2402 { 0x10880, 0x1089e, 1},
2403 { 0x108e0, 0x108f2, 1},
2404 { 0x108f4, 0x108f5, 1},
2405 { 0x10900, 0x10915, 1},
2406 { 0x10920, 0x10939, 1},
2407 { 0x10980, 0x109b7, 1},
2408 { 0x109be, 0x109bf, 1},
2409 { 0x10a00, 0x10a10, 16},
2410 { 0x10a11, 0x10a13, 1},
2411 { 0x10a15, 0x10a17, 1},
2412 { 0x10a19, 0x10a33, 1},
2413 { 0x10a60, 0x10a7c, 1},
2414 { 0x10a80, 0x10a9c, 1},
2415 { 0x10ac0, 0x10ac7, 1},
2416 { 0x10ac9, 0x10ae4, 1},
2417 { 0x10b00, 0x10b35, 1},
2418 { 0x10b40, 0x10b55, 1},
2419 { 0x10b60, 0x10b72, 1},
2420 { 0x10b80, 0x10b91, 1},
2421 { 0x10c00, 0x10c48, 1},
2422 { 0x10c80, 0x10cb2, 1},
2423 { 0x10cc0, 0x10cf2, 1},
2424 { 0x11003, 0x11037, 1},
2425 { 0x11083, 0x110af, 1},
2426 { 0x110d0, 0x110e8, 1},
2427 { 0x11103, 0x11126, 1},
2428 { 0x11150, 0x11172, 1},
2429 { 0x11176, 0x11183, 13},
2430 { 0x11184, 0x111b2, 1},
2431 { 0x111c1, 0x111c4, 1},
2432 { 0x111da, 0x111dc, 2},
2433 { 0x11200, 0x11211, 1},
2434 { 0x11213, 0x1122b, 1},
2435 { 0x11280, 0x11286, 1},
2436 { 0x11288, 0x1128a, 2},
2437 { 0x1128b, 0x1128d, 1},
2438 { 0x1128f, 0x1129d, 1},
2439 { 0x1129f, 0x112a8, 1},
2440 { 0x112b0, 0x112de, 1},
2441 { 0x11305, 0x1130c, 1},
2442 { 0x1130f, 0x11310, 1},
2443 { 0x11313, 0x11328, 1},
2444 { 0x1132a, 0x11330, 1},
2445 { 0x11332, 0x11333, 1},
2446 { 0x11335, 0x11339, 1},
2447 { 0x1133d, 0x11350, 19},
2448 { 0x1135d, 0x11361, 1},
2449 { 0x11480, 0x114af, 1},
2450 { 0x114c4, 0x114c5, 1},
2451 { 0x114c7, 0x11580, 185},
2452 { 0x11581, 0x115ae, 1},
2453 { 0x115d8, 0x115db, 1},
2454 { 0x11600, 0x1162f, 1},
2455 { 0x11644, 0x11680, 60},
2456 { 0x11681, 0x116aa, 1},
2457 { 0x11700, 0x11719, 1},
2458 { 0x118a0, 0x118df, 1},
2459 { 0x118ff, 0x11ac0, 449},
2460 { 0x11ac1, 0x11af8, 1},
2461 { 0x12000, 0x12399, 1},
2462 { 0x12480, 0x12543, 1},
2463 { 0x13000, 0x1342e, 1},
2464 { 0x14400, 0x14646, 1},
2465 { 0x16800, 0x16a38, 1},
2466 { 0x16a40, 0x16a5e, 1},
2467 { 0x16ad0, 0x16aed, 1},
2468 { 0x16b00, 0x16b2f, 1},
2469 { 0x16b40, 0x16b43, 1},
2470 { 0x16b63, 0x16b77, 1},
2471 { 0x16b7d, 0x16b8f, 1},
2472 { 0x16f00, 0x16f44, 1},
2473 { 0x16f50, 0x16f93, 67},
2474 { 0x16f94, 0x16f9f, 1},
2475 { 0x1b000, 0x1b001, 1},
2476 { 0x1bc00, 0x1bc6a, 1},
2477 { 0x1bc70, 0x1bc7c, 1},
2478 { 0x1bc80, 0x1bc88, 1},
2479 { 0x1bc90, 0x1bc99, 1},
2480 { 0x1d400, 0x1d454, 1},
2481 { 0x1d456, 0x1d49c, 1},
2482 { 0x1d49e, 0x1d49f, 1},
2483 { 0x1d4a2, 0x1d4a5, 3},
2484 { 0x1d4a6, 0x1d4a9, 3},
2485 { 0x1d4aa, 0x1d4ac, 1},
2486 { 0x1d4ae, 0x1d4b9, 1},
2487 { 0x1d4bb, 0x1d4bd, 2},
2488 { 0x1d4be, 0x1d4c3, 1},
2489 { 0x1d4c5, 0x1d505, 1},
2490 { 0x1d507, 0x1d50a, 1},
2491 { 0x1d50d, 0x1d514, 1},
2492 { 0x1d516, 0x1d51c, 1},
2493 { 0x1d51e, 0x1d539, 1},
2494 { 0x1d53b, 0x1d53e, 1},
2495 { 0x1d540, 0x1d544, 1},
2496 { 0x1d546, 0x1d54a, 4},
2497 { 0x1d54b, 0x1d550, 1},
2498 { 0x1d552, 0x1d6a5, 1},
2499 { 0x1d6a8, 0x1d6c0, 1},
2500 { 0x1d6c2, 0x1d6da, 1},
2501 { 0x1d6dc, 0x1d6fa, 1},
2502 { 0x1d6fc, 0x1d714, 1},
2503 { 0x1d716, 0x1d734, 1},
2504 { 0x1d736, 0x1d74e, 1},
2505 { 0x1d750, 0x1d76e, 1},
2506 { 0x1d770, 0x1d788, 1},
2507 { 0x1d78a, 0x1d7a8, 1},
2508 { 0x1d7aa, 0x1d7c2, 1},
2509 { 0x1d7c4, 0x1d7cb, 1},
2510 { 0x1e800, 0x1e8c4, 1},
2511 { 0x1ee00, 0x1ee03, 1},
2512 { 0x1ee05, 0x1ee1f, 1},
2513 { 0x1ee21, 0x1ee22, 1},
2514 { 0x1ee24, 0x1ee27, 3},
2515 { 0x1ee29, 0x1ee32, 1},
2516 { 0x1ee34, 0x1ee37, 1},
2517 { 0x1ee39, 0x1ee3b, 2},
2518 { 0x1ee42, 0x1ee47, 5},
2519 { 0x1ee49, 0x1ee4d, 2},
2520 { 0x1ee4e, 0x1ee4f, 1},
2521 { 0x1ee51, 0x1ee52, 1},
2522 { 0x1ee54, 0x1ee57, 3},
2523 { 0x1ee59, 0x1ee61, 2},
2524 { 0x1ee62, 0x1ee64, 2},
2525 { 0x1ee67, 0x1ee6a, 1},
2526 { 0x1ee6c, 0x1ee72, 1},
2527 { 0x1ee74, 0x1ee77, 1},
2528 { 0x1ee79, 0x1ee7c, 1},
2529 { 0x1ee7e, 0x1ee80, 2},
2530 { 0x1ee81, 0x1ee89, 1},
2531 { 0x1ee8b, 0x1ee9b, 1},
2532 { 0x1eea1, 0x1eea3, 1},
2533 { 0x1eea5, 0x1eea9, 1},
2534 { 0x1eeab, 0x1eebb, 1},
2535 { 0x20000, 0x2a6d6, 1},
2536 { 0x2a700, 0x2b734, 1},
2537 { 0x2b740, 0x2b81d, 1},
2538 { 0x2b820, 0x2cea1, 1},
2539 { 0x2f800, 0x2fa1d, 1},
2540 };
2541
2542 // A table of Unicode uppercase letters--Unicode code points
2543 // classified as "Letter, uppercase".
2544
2545 static const Unicode_range unicode_uppercase_letters[] =
2546 {
2547 { 0x0041, 0x005a, 1},
2548 { 0x00c0, 0x00d6, 1},
2549 { 0x00d8, 0x00de, 1},
2550 { 0x0100, 0x0136, 2},
2551 { 0x0139, 0x0147, 2},
2552 { 0x014a, 0x0178, 2},
2553 { 0x0179, 0x017d, 2},
2554 { 0x0181, 0x0182, 1},
2555 { 0x0184, 0x0186, 2},
2556 { 0x0187, 0x0189, 2},
2557 { 0x018a, 0x018b, 1},
2558 { 0x018e, 0x0191, 1},
2559 { 0x0193, 0x0194, 1},
2560 { 0x0196, 0x0198, 1},
2561 { 0x019c, 0x019d, 1},
2562 { 0x019f, 0x01a0, 1},
2563 { 0x01a2, 0x01a6, 2},
2564 { 0x01a7, 0x01a9, 2},
2565 { 0x01ac, 0x01ae, 2},
2566 { 0x01af, 0x01b1, 2},
2567 { 0x01b2, 0x01b3, 1},
2568 { 0x01b5, 0x01b7, 2},
2569 { 0x01b8, 0x01bc, 4},
2570 { 0x01c4, 0x01cd, 3},
2571 { 0x01cf, 0x01db, 2},
2572 { 0x01de, 0x01ee, 2},
2573 { 0x01f1, 0x01f4, 3},
2574 { 0x01f6, 0x01f8, 1},
2575 { 0x01fa, 0x0232, 2},
2576 { 0x023a, 0x023b, 1},
2577 { 0x023d, 0x023e, 1},
2578 { 0x0241, 0x0243, 2},
2579 { 0x0244, 0x0246, 1},
2580 { 0x0248, 0x024e, 2},
2581 { 0x0370, 0x0372, 2},
2582 { 0x0376, 0x037f, 9},
2583 { 0x0386, 0x0388, 2},
2584 { 0x0389, 0x038a, 1},
2585 { 0x038c, 0x038e, 2},
2586 { 0x038f, 0x0391, 2},
2587 { 0x0392, 0x03a1, 1},
2588 { 0x03a3, 0x03ab, 1},
2589 { 0x03cf, 0x03d2, 3},
2590 { 0x03d3, 0x03d4, 1},
2591 { 0x03d8, 0x03ee, 2},
2592 { 0x03f4, 0x03f7, 3},
2593 { 0x03f9, 0x03fa, 1},
2594 { 0x03fd, 0x042f, 1},
2595 { 0x0460, 0x0480, 2},
2596 { 0x048a, 0x04c0, 2},
2597 { 0x04c1, 0x04cd, 2},
2598 { 0x04d0, 0x052e, 2},
2599 { 0x0531, 0x0556, 1},
2600 { 0x10a0, 0x10c5, 1},
2601 { 0x10c7, 0x10cd, 6},
2602 { 0x1e00, 0x1e94, 2},
2603 { 0x1e9e, 0x1efe, 2},
2604 { 0x1f08, 0x1f0f, 1},
2605 { 0x1f18, 0x1f1d, 1},
2606 { 0x1f28, 0x1f2f, 1},
2607 { 0x1f38, 0x1f3f, 1},
2608 { 0x1f48, 0x1f4d, 1},
2609 { 0x1f59, 0x1f5f, 2},
2610 { 0x1f68, 0x1f6f, 1},
2611 { 0x1fb8, 0x1fbb, 1},
2612 { 0x1fc8, 0x1fcb, 1},
2613 { 0x1fd8, 0x1fdb, 1},
2614 { 0x1fe8, 0x1fec, 1},
2615 { 0x1ff8, 0x1ffb, 1},
2616 { 0x2102, 0x2107, 5},
2617 { 0x210b, 0x210d, 1},
2618 { 0x2110, 0x2112, 1},
2619 { 0x2115, 0x2119, 4},
2620 { 0x211a, 0x211d, 1},
2621 { 0x2124, 0x212a, 2},
2622 { 0x212b, 0x212d, 1},
2623 { 0x2130, 0x2133, 1},
2624 { 0x213e, 0x213f, 1},
2625 { 0x2145, 0x2183, 62},
2626 { 0x2c00, 0x2c2e, 1},
2627 { 0x2c60, 0x2c62, 2},
2628 { 0x2c63, 0x2c64, 1},
2629 { 0x2c67, 0x2c6d, 2},
2630 { 0x2c6e, 0x2c70, 1},
2631 { 0x2c72, 0x2c75, 3},
2632 { 0x2c7e, 0x2c80, 1},
2633 { 0x2c82, 0x2ce2, 2},
2634 { 0x2ceb, 0x2ced, 2},
2635 { 0x2cf2, 0xa640, 31054},
2636 { 0xa642, 0xa66c, 2},
2637 { 0xa680, 0xa69a, 2},
2638 { 0xa722, 0xa72e, 2},
2639 { 0xa732, 0xa76e, 2},
2640 { 0xa779, 0xa77d, 2},
2641 { 0xa77e, 0xa786, 2},
2642 { 0xa78b, 0xa78d, 2},
2643 { 0xa790, 0xa792, 2},
2644 { 0xa796, 0xa7aa, 2},
2645 { 0xa7ab, 0xa7ad, 1},
2646 { 0xa7b0, 0xa7b1, 1},
2647 { 0xff21, 0xff3a, 1},
2648 { 0x10400, 0x10427, 1},
2649 { 0x118a0, 0x118bf, 1},
2650 { 0x1d400, 0x1d419, 1},
2651 { 0x1d434, 0x1d44d, 1},
2652 { 0x1d468, 0x1d481, 1},
2653 { 0x1d49c, 0x1d49e, 2},
2654 { 0x1d49f, 0x1d4a5, 3},
2655 { 0x1d4a6, 0x1d4a9, 3},
2656 { 0x1d4aa, 0x1d4ac, 1},
2657 { 0x1d4ae, 0x1d4b5, 1},
2658 { 0x1d4d0, 0x1d4e9, 1},
2659 { 0x1d504, 0x1d505, 1},
2660 { 0x1d507, 0x1d50a, 1},
2661 { 0x1d50d, 0x1d514, 1},
2662 { 0x1d516, 0x1d51c, 1},
2663 { 0x1d538, 0x1d539, 1},
2664 { 0x1d53b, 0x1d53e, 1},
2665 { 0x1d540, 0x1d544, 1},
2666 { 0x1d546, 0x1d54a, 4},
2667 { 0x1d54b, 0x1d550, 1},
2668 { 0x1d56c, 0x1d585, 1},
2669 { 0x1d5a0, 0x1d5b9, 1},
2670 { 0x1d5d4, 0x1d5ed, 1},
2671 { 0x1d608, 0x1d621, 1},
2672 { 0x1d63c, 0x1d655, 1},
2673 { 0x1d670, 0x1d689, 1},
2674 { 0x1d6a8, 0x1d6c0, 1},
2675 { 0x1d6e2, 0x1d6fa, 1},
2676 { 0x1d71c, 0x1d734, 1},
2677 { 0x1d756, 0x1d76e, 1},
2678 { 0x1d790, 0x1d7a8, 1},
2679 { 0x1d7ca, 0x1d7ca, 1},
2680 };
2681
2682 // Return true if C is in RANGES.
2683
2684 bool
2685 Lex::is_in_unicode_range(unsigned int c, const Unicode_range* ranges,
2686 size_t range_size)
2687 {
2688 if (c < 0x100)
2689 {
2690 // The common case is a small value, and we know that it will be
2691 // in the first few entries of the table. Do a linear scan
2692 // rather than a binary search.
2693 for (size_t i = 0; i < range_size; ++i)
2694 {
2695 const Unicode_range* p = &ranges[i];
2696 if (c <= p->high)
2697 {
2698 if (c < p->low)
2699 return false;
2700 return (c - p->low) % p->stride == 0;
2701 }
2702 }
2703 return false;
2704 }
2705 else
2706 {
2707 size_t lo = 0;
2708 size_t hi = range_size;
2709 while (lo < hi)
2710 {
2711 size_t mid = lo + (hi - lo) / 2;
2712 const Unicode_range* p = &ranges[mid];
2713 if (c < p->low)
2714 hi = mid;
2715 else if (c > p->high)
2716 lo = mid + 1;
2717 else
2718 return (c - p->low) % p->stride == 0;
2719 }
2720 return false;
2721 }
2722 }
2723
2724 // Return whether C is a space character.
2725
2726 bool
2727 Lex::is_unicode_space(unsigned int c)
2728 {
2729 return Lex::is_in_unicode_range(c, unicode_space,
2730 ARRAY_SIZE(unicode_space));
2731 }
2732
2733 // Return whether C is a Unicode digit--a Unicode code point
2734 // classified as "Digit".
2735
2736 bool
2737 Lex::is_unicode_digit(unsigned int c)
2738 {
2739 return Lex::is_in_unicode_range(c, unicode_digits,
2740 ARRAY_SIZE(unicode_digits));
2741 }
2742
2743 // Return whether C is a Unicode letter--a Unicode code point
2744 // classified as "Letter".
2745
2746 bool
2747 Lex::is_unicode_letter(unsigned int c)
2748 {
2749 return Lex::is_in_unicode_range(c, unicode_letters,
2750 ARRAY_SIZE(unicode_letters));
2751 }
2752
2753 // Return whether C is a Unicode uppercase letter. a Unicode code
2754 // point classified as "Letter, uppercase".
2755
2756 bool
2757 Lex::is_unicode_uppercase(unsigned int c)
2758 {
2759 return Lex::is_in_unicode_range(c, unicode_uppercase_letters,
2760 ARRAY_SIZE(unicode_uppercase_letters));
2761 }
2762
2763 // Return whether the identifier NAME should be exported. NAME is a
2764 // mangled name which includes only ASCII characters.
2765
2766 bool
2767 Lex::is_exported_mangled_name(const std::string& name)
2768 {
2769 unsigned char c = name[0];
2770 if (c != '.')
2771 return c >= 'A' && c <= 'Z';
2772 else
2773 {
2774 const char* p = name.data();
2775 size_t len = name.length();
2776 if (len < 4 || p[1] != '.' || (p[2] != 'u' && p[2] != 'U'))
2777 return false;
2778 unsigned int ci = 0;
2779 size_t want = (p[2] == 'u' ? 4 : 8);
2780 if (len < want + 3)
2781 return false;
2782 for (size_t i = 3; i < want; ++i)
2783 {
2784 c = p[i];
2785 if (!Lex::is_hex_digit(c))
2786 return false;
2787 ci <<= 4;
2788 ci |= Lex::hex_val(c);
2789 }
2790 return Lex::is_unicode_uppercase(ci);
2791 }
2792 }
2793
2794 // Return whether the identifier NAME should be exported. NAME is a
2795 // an unmangled utf-8 string and may contain non-ASCII characters.
2796
2797 bool
2798 Lex::is_exported_name(const std::string& name)
2799 {
2800 unsigned int uchar;
2801 if (Lex::fetch_char(name.c_str(), &uchar) != 0)
2802 return Lex::is_unicode_letter(uchar) && Lex::is_unicode_uppercase(uchar);
2803 return false;
2804 }
2805
2806 // Return whether the identifier NAME contains an invalid character.
2807 // This is based on how we handle invalid characters in
2808 // gather_identifier.
2809
2810 bool
2811 Lex::is_invalid_identifier(const std::string& name)
2812 {
2813 return name.find("$INVALID$") != std::string::npos;
2814 }
Mirror of the offical gcc git repository hosted at gcc.gnu.org