]>
Commit | Line | Data |
---|---|---|
1 | /* | |
2 | * Copyright (c) 2015 Nicira, Inc. | |
3 | * | |
4 | * Licensed under the Apache License, Version 2.0 (the "License"); | |
5 | * you may not use this file except in compliance with the License. | |
6 | * You may obtain a copy of the License at: | |
7 | * | |
8 | * http://www.apache.org/licenses/LICENSE-2.0 | |
9 | * | |
10 | * Unless required by applicable law or agreed to in writing, software | |
11 | * distributed under the License is distributed on an "AS IS" BASIS, | |
12 | * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. | |
13 | * See the License for the specific language governing permissions and | |
14 | * limitations under the License. | |
15 | */ | |
16 | ||
17 | #include <config.h> | |
18 | #include "lex.h" | |
19 | #include <ctype.h> | |
20 | #include <errno.h> | |
21 | #include <stdarg.h> | |
22 | #include "dynamic-string.h" | |
23 | #include "json.h" | |
24 | #include "packets.h" | |
25 | #include "util.h" | |
26 | \f | |
27 | /* Returns a string that represents 'format'. */ | |
28 | const char * | |
29 | lex_format_to_string(enum lex_format format) | |
30 | { | |
31 | switch (format) { | |
32 | case LEX_F_DECIMAL: | |
33 | return "decimal"; | |
34 | case LEX_F_HEXADECIMAL: | |
35 | return "hexadecimal"; | |
36 | case LEX_F_IPV4: | |
37 | return "IPv4"; | |
38 | case LEX_F_IPV6: | |
39 | return "IPv6"; | |
40 | case LEX_F_ETHERNET: | |
41 | return "Ethernet"; | |
42 | default: | |
43 | abort(); | |
44 | } | |
45 | } | |
46 | \f | |
47 | /* Initializes 'token'. */ | |
48 | void | |
49 | lex_token_init(struct lex_token *token) | |
50 | { | |
51 | token->type = LEX_T_END; | |
52 | token->s = NULL; | |
53 | } | |
54 | ||
55 | /* Frees memory owned by 'token'. */ | |
56 | void | |
57 | lex_token_destroy(struct lex_token *token) | |
58 | { | |
59 | free(token->s); | |
60 | } | |
61 | ||
62 | /* Exchanges 'a' and 'b'. */ | |
63 | void | |
64 | lex_token_swap(struct lex_token *a, struct lex_token *b) | |
65 | { | |
66 | struct lex_token tmp = *a; | |
67 | *a = *b; | |
68 | *b = tmp; | |
69 | } | |
70 | \f | |
71 | /* lex_token_format(). */ | |
72 | ||
73 | static size_t | |
74 | lex_token_n_zeros(enum lex_format format) | |
75 | { | |
76 | switch (format) { | |
77 | case LEX_F_DECIMAL: return offsetof(union mf_subvalue, integer); | |
78 | case LEX_F_HEXADECIMAL: return 0; | |
79 | case LEX_F_IPV4: return offsetof(union mf_subvalue, ipv4); | |
80 | case LEX_F_IPV6: return offsetof(union mf_subvalue, ipv6); | |
81 | case LEX_F_ETHERNET: return offsetof(union mf_subvalue, mac); | |
82 | default: OVS_NOT_REACHED(); | |
83 | } | |
84 | } | |
85 | ||
86 | /* Returns the effective format for 'token', that is, the format in which it | |
87 | * should actually be printed. This is ordinarily the same as 'token->format', | |
88 | * but it's always possible that someone sets up a token with a format that | |
89 | * won't work for a value, e.g. 'token->value' is wider than 32 bits but the | |
90 | * format is LEX_F_IPV4. (The lexer itself won't do that; this is an attempt | |
91 | * to avoid confusion in the future.) */ | |
92 | static enum lex_format | |
93 | lex_token_get_format(const struct lex_token *token) | |
94 | { | |
95 | size_t n_zeros = lex_token_n_zeros(token->format); | |
96 | return (is_all_zeros(&token->value, n_zeros) | |
97 | && (token->type != LEX_T_MASKED_INTEGER | |
98 | || is_all_zeros(&token->mask, n_zeros)) | |
99 | ? token->format | |
100 | : LEX_F_HEXADECIMAL); | |
101 | } | |
102 | ||
103 | static void | |
104 | lex_token_format_value(const union mf_subvalue *value, | |
105 | enum lex_format format, struct ds *s) | |
106 | { | |
107 | switch (format) { | |
108 | case LEX_F_DECIMAL: | |
109 | ds_put_format(s, "%"PRIu64, ntohll(value->integer)); | |
110 | break; | |
111 | ||
112 | case LEX_F_HEXADECIMAL: | |
113 | mf_format_subvalue(value, s); | |
114 | break; | |
115 | ||
116 | case LEX_F_IPV4: | |
117 | ds_put_format(s, IP_FMT, IP_ARGS(value->ipv4)); | |
118 | break; | |
119 | ||
120 | case LEX_F_IPV6: | |
121 | ipv6_format_addr(&value->ipv6, s); | |
122 | break; | |
123 | ||
124 | case LEX_F_ETHERNET: | |
125 | ds_put_format(s, ETH_ADDR_FMT, ETH_ADDR_ARGS(value->mac)); | |
126 | break; | |
127 | ||
128 | default: | |
129 | OVS_NOT_REACHED(); | |
130 | } | |
131 | ||
132 | } | |
133 | ||
134 | static void | |
135 | lex_token_format_masked_integer(const struct lex_token *token, struct ds *s) | |
136 | { | |
137 | enum lex_format format = lex_token_get_format(token); | |
138 | ||
139 | lex_token_format_value(&token->value, format, s); | |
140 | ds_put_char(s, '/'); | |
141 | ||
142 | const union mf_subvalue *mask = &token->mask; | |
143 | if (format == LEX_F_IPV4 && ip_is_cidr(mask->ipv4)) { | |
144 | ds_put_format(s, "%d", ip_count_cidr_bits(mask->ipv4)); | |
145 | } else if (token->format == LEX_F_IPV6 && ipv6_is_cidr(&mask->ipv6)) { | |
146 | ds_put_format(s, "%d", ipv6_count_cidr_bits(&mask->ipv6)); | |
147 | } else { | |
148 | lex_token_format_value(&token->mask, format, s); | |
149 | } | |
150 | } | |
151 | ||
152 | /* Appends a string representation of 'token' to 's', in a format that can be | |
153 | * losslessly parsed back by the lexer. (LEX_T_END and LEX_T_ERROR can't be | |
154 | * parsed back.) */ | |
155 | void | |
156 | lex_token_format(const struct lex_token *token, struct ds *s) | |
157 | { | |
158 | switch (token->type) { | |
159 | case LEX_T_END: | |
160 | ds_put_cstr(s, "$"); | |
161 | break; | |
162 | ||
163 | case LEX_T_ID: | |
164 | ds_put_cstr(s, token->s); | |
165 | break; | |
166 | ||
167 | case LEX_T_ERROR: | |
168 | ds_put_cstr(s, "error("); | |
169 | json_string_escape(token->s, s); | |
170 | ds_put_char(s, ')'); | |
171 | break; | |
172 | ||
173 | case LEX_T_STRING: | |
174 | json_string_escape(token->s, s); | |
175 | break; | |
176 | ||
177 | case LEX_T_INTEGER: | |
178 | lex_token_format_value(&token->value, lex_token_get_format(token), s); | |
179 | break; | |
180 | ||
181 | case LEX_T_MASKED_INTEGER: | |
182 | lex_token_format_masked_integer(token, s); | |
183 | break; | |
184 | ||
185 | case LEX_T_LPAREN: | |
186 | ds_put_cstr(s, "("); | |
187 | break; | |
188 | case LEX_T_RPAREN: | |
189 | ds_put_cstr(s, ")"); | |
190 | break; | |
191 | case LEX_T_LCURLY: | |
192 | ds_put_cstr(s, "{"); | |
193 | break; | |
194 | case LEX_T_RCURLY: | |
195 | ds_put_cstr(s, "}"); | |
196 | break; | |
197 | case LEX_T_LSQUARE: | |
198 | ds_put_cstr(s, "["); | |
199 | break; | |
200 | case LEX_T_RSQUARE: | |
201 | ds_put_cstr(s, "]"); | |
202 | break; | |
203 | case LEX_T_EQ: | |
204 | ds_put_cstr(s, "=="); | |
205 | break; | |
206 | case LEX_T_NE: | |
207 | ds_put_cstr(s, "!="); | |
208 | break; | |
209 | case LEX_T_LT: | |
210 | ds_put_cstr(s, "<"); | |
211 | break; | |
212 | case LEX_T_LE: | |
213 | ds_put_cstr(s, "<="); | |
214 | break; | |
215 | case LEX_T_GT: | |
216 | ds_put_cstr(s, ">"); | |
217 | break; | |
218 | case LEX_T_GE: | |
219 | ds_put_cstr(s, ">="); | |
220 | break; | |
221 | case LEX_T_LOG_NOT: | |
222 | ds_put_cstr(s, "!"); | |
223 | break; | |
224 | case LEX_T_LOG_AND: | |
225 | ds_put_cstr(s, "&&"); | |
226 | break; | |
227 | case LEX_T_LOG_OR: | |
228 | ds_put_cstr(s, "||"); | |
229 | break; | |
230 | case LEX_T_ELLIPSIS: | |
231 | ds_put_cstr(s, ".."); | |
232 | break; | |
233 | case LEX_T_COMMA: | |
234 | ds_put_cstr(s, ","); | |
235 | break; | |
236 | case LEX_T_SEMICOLON: | |
237 | ds_put_cstr(s, ";"); | |
238 | break; | |
239 | case LEX_T_EQUALS: | |
240 | ds_put_cstr(s, "="); | |
241 | break; | |
242 | case LEX_T_EXCHANGE: | |
243 | ds_put_cstr(s, "<->"); | |
244 | break; | |
245 | case LEX_T_DECREMENT: | |
246 | ds_put_cstr(s, "--"); | |
247 | break; | |
248 | default: | |
249 | OVS_NOT_REACHED(); | |
250 | } | |
251 | ||
252 | } | |
253 | \f | |
254 | /* lex_token_parse(). */ | |
255 | ||
256 | static void OVS_PRINTF_FORMAT(2, 3) | |
257 | lex_error(struct lex_token *token, const char *message, ...) | |
258 | { | |
259 | ovs_assert(!token->s); | |
260 | token->type = LEX_T_ERROR; | |
261 | ||
262 | va_list args; | |
263 | va_start(args, message); | |
264 | token->s = xvasprintf(message, args); | |
265 | va_end(args); | |
266 | } | |
267 | ||
268 | static void | |
269 | lex_parse_hex_integer(const char *start, size_t len, struct lex_token *token) | |
270 | { | |
271 | const char *in = start + (len - 1); | |
272 | uint8_t *out = token->value.u8 + (sizeof token->value.u8 - 1); | |
273 | ||
274 | for (int i = 0; i < len; i++) { | |
275 | int hexit = hexit_value(in[-i]); | |
276 | if (hexit < 0) { | |
277 | lex_error(token, "Invalid syntax in hexadecimal constant."); | |
278 | return; | |
279 | } | |
280 | if (hexit && i / 2 >= sizeof token->value.u8) { | |
281 | lex_error(token, "Hexadecimal constant requires more than " | |
282 | "%"PRIuSIZE" bits.", 8 * sizeof token->value.u8); | |
283 | return; | |
284 | } | |
285 | out[-(i / 2)] |= i % 2 ? hexit << 4 : hexit; | |
286 | } | |
287 | token->format = LEX_F_HEXADECIMAL; | |
288 | } | |
289 | ||
290 | static const char * | |
291 | lex_parse_integer__(const char *p, struct lex_token *token) | |
292 | { | |
293 | lex_token_init(token); | |
294 | token->type = LEX_T_INTEGER; | |
295 | memset(&token->value, 0, sizeof token->value); | |
296 | const char *start = p; | |
297 | const char *end = start; | |
298 | while (isalnum((unsigned char) *end) || *end == ':' | |
299 | || (*end == '.' && end[1] != '.')) { | |
300 | end++; | |
301 | } | |
302 | size_t len = end - start; | |
303 | ||
304 | int n; | |
305 | struct eth_addr mac; | |
306 | ||
307 | if (!len) { | |
308 | lex_error(token, "Integer constant expected."); | |
309 | } else if (len == 17 | |
310 | && ovs_scan(start, ETH_ADDR_SCAN_FMT"%n", | |
311 | ETH_ADDR_SCAN_ARGS(mac), &n) | |
312 | && n == len) { | |
313 | token->value.mac = mac; | |
314 | token->format = LEX_F_ETHERNET; | |
315 | } else if (start + strspn(start, "0123456789") == end) { | |
316 | if (p[0] == '0' && len > 1) { | |
317 | lex_error(token, "Decimal constants must not have leading zeros."); | |
318 | } else { | |
319 | unsigned long long int integer; | |
320 | char *tail; | |
321 | ||
322 | errno = 0; | |
323 | integer = strtoull(p, &tail, 10); | |
324 | if (tail != end || errno == ERANGE) { | |
325 | lex_error(token, "Decimal constants must be less than 2**64."); | |
326 | } else { | |
327 | token->value.integer = htonll(integer); | |
328 | token->format = LEX_F_DECIMAL; | |
329 | } | |
330 | } | |
331 | } else if (p[0] == '0' && (p[1] == 'x' || p[1] == 'X')) { | |
332 | if (len > 2) { | |
333 | lex_parse_hex_integer(start + 2, len - 2, token); | |
334 | } else { | |
335 | lex_error(token, "Hex digits expected following 0%c.", p[1]); | |
336 | } | |
337 | } else if (len < INET6_ADDRSTRLEN) { | |
338 | char copy[INET6_ADDRSTRLEN]; | |
339 | memcpy(copy, p, len); | |
340 | copy[len] = '\0'; | |
341 | ||
342 | if (ip_parse(copy, &token->value.ipv4)) { | |
343 | token->format = LEX_F_IPV4; | |
344 | } else if (ipv6_parse(copy, &token->value.ipv6)) { | |
345 | token->format = LEX_F_IPV6; | |
346 | } else { | |
347 | lex_error(token, "Invalid numeric constant."); | |
348 | } | |
349 | } else { | |
350 | lex_error(token, "Invalid numeric constant."); | |
351 | } | |
352 | ||
353 | ovs_assert(token->type == LEX_T_INTEGER || token->type == LEX_T_ERROR); | |
354 | return end; | |
355 | } | |
356 | ||
357 | static const char * | |
358 | lex_parse_mask(const char *p, struct lex_token *token) | |
359 | { | |
360 | struct lex_token mask; | |
361 | ||
362 | /* Parse just past the '/' as a second integer. Handle errors. */ | |
363 | p = lex_parse_integer__(p + 1, &mask); | |
364 | if (mask.type == LEX_T_ERROR) { | |
365 | lex_token_swap(&mask, token); | |
366 | lex_token_destroy(&mask); | |
367 | return p; | |
368 | } | |
369 | ovs_assert(mask.type == LEX_T_INTEGER); | |
370 | ||
371 | /* Now convert the value and mask into a masked integer token. | |
372 | * We have a few special cases. */ | |
373 | token->type = LEX_T_MASKED_INTEGER; | |
374 | memset(&token->mask, 0, sizeof token->mask); | |
375 | uint32_t prefix_bits = ntohll(mask.value.integer); | |
376 | if (token->format == mask.format) { | |
377 | /* Same format value and mask is always OK. */ | |
378 | token->mask = mask.value; | |
379 | } else if (token->format == LEX_F_IPV4 | |
380 | && mask.format == LEX_F_DECIMAL | |
381 | && prefix_bits <= 32) { | |
382 | /* IPv4 address with decimal mask is a CIDR prefix. */ | |
383 | token->mask.integer = htonll(ntohl(be32_prefix_mask(prefix_bits))); | |
384 | } else if (token->format == LEX_F_IPV6 | |
385 | && mask.format == LEX_F_DECIMAL | |
386 | && prefix_bits <= 128) { | |
387 | /* IPv6 address with decimal mask is a CIDR prefix. */ | |
388 | token->mask.ipv6 = ipv6_create_mask(prefix_bits); | |
389 | } else if (token->format == LEX_F_DECIMAL | |
390 | && mask.format == LEX_F_HEXADECIMAL | |
391 | && token->value.integer == 0) { | |
392 | /* Special case for e.g. 0/0x1234. */ | |
393 | token->format = LEX_F_HEXADECIMAL; | |
394 | token->mask = mask.value; | |
395 | } else { | |
396 | lex_error(token, "Value and mask have incompatible formats."); | |
397 | return p; | |
398 | } | |
399 | ||
400 | /* Check invariant that a 1-bit in the value corresponds to a 1-bit in the | |
401 | * mask. */ | |
402 | for (int i = 0; i < ARRAY_SIZE(token->mask.be32); i++) { | |
403 | ovs_be32 v = token->value.be32[i]; | |
404 | ovs_be32 m = token->mask.be32[i]; | |
405 | ||
406 | if (v & ~m) { | |
407 | lex_error(token, "Value contains unmasked 1-bits."); | |
408 | break; | |
409 | } | |
410 | } | |
411 | ||
412 | /* Done! */ | |
413 | lex_token_destroy(&mask); | |
414 | return p; | |
415 | } | |
416 | ||
417 | static const char * | |
418 | lex_parse_integer(const char *p, struct lex_token *token) | |
419 | { | |
420 | p = lex_parse_integer__(p, token); | |
421 | if (token->type == LEX_T_INTEGER && *p == '/') { | |
422 | p = lex_parse_mask(p, token); | |
423 | } | |
424 | return p; | |
425 | } | |
426 | ||
427 | static const char * | |
428 | lex_parse_string(const char *p, struct lex_token *token) | |
429 | { | |
430 | const char *start = ++p; | |
431 | for (;;) { | |
432 | switch (*p) { | |
433 | case '\0': | |
434 | lex_error(token, "Input ends inside quoted string."); | |
435 | return p; | |
436 | ||
437 | case '"': | |
438 | token->type = (json_string_unescape(start, p - start, &token->s) | |
439 | ? LEX_T_STRING : LEX_T_ERROR); | |
440 | return p + 1; | |
441 | ||
442 | case '\\': | |
443 | p++; | |
444 | if (*p) { | |
445 | p++; | |
446 | } | |
447 | break; | |
448 | ||
449 | default: | |
450 | p++; | |
451 | break; | |
452 | } | |
453 | } | |
454 | } | |
455 | ||
456 | static bool | |
457 | lex_is_id1(unsigned char c) | |
458 | { | |
459 | return ((c >= 'a' && c <= 'z') || (c >= 'A' && c <= 'Z') | |
460 | || c == '_' || c == '.'); | |
461 | } | |
462 | ||
463 | static bool | |
464 | lex_is_idn(unsigned char c) | |
465 | { | |
466 | return lex_is_id1(c) || (c >= '0' && c <= '9'); | |
467 | } | |
468 | ||
469 | static const char * | |
470 | lex_parse_id(const char *p, struct lex_token *token) | |
471 | { | |
472 | const char *start = p; | |
473 | ||
474 | do { | |
475 | p++; | |
476 | } while (lex_is_idn(*p)); | |
477 | ||
478 | token->type = LEX_T_ID; | |
479 | token->s = xmemdup0(start, p - start); | |
480 | return p; | |
481 | } | |
482 | ||
483 | /* Initializes 'token' and parses the first token from the beginning of | |
484 | * null-terminated string 'p' into 'token'. Stores a pointer to the start of | |
485 | * the token (after skipping white space and comments, if any) into '*startp'. | |
486 | * Returns the character position at which to begin parsing the next token. */ | |
487 | const char * | |
488 | lex_token_parse(struct lex_token *token, const char *p, const char **startp) | |
489 | { | |
490 | lex_token_init(token); | |
491 | ||
492 | next: | |
493 | *startp = p; | |
494 | switch (*p) { | |
495 | case '\0': | |
496 | token->type = LEX_T_END; | |
497 | return p; | |
498 | ||
499 | case ' ': case '\t': case '\n': case '\r': | |
500 | p++; | |
501 | goto next; | |
502 | ||
503 | case '/': | |
504 | p++; | |
505 | if (*p == '/') { | |
506 | do { | |
507 | p++; | |
508 | } while (*p != '\0' && *p != '\n'); | |
509 | goto next; | |
510 | } else if (*p == '*') { | |
511 | p++; | |
512 | for (;;) { | |
513 | if (*p == '*' && p[1] == '/') { | |
514 | p += 2; | |
515 | goto next; | |
516 | } else if (*p == '\0' || *p == '\n') { | |
517 | lex_error(token, "`/*' without matching `*/'."); | |
518 | return p; | |
519 | } else { | |
520 | p++; | |
521 | } | |
522 | } | |
523 | goto next; | |
524 | } else { | |
525 | lex_error(token, | |
526 | "`/' is only valid as part of `//' or `/*'."); | |
527 | } | |
528 | break; | |
529 | ||
530 | case '(': | |
531 | token->type = LEX_T_LPAREN; | |
532 | p++; | |
533 | break; | |
534 | ||
535 | case ')': | |
536 | token->type = LEX_T_RPAREN; | |
537 | p++; | |
538 | break; | |
539 | ||
540 | case '{': | |
541 | token->type = LEX_T_LCURLY; | |
542 | p++; | |
543 | break; | |
544 | ||
545 | case '}': | |
546 | token->type = LEX_T_RCURLY; | |
547 | p++; | |
548 | break; | |
549 | ||
550 | case '[': | |
551 | token->type = LEX_T_LSQUARE; | |
552 | p++; | |
553 | break; | |
554 | ||
555 | case ']': | |
556 | token->type = LEX_T_RSQUARE; | |
557 | p++; | |
558 | break; | |
559 | ||
560 | case '=': | |
561 | p++; | |
562 | if (*p == '=') { | |
563 | token->type = LEX_T_EQ; | |
564 | p++; | |
565 | } else { | |
566 | token->type = LEX_T_EQUALS; | |
567 | } | |
568 | break; | |
569 | ||
570 | case '!': | |
571 | p++; | |
572 | if (*p == '=') { | |
573 | token->type = LEX_T_NE; | |
574 | p++; | |
575 | } else { | |
576 | token->type = LEX_T_LOG_NOT; | |
577 | } | |
578 | break; | |
579 | ||
580 | case '&': | |
581 | p++; | |
582 | if (*p == '&') { | |
583 | token->type = LEX_T_LOG_AND; | |
584 | p++; | |
585 | } else { | |
586 | lex_error(token, "`&' is only valid as part of `&&'."); | |
587 | } | |
588 | break; | |
589 | ||
590 | case '|': | |
591 | p++; | |
592 | if (*p == '|') { | |
593 | token->type = LEX_T_LOG_OR; | |
594 | p++; | |
595 | } else { | |
596 | lex_error(token, "`|' is only valid as part of `||'."); | |
597 | } | |
598 | break; | |
599 | ||
600 | case '<': | |
601 | p++; | |
602 | if (*p == '=') { | |
603 | token->type = LEX_T_LE; | |
604 | p++; | |
605 | } else if (*p == '-' && p[1] == '>') { | |
606 | token->type = LEX_T_EXCHANGE; | |
607 | p += 2; | |
608 | } else { | |
609 | token->type = LEX_T_LT; | |
610 | } | |
611 | break; | |
612 | ||
613 | case '>': | |
614 | p++; | |
615 | if (*p == '=') { | |
616 | token->type = LEX_T_GE; | |
617 | p++; | |
618 | } else { | |
619 | token->type = LEX_T_GT; | |
620 | } | |
621 | break; | |
622 | ||
623 | case '.': | |
624 | p++; | |
625 | if (*p == '.') { | |
626 | token->type = LEX_T_ELLIPSIS; | |
627 | p++; | |
628 | } else { | |
629 | lex_error(token, "`.' is only valid as part of `..' or a number."); | |
630 | } | |
631 | break; | |
632 | ||
633 | case ',': | |
634 | p++; | |
635 | token->type = LEX_T_COMMA; | |
636 | break; | |
637 | ||
638 | case ';': | |
639 | p++; | |
640 | token->type = LEX_T_SEMICOLON; | |
641 | break; | |
642 | ||
643 | case '-': | |
644 | p++; | |
645 | if (*p == '-') { | |
646 | token->type = LEX_T_DECREMENT; | |
647 | p++; | |
648 | } else { | |
649 | lex_error(token, "`-' is only valid as part of `--'."); | |
650 | } | |
651 | break; | |
652 | ||
653 | case '0': case '1': case '2': case '3': case '4': | |
654 | case '5': case '6': case '7': case '8': case '9': | |
655 | case ':': | |
656 | p = lex_parse_integer(p, token); | |
657 | break; | |
658 | ||
659 | case '"': | |
660 | p = lex_parse_string(p, token); | |
661 | break; | |
662 | ||
663 | case 'a': case 'b': case 'c': case 'd': case 'e': case 'f': | |
664 | case 'A': case 'B': case 'C': case 'D': case 'E': case 'F': | |
665 | /* We need to distinguish an Ethernet address or IPv6 address from an | |
666 | * identifier. Fortunately, Ethernet addresses and IPv6 addresses that | |
667 | * are ambiguous based on the first character, always start with hex | |
668 | * digits followed by a colon, but identifiers never do. */ | |
669 | p = (p[strspn(p, "0123456789abcdefABCDEF")] == ':' | |
670 | ? lex_parse_integer(p, token) | |
671 | : lex_parse_id(p, token)); | |
672 | break; | |
673 | ||
674 | default: | |
675 | if (lex_is_id1(*p)) { | |
676 | p = lex_parse_id(p, token); | |
677 | } else { | |
678 | if (isprint((unsigned char) *p)) { | |
679 | lex_error(token, "Invalid character `%c' in input.", *p); | |
680 | } else { | |
681 | lex_error(token, "Invalid byte 0x%d in input.", *p); | |
682 | } | |
683 | p++; | |
684 | } | |
685 | break; | |
686 | } | |
687 | ||
688 | return p; | |
689 | } | |
690 | \f | |
691 | /* Initializes 'lexer' for parsing 'input'. | |
692 | * | |
693 | * While the lexer is in use, 'input' must remain available, but the caller | |
694 | * otherwise retains ownership of 'input'. | |
695 | * | |
696 | * The caller must call lexer_get() to obtain the first token. */ | |
697 | void | |
698 | lexer_init(struct lexer *lexer, const char *input) | |
699 | { | |
700 | lexer->input = input; | |
701 | lexer->start = NULL; | |
702 | lex_token_init(&lexer->token); | |
703 | } | |
704 | ||
705 | /* Frees storage associated with 'lexer'. */ | |
706 | void | |
707 | lexer_destroy(struct lexer *lexer) | |
708 | { | |
709 | lex_token_destroy(&lexer->token); | |
710 | } | |
711 | ||
712 | /* Obtains the next token from 'lexer' into 'lexer->token', and returns the | |
713 | * token's type. The caller may examine 'lexer->token' directly to obtain full | |
714 | * information about the token. */ | |
715 | enum lex_type | |
716 | lexer_get(struct lexer *lexer) | |
717 | { | |
718 | lex_token_destroy(&lexer->token); | |
719 | lexer->input = lex_token_parse(&lexer->token, lexer->input, &lexer->start); | |
720 | return lexer->token.type; | |
721 | } | |
722 | ||
723 | /* Returns the type of the next token that will be fetched by lexer_get(), | |
724 | * without advancing 'lexer->token' to that token. */ | |
725 | enum lex_type | |
726 | lexer_lookahead(const struct lexer *lexer) | |
727 | { | |
728 | struct lex_token next; | |
729 | enum lex_type type; | |
730 | const char *start; | |
731 | ||
732 | lex_token_parse(&next, lexer->input, &start); | |
733 | type = next.type; | |
734 | lex_token_destroy(&next); | |
735 | return type; | |
736 | } | |
737 | ||
738 | /* If 'lexer''s current token has the given 'type', advances 'lexer' to the | |
739 | * next token and returns true. Otherwise returns false. */ | |
740 | bool | |
741 | lexer_match(struct lexer *lexer, enum lex_type type) | |
742 | { | |
743 | if (lexer->token.type == type) { | |
744 | lexer_get(lexer); | |
745 | return true; | |
746 | } else { | |
747 | return false; | |
748 | } | |
749 | } | |
750 | ||
751 | /* If 'lexer''s current token is the identifier given in 'id', advances 'lexer' | |
752 | * to the next token and returns true. Otherwise returns false. */ | |
753 | bool | |
754 | lexer_match_id(struct lexer *lexer, const char *id) | |
755 | { | |
756 | if (lexer->token.type == LEX_T_ID && !strcmp(lexer->token.s, id)) { | |
757 | lexer_get(lexer); | |
758 | return true; | |
759 | } else { | |
760 | return false; | |
761 | } | |
762 | } | |
763 | ||
764 | bool | |
765 | lexer_is_int(const struct lexer *lexer) | |
766 | { | |
767 | return (lexer->token.type == LEX_T_INTEGER | |
768 | && lexer->token.format == LEX_F_DECIMAL | |
769 | && ntohll(lexer->token.value.integer) <= INT_MAX); | |
770 | } | |
771 | ||
772 | bool | |
773 | lexer_get_int(struct lexer *lexer, int *value) | |
774 | { | |
775 | if (lexer_is_int(lexer)) { | |
776 | *value = ntohll(lexer->token.value.integer); | |
777 | lexer_get(lexer); | |
778 | return true; | |
779 | } else { | |
780 | *value = 0; | |
781 | return false; | |
782 | } | |
783 | } |