]>
Commit | Line | Data |
---|---|---|
10b1662b BP |
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 "util.h" | |
363b5330 BP |
25 | \f |
26 | /* Returns a string that represents 'format'. */ | |
27 | const char * | |
28 | lex_format_to_string(enum lex_format format) | |
29 | { | |
30 | switch (format) { | |
31 | case LEX_F_DECIMAL: | |
32 | return "decimal"; | |
33 | case LEX_F_HEXADECIMAL: | |
34 | return "hexadecimal"; | |
35 | case LEX_F_IPV4: | |
36 | return "IPv4"; | |
37 | case LEX_F_IPV6: | |
38 | return "IPv6"; | |
39 | case LEX_F_ETHERNET: | |
40 | return "Ethernet"; | |
41 | default: | |
42 | abort(); | |
43 | } | |
44 | } | |
45 | \f | |
10b1662b BP |
46 | /* Initializes 'token'. */ |
47 | void | |
48 | lex_token_init(struct lex_token *token) | |
49 | { | |
50 | token->type = LEX_T_END; | |
51 | token->s = NULL; | |
52 | } | |
53 | ||
54 | /* Frees memory owned by 'token'. */ | |
55 | void | |
56 | lex_token_destroy(struct lex_token *token) | |
57 | { | |
58 | free(token->s); | |
59 | } | |
60 | ||
61 | /* Exchanges 'a' and 'b'. */ | |
62 | void | |
63 | lex_token_swap(struct lex_token *a, struct lex_token *b) | |
64 | { | |
65 | struct lex_token tmp = *a; | |
66 | *a = *b; | |
67 | *b = tmp; | |
68 | } | |
69 | \f | |
70 | /* lex_token_format(). */ | |
71 | ||
72 | static size_t | |
73 | lex_token_n_zeros(enum lex_format format) | |
74 | { | |
75 | switch (format) { | |
76 | case LEX_F_DECIMAL: return offsetof(union mf_subvalue, integer); | |
77 | case LEX_F_HEXADECIMAL: return 0; | |
78 | case LEX_F_IPV4: return offsetof(union mf_subvalue, ipv4); | |
79 | case LEX_F_IPV6: return offsetof(union mf_subvalue, ipv6); | |
80 | case LEX_F_ETHERNET: return offsetof(union mf_subvalue, mac); | |
81 | default: OVS_NOT_REACHED(); | |
82 | } | |
83 | } | |
84 | ||
85 | /* Returns the effective format for 'token', that is, the format in which it | |
86 | * should actually be printed. This is ordinarily the same as 'token->format', | |
87 | * but it's always possible that someone sets up a token with a format that | |
88 | * won't work for a value, e.g. 'token->value' is wider than 32 bits but the | |
89 | * format is LEX_F_IPV4. (The lexer itself won't do that; this is an attempt | |
90 | * to avoid confusion in the future.) */ | |
91 | static enum lex_format | |
92 | lex_token_get_format(const struct lex_token *token) | |
93 | { | |
94 | size_t n_zeros = lex_token_n_zeros(token->format); | |
95 | return (is_all_zeros(&token->value, n_zeros) | |
96 | && (token->type != LEX_T_MASKED_INTEGER | |
97 | || is_all_zeros(&token->mask, n_zeros)) | |
98 | ? token->format | |
99 | : LEX_F_HEXADECIMAL); | |
100 | } | |
101 | ||
102 | static void | |
103 | lex_token_format_value(const union mf_subvalue *value, | |
104 | enum lex_format format, struct ds *s) | |
105 | { | |
106 | switch (format) { | |
107 | case LEX_F_DECIMAL: | |
108 | ds_put_format(s, "%"PRIu64, ntohll(value->integer)); | |
109 | break; | |
110 | ||
111 | case LEX_F_HEXADECIMAL: | |
112 | mf_format_subvalue(value, s); | |
113 | break; | |
114 | ||
115 | case LEX_F_IPV4: | |
116 | ds_put_format(s, IP_FMT, IP_ARGS(value->ipv4)); | |
117 | break; | |
118 | ||
119 | case LEX_F_IPV6: | |
120 | print_ipv6_addr(s, &value->ipv6); | |
121 | break; | |
122 | ||
123 | case LEX_F_ETHERNET: | |
124 | ds_put_format(s, ETH_ADDR_FMT, ETH_ADDR_ARGS(value->mac)); | |
125 | break; | |
126 | ||
127 | default: | |
128 | OVS_NOT_REACHED(); | |
129 | } | |
130 | ||
131 | } | |
132 | ||
133 | static void | |
134 | lex_token_format_masked_integer(const struct lex_token *token, struct ds *s) | |
135 | { | |
136 | enum lex_format format = lex_token_get_format(token); | |
137 | ||
138 | lex_token_format_value(&token->value, format, s); | |
139 | ds_put_char(s, '/'); | |
140 | ||
141 | const union mf_subvalue *mask = &token->mask; | |
142 | if (format == LEX_F_IPV4 && ip_is_cidr(mask->ipv4)) { | |
143 | ds_put_format(s, "%d", ip_count_cidr_bits(mask->ipv4)); | |
144 | } else if (token->format == LEX_F_IPV6 && ipv6_is_cidr(&mask->ipv6)) { | |
145 | ds_put_format(s, "%d", ipv6_count_cidr_bits(&mask->ipv6)); | |
146 | } else { | |
147 | lex_token_format_value(&token->mask, format, s); | |
148 | } | |
149 | } | |
150 | ||
10b1662b BP |
151 | /* Appends a string representation of 'token' to 's', in a format that can be |
152 | * losslessly parsed back by the lexer. (LEX_T_END and LEX_T_ERROR can't be | |
153 | * parsed back.) */ | |
154 | void | |
3d611299 | 155 | lex_token_format(const struct lex_token *token, struct ds *s) |
10b1662b BP |
156 | { |
157 | switch (token->type) { | |
158 | case LEX_T_END: | |
159 | ds_put_cstr(s, "$"); | |
160 | break; | |
161 | ||
162 | case LEX_T_ID: | |
163 | ds_put_cstr(s, token->s); | |
164 | break; | |
165 | ||
166 | case LEX_T_ERROR: | |
167 | ds_put_cstr(s, "error("); | |
3b626771 | 168 | json_string_escape(token->s, s); |
10b1662b BP |
169 | ds_put_char(s, ')'); |
170 | break; | |
171 | ||
172 | case LEX_T_STRING: | |
3b626771 | 173 | json_string_escape(token->s, s); |
10b1662b BP |
174 | break; |
175 | ||
176 | case LEX_T_INTEGER: | |
177 | lex_token_format_value(&token->value, lex_token_get_format(token), s); | |
178 | break; | |
179 | ||
180 | case LEX_T_MASKED_INTEGER: | |
181 | lex_token_format_masked_integer(token, s); | |
182 | break; | |
183 | ||
184 | case LEX_T_LPAREN: | |
185 | ds_put_cstr(s, "("); | |
186 | break; | |
187 | case LEX_T_RPAREN: | |
188 | ds_put_cstr(s, ")"); | |
189 | break; | |
190 | case LEX_T_LCURLY: | |
191 | ds_put_cstr(s, "{"); | |
192 | break; | |
193 | case LEX_T_RCURLY: | |
194 | ds_put_cstr(s, "}"); | |
195 | break; | |
196 | case LEX_T_LSQUARE: | |
197 | ds_put_cstr(s, "["); | |
198 | break; | |
199 | case LEX_T_RSQUARE: | |
200 | ds_put_cstr(s, "]"); | |
201 | break; | |
202 | case LEX_T_EQ: | |
203 | ds_put_cstr(s, "=="); | |
204 | break; | |
205 | case LEX_T_NE: | |
206 | ds_put_cstr(s, "!="); | |
207 | break; | |
208 | case LEX_T_LT: | |
209 | ds_put_cstr(s, "<"); | |
210 | break; | |
211 | case LEX_T_LE: | |
212 | ds_put_cstr(s, "<="); | |
213 | break; | |
214 | case LEX_T_GT: | |
215 | ds_put_cstr(s, ">"); | |
216 | break; | |
217 | case LEX_T_GE: | |
218 | ds_put_cstr(s, ">="); | |
219 | break; | |
220 | case LEX_T_LOG_NOT: | |
221 | ds_put_cstr(s, "!"); | |
222 | break; | |
223 | case LEX_T_LOG_AND: | |
224 | ds_put_cstr(s, "&&"); | |
225 | break; | |
226 | case LEX_T_LOG_OR: | |
227 | ds_put_cstr(s, "||"); | |
228 | break; | |
229 | case LEX_T_ELLIPSIS: | |
230 | ds_put_cstr(s, ".."); | |
231 | break; | |
232 | case LEX_T_COMMA: | |
233 | ds_put_cstr(s, ","); | |
234 | break; | |
235 | case LEX_T_SEMICOLON: | |
236 | ds_put_cstr(s, ";"); | |
237 | break; | |
238 | case LEX_T_EQUALS: | |
239 | ds_put_cstr(s, "="); | |
240 | break; | |
241 | default: | |
242 | OVS_NOT_REACHED(); | |
243 | } | |
244 | ||
245 | } | |
246 | \f | |
247 | /* lex_token_parse(). */ | |
248 | ||
249 | static void OVS_PRINTF_FORMAT(2, 3) | |
250 | lex_error(struct lex_token *token, const char *message, ...) | |
251 | { | |
252 | ovs_assert(!token->s); | |
253 | token->type = LEX_T_ERROR; | |
254 | ||
255 | va_list args; | |
256 | va_start(args, message); | |
257 | token->s = xvasprintf(message, args); | |
258 | va_end(args); | |
259 | } | |
260 | ||
261 | static void | |
262 | lex_parse_hex_integer(const char *start, size_t len, struct lex_token *token) | |
263 | { | |
264 | const char *in = start + (len - 1); | |
265 | uint8_t *out = token->value.u8 + (sizeof token->value.u8 - 1); | |
266 | ||
267 | for (int i = 0; i < len; i++) { | |
268 | int hexit = hexit_value(in[-i]); | |
269 | if (hexit < 0) { | |
270 | lex_error(token, "Invalid syntax in hexadecimal constant."); | |
271 | return; | |
272 | } | |
273 | if (hexit && i / 2 >= sizeof token->value.u8) { | |
274 | lex_error(token, "Hexadecimal constant requires more than " | |
275 | "%"PRIuSIZE" bits.", 8 * sizeof token->value.u8); | |
276 | return; | |
277 | } | |
278 | out[-(i / 2)] |= i % 2 ? hexit << 4 : hexit; | |
279 | } | |
280 | token->format = LEX_F_HEXADECIMAL; | |
281 | } | |
282 | ||
283 | static const char * | |
284 | lex_parse_integer__(const char *p, struct lex_token *token) | |
285 | { | |
286 | lex_token_init(token); | |
287 | token->type = LEX_T_INTEGER; | |
288 | memset(&token->value, 0, sizeof token->value); | |
289 | const char *start = p; | |
290 | const char *end = start; | |
291 | while (isalnum((unsigned char) *end) || *end == ':' | |
292 | || (*end == '.' && end[1] != '.')) { | |
293 | end++; | |
294 | } | |
295 | size_t len = end - start; | |
296 | ||
297 | int n; | |
298 | uint8_t mac[ETH_ADDR_LEN]; | |
299 | ||
300 | if (!len) { | |
301 | lex_error(token, "Integer constant expected."); | |
302 | } else if (len == 17 | |
303 | && ovs_scan(start, ETH_ADDR_SCAN_FMT"%n", | |
304 | ETH_ADDR_SCAN_ARGS(mac), &n) | |
305 | && n == len) { | |
306 | memcpy(token->value.mac, mac, sizeof token->value.mac); | |
307 | token->format = LEX_F_ETHERNET; | |
308 | } else if (start + strspn(start, "0123456789") == end) { | |
309 | if (p[0] == '0' && len > 1) { | |
310 | lex_error(token, "Decimal constants must not have leading zeros."); | |
311 | } else { | |
312 | unsigned long long int integer; | |
313 | char *tail; | |
314 | ||
315 | errno = 0; | |
316 | integer = strtoull(p, &tail, 10); | |
317 | if (tail != end || errno == ERANGE) { | |
318 | lex_error(token, "Decimal constants must be less than 2**64."); | |
319 | } else { | |
320 | token->value.integer = htonll(integer); | |
321 | token->format = LEX_F_DECIMAL; | |
322 | } | |
323 | } | |
324 | } else if (p[0] == '0' && (p[1] == 'x' || p[1] == 'X')) { | |
325 | if (len > 2) { | |
326 | lex_parse_hex_integer(start + 2, len - 2, token); | |
327 | } else { | |
328 | lex_error(token, "Hex digits expected following 0%c.", p[1]); | |
329 | } | |
330 | } else if (len < INET6_ADDRSTRLEN) { | |
331 | char copy[INET6_ADDRSTRLEN]; | |
332 | memcpy(copy, p, len); | |
333 | copy[len] = '\0'; | |
334 | ||
335 | struct in_addr ipv4; | |
336 | struct in6_addr ipv6; | |
337 | if (inet_pton(AF_INET, copy, &ipv4) == 1) { | |
338 | token->value.ipv4 = ipv4.s_addr; | |
339 | token->format = LEX_F_IPV4; | |
340 | } else if (inet_pton(AF_INET6, copy, &ipv6) == 1) { | |
341 | token->value.ipv6 = ipv6; | |
342 | token->format = LEX_F_IPV6; | |
343 | } else { | |
344 | lex_error(token, "Invalid numeric constant."); | |
345 | } | |
346 | } else { | |
347 | lex_error(token, "Invalid numeric constant."); | |
348 | } | |
349 | ||
350 | ovs_assert(token->type == LEX_T_INTEGER || token->type == LEX_T_ERROR); | |
351 | return end; | |
352 | } | |
353 | ||
354 | static const char * | |
355 | lex_parse_mask(const char *p, struct lex_token *token) | |
356 | { | |
357 | struct lex_token mask; | |
358 | ||
359 | /* Parse just past the '/' as a second integer. Handle errors. */ | |
360 | p = lex_parse_integer__(p + 1, &mask); | |
361 | if (mask.type == LEX_T_ERROR) { | |
362 | lex_token_swap(&mask, token); | |
363 | lex_token_destroy(&mask); | |
364 | return p; | |
365 | } | |
366 | ovs_assert(mask.type == LEX_T_INTEGER); | |
367 | ||
368 | /* Now convert the value and mask into a masked integer token. | |
369 | * We have a few special cases. */ | |
370 | token->type = LEX_T_MASKED_INTEGER; | |
371 | memset(&token->mask, 0, sizeof token->mask); | |
372 | uint32_t prefix_bits = ntohll(mask.value.integer); | |
373 | if (token->format == mask.format) { | |
374 | /* Same format value and mask is always OK. */ | |
375 | token->mask = mask.value; | |
376 | } else if (token->format == LEX_F_IPV4 | |
377 | && mask.format == LEX_F_DECIMAL | |
378 | && prefix_bits <= 32) { | |
379 | /* IPv4 address with decimal mask is a CIDR prefix. */ | |
380 | token->mask.integer = htonll(ntohl(be32_prefix_mask(prefix_bits))); | |
381 | } else if (token->format == LEX_F_IPV6 | |
382 | && mask.format == LEX_F_DECIMAL | |
383 | && prefix_bits <= 128) { | |
384 | /* IPv6 address with decimal mask is a CIDR prefix. */ | |
385 | token->mask.ipv6 = ipv6_create_mask(prefix_bits); | |
386 | } else if (token->format == LEX_F_DECIMAL | |
387 | && mask.format == LEX_F_HEXADECIMAL | |
388 | && token->value.integer == 0) { | |
389 | /* Special case for e.g. 0/0x1234. */ | |
390 | token->format = LEX_F_HEXADECIMAL; | |
391 | token->mask = mask.value; | |
392 | } else { | |
393 | lex_error(token, "Value and mask have incompatible formats."); | |
394 | return p; | |
395 | } | |
396 | ||
397 | /* Check invariant that a 1-bit in the value corresponds to a 1-bit in the | |
398 | * mask. */ | |
399 | for (int i = 0; i < ARRAY_SIZE(token->mask.be32); i++) { | |
400 | ovs_be32 v = token->value.be32[i]; | |
401 | ovs_be32 m = token->mask.be32[i]; | |
402 | ||
403 | if (v & ~m) { | |
404 | lex_error(token, "Value contains unmasked 1-bits."); | |
405 | break; | |
406 | } | |
407 | } | |
408 | ||
409 | /* Done! */ | |
410 | lex_token_destroy(&mask); | |
411 | return p; | |
412 | } | |
413 | ||
414 | static const char * | |
415 | lex_parse_integer(const char *p, struct lex_token *token) | |
416 | { | |
417 | p = lex_parse_integer__(p, token); | |
418 | if (token->type == LEX_T_INTEGER && *p == '/') { | |
419 | p = lex_parse_mask(p, token); | |
420 | } | |
421 | return p; | |
422 | } | |
423 | ||
424 | static const char * | |
425 | lex_parse_string(const char *p, struct lex_token *token) | |
426 | { | |
427 | const char *start = ++p; | |
428 | for (;;) { | |
429 | switch (*p) { | |
430 | case '\0': | |
431 | lex_error(token, "Input ends inside quoted string."); | |
432 | return p; | |
433 | ||
434 | case '"': | |
435 | token->type = (json_string_unescape(start, p - start, &token->s) | |
436 | ? LEX_T_STRING : LEX_T_ERROR); | |
437 | return p + 1; | |
438 | ||
439 | case '\\': | |
440 | p++; | |
441 | if (*p) { | |
442 | p++; | |
443 | } | |
444 | break; | |
445 | ||
446 | default: | |
447 | p++; | |
448 | break; | |
449 | } | |
450 | } | |
451 | } | |
452 | ||
453 | static bool | |
454 | lex_is_id1(unsigned char c) | |
455 | { | |
456 | return ((c >= 'a' && c <= 'z') || (c >= 'A' && c <= 'Z') | |
457 | || c == '_' || c == '.'); | |
458 | } | |
459 | ||
460 | static bool | |
461 | lex_is_idn(unsigned char c) | |
462 | { | |
463 | return lex_is_id1(c) || (c >= '0' && c <= '9'); | |
464 | } | |
465 | ||
466 | static const char * | |
467 | lex_parse_id(const char *p, struct lex_token *token) | |
468 | { | |
469 | const char *start = p; | |
470 | ||
471 | do { | |
472 | p++; | |
473 | } while (lex_is_idn(*p)); | |
474 | ||
475 | token->type = LEX_T_ID; | |
476 | token->s = xmemdup0(start, p - start); | |
477 | return p; | |
478 | } | |
479 | ||
480 | /* Initializes 'token' and parses the first token from the beginning of | |
481 | * null-terminated string 'p' into 'token'. Stores a pointer to the start of | |
482 | * the token (after skipping white space and comments, if any) into '*startp'. | |
483 | * Returns the character position at which to begin parsing the next token. */ | |
484 | const char * | |
485 | lex_token_parse(struct lex_token *token, const char *p, const char **startp) | |
486 | { | |
487 | lex_token_init(token); | |
488 | ||
489 | next: | |
490 | *startp = p; | |
491 | switch (*p) { | |
492 | case '\0': | |
493 | token->type = LEX_T_END; | |
494 | return p; | |
495 | ||
496 | case ' ': case '\t': case '\n': case '\r': | |
497 | p++; | |
498 | goto next; | |
499 | ||
500 | case '/': | |
501 | p++; | |
502 | if (*p == '/') { | |
503 | do { | |
504 | p++; | |
505 | } while (*p != '\0' && *p != '\n'); | |
506 | goto next; | |
507 | } else if (*p == '*') { | |
508 | p++; | |
509 | for (;;) { | |
510 | if (*p == '*' && p[1] == '/') { | |
511 | p += 2; | |
512 | goto next; | |
513 | } else if (*p == '\0' || *p == '\n') { | |
514 | lex_error(token, "`/*' without matching `*/'."); | |
515 | return p; | |
516 | } else { | |
517 | p++; | |
518 | } | |
519 | } | |
520 | goto next; | |
521 | } else { | |
522 | lex_error(token, | |
523 | "`/' is only valid as part of `//' or `/*'."); | |
524 | } | |
525 | break; | |
526 | ||
527 | case '(': | |
528 | token->type = LEX_T_LPAREN; | |
529 | p++; | |
530 | break; | |
531 | ||
532 | case ')': | |
533 | token->type = LEX_T_RPAREN; | |
534 | p++; | |
535 | break; | |
536 | ||
537 | case '{': | |
538 | token->type = LEX_T_LCURLY; | |
539 | p++; | |
540 | break; | |
541 | ||
542 | case '}': | |
543 | token->type = LEX_T_RCURLY; | |
544 | p++; | |
545 | break; | |
546 | ||
547 | case '[': | |
548 | token->type = LEX_T_LSQUARE; | |
549 | p++; | |
550 | break; | |
551 | ||
552 | case ']': | |
553 | token->type = LEX_T_RSQUARE; | |
554 | p++; | |
555 | break; | |
556 | ||
557 | case '=': | |
558 | p++; | |
559 | if (*p == '=') { | |
560 | token->type = LEX_T_EQ; | |
561 | p++; | |
562 | } else { | |
563 | token->type = LEX_T_EQUALS; | |
564 | } | |
565 | break; | |
566 | ||
567 | case '!': | |
568 | p++; | |
569 | if (*p == '=') { | |
570 | token->type = LEX_T_NE; | |
571 | p++; | |
572 | } else { | |
573 | token->type = LEX_T_LOG_NOT; | |
574 | } | |
575 | break; | |
576 | ||
577 | case '&': | |
578 | p++; | |
579 | if (*p == '&') { | |
580 | token->type = LEX_T_LOG_AND; | |
581 | p++; | |
582 | } else { | |
583 | lex_error(token, "`&' is only valid as part of `&&'."); | |
584 | } | |
585 | break; | |
586 | ||
587 | case '|': | |
588 | p++; | |
589 | if (*p == '|') { | |
590 | token->type = LEX_T_LOG_OR; | |
591 | p++; | |
592 | } else { | |
593 | lex_error(token, "`|' is only valid as part of `||'."); | |
594 | } | |
595 | break; | |
596 | ||
597 | case '<': | |
598 | p++; | |
599 | if (*p == '=') { | |
600 | token->type = LEX_T_LE; | |
601 | p++; | |
602 | } else { | |
603 | token->type = LEX_T_LT; | |
604 | } | |
605 | break; | |
606 | ||
607 | case '>': | |
608 | p++; | |
609 | if (*p == '=') { | |
610 | token->type = LEX_T_GE; | |
611 | p++; | |
612 | } else { | |
613 | token->type = LEX_T_GT; | |
614 | } | |
615 | break; | |
616 | ||
617 | case '.': | |
618 | p++; | |
619 | if (*p == '.') { | |
620 | token->type = LEX_T_ELLIPSIS; | |
621 | p++; | |
622 | } else { | |
623 | lex_error(token, "`.' is only valid as part of `..' or a number."); | |
624 | } | |
625 | break; | |
626 | ||
627 | case ',': | |
628 | p++; | |
629 | token->type = LEX_T_COMMA; | |
630 | break; | |
631 | ||
632 | case ';': | |
633 | p++; | |
634 | token->type = LEX_T_SEMICOLON; | |
635 | break; | |
636 | ||
637 | case '0': case '1': case '2': case '3': case '4': | |
638 | case '5': case '6': case '7': case '8': case '9': | |
639 | case ':': | |
640 | p = lex_parse_integer(p, token); | |
641 | break; | |
642 | ||
643 | case '"': | |
644 | p = lex_parse_string(p, token); | |
645 | break; | |
646 | ||
647 | case 'a': case 'b': case 'c': case 'd': case 'e': case 'f': | |
648 | case 'A': case 'B': case 'C': case 'D': case 'E': case 'F': | |
649 | /* We need to distinguish an Ethernet address or IPv6 address from an | |
650 | * identifier. Fortunately, Ethernet addresses and IPv6 addresses that | |
651 | * are ambiguous based on the first character, always start with hex | |
652 | * digits followed by a colon, but identifiers never do. */ | |
653 | p = (p[strspn(p, "0123456789abcdefABCDEF")] == ':' | |
654 | ? lex_parse_integer(p, token) | |
655 | : lex_parse_id(p, token)); | |
656 | break; | |
657 | ||
658 | default: | |
659 | if (lex_is_id1(*p)) { | |
660 | p = lex_parse_id(p, token); | |
661 | } else { | |
662 | if (isprint((unsigned char) *p)) { | |
663 | lex_error(token, "Invalid character `%c' in input.", *p); | |
664 | } else { | |
665 | lex_error(token, "Invalid byte 0x%d in input.", *p); | |
666 | } | |
667 | p++; | |
668 | } | |
669 | break; | |
670 | } | |
671 | ||
672 | return p; | |
673 | } | |
674 | \f | |
675 | /* Initializes 'lexer' for parsing 'input'. | |
676 | * | |
677 | * While the lexer is in use, 'input' must remain available, but the caller | |
678 | * otherwise retains ownership of 'input'. | |
679 | * | |
680 | * The caller must call lexer_get() to obtain the first token. */ | |
681 | void | |
682 | lexer_init(struct lexer *lexer, const char *input) | |
683 | { | |
684 | lexer->input = input; | |
685 | lexer->start = NULL; | |
686 | lex_token_init(&lexer->token); | |
687 | } | |
688 | ||
689 | /* Frees storage associated with 'lexer'. */ | |
690 | void | |
691 | lexer_destroy(struct lexer *lexer) | |
692 | { | |
693 | lex_token_destroy(&lexer->token); | |
694 | } | |
695 | ||
696 | /* Obtains the next token from 'lexer' into 'lexer->token', and returns the | |
697 | * token's type. The caller may examine 'lexer->token' directly to obtain full | |
698 | * information about the token. */ | |
699 | enum lex_type | |
700 | lexer_get(struct lexer *lexer) | |
701 | { | |
702 | lex_token_destroy(&lexer->token); | |
703 | lexer->input = lex_token_parse(&lexer->token, lexer->input, &lexer->start); | |
704 | return lexer->token.type; | |
705 | } | |
706 | ||
27912fdb BP |
707 | /* Returns the type of the next token that will be fetched by lexer_get(), |
708 | * without advancing 'lexer->token' to that token. */ | |
709 | enum lex_type | |
710 | lexer_lookahead(const struct lexer *lexer) | |
711 | { | |
712 | struct lex_token next; | |
713 | enum lex_type type; | |
714 | const char *start; | |
715 | ||
716 | lex_token_parse(&next, lexer->input, &start); | |
717 | type = next.type; | |
718 | lex_token_destroy(&next); | |
719 | return type; | |
720 | } | |
721 | ||
10b1662b BP |
722 | /* If 'lexer''s current token has the given 'type', advances 'lexer' to the |
723 | * next token and returns true. Otherwise returns false. */ | |
724 | bool | |
725 | lexer_match(struct lexer *lexer, enum lex_type type) | |
726 | { | |
727 | if (lexer->token.type == type) { | |
728 | lexer_get(lexer); | |
729 | return true; | |
730 | } else { | |
731 | return false; | |
732 | } | |
733 | } | |
27912fdb BP |
734 | |
735 | /* If 'lexer''s current token is the identifier given in 'id', advances 'lexer' | |
736 | * to the next token and returns true. Otherwise returns false. */ | |
737 | bool | |
738 | lexer_match_id(struct lexer *lexer, const char *id) | |
739 | { | |
740 | if (lexer->token.type == LEX_T_ID && !strcmp(lexer->token.s, id)) { | |
741 | lexer_get(lexer); | |
742 | return true; | |
743 | } else { | |
744 | return false; | |
745 | } | |
746 | } |