xref: /freebsd/contrib/ldns/parse.c (revision 3a0793336edfc21cb6d4c8c5c5d7f1665f3e6c5a) 
1 /*
2  * a generic (simple) parser. Use to parse rr's, private key
3  * information and /etc/resolv.conf files
4  *
5  * a Net::DNS like library for C
6  * LibDNS Team @ NLnet Labs
7  * (c) NLnet Labs, 2005-2006
8  * See the file LICENSE for the license
9  */
10 #include <ldns/config.h>
11 #include <ldns/ldns.h>
12 
13 #include <limits.h>
14 #include <strings.h>
15 
16 ldns_lookup_table ldns_directive_types[] = {
17         { LDNS_DIR_TTL, "$TTL" },
18         { LDNS_DIR_ORIGIN, "$ORIGIN" },
19         { LDNS_DIR_INCLUDE, "$INCLUDE" },
20         { 0, NULL }
21 };
22 
23 /* add max_limit here? */
24 ssize_t
25 ldns_fget_token(FILE *f, char *token, const char *delim, size_t limit)
26 {
27 	return ldns_fget_token_l(f, token, delim, limit, NULL);
28 }
29 
30 enum file_type2parse {
31 	zone_file_type, resolv_conf_file_type
32 };
33 
34 static ldns_status
35 ldns_fget_token_l_st_file_type(FILE *f, char **token, size_t *limit,
36 		bool fixed, const char *delim, int *line_nr,
37 		enum file_type2parse file_type)
38 {
39 	int c, prev_c;
40 	int p; /* 0 -> no parentheses seen, >0 nr of ( seen */
41 	int com, quoted;
42 	char *t, *old_token;
43 	size_t i;
44 	const char *d;
45 	const char *del;
46 
47 	/* standard delimiters */
48 	if (!delim) {
49 		/* from isspace(3) */
50 		del = LDNS_PARSE_NORMAL;
51 	} else {
52 		del = delim;
53 	}
54 	if (!token || !limit)
55 		return LDNS_STATUS_NULL;
56 
57 	if (fixed) {
58 		if (*token == NULL || *limit == 0)
59 			return LDNS_STATUS_NULL;
60 
61 	} else if (*token == NULL) {
62 		*limit = LDNS_MAX_LINELEN;
63 		if (!(*token = LDNS_XMALLOC(char, *limit + 1)))
64 			return LDNS_STATUS_MEM_ERR;
65 
66 	} else if (*limit == 0)
67 		return LDNS_STATUS_ERR;
68 	p = 0;
69 	i = 0;
70 	com = 0;
71 	quoted = 0;
72 	prev_c = 0;
73 	t = *token;
74 	if (del[0] == '"') {
75 		quoted = 1;
76 	}
77 	while ((c = getc(f)) != EOF) {
78 		if (c == '\r') /* carriage return */
79 			c = ' ';
80 		if (c == '(' && prev_c != '\\' && !quoted) {
81 			/* this only counts for non-comments */
82 			if (com == 0) {
83 				p++;
84 			}
85 			prev_c = c;
86 			continue;
87 		}
88 
89 		if (c == ')' && prev_c != '\\' && !quoted) {
90 			/* this only counts for non-comments */
91 			if (com == 0) {
92 				p--;
93 			}
94 			prev_c = c;
95 			continue;
96 		}
97 
98 		if (p < 0) {
99 			/* more ) then ( - close off the string */
100 			*t = '\0';
101 			return i == 0 ? LDNS_STATUS_SYNTAX_EMPTY
102 			              : LDNS_STATUS_OK;
103 		}
104 
105 		/* do something with comments ; */
106 		if ((c == ';'
107 		||  (c == '#' && file_type == resolv_conf_file_type))
108 				&& quoted == 0) {
109 			if (prev_c != '\\') {
110 				com = 1;
111 			}
112 		}
113 		if (c == '\"' && com == 0 && prev_c != '\\') {
114 			quoted = 1 - quoted;
115 		}
116 
117 		if (c == '\n' && com != 0) {
118 			/* comments */
119 			com = 0;
120 			*t = ' ';
121 			if (line_nr) {
122 				*line_nr = *line_nr + 1;
123 			}
124 			if (p == 0 && i > 0) {
125 				goto tokenread;
126 			} else {
127 				prev_c = c;
128 				continue;
129 			}
130 		}
131 
132 		if (com == 1) {
133 			*t = ' ';
134 			prev_c = c;
135 			continue;
136 		}
137 
138 		if (c == '\n' && p != 0 && t > *token) {
139 			/* in parentheses */
140 			if (line_nr) {
141 				*line_nr = *line_nr + 1;
142 			}
143 			if (*limit > 0
144 			&&  (i >= *limit || (size_t)(t - *token) >= *limit)) {
145 				if (fixed) {
146 					*t = '\0';
147 					return LDNS_STATUS_SYNTAX_ERR;
148 				}
149 				old_token = *token;
150 				*limit *= 2;
151 				*token = LDNS_XREALLOC(*token, char, *limit + 1);
152 				if (*token == NULL) {
153 					*token = old_token;
154 					*t = '\0';
155 					return LDNS_STATUS_MEM_ERR;
156 				}
157 				if (*token != old_token)
158 					t = *token + (t - old_token);
159 			}
160 			*t++ = ' ';
161 			prev_c = c;
162 			continue;
163 		}
164 
165 		/* check if we hit the delim */
166 		for (d = del; *d; d++) {
167 			if (c == *d && i > 0 && prev_c != '\\' && p == 0) {
168 				if (c == '\n' && line_nr) {
169 					*line_nr = *line_nr + 1;
170 				}
171 				goto tokenread;
172 			}
173 		}
174 		if (c != '\0' && c != '\n') {
175 			i++;
176 		}
177 		if (*limit > 0
178 		&&  (i >= *limit || (size_t)(t - *token) >= *limit)) {
179 			if (fixed) {
180 				*t = '\0';
181 				return LDNS_STATUS_SYNTAX_ERR;
182 			}
183 			old_token = *token;
184 			*limit *= 2;
185 			*token = LDNS_XREALLOC(*token, char, *limit + 1);
186 			if (*token == NULL) {
187 				*token = old_token;
188 				*t = '\0';
189 				return LDNS_STATUS_MEM_ERR;
190 			}
191 			if (*token != old_token)
192 				t = *token + (t - old_token);
193 		}
194 		if (c != '\0' && c != '\n') {
195 			*t++ = c;
196 		}
197 		if (c == '\n' && line_nr) {
198 			*line_nr = *line_nr + 1;
199 		}
200 		if (c == '\\' && prev_c == '\\')
201 			prev_c = 0;
202 		else	prev_c = c;
203 	}
204 	*t = '\0';
205 	if (c == EOF) {
206 		return i == 0 ? LDNS_STATUS_SYNTAX_EMPTY : LDNS_STATUS_OK;
207 	}
208 
209 	if (p != 0) {
210 		return LDNS_STATUS_SYNTAX_ERR;
211 	}
212 	return i == 0 ? LDNS_STATUS_SYNTAX_EMPTY : LDNS_STATUS_OK;
213 
214 tokenread:
215 	if(*del == '"') /* do not skip over quotes, they are significant */
216 		ldns_fskipcs_l(f, del+1, line_nr);
217 	else	ldns_fskipcs_l(f, del, line_nr);
218 	*t = '\0';
219 	if (p != 0) {
220 		return LDNS_STATUS_SYNTAX_ERR;
221 	}
222 	return i == 0 ? LDNS_STATUS_SYNTAX_EMPTY : LDNS_STATUS_OK;
223 }
224 
225 ldns_status
226 ldns_fget_token_l_st(FILE *f, char **token, size_t *limit, bool fixed
227                     , const char *delim, int *line_nr)
228 {
229 	return ldns_fget_token_l_st_file_type(
230 		f, token, limit, fixed, delim, line_nr, zone_file_type);
231 }
232 
233 ssize_t
234 ldns_fget_token_l_resolv_conf(FILE *f, char *token, const char *delim,
235 		size_t limit, int *line_nr)
236 {
237 	if (limit == 0)
238 		limit = LDNS_MAX_LINELEN;
239 	if (ldns_fget_token_l_st_file_type(f, &token, &limit, true, delim,
240 				line_nr, resolv_conf_file_type))
241 		return -1;
242 	else
243 		return (ssize_t)strlen(token);
244 }
245 
246 ssize_t
247 ldns_fget_token_l(FILE *f, char *token, const char *delim, size_t limit, int *line_nr)
248 {
249 	if (limit == 0)
250 		limit = LDNS_MAX_LINELEN;
251 	if (ldns_fget_token_l_st(f, &token, &limit, true, delim, line_nr))
252 		return -1;
253 	else
254 		return (ssize_t)strlen(token);
255 }
256 
257 ssize_t
258 ldns_fget_keyword_data(FILE *f, const char *keyword, const char *k_del, char *data,
259                const char *d_del, size_t data_limit)
260 {
261        return ldns_fget_keyword_data_l(f, keyword, k_del, data, d_del,
262 		       data_limit, NULL);
263 }
264 
265 ssize_t
266 ldns_fget_keyword_data_l(FILE *f, const char *keyword, const char *k_del, char *data,
267                const char *d_del, size_t data_limit, int *line_nr)
268 {
269        /* we assume: keyword|sep|data */
270        char *fkeyword;
271        ssize_t i;
272 
273        if(strlen(keyword) >= LDNS_MAX_KEYWORDLEN)
274                return -1;
275        fkeyword = LDNS_XMALLOC(char, LDNS_MAX_KEYWORDLEN);
276        if(!fkeyword)
277                return -1;
278 
279        i = ldns_fget_token(f, fkeyword, k_del, LDNS_MAX_KEYWORDLEN);
280        if(i==0 || i==-1) {
281                LDNS_FREE(fkeyword);
282                return -1;
283        }
284 
285        /* case??? i instead of strlen? */
286        if (strncmp(fkeyword, keyword, LDNS_MAX_KEYWORDLEN - 1) == 0) {
287                /* whee! */
288                /* printf("%s\n%s\n", "Matching keyword", fkeyword); */
289                i = ldns_fget_token_l(f, data, d_del, data_limit, line_nr);
290                LDNS_FREE(fkeyword);
291                return i;
292        } else {
293                /*printf("no match for %s (read: %s)\n", keyword, fkeyword);*/
294                LDNS_FREE(fkeyword);
295                return -1;
296        }
297 }
298 
299 
300 ssize_t
301 ldns_bget_token(ldns_buffer *b, char *token, const char *delim, size_t limit)
302 {
303 	int c, lc;
304 	int p; /* 0 -> no parentheses seen, >0 nr of ( seen */
305 	int com, quoted;
306 	char *t;
307 	size_t i;
308 	const char *d;
309 	const char *del;
310 
311 	/* standard delimiters */
312 	if (!delim) {
313 		/* from isspace(3) */
314 		del = LDNS_PARSE_NORMAL;
315 	} else {
316 		del = delim;
317 	}
318 
319 	p = 0;
320 	i = 0;
321 	com = 0;
322 	quoted = 0;
323 	t = token;
324 	lc = 0;
325 	if (del[0] == '"') {
326 		quoted = 1;
327 	}
328 
329 	while ((c = ldns_bgetc(b)) != EOF) {
330 		if (c == '\r') /* carriage return */
331 			c = ' ';
332 		if (c == '(' && lc != '\\' && !quoted) {
333 			/* this only counts for non-comments */
334 			if (com == 0) {
335 				p++;
336 			}
337 			lc = c;
338 			continue;
339 		}
340 
341 		if (c == ')' && lc != '\\' && !quoted) {
342 			/* this only counts for non-comments */
343 			if (com == 0) {
344 				p--;
345 			}
346 			lc = c;
347 			continue;
348 		}
349 
350 		if (p < 0) {
351 			/* more ) then ( */
352 			*t = '\0';
353 			return 0;
354 		}
355 
356 		/* do something with comments ; */
357 		if (c == ';' && quoted == 0) {
358 			if (lc != '\\') {
359 				com = 1;
360 			}
361 		}
362 		if (c == '"' && com == 0 && lc != '\\') {
363 			quoted = 1 - quoted;
364 		}
365 
366 		if (c == '\n' && com != 0) {
367 			/* comments */
368 			com = 0;
369 			*t = ' ';
370 			lc = c;
371 			continue;
372 		}
373 
374 		if (com == 1) {
375 			*t = ' ';
376 			lc = c;
377 			continue;
378 		}
379 
380 		if (c == '\n' && p != 0) {
381 			/* in parentheses */
382 			*t++ = ' ';
383 			lc = c;
384 			continue;
385 		}
386 
387 		/* check if we hit the delim */
388 		for (d = del; *d; d++) {
389                         if (c == *d && lc != '\\' && p == 0) {
390 				goto tokenread;
391                         }
392 		}
393 
394 		i++;
395 		if (limit > 0 && (i >= limit || (size_t)(t-token) >= limit)) {
396 			*t = '\0';
397 			return -1;
398 		}
399 		*t++ = c;
400 
401 		if (c == '\\' && lc == '\\') {
402 			lc = 0;
403 		} else {
404 			lc = c;
405 		}
406 	}
407 	*t = '\0';
408 	if (i == 0) {
409 		/* nothing read */
410 		return -1;
411 	}
412 	if (p != 0) {
413 		return -1;
414 	}
415 	return (ssize_t)i;
416 
417 tokenread:
418 	if(*del == '"') /* do not skip over quotes, they are significant */
419 		ldns_bskipcs(b, del+1);
420 	else	ldns_bskipcs(b, del);
421 	*t = '\0';
422 
423 	if (p != 0) {
424 		return -1;
425 	}
426 	return (ssize_t)i;
427 }
428 
429 
430 void
431 ldns_bskipcs(ldns_buffer *buffer, const char *s)
432 {
433         bool found;
434         char c;
435         const char *d;
436 
437         while(ldns_buffer_available_at(buffer, buffer->_position, sizeof(char))) {
438                 c = (char) ldns_buffer_read_u8_at(buffer, buffer->_position);
439                 found = false;
440                 for (d = s; *d; d++) {
441                         if (*d == c) {
442                                 found = true;
443                         }
444                 }
445                 if (found && buffer->_limit > buffer->_position) {
446                         buffer->_position += sizeof(char);
447                 } else {
448                         return;
449                 }
450         }
451 }
452 
453 void
454 ldns_fskipcs(FILE *fp, const char *s)
455 {
456 	ldns_fskipcs_l(fp, s, NULL);
457 }
458 
459 void
460 ldns_fskipcs_l(FILE *fp, const char *s, int *line_nr)
461 {
462         bool found;
463         int c;
464         const char *d;
465 
466 	while ((c = fgetc(fp)) != EOF) {
467 		if (line_nr && c == '\n') {
468 			*line_nr = *line_nr + 1;
469 		}
470                 found = false;
471                 for (d = s; *d; d++) {
472                         if (*d == c) {
473                                 found = true;
474                         }
475                 }
476 		if (!found) {
477 			/* with getc, we've read too far */
478 			ungetc(c, fp);
479 			return;
480 		}
481 	}
482 }
483 
484 ssize_t
485 ldns_bget_keyword_data(ldns_buffer *b, const char *keyword, const char *k_del, char
486 *data, const char *d_del, size_t data_limit)
487 {
488        /* we assume: keyword|sep|data */
489        char *fkeyword;
490        ssize_t i;
491 
492        if(strlen(keyword) >= LDNS_MAX_KEYWORDLEN)
493                return -1;
494        fkeyword = LDNS_XMALLOC(char, LDNS_MAX_KEYWORDLEN);
495        if(!fkeyword)
496                return -1; /* out of memory */
497 
498        i = ldns_bget_token(b, fkeyword, k_del, data_limit);
499        if(i==0 || i==-1) {
500                LDNS_FREE(fkeyword);
501                return -1; /* nothing read */
502        }
503 
504        /* case??? */
505        if (strncmp(fkeyword, keyword, strlen(keyword)) == 0) {
506                LDNS_FREE(fkeyword);
507                /* whee, the match! */
508                /* retrieve it's data */
509                i = ldns_bget_token(b, data, d_del, 0);
510                return i;
511        } else {
512                LDNS_FREE(fkeyword);
513                return -1;
514        }
515 }
516 
517