xref: /freebsd/contrib/libpcap/scanner.l (revision a7dea1671b87c07d2d266f836bfa8b58efc7c134)
1 %top {
2 /* Must come first for _LARGE_FILE_API on AIX. */
3 #ifdef HAVE_CONFIG_H
4 #include <config.h>
5 #endif
6 
7 /*
8  * Must come first to avoid warnings on Windows.
9  *
10  * Flex-generated scanners may only include <inttypes.h> if __STDC_VERSION__
11  * is defined with a value >= 199901, meaning "full C99", and MSVC may not
12  * define it with that value, because it isn't 100% C99-compliant, even
13  * though it has an <inttypes.h> capable of defining everything the Flex
14  * scanner needs.
15  *
16  * We, however, will include it if we know we have an MSVC version that has
17  * it; this means that we may define the INTn_MAX and UINTn_MAX values in
18  * scanner.c, and then include <stdint.h>, which may define them differently
19  * (same value, but different string of characters), causing compiler warnings.
20  *
21  * If we include it here, and they're defined, that'll prevent scanner.c
22  * from defining them.  So we include <pcap/pcap-inttypes.h>, to get
23  * <inttypes.h> if we have it.
24  */
25 #include <pcap/pcap-inttypes.h>
26 
27 #include "diag-control.h"
28 }
29 
30 /*
31  * We want a reentrant scanner.
32  */
33 %option reentrant
34 
35 /*
36  * And we need to pass the compiler state to the scanner.
37  */
38 %option extra-type="compiler_state_t *"
39 
40 /*
41  * We don't use input, so don't generate code for it.
42  */
43 %option noinput
44 
45 /*
46  * We don't use unput, so don't generate code for it.
47  */
48 %option nounput
49 
50 /*
51  * We don't read from the terminal.
52  */
53 %option never-interactive
54 
55 /*
56  * We want to stop processing when we get to the end of the input.
57  */
58 %option noyywrap
59 
60 /*
61  * We want to generate code that can be used by a reentrant parser
62  * generated by Bison or Berkeley YACC.
63  */
64 %option bison-bridge
65 
66 %{
67 /*
68  * Copyright (c) 1988, 1989, 1990, 1991, 1992, 1993, 1994, 1995, 1996, 1997
69  *	The Regents of the University of California.  All rights reserved.
70  *
71  * Redistribution and use in source and binary forms, with or without
72  * modification, are permitted provided that: (1) source code distributions
73  * retain the above copyright notice and this paragraph in its entirety, (2)
74  * distributions including binary code include the above copyright notice and
75  * this paragraph in its entirety in the documentation or other materials
76  * provided with the distribution, and (3) all advertising materials mentioning
77  * features or use of this software display the following acknowledgement:
78  * ``This product includes software developed by the University of California,
79  * Lawrence Berkeley Laboratory and its contributors.'' Neither the name of
80  * the University nor the names of its contributors may be used to endorse
81  * or promote products derived from this software without specific prior
82  * written permission.
83  * THIS SOFTWARE IS PROVIDED ``AS IS'' AND WITHOUT ANY EXPRESS OR IMPLIED
84  * WARRANTIES, INCLUDING, WITHOUT LIMITATION, THE IMPLIED WARRANTIES OF
85  * MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE.
86  */
87 
88 #include <ctype.h>
89 #include <string.h>
90 
91 #include "pcap-int.h"
92 
93 #include "gencode.h"
94 
95 #include "grammar.h"
96 
97 /*
98  * Earlier versions of Flex don't declare these, so we declare them
99  * ourselves to squelch warnings.
100  */
101 int pcap_get_column(yyscan_t);
102 void pcap_set_column(int, yyscan_t);
103 
104 #ifdef INET6
105 
106 #ifdef _WIN32
107 #include <winsock2.h>
108 #include <ws2tcpip.h>
109 /*
110  * To quote the MSDN page for getaddrinfo() at
111  *
112  *    https://msdn.microsoft.com/en-us/library/windows/desktop/ms738520(v=vs.85).aspx
113  *
114  * "Support for getaddrinfo on Windows 2000 and older versions
115  * The getaddrinfo function was added to the Ws2_32.dll on Windows XP and
116  * later. To execute an application that uses this function on earlier
117  * versions of Windows, then you need to include the Ws2tcpip.h and
118  * Wspiapi.h files. When the Wspiapi.h include file is added, the
119  * getaddrinfo function is defined to the WspiapiGetAddrInfo inline
120  * function in the Wspiapi.h file. At runtime, the WspiapiGetAddrInfo
121  * function is implemented in such a way that if the Ws2_32.dll or the
122  * Wship6.dll (the file containing getaddrinfo in the IPv6 Technology
123  * Preview for Windows 2000) does not include getaddrinfo, then a
124  * version of getaddrinfo is implemented inline based on code in the
125  * Wspiapi.h header file. This inline code will be used on older Windows
126  * platforms that do not natively support the getaddrinfo function."
127  *
128  * We use getaddrinfo(), so we include Wspiapi.h here.
129  */
130 #include <wspiapi.h>
131 #else /* _WIN32 */
132 #include <sys/socket.h>	/* for "struct sockaddr" in "struct addrinfo" */
133 #include <netdb.h>	/* for "struct addrinfo" */
134 #endif /* _WIN32 */
135 
136 /* Workaround for AIX 4.3 */
137 #if !defined(AI_NUMERICHOST)
138 #define AI_NUMERICHOST 0x04
139 #endif
140 
141 #endif /*INET6*/
142 
143 #include <pcap/namedb.h>
144 #include "grammar.h"
145 
146 #ifdef HAVE_OS_PROTO_H
147 #include "os-proto.h"
148 #endif
149 
150 static int stoi(char *);
151 static inline int xdtoi(int);
152 
153 /*
154  * Disable diagnostics in the code generated by Flex.
155  */
156 DIAG_OFF_FLEX
157 
158 %}
159 
160 N		([0-9]+|(0X|0x)[0-9A-Fa-f]+)
161 B		([0-9A-Fa-f][0-9A-Fa-f]?)
162 B2		([0-9A-Fa-f][0-9A-Fa-f][0-9A-Fa-f][0-9A-Fa-f])
163 W		([0-9A-Fa-f][0-9A-Fa-f]?[0-9A-Fa-f]?[0-9A-Fa-f]?)
164 
165 %a 18400
166 %o 21500
167 %e 7600
168 %k 4550
169 %p 27600
170 %n 2000
171 
172 V680		{W}:{W}:{W}:{W}:{W}:{W}:{W}:{W}
173 
174 V670		::{W}:{W}:{W}:{W}:{W}:{W}:{W}
175 V671		{W}::{W}:{W}:{W}:{W}:{W}:{W}
176 V672		{W}:{W}::{W}:{W}:{W}:{W}:{W}
177 V673		{W}:{W}:{W}::{W}:{W}:{W}:{W}
178 V674		{W}:{W}:{W}:{W}::{W}:{W}:{W}
179 V675		{W}:{W}:{W}:{W}:{W}::{W}:{W}
180 V676		{W}:{W}:{W}:{W}:{W}:{W}::{W}
181 V677		{W}:{W}:{W}:{W}:{W}:{W}:{W}::
182 
183 V660		::{W}:{W}:{W}:{W}:{W}:{W}
184 V661		{W}::{W}:{W}:{W}:{W}:{W}
185 V662		{W}:{W}::{W}:{W}:{W}:{W}
186 V663		{W}:{W}:{W}::{W}:{W}:{W}
187 V664		{W}:{W}:{W}:{W}::{W}:{W}
188 V665		{W}:{W}:{W}:{W}:{W}::{W}
189 V666		{W}:{W}:{W}:{W}:{W}:{W}::
190 
191 V650		::{W}:{W}:{W}:{W}:{W}
192 V651		{W}::{W}:{W}:{W}:{W}
193 V652		{W}:{W}::{W}:{W}:{W}
194 V653		{W}:{W}:{W}::{W}:{W}
195 V654		{W}:{W}:{W}:{W}::{W}
196 V655		{W}:{W}:{W}:{W}:{W}::
197 
198 V640		::{W}:{W}:{W}:{W}
199 V641		{W}::{W}:{W}:{W}
200 V642		{W}:{W}::{W}:{W}
201 V643		{W}:{W}:{W}::{W}
202 V644		{W}:{W}:{W}:{W}::
203 
204 V630		::{W}:{W}:{W}
205 V631		{W}::{W}:{W}
206 V632		{W}:{W}::{W}
207 V633		{W}:{W}:{W}::
208 
209 V620		::{W}:{W}
210 V621		{W}::{W}
211 V622		{W}:{W}::
212 
213 V610		::{W}
214 V611		{W}::
215 
216 V600		::
217 
218 V6604		{W}:{W}:{W}:{W}:{W}:{W}:{N}\.{N}\.{N}\.{N}
219 
220 V6504		::{W}:{W}:{W}:{W}:{W}:{N}\.{N}\.{N}\.{N}
221 V6514		{W}::{W}:{W}:{W}:{W}:{N}\.{N}\.{N}\.{N}
222 V6524		{W}:{W}::{W}:{W}:{W}:{N}\.{N}\.{N}\.{N}
223 V6534		{W}:{W}:{W}::{W}:{W}:{N}\.{N}\.{N}\.{N}
224 V6544		{W}:{W}:{W}:{W}::{W}:{N}\.{N}\.{N}\.{N}
225 V6554		{W}:{W}:{W}:{W}:{W}::{N}\.{N}\.{N}\.{N}
226 
227 V6404		::{W}:{W}:{W}:{W}:{N}\.{N}\.{N}\.{N}
228 V6414		{W}::{W}:{W}:{W}:{N}\.{N}\.{N}\.{N}
229 V6424		{W}:{W}::{W}:{W}:{N}\.{N}\.{N}\.{N}
230 V6434		{W}:{W}:{W}::{W}:{N}\.{N}\.{N}\.{N}
231 V6444		{W}:{W}:{W}:{W}::{N}\.{N}\.{N}\.{N}
232 
233 V6304		::{W}:{W}:{W}:{N}\.{N}\.{N}\.{N}
234 V6314		{W}::{W}:{W}:{N}\.{N}\.{N}\.{N}
235 V6324		{W}:{W}::{W}:{N}\.{N}\.{N}\.{N}
236 V6334		{W}:{W}:{W}::{N}\.{N}\.{N}\.{N}
237 
238 V6204		::{W}:{W}:{N}\.{N}\.{N}\.{N}
239 V6214		{W}::{W}:{N}\.{N}\.{N}\.{N}
240 V6224		{W}:{W}::{N}\.{N}\.{N}\.{N}
241 
242 V6104		::{W}:{N}\.{N}\.{N}\.{N}
243 V6114		{W}::{N}\.{N}\.{N}\.{N}
244 
245 V6004		::{N}\.{N}\.{N}\.{N}
246 
247 
248 V6		({V680}|{V670}|{V671}|{V672}|{V673}|{V674}|{V675}|{V676}|{V677}|{V660}|{V661}|{V662}|{V663}|{V664}|{V665}|{V666}|{V650}|{V651}|{V652}|{V653}|{V654}|{V655}|{V640}|{V641}|{V642}|{V643}|{V644}|{V630}|{V631}|{V632}|{V633}|{V620}|{V621}|{V622}|{V610}|{V611}|{V600}|{V6604}|{V6504}|{V6514}|{V6524}|{V6534}|{V6544}|{V6554}|{V6404}|{V6414}|{V6424}|{V6434}|{V6444}|{V6304}|{V6314}|{V6324}|{V6334}|{V6204}|{V6214}|{V6224}|{V6104}|{V6114}|{V6004})
249 
250 MAC		({B}:{B}:{B}:{B}:{B}:{B}|{B}\-{B}\-{B}\-{B}\-{B}\-{B}|{B}\.{B}\.{B}\.{B}\.{B}\.{B}|{B2}\.{B2}\.{B2}|{B2}{3})
251 
252 
253 
254 %%
255 dst		return DST;
256 src		return SRC;
257 
258 link|ether|ppp|slip  return LINK;
259 fddi|tr|wlan	return LINK;
260 arp		return ARP;
261 rarp		return RARP;
262 ip		return IP;
263 sctp		return SCTP;
264 tcp		return TCP;
265 udp		return UDP;
266 icmp		return ICMP;
267 igmp		return IGMP;
268 igrp		return IGRP;
269 pim		return PIM;
270 vrrp		return VRRP;
271 carp		return CARP;
272 radio		return RADIO;
273 
274 ip6		return IPV6;
275 icmp6		return ICMPV6;
276 ah		return AH;
277 esp		return ESP;
278 
279 atalk		return ATALK;
280 aarp		return AARP;
281 decnet		return DECNET;
282 lat		return LAT;
283 sca		return SCA;
284 moprc		return MOPRC;
285 mopdl		return MOPDL;
286 
287 iso		return ISO;
288 esis		return ESIS;
289 es-is		return ESIS;
290 isis		return ISIS;
291 is-is		return ISIS;
292 l1              return L1;
293 l2              return L2;
294 iih             return IIH;
295 lsp             return LSP;
296 snp             return SNP;
297 csnp            return CSNP;
298 psnp            return PSNP;
299 
300 clnp		return CLNP;
301 
302 stp		return STP;
303 
304 ipx		return IPX;
305 
306 netbeui		return NETBEUI;
307 
308 host		return HOST;
309 net		return NET;
310 mask		return NETMASK;
311 port		return PORT;
312 portrange	return PORTRANGE;
313 proto		return PROTO;
314 protochain	return PROTOCHAIN;
315 
316 gateway		return GATEWAY;
317 
318 type		return TYPE;
319 subtype		return SUBTYPE;
320 direction|dir	return DIR;
321 address1|addr1	return ADDR1;
322 address2|addr2	return ADDR2;
323 address3|addr3	return ADDR3;
324 address4|addr4	return ADDR4;
325 ra		return RA;
326 ta		return TA;
327 
328 less		return LESS;
329 greater		return GREATER;
330 byte		return CBYTE;
331 broadcast	return TK_BROADCAST;
332 multicast	return TK_MULTICAST;
333 
334 and|"&&"	return AND;
335 or|"||"		return OR;
336 not		return '!';
337 
338 len|length	return LEN;
339 inbound		return INBOUND;
340 outbound	return OUTBOUND;
341 
342 vlan		return VLAN;
343 mpls		return MPLS;
344 pppoed		return PPPOED;
345 pppoes		return PPPOES;
346 geneve		return GENEVE;
347 
348 lane		return LANE;
349 llc		return LLC;
350 metac		return METAC;
351 bcc		return BCC;
352 oam		return OAM;
353 oamf4		return OAMF4;
354 oamf4ec		return OAMF4EC;
355 oamf4sc		return OAMF4SC;
356 sc		return SC;
357 ilmic		return ILMIC;
358 vpi		return VPI;
359 vci		return VCI;
360 connectmsg	return CONNECTMSG;
361 metaconnect	return METACONNECT;
362 
363 on|ifname	return PF_IFNAME;
364 rset|ruleset	return PF_RSET;
365 rnr|rulenum	return PF_RNR;
366 srnr|subrulenum	return PF_SRNR;
367 reason		return PF_REASON;
368 action		return PF_ACTION;
369 
370 fisu		return FISU;
371 lssu		return LSSU;
372 lsu		return LSSU;
373 msu		return MSU;
374 hfisu		return HFISU;
375 hlssu		return HLSSU;
376 hmsu		return HMSU;
377 sio		return SIO;
378 opc		return OPC;
379 dpc		return DPC;
380 sls		return SLS;
381 hsio		return HSIO;
382 hopc		return HOPC;
383 hdpc		return HDPC;
384 hsls		return HSLS;
385 
386 [ \r\n\t]		;
387 [+\-*/%:\[\]!<>()&|\^=]	return yytext[0];
388 ">="			return GEQ;
389 "<="			return LEQ;
390 "!="			return NEQ;
391 "=="			return '=';
392 "<<"			return LSH;
393 ">>"			return RSH;
394 ${B}			{ yylval->s = sdup(yyextra, yytext); return AID; }
395 {MAC}			{ yylval->s = sdup(yyextra, yytext); return EID; }
396 {N}			{ yylval->i = stoi((char *)yytext); return NUM; }
397 ({N}\.{N})|({N}\.{N}\.{N})|({N}\.{N}\.{N}\.{N})	{
398 			yylval->s = sdup(yyextra, (char *)yytext); return HID; }
399 {V6}			{
400 #ifdef INET6
401 			  struct addrinfo hints, *res;
402 			  memset(&hints, 0, sizeof(hints));
403 			  hints.ai_family = AF_INET6;
404 			  hints.ai_flags = AI_NUMERICHOST;
405 			  if (getaddrinfo(yytext, NULL, &hints, &res)) {
406 				bpf_set_error(yyextra, "bogus IPv6 address %s", yytext);
407 				yylval->s = NULL;
408 			  } else {
409 				freeaddrinfo(res);
410 				yylval->s = sdup(yyextra, (char *)yytext);
411 			  }
412 #else
413 			  bpf_set_error(yyextra, "IPv6 address %s not supported", yytext);
414 			  yylval->s = NULL;
415 #endif /*INET6*/
416 			  return HID6;
417 			}
418 {B}:+({B}:+)+		{ bpf_set_error(yyextra, "bogus ethernet address %s", yytext); yylval->s = NULL; return EID; }
419 icmptype		{ yylval->i = 0; return NUM; }
420 icmpcode		{ yylval->i = 1; return NUM; }
421 icmp-echoreply		{ yylval->i = 0; return NUM; }
422 icmp-unreach		{ yylval->i = 3; return NUM; }
423 icmp-sourcequench	{ yylval->i = 4; return NUM; }
424 icmp-redirect		{ yylval->i = 5; return NUM; }
425 icmp-echo		{ yylval->i = 8; return NUM; }
426 icmp-routeradvert	{ yylval->i = 9; return NUM; }
427 icmp-routersolicit	{ yylval->i = 10; return NUM; }
428 icmp-timxceed		{ yylval->i = 11; return NUM; }
429 icmp-paramprob		{ yylval->i = 12; return NUM; }
430 icmp-tstamp		{ yylval->i = 13; return NUM; }
431 icmp-tstampreply	{ yylval->i = 14; return NUM; }
432 icmp-ireq		{ yylval->i = 15; return NUM; }
433 icmp-ireqreply		{ yylval->i = 16; return NUM; }
434 icmp-maskreq		{ yylval->i = 17; return NUM; }
435 icmp-maskreply		{ yylval->i = 18; return NUM; }
436 
437 icmp6type       { yylval->i = 0; return NUM; }
438 icmp6code       { yylval->i = 1; return NUM; }
439 
440 icmp6-echo      { yylval->i = 128; return NUM; }
441 icmp6-echoreply { yylval->i = 129; return NUM; }
442 icmp6-multicastlistenerquery    { yylval->i = 130; return NUM; }
443 icmp6-multicastlistenerreportv1 { yylval->i = 131; return NUM; }
444 icmp6-multicastlistenerdone     { yylval->i = 132; return NUM; }
445 icmp6-routersolicit   { yylval->i = 133; return NUM; }
446 icmp6-routeradvert    { yylval->i = 134; return NUM; }
447 icmp6-neighborsolicit { yylval->i = 135; return NUM; }
448 icmp6-neighboradvert  { yylval->i = 136; return NUM; }
449 icmp6-redirect    { yylval->i = 137; return NUM; }
450 icmp6-routerrenum { yylval->i = 138; return NUM; }
451 icmp6-nodeinformationquery      { yylval->i = 139; return NUM; }
452 icmp6-nodeinformationresponse   { yylval->i = 140; return NUM; }
453 icmp6-ineighbordiscoverysolicit { yylval->i = 141; return NUM; }
454 icmp6-ineighbordiscoveryadvert  { yylval->i = 142; return NUM; }
455 icmp6-multicastlistenerreportv2 { yylval->i = 143; return NUM; }
456 icmp6-homeagentdiscoveryrequest { yylval->i = 144; return NUM; }
457 icmp6-homeagentdiscoveryreply   { yylval->i = 145; return NUM; }
458 icmp6-mobileprefixsolicit       { yylval->i = 146; return NUM; }
459 icmp6-mobileprefixadvert        { yylval->i = 147; return NUM; }
460 icmp6-certpathsolicit           { yylval->i = 148; return NUM; }
461 icmp6-certpathadvert            { yylval->i = 149; return NUM; }
462 icmp6-multicastrouteradvert     { yylval->i = 151; return NUM; }
463 icmp6-multicastroutersolicit    { yylval->i = 152; return NUM; }
464 icmp6-multicastrouterterm       { yylval->i = 153; return NUM; }
465 
466 tcpflags		{ yylval->i = 13; return NUM; }
467 tcp-fin			{ yylval->i = 0x01; return NUM; }
468 tcp-syn			{ yylval->i = 0x02; return NUM; }
469 tcp-rst			{ yylval->i = 0x04; return NUM; }
470 tcp-push		{ yylval->i = 0x08; return NUM; }
471 tcp-ack			{ yylval->i = 0x10; return NUM; }
472 tcp-urg			{ yylval->i = 0x20; return NUM; }
473 tcp-ece			{ yylval->i = 0x40; return NUM; }
474 tcp-cwr			{ yylval->i = 0x80; return NUM; }
475 [A-Za-z0-9]([-_.A-Za-z0-9]*[.A-Za-z0-9])? {
476 			 yylval->s = sdup(yyextra, (char *)yytext); return ID; }
477 "\\"[^ !()\n\t]+	{ yylval->s = sdup(yyextra, (char *)yytext + 1); return ID; }
478 .			{ return LEX_ERROR; }
479 %%
480 
481 /*
482  * Turn diagnostics back on, so we check the code that we've written.
483  */
484 DIAG_ON_FLEX
485 
486 /* Hex digit to integer. */
487 static inline int
488 xdtoi(int c)
489 {
490 	if (isdigit(c))
491 		return c - '0';
492 	else if (islower(c))
493 		return c - 'a' + 10;
494 	else
495 		return c - 'A' + 10;
496 }
497 
498 /*
499  * Convert string to integer.  Just like atoi(), but checks for
500  * preceding 0x or 0 and uses hex or octal instead of decimal.
501  */
502 static int
503 stoi(char *s)
504 {
505 	int base = 10;
506 	int n = 0;
507 
508 	if (*s == '0') {
509 		if (s[1] == 'x' || s[1] == 'X') {
510 			s += 2;
511 			base = 16;
512 		}
513 		else {
514 			base = 8;
515 			s += 1;
516 		}
517 	}
518 	while (*s)
519 		n = n * base + xdtoi(*s++);
520 
521 	return n;
522 }
523