compile.c revision 1.40 1 1.40 christos /* $NetBSD: compile.c,v 1.40 2014/05/05 17:12:11 christos Exp $ */
2 1.15 tls
3 1.1 alm /*-
4 1.9 cgd * Copyright (c) 1992, 1993
5 1.9 cgd * The Regents of the University of California. All rights reserved.
6 1.1 alm *
7 1.1 alm * This code is derived from software contributed to Berkeley by
8 1.1 alm * Diomidis Spinellis of Imperial College, University of London.
9 1.1 alm *
10 1.1 alm * Redistribution and use in source and binary forms, with or without
11 1.1 alm * modification, are permitted provided that the following conditions
12 1.1 alm * are met:
13 1.1 alm * 1. Redistributions of source code must retain the above copyright
14 1.1 alm * notice, this list of conditions and the following disclaimer.
15 1.1 alm * 2. Redistributions in binary form must reproduce the above copyright
16 1.1 alm * notice, this list of conditions and the following disclaimer in the
17 1.1 alm * documentation and/or other materials provided with the distribution.
18 1.25 agc * 3. Neither the name of the University nor the names of its contributors
19 1.25 agc * may be used to endorse or promote products derived from this software
20 1.25 agc * without specific prior written permission.
21 1.25 agc *
22 1.25 agc * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
23 1.25 agc * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
24 1.25 agc * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
25 1.25 agc * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
26 1.25 agc * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
27 1.25 agc * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
28 1.25 agc * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
29 1.25 agc * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
30 1.25 agc * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
31 1.25 agc * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
32 1.25 agc * SUCH DAMAGE.
33 1.25 agc */
34 1.25 agc
35 1.25 agc /*-
36 1.25 agc * Copyright (c) 1992 Diomidis Spinellis.
37 1.25 agc *
38 1.25 agc * This code is derived from software contributed to Berkeley by
39 1.25 agc * Diomidis Spinellis of Imperial College, University of London.
40 1.25 agc *
41 1.25 agc * Redistribution and use in source and binary forms, with or without
42 1.25 agc * modification, are permitted provided that the following conditions
43 1.25 agc * are met:
44 1.25 agc * 1. Redistributions of source code must retain the above copyright
45 1.25 agc * notice, this list of conditions and the following disclaimer.
46 1.25 agc * 2. Redistributions in binary form must reproduce the above copyright
47 1.25 agc * notice, this list of conditions and the following disclaimer in the
48 1.25 agc * documentation and/or other materials provided with the distribution.
49 1.1 alm * 3. All advertising materials mentioning features or use of this software
50 1.1 alm * must display the following acknowledgement:
51 1.1 alm * This product includes software developed by the University of
52 1.1 alm * California, Berkeley and its contributors.
53 1.1 alm * 4. Neither the name of the University nor the names of its contributors
54 1.1 alm * may be used to endorse or promote products derived from this software
55 1.1 alm * without specific prior written permission.
56 1.1 alm *
57 1.1 alm * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
58 1.1 alm * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
59 1.1 alm * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
60 1.1 alm * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
61 1.1 alm * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
62 1.1 alm * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
63 1.1 alm * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
64 1.1 alm * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
65 1.1 alm * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
66 1.1 alm * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
67 1.1 alm * SUCH DAMAGE.
68 1.1 alm */
69 1.1 alm
70 1.34 gdamore #if HAVE_NBTOOL_CONFIG_H
71 1.34 gdamore #include "nbtool_config.h"
72 1.34 gdamore #endif
73 1.34 gdamore
74 1.17 lukem #include <sys/cdefs.h>
75 1.1 alm #ifndef lint
76 1.16 mrg #if 0
77 1.16 mrg static char sccsid[] = "@(#)compile.c 8.2 (Berkeley) 4/28/95";
78 1.16 mrg #else
79 1.40 christos __RCSID("$NetBSD: compile.c,v 1.40 2014/05/05 17:12:11 christos Exp $");
80 1.16 mrg #endif
81 1.1 alm #endif /* not lint */
82 1.1 alm
83 1.1 alm #include <sys/types.h>
84 1.1 alm #include <sys/stat.h>
85 1.1 alm
86 1.1 alm #include <ctype.h>
87 1.1 alm #include <errno.h>
88 1.1 alm #include <fcntl.h>
89 1.1 alm #include <limits.h>
90 1.1 alm #include <regex.h>
91 1.1 alm #include <stdio.h>
92 1.1 alm #include <stdlib.h>
93 1.1 alm #include <string.h>
94 1.1 alm
95 1.1 alm #include "defs.h"
96 1.1 alm #include "extern.h"
97 1.1 alm
98 1.35 christos #ifndef _POSIX2_LINE_MAX
99 1.35 christos #define _POSIX2_LINE_MAX (2 * BUFSIZ)
100 1.35 christos #endif
101 1.35 christos
102 1.9 cgd #define LHSZ 128
103 1.9 cgd #define LHMASK (LHSZ - 1)
104 1.9 cgd static struct labhash {
105 1.9 cgd struct labhash *lh_next;
106 1.9 cgd u_int lh_hash;
107 1.9 cgd struct s_command *lh_cmd;
108 1.9 cgd int lh_ref;
109 1.9 cgd } *labels[LHSZ];
110 1.9 cgd
111 1.24 wiz static char *compile_addr(char *, struct s_addr *);
112 1.24 wiz static char *compile_ccl(char **, char *);
113 1.24 wiz static char *compile_delimited(char *, char *);
114 1.24 wiz static char *compile_flags(char *, struct s_subst *);
115 1.24 wiz static char *compile_re(char *, regex_t **);
116 1.24 wiz static char *compile_subst(char *, struct s_subst *);
117 1.24 wiz static char *compile_text(void);
118 1.24 wiz static char *compile_tr(char *, char **);
119 1.1 alm static struct s_command
120 1.24 wiz **compile_stream(struct s_command **);
121 1.36 lukem static char *duptoeol(char *, const char *);
122 1.24 wiz static void enterlabel(struct s_command *);
123 1.1 alm static struct s_command
124 1.24 wiz *findlabel(char *);
125 1.24 wiz static void fixuplabel(struct s_command *, struct s_command *);
126 1.24 wiz static void uselabel(void);
127 1.1 alm
128 1.1 alm /*
129 1.1 alm * Command specification. This is used to drive the command parser.
130 1.1 alm */
131 1.1 alm struct s_format {
132 1.1 alm char code; /* Command code */
133 1.1 alm int naddr; /* Number of address args */
134 1.1 alm enum e_args args; /* Argument type */
135 1.1 alm };
136 1.1 alm
137 1.1 alm static struct s_format cmd_fmts[] = {
138 1.1 alm {'{', 2, GROUP},
139 1.14 mycroft {'}', 0, ENDGROUP},
140 1.1 alm {'a', 1, TEXT},
141 1.1 alm {'b', 2, BRANCH},
142 1.1 alm {'c', 2, TEXT},
143 1.1 alm {'d', 2, EMPTY},
144 1.1 alm {'D', 2, EMPTY},
145 1.1 alm {'g', 2, EMPTY},
146 1.1 alm {'G', 2, EMPTY},
147 1.1 alm {'h', 2, EMPTY},
148 1.1 alm {'H', 2, EMPTY},
149 1.1 alm {'i', 1, TEXT},
150 1.1 alm {'l', 2, EMPTY},
151 1.1 alm {'n', 2, EMPTY},
152 1.1 alm {'N', 2, EMPTY},
153 1.1 alm {'p', 2, EMPTY},
154 1.1 alm {'P', 2, EMPTY},
155 1.1 alm {'q', 1, EMPTY},
156 1.1 alm {'r', 1, RFILE},
157 1.1 alm {'s', 2, SUBST},
158 1.1 alm {'t', 2, BRANCH},
159 1.1 alm {'w', 2, WFILE},
160 1.1 alm {'x', 2, EMPTY},
161 1.1 alm {'y', 2, TR},
162 1.1 alm {'!', 2, NONSEL},
163 1.1 alm {':', 0, LABEL},
164 1.1 alm {'#', 0, COMMENT},
165 1.1 alm {'=', 1, EMPTY},
166 1.1 alm {'\0', 0, COMMENT},
167 1.1 alm };
168 1.1 alm
169 1.1 alm /* The compiled program. */
170 1.1 alm struct s_command *prog;
171 1.1 alm
172 1.1 alm /*
173 1.1 alm * Compile the program into prog.
174 1.1 alm * Initialise appends.
175 1.1 alm */
176 1.1 alm void
177 1.24 wiz compile(void)
178 1.1 alm {
179 1.14 mycroft *compile_stream(&prog) = NULL;
180 1.9 cgd fixuplabel(prog, NULL);
181 1.9 cgd uselabel();
182 1.18 drochner if (appendnum > 0)
183 1.18 drochner appends = xmalloc(sizeof(struct s_appends) * appendnum);
184 1.1 alm match = xmalloc((maxnsub + 1) * sizeof(regmatch_t));
185 1.1 alm }
186 1.1 alm
187 1.32 christos #define EATSPACE() \
188 1.32 christos while (*p && isascii((unsigned char)*p) && \
189 1.32 christos isspace((unsigned char)*p)) \
190 1.32 christos p++ \
191 1.1 alm
192 1.1 alm static struct s_command **
193 1.24 wiz compile_stream(struct s_command **link)
194 1.14 mycroft {
195 1.17 lukem char *p;
196 1.37 tnn static char *lbuf; /* To avoid excessive malloc calls */
197 1.37 tnn static size_t bufsize;
198 1.14 mycroft struct s_command *cmd, *cmd2, *stack;
199 1.1 alm struct s_format *fp;
200 1.1 alm int naddr; /* Number of addresses */
201 1.1 alm
202 1.14 mycroft stack = 0;
203 1.1 alm for (;;) {
204 1.37 tnn if ((p = cu_fgets(&lbuf, &bufsize)) == NULL) {
205 1.14 mycroft if (stack != 0)
206 1.1 alm err(COMPILE, "unexpected EOF (pending }'s)");
207 1.1 alm return (link);
208 1.1 alm }
209 1.1 alm
210 1.1 alm semicolon: EATSPACE();
211 1.32 christos if (*p == '#' || *p == '\0')
212 1.32 christos continue;
213 1.32 christos else if (*p == ';') {
214 1.32 christos p++;
215 1.32 christos goto semicolon;
216 1.20 kleink }
217 1.1 alm *link = cmd = xmalloc(sizeof(struct s_command));
218 1.1 alm link = &cmd->next;
219 1.1 alm cmd->nonsel = cmd->inrange = 0;
220 1.1 alm /* First parse the addresses */
221 1.1 alm naddr = 0;
222 1.1 alm
223 1.1 alm /* Valid characters to start an address */
224 1.1 alm #define addrchar(c) (strchr("0123456789/\\$", (c)))
225 1.1 alm if (addrchar(*p)) {
226 1.1 alm naddr++;
227 1.1 alm cmd->a1 = xmalloc(sizeof(struct s_addr));
228 1.1 alm p = compile_addr(p, cmd->a1);
229 1.1 alm EATSPACE(); /* EXTENSION */
230 1.1 alm if (*p == ',') {
231 1.1 alm p++;
232 1.1 alm EATSPACE(); /* EXTENSION */
233 1.13 mycroft naddr++;
234 1.1 alm cmd->a2 = xmalloc(sizeof(struct s_addr));
235 1.1 alm p = compile_addr(p, cmd->a2);
236 1.13 mycroft EATSPACE();
237 1.12 mycroft } else
238 1.12 mycroft cmd->a2 = 0;
239 1.12 mycroft } else
240 1.13 mycroft cmd->a1 = cmd->a2 = 0;
241 1.1 alm
242 1.1 alm nonsel: /* Now parse the command */
243 1.1 alm if (!*p)
244 1.1 alm err(COMPILE, "command expected");
245 1.1 alm cmd->code = *p;
246 1.1 alm for (fp = cmd_fmts; fp->code; fp++)
247 1.1 alm if (fp->code == *p)
248 1.1 alm break;
249 1.1 alm if (!fp->code)
250 1.1 alm err(COMPILE, "invalid command code %c", *p);
251 1.1 alm if (naddr > fp->naddr)
252 1.1 alm err(COMPILE,
253 1.1 alm "command %c expects up to %d address(es), found %d", *p, fp->naddr, naddr);
254 1.1 alm switch (fp->args) {
255 1.1 alm case NONSEL: /* ! */
256 1.13 mycroft p++;
257 1.13 mycroft EATSPACE();
258 1.1 alm cmd->nonsel = ! cmd->nonsel;
259 1.1 alm goto nonsel;
260 1.1 alm case GROUP: /* { */
261 1.1 alm p++;
262 1.1 alm EATSPACE();
263 1.14 mycroft cmd->next = stack;
264 1.14 mycroft stack = cmd;
265 1.14 mycroft link = &cmd->u.c;
266 1.14 mycroft if (*p)
267 1.14 mycroft goto semicolon;
268 1.14 mycroft break;
269 1.14 mycroft case ENDGROUP:
270 1.12 mycroft /*
271 1.12 mycroft * Short-circuit command processing, since end of
272 1.12 mycroft * group is really just a noop.
273 1.12 mycroft */
274 1.14 mycroft cmd->nonsel = 1;
275 1.14 mycroft if (stack == 0)
276 1.14 mycroft err(COMPILE, "unexpected }");
277 1.14 mycroft cmd2 = stack;
278 1.14 mycroft stack = cmd2->next;
279 1.14 mycroft cmd2->next = cmd;
280 1.14 mycroft /*FALLTHROUGH*/
281 1.1 alm case EMPTY: /* d D g G h H l n N p P q x = \0 */
282 1.1 alm p++;
283 1.1 alm EATSPACE();
284 1.40 christos switch (*p) {
285 1.40 christos case ';':
286 1.1 alm p++;
287 1.1 alm link = &cmd->next;
288 1.1 alm goto semicolon;
289 1.40 christos case '}':
290 1.40 christos goto semicolon;
291 1.40 christos case '\0':
292 1.40 christos break;
293 1.40 christos default:
294 1.1 alm err(COMPILE,
295 1.1 alm "extra characters at the end of %c command", cmd->code);
296 1.40 christos }
297 1.1 alm break;
298 1.1 alm case TEXT: /* a c i */
299 1.1 alm p++;
300 1.1 alm EATSPACE();
301 1.1 alm if (*p != '\\')
302 1.1 alm err(COMPILE,
303 1.1 alm "command %c expects \\ followed by text", cmd->code);
304 1.1 alm p++;
305 1.1 alm EATSPACE();
306 1.1 alm if (*p)
307 1.1 alm err(COMPILE,
308 1.1 alm "extra characters after \\ at the end of %c command", cmd->code);
309 1.1 alm cmd->t = compile_text();
310 1.1 alm break;
311 1.1 alm case COMMENT: /* \0 # */
312 1.1 alm break;
313 1.1 alm case WFILE: /* w */
314 1.1 alm p++;
315 1.1 alm EATSPACE();
316 1.1 alm if (*p == '\0')
317 1.1 alm err(COMPILE, "filename expected");
318 1.9 cgd cmd->t = duptoeol(p, "w command");
319 1.1 alm if (aflag)
320 1.1 alm cmd->u.fd = -1;
321 1.1 alm else if ((cmd->u.fd = open(p,
322 1.1 alm O_WRONLY|O_APPEND|O_CREAT|O_TRUNC,
323 1.1 alm DEFFILEMODE)) == -1)
324 1.23 itojun err(FATAL, "%s: %s", p, strerror(errno));
325 1.1 alm break;
326 1.1 alm case RFILE: /* r */
327 1.1 alm p++;
328 1.1 alm EATSPACE();
329 1.1 alm if (*p == '\0')
330 1.1 alm err(COMPILE, "filename expected");
331 1.1 alm else
332 1.9 cgd cmd->t = duptoeol(p, "read command");
333 1.1 alm break;
334 1.1 alm case BRANCH: /* b t */
335 1.1 alm p++;
336 1.1 alm EATSPACE();
337 1.1 alm if (*p == '\0')
338 1.1 alm cmd->t = NULL;
339 1.1 alm else
340 1.9 cgd cmd->t = duptoeol(p, "branch");
341 1.1 alm break;
342 1.1 alm case LABEL: /* : */
343 1.1 alm p++;
344 1.1 alm EATSPACE();
345 1.9 cgd cmd->t = duptoeol(p, "label");
346 1.1 alm if (strlen(p) == 0)
347 1.1 alm err(COMPILE, "empty label");
348 1.9 cgd enterlabel(cmd);
349 1.1 alm break;
350 1.1 alm case SUBST: /* s */
351 1.1 alm p++;
352 1.1 alm if (*p == '\0' || *p == '\\')
353 1.1 alm err(COMPILE,
354 1.1 alm "substitute pattern can not be delimited by newline or backslash");
355 1.1 alm cmd->u.s = xmalloc(sizeof(struct s_subst));
356 1.1 alm p = compile_re(p, &cmd->u.s->re);
357 1.1 alm if (p == NULL)
358 1.1 alm err(COMPILE, "unterminated substitute pattern");
359 1.1 alm --p;
360 1.1 alm p = compile_subst(p, cmd->u.s);
361 1.1 alm p = compile_flags(p, cmd->u.s);
362 1.1 alm EATSPACE();
363 1.1 alm if (*p == ';') {
364 1.1 alm p++;
365 1.1 alm link = &cmd->next;
366 1.1 alm goto semicolon;
367 1.1 alm }
368 1.1 alm break;
369 1.1 alm case TR: /* y */
370 1.1 alm p++;
371 1.33 mrg p = compile_tr(p, (char **)(void *)&cmd->u.y);
372 1.1 alm EATSPACE();
373 1.40 christos switch (*p) {
374 1.40 christos case ';':
375 1.1 alm p++;
376 1.1 alm link = &cmd->next;
377 1.1 alm goto semicolon;
378 1.40 christos case '}':
379 1.40 christos goto semicolon;
380 1.40 christos case '\0':
381 1.40 christos break;
382 1.40 christos default:
383 1.1 alm err(COMPILE,
384 1.1 alm "extra text at the end of a transform command");
385 1.40 christos }
386 1.1 alm break;
387 1.1 alm }
388 1.1 alm }
389 1.1 alm }
390 1.1 alm
391 1.1 alm /*
392 1.22 wiz * Get a delimited string. P points to the delimiter of the string; d points
393 1.1 alm * to a buffer area. Newline and delimiter escapes are processed; other
394 1.1 alm * escapes are ignored.
395 1.1 alm *
396 1.1 alm * Returns a pointer to the first character after the final delimiter or NULL
397 1.1 alm * in the case of a non-terminated string. The character array d is filled
398 1.1 alm * with the processed string.
399 1.1 alm */
400 1.1 alm static char *
401 1.24 wiz compile_delimited(char *p, char *d)
402 1.1 alm {
403 1.1 alm char c;
404 1.1 alm
405 1.1 alm c = *p++;
406 1.1 alm if (c == '\0')
407 1.1 alm return (NULL);
408 1.1 alm else if (c == '\\')
409 1.1 alm err(COMPILE, "\\ can not be used as a string delimiter");
410 1.1 alm else if (c == '\n')
411 1.1 alm err(COMPILE, "newline can not be used as a string delimiter");
412 1.1 alm while (*p) {
413 1.11 alm if (*p == '[') {
414 1.11 alm if ((d = compile_ccl(&p, d)) == NULL)
415 1.11 alm err(COMPILE, "unbalanced brackets ([])");
416 1.11 alm continue;
417 1.11 alm } else if (*p == '\\' && p[1] == '[') {
418 1.11 alm *d++ = *p++;
419 1.11 alm } else if (*p == '\\' && p[1] == c)
420 1.1 alm p++;
421 1.1 alm else if (*p == '\\' && p[1] == 'n') {
422 1.1 alm *d++ = '\n';
423 1.1 alm p += 2;
424 1.1 alm continue;
425 1.1 alm } else if (*p == '\\' && p[1] == '\\')
426 1.1 alm *d++ = *p++;
427 1.1 alm else if (*p == c) {
428 1.1 alm *d = '\0';
429 1.1 alm return (p + 1);
430 1.1 alm }
431 1.1 alm *d++ = *p++;
432 1.1 alm }
433 1.1 alm return (NULL);
434 1.11 alm }
435 1.11 alm
436 1.11 alm
437 1.11 alm /* compile_ccl: expand a POSIX character class */
438 1.11 alm static char *
439 1.24 wiz compile_ccl(char **sp, char *t)
440 1.11 alm {
441 1.11 alm int c, d;
442 1.11 alm char *s = *sp;
443 1.11 alm
444 1.11 alm *t++ = *s++;
445 1.11 alm if (*s == '^')
446 1.11 alm *t++ = *s++;
447 1.11 alm if (*s == ']')
448 1.11 alm *t++ = *s++;
449 1.11 alm for (; *s && (*t = *s) != ']'; s++, t++)
450 1.11 alm if (*s == '[' && ((d = *(s+1)) == '.' || d == ':' || d == '=')) {
451 1.11 alm *++t = *++s, t++, s++;
452 1.11 alm for (c = *s; (*t = *s) != ']' || c != d; s++, t++)
453 1.11 alm if ((c = *s) == '\0')
454 1.11 alm return NULL;
455 1.11 alm } else if (*s == '\\' && s[1] == 'n')
456 1.11 alm *t = '\n', s++;
457 1.11 alm return (*s == ']') ? *sp = ++s, ++t : NULL;
458 1.2 alm }
459 1.2 alm
460 1.1 alm /*
461 1.1 alm * Get a regular expression. P points to the delimiter of the regular
462 1.1 alm * expression; repp points to the address of a regexp pointer. Newline
463 1.1 alm * and delimiter escapes are processed; other escapes are ignored.
464 1.1 alm * Returns a pointer to the first character after the final delimiter
465 1.1 alm * or NULL in the case of a non terminated regular expression. The regexp
466 1.1 alm * pointer is set to the compiled regular expression.
467 1.1 alm * Cflags are passed to regcomp.
468 1.1 alm */
469 1.1 alm static char *
470 1.24 wiz compile_re(char *p, regex_t **repp)
471 1.1 alm {
472 1.1 alm int eval;
473 1.37 tnn char *re;
474 1.1 alm
475 1.37 tnn re = xmalloc(strlen(p) + 1); /* strlen(re) <= strlen(p) */
476 1.1 alm p = compile_delimited(p, re);
477 1.1 alm if (p && strlen(re) == 0) {
478 1.1 alm *repp = NULL;
479 1.37 tnn free(re);
480 1.1 alm return (p);
481 1.1 alm }
482 1.1 alm *repp = xmalloc(sizeof(regex_t));
483 1.21 atatat if (p && (eval = regcomp(*repp, re, ere)) != 0)
484 1.1 alm err(COMPILE, "RE error: %s", strregerror(eval, *repp));
485 1.1 alm if (maxnsub < (*repp)->re_nsub)
486 1.1 alm maxnsub = (*repp)->re_nsub;
487 1.38 msaitoh free(re);
488 1.1 alm return (p);
489 1.1 alm }
490 1.1 alm
491 1.1 alm /*
492 1.1 alm * Compile the substitution string of a regular expression and set res to
493 1.1 alm * point to a saved copy of it. Nsub is the number of parenthesized regular
494 1.1 alm * expressions.
495 1.1 alm */
496 1.1 alm static char *
497 1.24 wiz compile_subst(char *p, struct s_subst *s)
498 1.1 alm {
499 1.37 tnn static char *lbuf;
500 1.37 tnn static size_t bufsize;
501 1.37 tnn int asize, ref, size, len;
502 1.1 alm char c, *text, *op, *sp;
503 1.26 minskim int sawesc = 0;
504 1.1 alm
505 1.1 alm c = *p++; /* Terminator character */
506 1.1 alm if (c == '\0')
507 1.1 alm return (NULL);
508 1.1 alm
509 1.1 alm s->maxbref = 0;
510 1.1 alm s->linenum = linenum;
511 1.37 tnn text = NULL;
512 1.37 tnn asize = size = 0;
513 1.1 alm do {
514 1.37 tnn len = ROUNDLEN(strlen(p) + 1);
515 1.37 tnn if (asize - size < len) {
516 1.37 tnn do {
517 1.37 tnn asize += len;
518 1.37 tnn } while (asize - size < len);
519 1.37 tnn text = xrealloc(text, asize);
520 1.37 tnn }
521 1.1 alm op = sp = text + size;
522 1.1 alm for (; *p; p++) {
523 1.26 minskim if (*p == '\\' || sawesc) {
524 1.26 minskim /*
525 1.26 minskim * If this is a continuation from the last
526 1.26 minskim * buffer, we won't have a character to
527 1.26 minskim * skip over.
528 1.26 minskim */
529 1.26 minskim if (sawesc)
530 1.26 minskim sawesc = 0;
531 1.26 minskim else
532 1.26 minskim p++;
533 1.26 minskim
534 1.26 minskim if (*p == '\0') {
535 1.26 minskim /*
536 1.26 minskim * This escaped character is continued
537 1.26 minskim * in the next part of the line. Note
538 1.26 minskim * this fact, then cause the loop to
539 1.26 minskim * exit w/ normal EOL case and reenter
540 1.26 minskim * above with the new buffer.
541 1.26 minskim */
542 1.26 minskim sawesc = 1;
543 1.26 minskim p--;
544 1.26 minskim continue;
545 1.26 minskim } else if (strchr("123456789", *p) != NULL) {
546 1.1 alm *sp++ = '\\';
547 1.1 alm ref = *p - '0';
548 1.1 alm if (s->re != NULL &&
549 1.36 lukem (size_t)ref > s->re->re_nsub)
550 1.1 alm err(COMPILE,
551 1.1 alm "\\%c not defined in the RE", *p);
552 1.1 alm if (s->maxbref < ref)
553 1.1 alm s->maxbref = ref;
554 1.1 alm } else if (*p == '&' || *p == '\\')
555 1.1 alm *sp++ = '\\';
556 1.1 alm } else if (*p == c) {
557 1.1 alm p++;
558 1.1 alm *sp++ = '\0';
559 1.1 alm size += sp - op;
560 1.1 alm s->new = xrealloc(text, size);
561 1.1 alm return (p);
562 1.1 alm } else if (*p == '\n') {
563 1.1 alm err(COMPILE,
564 1.1 alm "unescaped newline inside substitute pattern");
565 1.1 alm /* NOTREACHED */
566 1.1 alm }
567 1.1 alm *sp++ = *p;
568 1.1 alm }
569 1.1 alm size += sp - op;
570 1.37 tnn } while ((p = cu_fgets(&lbuf, &bufsize)));
571 1.1 alm err(COMPILE, "unterminated substitute in regular expression");
572 1.1 alm /* NOTREACHED */
573 1.17 lukem return (NULL);
574 1.1 alm }
575 1.1 alm
576 1.1 alm /*
577 1.1 alm * Compile the flags of the s command
578 1.1 alm */
579 1.1 alm static char *
580 1.24 wiz compile_flags(char *p, struct s_subst *s)
581 1.1 alm {
582 1.1 alm int gn; /* True if we have seen g or n */
583 1.37 tnn char wfile[PATH_MAX], *q;
584 1.1 alm
585 1.1 alm s->n = 1; /* Default */
586 1.1 alm s->p = 0;
587 1.1 alm s->wfile = NULL;
588 1.1 alm s->wfd = -1;
589 1.1 alm for (gn = 0;;) {
590 1.1 alm EATSPACE(); /* EXTENSION */
591 1.1 alm switch (*p) {
592 1.1 alm case 'g':
593 1.1 alm if (gn)
594 1.1 alm err(COMPILE,
595 1.1 alm "more than one number or 'g' in substitute flags");
596 1.1 alm gn = 1;
597 1.1 alm s->n = 0;
598 1.1 alm break;
599 1.1 alm case '\0':
600 1.1 alm case '\n':
601 1.1 alm case ';':
602 1.1 alm return (p);
603 1.1 alm case 'p':
604 1.1 alm s->p = 1;
605 1.1 alm break;
606 1.1 alm case '1': case '2': case '3':
607 1.1 alm case '4': case '5': case '6':
608 1.1 alm case '7': case '8': case '9':
609 1.1 alm if (gn)
610 1.1 alm err(COMPILE,
611 1.1 alm "more than one number or 'g' in substitute flags");
612 1.1 alm gn = 1;
613 1.1 alm /* XXX Check for overflow */
614 1.1 alm s->n = (int)strtol(p, &p, 10);
615 1.27 grant p--;
616 1.1 alm break;
617 1.1 alm case 'w':
618 1.1 alm p++;
619 1.9 cgd #ifdef HISTORIC_PRACTICE
620 1.1 alm if (*p != ' ') {
621 1.1 alm err(WARNING, "space missing before w wfile");
622 1.1 alm return (p);
623 1.1 alm }
624 1.1 alm #endif
625 1.1 alm EATSPACE();
626 1.1 alm q = wfile;
627 1.1 alm while (*p) {
628 1.1 alm if (*p == '\n')
629 1.1 alm break;
630 1.1 alm *q++ = *p++;
631 1.1 alm }
632 1.1 alm *q = '\0';
633 1.1 alm if (q == wfile)
634 1.1 alm err(COMPILE, "no wfile specified");
635 1.1 alm s->wfile = strdup(wfile);
636 1.1 alm if (!aflag && (s->wfd = open(wfile,
637 1.1 alm O_WRONLY|O_APPEND|O_CREAT|O_TRUNC,
638 1.1 alm DEFFILEMODE)) == -1)
639 1.23 itojun err(FATAL, "%s: %s", wfile, strerror(errno));
640 1.1 alm return (p);
641 1.1 alm default:
642 1.1 alm err(COMPILE,
643 1.1 alm "bad flag in substitute command: '%c'", *p);
644 1.1 alm break;
645 1.1 alm }
646 1.1 alm p++;
647 1.1 alm }
648 1.1 alm }
649 1.1 alm
650 1.1 alm /*
651 1.1 alm * Compile a translation set of strings into a lookup table.
652 1.1 alm */
653 1.1 alm static char *
654 1.24 wiz compile_tr(char *p, char **transtab)
655 1.1 alm {
656 1.1 alm int i;
657 1.1 alm char *lt, *op, *np;
658 1.37 tnn char *old = NULL, *new = NULL;
659 1.1 alm
660 1.1 alm if (*p == '\0' || *p == '\\')
661 1.1 alm err(COMPILE,
662 1.1 alm "transform pattern can not be delimited by newline or backslash");
663 1.37 tnn old = xmalloc(strlen(p) + 1);
664 1.1 alm p = compile_delimited(p, old);
665 1.1 alm if (p == NULL) {
666 1.1 alm err(COMPILE, "unterminated transform source string");
667 1.37 tnn goto bad;
668 1.1 alm }
669 1.37 tnn new = xmalloc(strlen(p) + 1);
670 1.39 joerg p = compile_delimited(p - 1, new);
671 1.1 alm if (p == NULL) {
672 1.1 alm err(COMPILE, "unterminated transform target string");
673 1.37 tnn goto bad;
674 1.1 alm }
675 1.1 alm EATSPACE();
676 1.1 alm if (strlen(new) != strlen(old)) {
677 1.1 alm err(COMPILE, "transform strings are not the same length");
678 1.37 tnn goto bad;
679 1.1 alm }
680 1.1 alm /* We assume characters are 8 bits */
681 1.28 grant lt = xmalloc(UCHAR_MAX+1);
682 1.1 alm for (i = 0; i <= UCHAR_MAX; i++)
683 1.1 alm lt[i] = (char)i;
684 1.1 alm for (op = old, np = new; *op; op++, np++)
685 1.1 alm lt[(u_char)*op] = *np;
686 1.1 alm *transtab = lt;
687 1.37 tnn free(old);
688 1.37 tnn free(new);
689 1.1 alm return (p);
690 1.37 tnn bad:
691 1.37 tnn free(old);
692 1.37 tnn free(new);
693 1.37 tnn return (NULL);
694 1.1 alm }
695 1.1 alm
696 1.1 alm /*
697 1.31 matt * Compile the text following an a, c, or i command.
698 1.1 alm */
699 1.1 alm static char *
700 1.24 wiz compile_text(void)
701 1.1 alm {
702 1.37 tnn int asize, size, len;
703 1.37 tnn char *lbuf, *text, *p, *op, *s;
704 1.37 tnn size_t bufsize;
705 1.37 tnn
706 1.37 tnn lbuf = text = NULL;
707 1.37 tnn asize = size = 0;
708 1.37 tnn while ((p = cu_fgets(&lbuf, &bufsize))) {
709 1.37 tnn len = ROUNDLEN(strlen(p) + 1);
710 1.37 tnn if (asize - size < len) {
711 1.37 tnn do {
712 1.37 tnn asize += len;
713 1.37 tnn } while (asize - size < len);
714 1.37 tnn text = xrealloc(text, asize);
715 1.37 tnn }
716 1.1 alm op = s = text + size;
717 1.1 alm for (; *p; p++) {
718 1.1 alm if (*p == '\\')
719 1.1 alm p++;
720 1.1 alm *s++ = *p;
721 1.1 alm }
722 1.1 alm size += s - op;
723 1.1 alm if (p[-2] != '\\') {
724 1.1 alm *s = '\0';
725 1.1 alm break;
726 1.1 alm }
727 1.1 alm }
728 1.37 tnn free(lbuf);
729 1.1 alm return (xrealloc(text, size + 1));
730 1.1 alm }
731 1.1 alm
732 1.1 alm /*
733 1.1 alm * Get an address and return a pointer to the first character after
734 1.1 alm * it. Fill the structure pointed to according to the address.
735 1.1 alm */
736 1.1 alm static char *
737 1.24 wiz compile_addr(char *p, struct s_addr *a)
738 1.1 alm {
739 1.1 alm char *end;
740 1.1 alm
741 1.1 alm switch (*p) {
742 1.1 alm case '\\': /* Context address */
743 1.1 alm ++p;
744 1.1 alm /* FALLTHROUGH */
745 1.1 alm case '/': /* Context address */
746 1.1 alm p = compile_re(p, &a->u.r);
747 1.1 alm if (p == NULL)
748 1.1 alm err(COMPILE, "unterminated regular expression");
749 1.1 alm a->type = AT_RE;
750 1.1 alm return (p);
751 1.1 alm
752 1.1 alm case '$': /* Last line */
753 1.1 alm a->type = AT_LAST;
754 1.1 alm return (p + 1);
755 1.1 alm /* Line number */
756 1.1 alm case '0': case '1': case '2': case '3': case '4':
757 1.1 alm case '5': case '6': case '7': case '8': case '9':
758 1.1 alm a->type = AT_LINE;
759 1.1 alm a->u.l = strtol(p, &end, 10);
760 1.1 alm return (end);
761 1.1 alm default:
762 1.1 alm err(COMPILE, "expected context address");
763 1.1 alm return (NULL);
764 1.1 alm }
765 1.1 alm }
766 1.1 alm
767 1.1 alm /*
768 1.9 cgd * duptoeol --
769 1.9 cgd * Return a copy of all the characters up to \n or \0.
770 1.1 alm */
771 1.1 alm static char *
772 1.36 lukem duptoeol(char *s, const char *ctype)
773 1.1 alm {
774 1.1 alm size_t len;
775 1.9 cgd int ws;
776 1.1 alm char *start;
777 1.1 alm
778 1.9 cgd ws = 0;
779 1.9 cgd for (start = s; *s != '\0' && *s != '\n'; ++s)
780 1.19 christos ws = isspace((unsigned char)*s);
781 1.1 alm *s = '\0';
782 1.9 cgd if (ws)
783 1.9 cgd err(WARNING, "whitespace after %s", ctype);
784 1.1 alm len = s - start + 1;
785 1.1 alm return (memmove(xmalloc(len), start, len));
786 1.1 alm }
787 1.1 alm
788 1.1 alm /*
789 1.9 cgd * Convert goto label names to addresses, and count a and r commands, in
790 1.9 cgd * the given subset of the script. Free the memory used by labels in b
791 1.9 cgd * and t commands (but not by :).
792 1.9 cgd *
793 1.1 alm * TODO: Remove } nodes
794 1.1 alm */
795 1.1 alm static void
796 1.24 wiz fixuplabel(struct s_command *cp, struct s_command *end)
797 1.1 alm {
798 1.1 alm
799 1.1 alm for (; cp != end; cp = cp->next)
800 1.1 alm switch (cp->code) {
801 1.1 alm case 'a':
802 1.1 alm case 'r':
803 1.1 alm appendnum++;
804 1.1 alm break;
805 1.1 alm case 'b':
806 1.1 alm case 't':
807 1.9 cgd /* Resolve branch target. */
808 1.1 alm if (cp->t == NULL) {
809 1.1 alm cp->u.c = NULL;
810 1.1 alm break;
811 1.1 alm }
812 1.9 cgd if ((cp->u.c = findlabel(cp->t)) == NULL)
813 1.1 alm err(COMPILE2, "undefined label '%s'", cp->t);
814 1.1 alm free(cp->t);
815 1.1 alm break;
816 1.1 alm case '{':
817 1.9 cgd /* Do interior commands. */
818 1.9 cgd fixuplabel(cp->u.c, cp->next);
819 1.1 alm break;
820 1.1 alm }
821 1.9 cgd }
822 1.9 cgd
823 1.9 cgd /*
824 1.9 cgd * Associate the given command label for later lookup.
825 1.9 cgd */
826 1.9 cgd static void
827 1.24 wiz enterlabel(struct s_command *cp)
828 1.9 cgd {
829 1.17 lukem struct labhash **lhp, *lh;
830 1.17 lukem u_char *p;
831 1.17 lukem u_int h, c;
832 1.9 cgd
833 1.9 cgd for (h = 0, p = (u_char *)cp->t; (c = *p) != 0; p++)
834 1.9 cgd h = (h << 5) + h + c;
835 1.9 cgd lhp = &labels[h & LHMASK];
836 1.9 cgd for (lh = *lhp; lh != NULL; lh = lh->lh_next)
837 1.9 cgd if (lh->lh_hash == h && strcmp(cp->t, lh->lh_cmd->t) == 0)
838 1.9 cgd err(COMPILE2, "duplicate label '%s'", cp->t);
839 1.9 cgd lh = xmalloc(sizeof *lh);
840 1.9 cgd lh->lh_next = *lhp;
841 1.9 cgd lh->lh_hash = h;
842 1.9 cgd lh->lh_cmd = cp;
843 1.9 cgd lh->lh_ref = 0;
844 1.9 cgd *lhp = lh;
845 1.9 cgd }
846 1.9 cgd
847 1.9 cgd /*
848 1.9 cgd * Find the label contained in the command l in the command linked
849 1.9 cgd * list cp. L is excluded from the search. Return NULL if not found.
850 1.9 cgd */
851 1.9 cgd static struct s_command *
852 1.24 wiz findlabel(char *name)
853 1.9 cgd {
854 1.17 lukem struct labhash *lh;
855 1.17 lukem u_char *p;
856 1.17 lukem u_int h, c;
857 1.9 cgd
858 1.9 cgd for (h = 0, p = (u_char *)name; (c = *p) != 0; p++)
859 1.9 cgd h = (h << 5) + h + c;
860 1.9 cgd for (lh = labels[h & LHMASK]; lh != NULL; lh = lh->lh_next) {
861 1.9 cgd if (lh->lh_hash == h && strcmp(name, lh->lh_cmd->t) == 0) {
862 1.9 cgd lh->lh_ref = 1;
863 1.9 cgd return (lh->lh_cmd);
864 1.9 cgd }
865 1.9 cgd }
866 1.9 cgd return (NULL);
867 1.9 cgd }
868 1.9 cgd
869 1.9 cgd /*
870 1.9 cgd * Warn about any unused labels. As a side effect, release the label hash
871 1.9 cgd * table space.
872 1.9 cgd */
873 1.9 cgd static void
874 1.24 wiz uselabel(void)
875 1.9 cgd {
876 1.17 lukem struct labhash *lh, *next;
877 1.17 lukem int i;
878 1.9 cgd
879 1.9 cgd for (i = 0; i < LHSZ; i++) {
880 1.9 cgd for (lh = labels[i]; lh != NULL; lh = next) {
881 1.9 cgd next = lh->lh_next;
882 1.9 cgd if (!lh->lh_ref)
883 1.9 cgd err(WARNING, "unused label '%s'",
884 1.9 cgd lh->lh_cmd->t);
885 1.9 cgd free(lh);
886 1.9 cgd }
887 1.9 cgd }
888 1.1 alm }
889