compile.c revision 1.25 1 1.25 agc /* $NetBSD: compile.c,v 1.25 2003/08/07 11:15:49 agc 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 1.17 lukem #include <sys/cdefs.h>
71 1.1 alm #ifndef lint
72 1.16 mrg #if 0
73 1.16 mrg static char sccsid[] = "@(#)compile.c 8.2 (Berkeley) 4/28/95";
74 1.16 mrg #else
75 1.25 agc __RCSID("$NetBSD: compile.c,v 1.25 2003/08/07 11:15:49 agc Exp $");
76 1.16 mrg #endif
77 1.1 alm #endif /* not lint */
78 1.1 alm
79 1.1 alm #include <sys/types.h>
80 1.1 alm #include <sys/stat.h>
81 1.1 alm
82 1.1 alm #include <ctype.h>
83 1.1 alm #include <errno.h>
84 1.1 alm #include <fcntl.h>
85 1.1 alm #include <limits.h>
86 1.1 alm #include <regex.h>
87 1.1 alm #include <stdio.h>
88 1.1 alm #include <stdlib.h>
89 1.1 alm #include <string.h>
90 1.1 alm
91 1.1 alm #include "defs.h"
92 1.1 alm #include "extern.h"
93 1.1 alm
94 1.9 cgd #define LHSZ 128
95 1.9 cgd #define LHMASK (LHSZ - 1)
96 1.9 cgd static struct labhash {
97 1.9 cgd struct labhash *lh_next;
98 1.9 cgd u_int lh_hash;
99 1.9 cgd struct s_command *lh_cmd;
100 1.9 cgd int lh_ref;
101 1.9 cgd } *labels[LHSZ];
102 1.9 cgd
103 1.24 wiz static char *compile_addr(char *, struct s_addr *);
104 1.24 wiz static char *compile_ccl(char **, char *);
105 1.24 wiz static char *compile_delimited(char *, char *);
106 1.24 wiz static char *compile_flags(char *, struct s_subst *);
107 1.24 wiz static char *compile_re(char *, regex_t **);
108 1.24 wiz static char *compile_subst(char *, struct s_subst *);
109 1.24 wiz static char *compile_text(void);
110 1.24 wiz static char *compile_tr(char *, char **);
111 1.1 alm static struct s_command
112 1.24 wiz **compile_stream(struct s_command **);
113 1.24 wiz static char *duptoeol(char *, char *);
114 1.24 wiz static void enterlabel(struct s_command *);
115 1.1 alm static struct s_command
116 1.24 wiz *findlabel(char *);
117 1.24 wiz static void fixuplabel(struct s_command *, struct s_command *);
118 1.24 wiz static void uselabel(void);
119 1.1 alm
120 1.1 alm /*
121 1.1 alm * Command specification. This is used to drive the command parser.
122 1.1 alm */
123 1.1 alm struct s_format {
124 1.1 alm char code; /* Command code */
125 1.1 alm int naddr; /* Number of address args */
126 1.1 alm enum e_args args; /* Argument type */
127 1.1 alm };
128 1.1 alm
129 1.1 alm static struct s_format cmd_fmts[] = {
130 1.1 alm {'{', 2, GROUP},
131 1.14 mycroft {'}', 0, ENDGROUP},
132 1.1 alm {'a', 1, TEXT},
133 1.1 alm {'b', 2, BRANCH},
134 1.1 alm {'c', 2, TEXT},
135 1.1 alm {'d', 2, EMPTY},
136 1.1 alm {'D', 2, EMPTY},
137 1.1 alm {'g', 2, EMPTY},
138 1.1 alm {'G', 2, EMPTY},
139 1.1 alm {'h', 2, EMPTY},
140 1.1 alm {'H', 2, EMPTY},
141 1.1 alm {'i', 1, TEXT},
142 1.1 alm {'l', 2, EMPTY},
143 1.1 alm {'n', 2, EMPTY},
144 1.1 alm {'N', 2, EMPTY},
145 1.1 alm {'p', 2, EMPTY},
146 1.1 alm {'P', 2, EMPTY},
147 1.1 alm {'q', 1, EMPTY},
148 1.1 alm {'r', 1, RFILE},
149 1.1 alm {'s', 2, SUBST},
150 1.1 alm {'t', 2, BRANCH},
151 1.1 alm {'w', 2, WFILE},
152 1.1 alm {'x', 2, EMPTY},
153 1.1 alm {'y', 2, TR},
154 1.1 alm {'!', 2, NONSEL},
155 1.1 alm {':', 0, LABEL},
156 1.1 alm {'#', 0, COMMENT},
157 1.1 alm {'=', 1, EMPTY},
158 1.1 alm {'\0', 0, COMMENT},
159 1.1 alm };
160 1.1 alm
161 1.1 alm /* The compiled program. */
162 1.1 alm struct s_command *prog;
163 1.1 alm
164 1.1 alm /*
165 1.1 alm * Compile the program into prog.
166 1.1 alm * Initialise appends.
167 1.1 alm */
168 1.1 alm void
169 1.24 wiz compile(void)
170 1.1 alm {
171 1.14 mycroft *compile_stream(&prog) = NULL;
172 1.9 cgd fixuplabel(prog, NULL);
173 1.9 cgd uselabel();
174 1.18 drochner if (appendnum > 0)
175 1.18 drochner appends = xmalloc(sizeof(struct s_appends) * appendnum);
176 1.1 alm match = xmalloc((maxnsub + 1) * sizeof(regmatch_t));
177 1.1 alm }
178 1.1 alm
179 1.1 alm #define EATSPACE() do { \
180 1.1 alm if (p) \
181 1.19 christos while (*p && isascii((unsigned char)*p) && \
182 1.19 christos isspace((unsigned char)*p)) \
183 1.1 alm p++; \
184 1.1 alm } while (0)
185 1.1 alm
186 1.1 alm static struct s_command **
187 1.24 wiz compile_stream(struct s_command **link)
188 1.14 mycroft {
189 1.17 lukem char *p;
190 1.1 alm static char lbuf[_POSIX2_LINE_MAX + 1]; /* To save stack */
191 1.14 mycroft struct s_command *cmd, *cmd2, *stack;
192 1.1 alm struct s_format *fp;
193 1.1 alm int naddr; /* Number of addresses */
194 1.1 alm
195 1.14 mycroft stack = 0;
196 1.1 alm for (;;) {
197 1.1 alm if ((p = cu_fgets(lbuf, sizeof(lbuf))) == NULL) {
198 1.14 mycroft if (stack != 0)
199 1.1 alm err(COMPILE, "unexpected EOF (pending }'s)");
200 1.1 alm return (link);
201 1.1 alm }
202 1.1 alm
203 1.1 alm semicolon: EATSPACE();
204 1.20 kleink if (p) {
205 1.20 kleink if (*p == '#' || *p == '\0')
206 1.20 kleink continue;
207 1.20 kleink else if (*p == ';') {
208 1.20 kleink p++;
209 1.20 kleink goto semicolon;
210 1.20 kleink }
211 1.20 kleink }
212 1.1 alm *link = cmd = xmalloc(sizeof(struct s_command));
213 1.1 alm link = &cmd->next;
214 1.1 alm cmd->nonsel = cmd->inrange = 0;
215 1.1 alm /* First parse the addresses */
216 1.1 alm naddr = 0;
217 1.1 alm
218 1.1 alm /* Valid characters to start an address */
219 1.1 alm #define addrchar(c) (strchr("0123456789/\\$", (c)))
220 1.1 alm if (addrchar(*p)) {
221 1.1 alm naddr++;
222 1.1 alm cmd->a1 = xmalloc(sizeof(struct s_addr));
223 1.1 alm p = compile_addr(p, cmd->a1);
224 1.1 alm EATSPACE(); /* EXTENSION */
225 1.1 alm if (*p == ',') {
226 1.1 alm p++;
227 1.1 alm EATSPACE(); /* EXTENSION */
228 1.13 mycroft naddr++;
229 1.1 alm cmd->a2 = xmalloc(sizeof(struct s_addr));
230 1.1 alm p = compile_addr(p, cmd->a2);
231 1.13 mycroft EATSPACE();
232 1.12 mycroft } else
233 1.12 mycroft cmd->a2 = 0;
234 1.12 mycroft } else
235 1.13 mycroft cmd->a1 = cmd->a2 = 0;
236 1.1 alm
237 1.1 alm nonsel: /* Now parse the command */
238 1.1 alm if (!*p)
239 1.1 alm err(COMPILE, "command expected");
240 1.1 alm cmd->code = *p;
241 1.1 alm for (fp = cmd_fmts; fp->code; fp++)
242 1.1 alm if (fp->code == *p)
243 1.1 alm break;
244 1.1 alm if (!fp->code)
245 1.1 alm err(COMPILE, "invalid command code %c", *p);
246 1.1 alm if (naddr > fp->naddr)
247 1.1 alm err(COMPILE,
248 1.1 alm "command %c expects up to %d address(es), found %d", *p, fp->naddr, naddr);
249 1.1 alm switch (fp->args) {
250 1.1 alm case NONSEL: /* ! */
251 1.13 mycroft p++;
252 1.13 mycroft EATSPACE();
253 1.1 alm cmd->nonsel = ! cmd->nonsel;
254 1.1 alm goto nonsel;
255 1.1 alm case GROUP: /* { */
256 1.1 alm p++;
257 1.1 alm EATSPACE();
258 1.14 mycroft cmd->next = stack;
259 1.14 mycroft stack = cmd;
260 1.14 mycroft link = &cmd->u.c;
261 1.14 mycroft if (*p)
262 1.14 mycroft goto semicolon;
263 1.14 mycroft break;
264 1.14 mycroft case ENDGROUP:
265 1.12 mycroft /*
266 1.12 mycroft * Short-circuit command processing, since end of
267 1.12 mycroft * group is really just a noop.
268 1.12 mycroft */
269 1.14 mycroft cmd->nonsel = 1;
270 1.14 mycroft if (stack == 0)
271 1.14 mycroft err(COMPILE, "unexpected }");
272 1.14 mycroft cmd2 = stack;
273 1.14 mycroft stack = cmd2->next;
274 1.14 mycroft cmd2->next = cmd;
275 1.14 mycroft /*FALLTHROUGH*/
276 1.1 alm case EMPTY: /* d D g G h H l n N p P q x = \0 */
277 1.1 alm p++;
278 1.1 alm EATSPACE();
279 1.1 alm if (*p == ';') {
280 1.1 alm p++;
281 1.1 alm link = &cmd->next;
282 1.1 alm goto semicolon;
283 1.1 alm }
284 1.1 alm if (*p)
285 1.1 alm err(COMPILE,
286 1.1 alm "extra characters at the end of %c command", cmd->code);
287 1.1 alm break;
288 1.1 alm case TEXT: /* a c i */
289 1.1 alm p++;
290 1.1 alm EATSPACE();
291 1.1 alm if (*p != '\\')
292 1.1 alm err(COMPILE,
293 1.1 alm "command %c expects \\ followed by text", cmd->code);
294 1.1 alm p++;
295 1.1 alm EATSPACE();
296 1.1 alm if (*p)
297 1.1 alm err(COMPILE,
298 1.1 alm "extra characters after \\ at the end of %c command", cmd->code);
299 1.1 alm cmd->t = compile_text();
300 1.1 alm break;
301 1.1 alm case COMMENT: /* \0 # */
302 1.1 alm break;
303 1.1 alm case WFILE: /* w */
304 1.1 alm p++;
305 1.1 alm EATSPACE();
306 1.1 alm if (*p == '\0')
307 1.1 alm err(COMPILE, "filename expected");
308 1.9 cgd cmd->t = duptoeol(p, "w command");
309 1.1 alm if (aflag)
310 1.1 alm cmd->u.fd = -1;
311 1.1 alm else if ((cmd->u.fd = open(p,
312 1.1 alm O_WRONLY|O_APPEND|O_CREAT|O_TRUNC,
313 1.1 alm DEFFILEMODE)) == -1)
314 1.23 itojun err(FATAL, "%s: %s", p, strerror(errno));
315 1.1 alm break;
316 1.1 alm case RFILE: /* r */
317 1.1 alm p++;
318 1.1 alm EATSPACE();
319 1.1 alm if (*p == '\0')
320 1.1 alm err(COMPILE, "filename expected");
321 1.1 alm else
322 1.9 cgd cmd->t = duptoeol(p, "read command");
323 1.1 alm break;
324 1.1 alm case BRANCH: /* b t */
325 1.1 alm p++;
326 1.1 alm EATSPACE();
327 1.1 alm if (*p == '\0')
328 1.1 alm cmd->t = NULL;
329 1.1 alm else
330 1.9 cgd cmd->t = duptoeol(p, "branch");
331 1.1 alm break;
332 1.1 alm case LABEL: /* : */
333 1.1 alm p++;
334 1.1 alm EATSPACE();
335 1.9 cgd cmd->t = duptoeol(p, "label");
336 1.1 alm if (strlen(p) == 0)
337 1.1 alm err(COMPILE, "empty label");
338 1.9 cgd enterlabel(cmd);
339 1.1 alm break;
340 1.1 alm case SUBST: /* s */
341 1.1 alm p++;
342 1.1 alm if (*p == '\0' || *p == '\\')
343 1.1 alm err(COMPILE,
344 1.1 alm "substitute pattern can not be delimited by newline or backslash");
345 1.1 alm cmd->u.s = xmalloc(sizeof(struct s_subst));
346 1.1 alm p = compile_re(p, &cmd->u.s->re);
347 1.1 alm if (p == NULL)
348 1.1 alm err(COMPILE, "unterminated substitute pattern");
349 1.1 alm --p;
350 1.1 alm p = compile_subst(p, cmd->u.s);
351 1.1 alm p = compile_flags(p, cmd->u.s);
352 1.1 alm EATSPACE();
353 1.1 alm if (*p == ';') {
354 1.1 alm p++;
355 1.1 alm link = &cmd->next;
356 1.1 alm goto semicolon;
357 1.1 alm }
358 1.1 alm break;
359 1.1 alm case TR: /* y */
360 1.1 alm p++;
361 1.1 alm p = compile_tr(p, (char **)&cmd->u.y);
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 if (*p)
369 1.1 alm err(COMPILE,
370 1.1 alm "extra text at the end of a transform command");
371 1.1 alm break;
372 1.1 alm }
373 1.1 alm }
374 1.1 alm }
375 1.1 alm
376 1.1 alm /*
377 1.22 wiz * Get a delimited string. P points to the delimiter of the string; d points
378 1.1 alm * to a buffer area. Newline and delimiter escapes are processed; other
379 1.1 alm * escapes are ignored.
380 1.1 alm *
381 1.1 alm * Returns a pointer to the first character after the final delimiter or NULL
382 1.1 alm * in the case of a non-terminated string. The character array d is filled
383 1.1 alm * with the processed string.
384 1.1 alm */
385 1.1 alm static char *
386 1.24 wiz compile_delimited(char *p, char *d)
387 1.1 alm {
388 1.1 alm char c;
389 1.1 alm
390 1.1 alm c = *p++;
391 1.1 alm if (c == '\0')
392 1.1 alm return (NULL);
393 1.1 alm else if (c == '\\')
394 1.1 alm err(COMPILE, "\\ can not be used as a string delimiter");
395 1.1 alm else if (c == '\n')
396 1.1 alm err(COMPILE, "newline can not be used as a string delimiter");
397 1.1 alm while (*p) {
398 1.11 alm if (*p == '[') {
399 1.11 alm if ((d = compile_ccl(&p, d)) == NULL)
400 1.11 alm err(COMPILE, "unbalanced brackets ([])");
401 1.11 alm continue;
402 1.11 alm } else if (*p == '\\' && p[1] == '[') {
403 1.11 alm *d++ = *p++;
404 1.11 alm } else if (*p == '\\' && p[1] == c)
405 1.1 alm p++;
406 1.1 alm else if (*p == '\\' && p[1] == 'n') {
407 1.1 alm *d++ = '\n';
408 1.1 alm p += 2;
409 1.1 alm continue;
410 1.1 alm } else if (*p == '\\' && p[1] == '\\')
411 1.1 alm *d++ = *p++;
412 1.1 alm else if (*p == c) {
413 1.1 alm *d = '\0';
414 1.1 alm return (p + 1);
415 1.1 alm }
416 1.1 alm *d++ = *p++;
417 1.1 alm }
418 1.1 alm return (NULL);
419 1.11 alm }
420 1.11 alm
421 1.11 alm
422 1.11 alm /* compile_ccl: expand a POSIX character class */
423 1.11 alm static char *
424 1.24 wiz compile_ccl(char **sp, char *t)
425 1.11 alm {
426 1.11 alm int c, d;
427 1.11 alm char *s = *sp;
428 1.11 alm
429 1.11 alm *t++ = *s++;
430 1.11 alm if (*s == '^')
431 1.11 alm *t++ = *s++;
432 1.11 alm if (*s == ']')
433 1.11 alm *t++ = *s++;
434 1.11 alm for (; *s && (*t = *s) != ']'; s++, t++)
435 1.11 alm if (*s == '[' && ((d = *(s+1)) == '.' || d == ':' || d == '=')) {
436 1.11 alm *++t = *++s, t++, s++;
437 1.11 alm for (c = *s; (*t = *s) != ']' || c != d; s++, t++)
438 1.11 alm if ((c = *s) == '\0')
439 1.11 alm return NULL;
440 1.11 alm } else if (*s == '\\' && s[1] == 'n')
441 1.11 alm *t = '\n', s++;
442 1.11 alm return (*s == ']') ? *sp = ++s, ++t : NULL;
443 1.2 alm }
444 1.2 alm
445 1.1 alm /*
446 1.1 alm * Get a regular expression. P points to the delimiter of the regular
447 1.1 alm * expression; repp points to the address of a regexp pointer. Newline
448 1.1 alm * and delimiter escapes are processed; other escapes are ignored.
449 1.1 alm * Returns a pointer to the first character after the final delimiter
450 1.1 alm * or NULL in the case of a non terminated regular expression. The regexp
451 1.1 alm * pointer is set to the compiled regular expression.
452 1.1 alm * Cflags are passed to regcomp.
453 1.1 alm */
454 1.1 alm static char *
455 1.24 wiz compile_re(char *p, regex_t **repp)
456 1.1 alm {
457 1.1 alm int eval;
458 1.1 alm char re[_POSIX2_LINE_MAX + 1];
459 1.1 alm
460 1.1 alm p = compile_delimited(p, re);
461 1.1 alm if (p && strlen(re) == 0) {
462 1.1 alm *repp = NULL;
463 1.1 alm return (p);
464 1.1 alm }
465 1.1 alm *repp = xmalloc(sizeof(regex_t));
466 1.21 atatat if (p && (eval = regcomp(*repp, re, ere)) != 0)
467 1.1 alm err(COMPILE, "RE error: %s", strregerror(eval, *repp));
468 1.1 alm if (maxnsub < (*repp)->re_nsub)
469 1.1 alm maxnsub = (*repp)->re_nsub;
470 1.1 alm return (p);
471 1.1 alm }
472 1.1 alm
473 1.1 alm /*
474 1.1 alm * Compile the substitution string of a regular expression and set res to
475 1.1 alm * point to a saved copy of it. Nsub is the number of parenthesized regular
476 1.1 alm * expressions.
477 1.1 alm */
478 1.1 alm static char *
479 1.24 wiz compile_subst(char *p, struct s_subst *s)
480 1.1 alm {
481 1.1 alm static char lbuf[_POSIX2_LINE_MAX + 1];
482 1.1 alm int asize, ref, size;
483 1.1 alm char c, *text, *op, *sp;
484 1.1 alm
485 1.1 alm c = *p++; /* Terminator character */
486 1.1 alm if (c == '\0')
487 1.1 alm return (NULL);
488 1.1 alm
489 1.1 alm s->maxbref = 0;
490 1.1 alm s->linenum = linenum;
491 1.1 alm asize = 2 * _POSIX2_LINE_MAX + 1;
492 1.1 alm text = xmalloc(asize);
493 1.1 alm size = 0;
494 1.1 alm do {
495 1.1 alm op = sp = text + size;
496 1.1 alm for (; *p; p++) {
497 1.1 alm if (*p == '\\') {
498 1.1 alm p++;
499 1.1 alm if (strchr("123456789", *p) != NULL) {
500 1.1 alm *sp++ = '\\';
501 1.1 alm ref = *p - '0';
502 1.1 alm if (s->re != NULL &&
503 1.1 alm ref > s->re->re_nsub)
504 1.1 alm err(COMPILE,
505 1.1 alm "\\%c not defined in the RE", *p);
506 1.1 alm if (s->maxbref < ref)
507 1.1 alm s->maxbref = ref;
508 1.1 alm } else if (*p == '&' || *p == '\\')
509 1.1 alm *sp++ = '\\';
510 1.1 alm } else if (*p == c) {
511 1.1 alm p++;
512 1.1 alm *sp++ = '\0';
513 1.1 alm size += sp - op;
514 1.1 alm s->new = xrealloc(text, size);
515 1.1 alm return (p);
516 1.1 alm } else if (*p == '\n') {
517 1.1 alm err(COMPILE,
518 1.1 alm "unescaped newline inside substitute pattern");
519 1.1 alm /* NOTREACHED */
520 1.1 alm }
521 1.1 alm *sp++ = *p;
522 1.1 alm }
523 1.1 alm size += sp - op;
524 1.1 alm if (asize - size < _POSIX2_LINE_MAX + 1) {
525 1.1 alm asize *= 2;
526 1.1 alm text = xmalloc(asize);
527 1.1 alm }
528 1.1 alm } while (cu_fgets(p = lbuf, sizeof(lbuf)));
529 1.1 alm err(COMPILE, "unterminated substitute in regular expression");
530 1.1 alm /* NOTREACHED */
531 1.17 lukem return (NULL);
532 1.1 alm }
533 1.1 alm
534 1.1 alm /*
535 1.1 alm * Compile the flags of the s command
536 1.1 alm */
537 1.1 alm static char *
538 1.24 wiz compile_flags(char *p, struct s_subst *s)
539 1.1 alm {
540 1.1 alm int gn; /* True if we have seen g or n */
541 1.1 alm char wfile[_POSIX2_LINE_MAX + 1], *q;
542 1.1 alm
543 1.1 alm s->n = 1; /* Default */
544 1.1 alm s->p = 0;
545 1.1 alm s->wfile = NULL;
546 1.1 alm s->wfd = -1;
547 1.1 alm for (gn = 0;;) {
548 1.1 alm EATSPACE(); /* EXTENSION */
549 1.1 alm switch (*p) {
550 1.1 alm case 'g':
551 1.1 alm if (gn)
552 1.1 alm err(COMPILE,
553 1.1 alm "more than one number or 'g' in substitute flags");
554 1.1 alm gn = 1;
555 1.1 alm s->n = 0;
556 1.1 alm break;
557 1.1 alm case '\0':
558 1.1 alm case '\n':
559 1.1 alm case ';':
560 1.1 alm return (p);
561 1.1 alm case 'p':
562 1.1 alm s->p = 1;
563 1.1 alm break;
564 1.1 alm case '1': case '2': case '3':
565 1.1 alm case '4': case '5': case '6':
566 1.1 alm case '7': case '8': case '9':
567 1.1 alm if (gn)
568 1.1 alm err(COMPILE,
569 1.1 alm "more than one number or 'g' in substitute flags");
570 1.1 alm gn = 1;
571 1.1 alm /* XXX Check for overflow */
572 1.1 alm s->n = (int)strtol(p, &p, 10);
573 1.1 alm break;
574 1.1 alm case 'w':
575 1.1 alm p++;
576 1.9 cgd #ifdef HISTORIC_PRACTICE
577 1.1 alm if (*p != ' ') {
578 1.1 alm err(WARNING, "space missing before w wfile");
579 1.1 alm return (p);
580 1.1 alm }
581 1.1 alm #endif
582 1.1 alm EATSPACE();
583 1.1 alm q = wfile;
584 1.1 alm while (*p) {
585 1.1 alm if (*p == '\n')
586 1.1 alm break;
587 1.1 alm *q++ = *p++;
588 1.1 alm }
589 1.1 alm *q = '\0';
590 1.1 alm if (q == wfile)
591 1.1 alm err(COMPILE, "no wfile specified");
592 1.1 alm s->wfile = strdup(wfile);
593 1.1 alm if (!aflag && (s->wfd = open(wfile,
594 1.1 alm O_WRONLY|O_APPEND|O_CREAT|O_TRUNC,
595 1.1 alm DEFFILEMODE)) == -1)
596 1.23 itojun err(FATAL, "%s: %s", wfile, strerror(errno));
597 1.1 alm return (p);
598 1.1 alm default:
599 1.1 alm err(COMPILE,
600 1.1 alm "bad flag in substitute command: '%c'", *p);
601 1.1 alm break;
602 1.1 alm }
603 1.1 alm p++;
604 1.1 alm }
605 1.1 alm }
606 1.1 alm
607 1.1 alm /*
608 1.1 alm * Compile a translation set of strings into a lookup table.
609 1.1 alm */
610 1.1 alm static char *
611 1.24 wiz compile_tr(char *p, char **transtab)
612 1.1 alm {
613 1.1 alm int i;
614 1.1 alm char *lt, *op, *np;
615 1.1 alm char old[_POSIX2_LINE_MAX + 1];
616 1.1 alm char new[_POSIX2_LINE_MAX + 1];
617 1.1 alm
618 1.1 alm if (*p == '\0' || *p == '\\')
619 1.1 alm err(COMPILE,
620 1.1 alm "transform pattern can not be delimited by newline or backslash");
621 1.1 alm p = compile_delimited(p, old);
622 1.1 alm if (p == NULL) {
623 1.1 alm err(COMPILE, "unterminated transform source string");
624 1.1 alm return (NULL);
625 1.1 alm }
626 1.1 alm p = compile_delimited(--p, new);
627 1.1 alm if (p == NULL) {
628 1.1 alm err(COMPILE, "unterminated transform target string");
629 1.1 alm return (NULL);
630 1.1 alm }
631 1.1 alm EATSPACE();
632 1.1 alm if (strlen(new) != strlen(old)) {
633 1.1 alm err(COMPILE, "transform strings are not the same length");
634 1.1 alm return (NULL);
635 1.1 alm }
636 1.1 alm /* We assume characters are 8 bits */
637 1.1 alm lt = xmalloc(UCHAR_MAX);
638 1.1 alm for (i = 0; i <= UCHAR_MAX; i++)
639 1.1 alm lt[i] = (char)i;
640 1.1 alm for (op = old, np = new; *op; op++, np++)
641 1.1 alm lt[(u_char)*op] = *np;
642 1.1 alm *transtab = lt;
643 1.1 alm return (p);
644 1.1 alm }
645 1.1 alm
646 1.1 alm /*
647 1.1 alm * Compile the text following an a or i command.
648 1.1 alm */
649 1.1 alm static char *
650 1.24 wiz compile_text(void)
651 1.1 alm {
652 1.1 alm int asize, size;
653 1.1 alm char *text, *p, *op, *s;
654 1.1 alm char lbuf[_POSIX2_LINE_MAX + 1];
655 1.1 alm
656 1.1 alm asize = 2 * _POSIX2_LINE_MAX + 1;
657 1.1 alm text = xmalloc(asize);
658 1.1 alm size = 0;
659 1.1 alm while (cu_fgets(lbuf, sizeof(lbuf))) {
660 1.1 alm op = s = text + size;
661 1.1 alm p = lbuf;
662 1.1 alm EATSPACE();
663 1.1 alm for (; *p; p++) {
664 1.1 alm if (*p == '\\')
665 1.1 alm p++;
666 1.1 alm *s++ = *p;
667 1.1 alm }
668 1.1 alm size += s - op;
669 1.1 alm if (p[-2] != '\\') {
670 1.1 alm *s = '\0';
671 1.1 alm break;
672 1.1 alm }
673 1.1 alm if (asize - size < _POSIX2_LINE_MAX + 1) {
674 1.1 alm asize *= 2;
675 1.1 alm text = xmalloc(asize);
676 1.1 alm }
677 1.1 alm }
678 1.1 alm return (xrealloc(text, size + 1));
679 1.1 alm }
680 1.1 alm
681 1.1 alm /*
682 1.1 alm * Get an address and return a pointer to the first character after
683 1.1 alm * it. Fill the structure pointed to according to the address.
684 1.1 alm */
685 1.1 alm static char *
686 1.24 wiz compile_addr(char *p, struct s_addr *a)
687 1.1 alm {
688 1.1 alm char *end;
689 1.1 alm
690 1.1 alm switch (*p) {
691 1.1 alm case '\\': /* Context address */
692 1.1 alm ++p;
693 1.1 alm /* FALLTHROUGH */
694 1.1 alm case '/': /* Context address */
695 1.1 alm p = compile_re(p, &a->u.r);
696 1.1 alm if (p == NULL)
697 1.1 alm err(COMPILE, "unterminated regular expression");
698 1.1 alm a->type = AT_RE;
699 1.1 alm return (p);
700 1.1 alm
701 1.1 alm case '$': /* Last line */
702 1.1 alm a->type = AT_LAST;
703 1.1 alm return (p + 1);
704 1.1 alm /* Line number */
705 1.1 alm case '0': case '1': case '2': case '3': case '4':
706 1.1 alm case '5': case '6': case '7': case '8': case '9':
707 1.1 alm a->type = AT_LINE;
708 1.1 alm a->u.l = strtol(p, &end, 10);
709 1.1 alm return (end);
710 1.1 alm default:
711 1.1 alm err(COMPILE, "expected context address");
712 1.1 alm return (NULL);
713 1.1 alm }
714 1.1 alm }
715 1.1 alm
716 1.1 alm /*
717 1.9 cgd * duptoeol --
718 1.9 cgd * Return a copy of all the characters up to \n or \0.
719 1.1 alm */
720 1.1 alm static char *
721 1.24 wiz duptoeol(char *s, char *ctype)
722 1.1 alm {
723 1.1 alm size_t len;
724 1.9 cgd int ws;
725 1.1 alm char *start;
726 1.1 alm
727 1.9 cgd ws = 0;
728 1.9 cgd for (start = s; *s != '\0' && *s != '\n'; ++s)
729 1.19 christos ws = isspace((unsigned char)*s);
730 1.1 alm *s = '\0';
731 1.9 cgd if (ws)
732 1.9 cgd err(WARNING, "whitespace after %s", ctype);
733 1.1 alm len = s - start + 1;
734 1.1 alm return (memmove(xmalloc(len), start, len));
735 1.1 alm }
736 1.1 alm
737 1.1 alm /*
738 1.9 cgd * Convert goto label names to addresses, and count a and r commands, in
739 1.9 cgd * the given subset of the script. Free the memory used by labels in b
740 1.9 cgd * and t commands (but not by :).
741 1.9 cgd *
742 1.1 alm * TODO: Remove } nodes
743 1.1 alm */
744 1.1 alm static void
745 1.24 wiz fixuplabel(struct s_command *cp, struct s_command *end)
746 1.1 alm {
747 1.1 alm
748 1.1 alm for (; cp != end; cp = cp->next)
749 1.1 alm switch (cp->code) {
750 1.1 alm case 'a':
751 1.1 alm case 'r':
752 1.1 alm appendnum++;
753 1.1 alm break;
754 1.1 alm case 'b':
755 1.1 alm case 't':
756 1.9 cgd /* Resolve branch target. */
757 1.1 alm if (cp->t == NULL) {
758 1.1 alm cp->u.c = NULL;
759 1.1 alm break;
760 1.1 alm }
761 1.9 cgd if ((cp->u.c = findlabel(cp->t)) == NULL)
762 1.1 alm err(COMPILE2, "undefined label '%s'", cp->t);
763 1.1 alm free(cp->t);
764 1.1 alm break;
765 1.1 alm case '{':
766 1.9 cgd /* Do interior commands. */
767 1.9 cgd fixuplabel(cp->u.c, cp->next);
768 1.1 alm break;
769 1.1 alm }
770 1.9 cgd }
771 1.9 cgd
772 1.9 cgd /*
773 1.9 cgd * Associate the given command label for later lookup.
774 1.9 cgd */
775 1.9 cgd static void
776 1.24 wiz enterlabel(struct s_command *cp)
777 1.9 cgd {
778 1.17 lukem struct labhash **lhp, *lh;
779 1.17 lukem u_char *p;
780 1.17 lukem u_int h, c;
781 1.9 cgd
782 1.9 cgd for (h = 0, p = (u_char *)cp->t; (c = *p) != 0; p++)
783 1.9 cgd h = (h << 5) + h + c;
784 1.9 cgd lhp = &labels[h & LHMASK];
785 1.9 cgd for (lh = *lhp; lh != NULL; lh = lh->lh_next)
786 1.9 cgd if (lh->lh_hash == h && strcmp(cp->t, lh->lh_cmd->t) == 0)
787 1.9 cgd err(COMPILE2, "duplicate label '%s'", cp->t);
788 1.9 cgd lh = xmalloc(sizeof *lh);
789 1.9 cgd lh->lh_next = *lhp;
790 1.9 cgd lh->lh_hash = h;
791 1.9 cgd lh->lh_cmd = cp;
792 1.9 cgd lh->lh_ref = 0;
793 1.9 cgd *lhp = lh;
794 1.9 cgd }
795 1.9 cgd
796 1.9 cgd /*
797 1.9 cgd * Find the label contained in the command l in the command linked
798 1.9 cgd * list cp. L is excluded from the search. Return NULL if not found.
799 1.9 cgd */
800 1.9 cgd static struct s_command *
801 1.24 wiz findlabel(char *name)
802 1.9 cgd {
803 1.17 lukem struct labhash *lh;
804 1.17 lukem u_char *p;
805 1.17 lukem u_int h, c;
806 1.9 cgd
807 1.9 cgd for (h = 0, p = (u_char *)name; (c = *p) != 0; p++)
808 1.9 cgd h = (h << 5) + h + c;
809 1.9 cgd for (lh = labels[h & LHMASK]; lh != NULL; lh = lh->lh_next) {
810 1.9 cgd if (lh->lh_hash == h && strcmp(name, lh->lh_cmd->t) == 0) {
811 1.9 cgd lh->lh_ref = 1;
812 1.9 cgd return (lh->lh_cmd);
813 1.9 cgd }
814 1.9 cgd }
815 1.9 cgd return (NULL);
816 1.9 cgd }
817 1.9 cgd
818 1.9 cgd /*
819 1.9 cgd * Warn about any unused labels. As a side effect, release the label hash
820 1.9 cgd * table space.
821 1.9 cgd */
822 1.9 cgd static void
823 1.24 wiz uselabel(void)
824 1.9 cgd {
825 1.17 lukem struct labhash *lh, *next;
826 1.17 lukem int i;
827 1.9 cgd
828 1.9 cgd for (i = 0; i < LHSZ; i++) {
829 1.9 cgd for (lh = labels[i]; lh != NULL; lh = next) {
830 1.9 cgd next = lh->lh_next;
831 1.9 cgd if (!lh->lh_ref)
832 1.9 cgd err(WARNING, "unused label '%s'",
833 1.9 cgd lh->lh_cmd->t);
834 1.9 cgd free(lh);
835 1.9 cgd }
836 1.9 cgd }
837 1.1 alm }
838