GCC Code Coverage Report
Directory: ./ Exec Total Coverage
File: usr.bin/sed/compile.c Lines: 312 417 74.8 %
Date: 2016-12-06 Branches: 217 381 57.0 %

Line Branch Exec Source
1
/*	$OpenBSD: compile.c,v 1.40 2015/10/26 22:24:44 jca Exp $	*/
2
3
/*-
4
 * Copyright (c) 1992 Diomidis Spinellis.
5
 * Copyright (c) 1992, 1993
6
 *	The Regents of the University of California.  All rights reserved.
7
 *
8
 * This code is derived from software contributed to Berkeley by
9
 * Diomidis Spinellis of Imperial College, University of London.
10
 *
11
 * Redistribution and use in source and binary forms, with or without
12
 * modification, are permitted provided that the following conditions
13
 * are met:
14
 * 1. Redistributions of source code must retain the above copyright
15
 *    notice, this list of conditions and the following disclaimer.
16
 * 2. Redistributions in binary form must reproduce the above copyright
17
 *    notice, this list of conditions and the following disclaimer in the
18
 *    documentation and/or other materials provided with the distribution.
19
 * 3. Neither the name of the University nor the names of its contributors
20
 *    may be used to endorse or promote products derived from this software
21
 *    without specific prior written permission.
22
 *
23
 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
24
 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
25
 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
26
 * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
27
 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
28
 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
29
 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
30
 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
31
 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
32
 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
33
 * SUCH DAMAGE.
34
 */
35
36
#include <sys/types.h>
37
#include <sys/stat.h>
38
39
#include <ctype.h>
40
#include <errno.h>
41
#include <fcntl.h>
42
#include <limits.h>
43
#include <regex.h>
44
#include <stdio.h>
45
#include <stdlib.h>
46
#include <string.h>
47
48
#include "defs.h"
49
#include "extern.h"
50
51
#define LHSZ	128
52
#define	LHMASK	(LHSZ - 1)
53
static struct labhash {
54
	struct	labhash *lh_next;
55
	u_int	lh_hash;
56
	struct	s_command *lh_cmd;
57
	int	lh_ref;
58
} *labels[LHSZ];
59
60
static char	 *compile_addr(char *, struct s_addr *);
61
static char	 *compile_ccl(char **, char *);
62
static char	 *compile_delimited(char *, char *, int);
63
static char	 *compile_flags(char *, struct s_subst *);
64
static char	 *compile_re(char *, regex_t **);
65
static char	 *compile_subst(char *, struct s_subst *);
66
static char	 *compile_text(void);
67
static char	 *compile_tr(char *, char **);
68
static struct s_command
69
		**compile_stream(struct s_command **);
70
static char	 *duptoeol(char *, char *, char **);
71
static void	  enterlabel(struct s_command *);
72
static struct s_command
73
		 *findlabel(char *);
74
static void	  fixuplabel(struct s_command *, struct s_command *);
75
static void	  uselabel(void);
76
77
/*
78
 * Command specification.  This is used to drive the command parser.
79
 */
80
struct s_format {
81
	char code;				/* Command code */
82
	int naddr;				/* Number of address args */
83
	enum e_args args;			/* Argument type */
84
};
85
86
static struct s_format cmd_fmts[] = {
87
	{'{', 2, GROUP},
88
	{'}', 0, ENDGROUP},
89
	{'a', 1, TEXT},
90
	{'b', 2, BRANCH},
91
	{'c', 2, TEXT},
92
	{'d', 2, EMPTY},
93
	{'D', 2, EMPTY},
94
	{'g', 2, EMPTY},
95
	{'G', 2, EMPTY},
96
	{'h', 2, EMPTY},
97
	{'H', 2, EMPTY},
98
	{'i', 1, TEXT},
99
	{'l', 2, EMPTY},
100
	{'n', 2, EMPTY},
101
	{'N', 2, EMPTY},
102
	{'p', 2, EMPTY},
103
	{'P', 2, EMPTY},
104
	{'q', 1, EMPTY},
105
	{'r', 1, RFILE},
106
	{'s', 2, SUBST},
107
	{'t', 2, BRANCH},
108
	{'w', 2, WFILE},
109
	{'x', 2, EMPTY},
110
	{'y', 2, TR},
111
	{'!', 2, NONSEL},
112
	{':', 0, LABEL},
113
	{'#', 0, COMMENT},
114
	{'=', 1, EMPTY},
115
	{'\0', 0, COMMENT},
116
};
117
118
/* The compiled program. */
119
struct s_command *prog;
120
121
/*
122
 * Compile the program into prog.
123
 * Initialise appends.
124
 */
125
void
126
compile(void)
127
781
{
128
781
	*compile_stream(&prog) = NULL;
129
781
	fixuplabel(prog, NULL);
130
781
	uselabel();
131
781
	appends = xreallocarray(NULL, appendnum, sizeof(struct s_appends));
132
781
	match = xreallocarray(NULL, maxnsub + 1, sizeof(regmatch_t));
133
781
}
134
135
#define EATSPACE() do {							\
136
	if (p)								\
137
		while (isascii((unsigned char)*p) &&			\
138
		    isspace((unsigned char)*p))				\
139
			p++;						\
140
	} while (0)
141
142
static struct s_command **
143
compile_stream(struct s_command **link)
144
781
{
145
	char *p;
146
	static char *lbuf;	/* To avoid excessive malloc calls */
147
	static size_t bufsize;
148
	struct s_command *cmd, *cmd2, *stack;
149
	struct s_format *fp;
150
	int naddr;				/* Number of addresses */
151
152
781
	stack = 0;
153
	for (;;) {
154
14716
		if ((p = cu_fgets(&lbuf, &bufsize)) == NULL) {
155
781
			if (stack != 0)
156
				error(COMPILE, "unexpected EOF (pending }'s)");
157
781
			return (link);
158
		}
159
160

29963
semicolon:	EATSPACE();
161
14732
		if (*p == '#' || *p == '\0')
162
201
			continue;
163
14531
		if (*p == ';') {
164
			p++;
165
			goto semicolon;
166
		}
167
14531
		*link = cmd = xmalloc(sizeof(struct s_command));
168
14531
		link = &cmd->next;
169
14531
		cmd->nonsel = cmd->inrange = 0;
170
		/* First parse the addresses */
171
14531
		naddr = 0;
172
173
/* Valid characters to start an address */
174
#define	addrchar(c)	(strchr("0123456789/\\$", (c)))
175
14531
		if (addrchar(*p)) {
176
1008
			naddr++;
177
1008
			cmd->a1 = xmalloc(sizeof(struct s_addr));
178
1008
			p = compile_addr(p, cmd->a1);
179

2062
			EATSPACE();				/* EXTENSION */
180
1008
			if (*p == ',') {
181
1
				p++;
182

2
				EATSPACE();			/* EXTENSION */
183
1
				naddr++;
184
1
				cmd->a2 = xmalloc(sizeof(struct s_addr));
185
1
				p = compile_addr(p, cmd->a2);
186

2
				EATSPACE();
187
			} else {
188
1007
				cmd->a2 = 0;
189
			}
190
		} else {
191
13523
			cmd->a1 = cmd->a2 = 0;
192
		}
193
194
14711
nonsel:		/* Now parse the command */
195
14711
		if (!*p)
196
			error(COMPILE, "command expected");
197
14711
		cmd->code = *p;
198
267627
		for (fp = cmd_fmts; fp->code; fp++)
199
267627
			if (fp->code == *p)
200
14711
				break;
201
14711
		if (!fp->code)
202
			error(COMPILE, "invalid command code %c", *p);
203
14711
		if (naddr > fp->naddr)
204
			error(COMPILE,
205
			    "command %c expects up to %d address(es), found %d",
206
			    *p, fp->naddr, naddr);
207



14711
		switch (fp->args) {
208
		case NONSEL:			/* ! */
209
180
			p++;
210

360
			EATSPACE();
211
180
			cmd->nonsel = 1;
212
180
			goto nonsel;
213
		case GROUP:			/* { */
214
498
			p++;
215

1494
			EATSPACE();
216
498
			cmd->next = stack;
217
498
			stack = cmd;
218
498
			link = &cmd->u.c;
219
498
			if (*p)
220
				goto semicolon;
221
			break;
222
		case ENDGROUP:
223
			/*
224
			 * Short-circuit command processing, since end of
225
			 * group is really just a noop.
226
			 */
227
498
			cmd->nonsel = 1;
228
498
			if (stack == 0)
229
				error(COMPILE, "unexpected }");
230
498
			cmd2 = stack;
231
498
			stack = cmd2->next;
232
498
			cmd2->next = cmd;
233
			/*FALLTHROUGH*/
234
		case EMPTY:		/* d D g G h H l n N p P q x = \0 */
235
1348
			p++;
236

3945
			EATSPACE();
237
1348
			if (*p == ';') {
238
11
				p++;
239
11
				link = &cmd->next;
240
11
				goto semicolon;
241
			}
242
1337
			if (*p)
243
				error(COMPILE,
244
"extra characters at the end of %c command", cmd->code);
245
			break;
246
		case TEXT:			/* a c i */
247
8
			p++;
248

16
			EATSPACE();
249
8
			if (*p != '\\')
250
				error(COMPILE, "command %c expects \\ followed by"
251
				    " text", cmd->code);
252
8
			p++;
253

24
			EATSPACE();
254
8
			if (*p)
255
				error(COMPILE, "extra characters after \\ at the"
256
				    " end of %c command", cmd->code);
257
8
			cmd->t = compile_text();
258
8
			break;
259
		case COMMENT:			/* \0 # */
260
			break;
261
		case WFILE:			/* w */
262
			p++;
263
			EATSPACE();
264
			if (*p == '\0')
265
				error(COMPILE, "filename expected");
266
			cmd->t = duptoeol(p, "w command", NULL);
267
			if (aflag)
268
				cmd->u.fd = -1;
269
			else if ((cmd->u.fd = open(p,
270
			    O_WRONLY|O_APPEND|O_CREAT|O_TRUNC,
271
			    DEFFILEMODE)) == -1)
272
				error(FATAL, "%s: %s", p, strerror(errno));
273
			break;
274
		case RFILE:			/* r */
275
405
			p++;
276

1215
			EATSPACE();
277
405
			cmd->t = duptoeol(p, "read command", NULL);
278
405
			break;
279
		case BRANCH:			/* b t */
280
677
			p++;
281

2031
			EATSPACE();
282
677
			if (*p == '\0')
283
140
				cmd->t = NULL;
284
			else
285
537
				cmd->t = duptoeol(p, "branch", &p);
286
677
			if (*p == ';') {
287
				p++;
288
				goto semicolon;
289
			}
290
			break;
291
		case LABEL:			/* : */
292
112
			p++;
293

224
			EATSPACE();
294
112
			cmd->t = duptoeol(p, "label", &p);
295
112
			if (strlen(cmd->t) == 0)
296
				error(COMPILE, "empty label");
297
112
			enterlabel(cmd);
298
112
			if (*p == ';') {
299
				p++;
300
				goto semicolon;
301
			}
302
			break;
303
		case SUBST:			/* s */
304
11293
			p++;
305
11293
			if (*p == '\0' || *p == '\\')
306
				error(COMPILE, "substitute pattern can not be"
307
				    " delimited by newline or backslash");
308
11293
			cmd->u.s = xmalloc(sizeof(struct s_subst));
309
11293
			p = compile_re(p, &cmd->u.s->re);
310
11293
			if (p == NULL)
311
				error(COMPILE, "unterminated substitute pattern");
312
11293
			--p;
313
11293
			p = compile_subst(p, cmd->u.s);
314
11293
			p = compile_flags(p, cmd->u.s);
315

22586
			EATSPACE();
316
11293
			if (*p == ';') {
317
596
				p++;
318
596
				link = &cmd->next;
319
596
				goto semicolon;
320
			}
321
			break;
322
		case TR:			/* y */
323
190
			p++;
324
190
			p = compile_tr(p, (char **)&cmd->u.y);
325

380
			EATSPACE();
326
190
			if (*p == ';') {
327
190
				p++;
328
190
				link = &cmd->next;
329
190
				goto semicolon;
330
			}
331
			if (*p)
332
				error(COMPILE, "extra text at the end of a"
333
				    " transform command");
334
			break;
335
		}
336
	}
337
}
338
339
/*
340
 * Get a delimited string.  P points to the delimeter of the string; d points
341
 * to a buffer area.  Newline and delimiter escapes are processed; other
342
 * escapes are ignored.
343
 *
344
 * Returns a pointer to the first character after the final delimiter or NULL
345
 * in the case of a non-terminated string.  The character array d is filled
346
 * with the processed string.
347
 */
348
static char *
349
compile_delimited(char *p, char *d, int is_tr)
350
12637
{
351
	char c;
352
353
12637
	c = *p++;
354
12637
	if (c == '\0')
355
		return (NULL);
356
12637
	else if (c == '\\')
357
		error(COMPILE, "\\ can not be used as a string delimiter");
358
12637
	else if (c == '\n')
359
		error(COMPILE, "newline can not be used as a string delimiter");
360
214448
	while (*p) {
361

214448
		if (*p == '[' && *p != c) {
362
2122
			if ((d = compile_ccl(&p, d)) == NULL)
363
				error(COMPILE, "unbalanced brackets ([])");
364
			continue;
365

212326
		} else if (*p == '\\' && p[1] == '[') {
366
			*d++ = *p++;
367

212326
		} else if (*p == '\\' && p[1] == c) {
368
			p++;
369

212326
		} else if (*p == '\\' && p[1] == 'n') {
370
5
			*d++ = '\n';
371
5
			p += 2;
372
5
			continue;
373

212321
		} else if (*p == '\\' && p[1] == '\\') {
374
3
			if (is_tr)
375
				p++;
376
			else
377
3
				*d++ = *p++;
378
212318
		} else if (*p == c) {
379
12637
			*d = '\0';
380
12637
			return (p + 1);
381
		}
382
199684
		*d++ = *p++;
383
	}
384
	return (NULL);
385
}
386
387
388
/* compile_ccl: expand a POSIX character class */
389
static char *
390
compile_ccl(char **sp, char *t)
391
2122
{
392
	int c, d;
393
2122
	char *s = *sp;
394
395
2122
	*t++ = *s++;
396
2122
	if (*s == '^')
397
485
		*t++ = *s++;
398
2122
	if (*s == ']')
399
		*t++ = *s++;
400

19560
	for (; *s && (*t = *s) != ']'; s++, t++)
401

19560
		if (*s == '[' && ((d = *(s+1)) == '.' || d == ':' || d == '=')) {
402
			*++t = *++s, t++, s++;
403
			for (c = *s; (*t = *s) != ']' || c != d; s++, t++)
404
				if ((c = *s) == '\0')
405
					return NULL;
406

19560
		} else if (*s == '\\' && s[1] == 'n') {
407
			*t = '\n';
408
			s++;
409
		}
410
2122
	if (*s == ']') {
411
2122
		*sp = ++s;
412
2122
		return (++t);
413
	} else {
414
		return (NULL);
415
	}
416
}
417
418
/*
419
 * Get a regular expression.  P points to the delimiter of the regular
420
 * expression; repp points to the address of a regexp pointer.  Newline
421
 * and delimiter escapes are processed; other escapes are ignored.
422
 * Returns a pointer to the first character after the final delimiter
423
 * or NULL in the case of a non terminated regular expression.  The regexp
424
 * pointer is set to the compiled regular expression.
425
 * Cflags are passed to regcomp.
426
 */
427
static char *
428
compile_re(char *p, regex_t **repp)
429
12257
{
430
	int eval;
431
	char *re;
432
433
12257
	re = xmalloc(strlen(p) + 1); /* strlen(re) <= strlen(p) */
434
12257
	p = compile_delimited(p, re, 0);
435

12257
	if (p && strlen(re) == 0) {
436
1
		*repp = NULL;
437
1
		free(re);
438
1
		return (p);
439
	}
440
12256
	*repp = xmalloc(sizeof(regex_t));
441

12256
	if (p && (eval = regcomp(*repp, re, Eflag ? REG_EXTENDED : 0)) != 0)
442
		error(COMPILE, "RE error: %s", strregerror(eval, *repp));
443
12256
	if (maxnsub < (*repp)->re_nsub)
444
62
		maxnsub = (*repp)->re_nsub;
445
12256
	free(re);
446
12256
	return (p);
447
}
448
449
/*
450
 * Compile the substitution string of a regular expression and set res to
451
 * point to a saved copy of it.  Nsub is the number of parenthesized regular
452
 * expressions.
453
 */
454
static char *
455
compile_subst(char *p, struct s_subst *s)
456
11293
{
457
	static char *lbuf;
458
	static size_t bufsize;
459
	int asize, ref, size;
460
	char c, *text, *op, *sp;
461
11293
	int sawesc = 0;
462
463
11293
	c = *p++;			/* Terminator character */
464
11293
	if (c == '\0')
465
		return (NULL);
466
467
11293
	s->maxbref = 0;
468
11293
	s->linenum = linenum;
469
11293
	text = NULL;
470
11293
	asize = size = 0;
471
	do {
472
11296
		size_t len = ROUNDLEN(strlen(p) + 1);
473
11296
		if (asize - size < len) {
474
			do {
475
11296
				asize += len;
476
11296
			} while (asize - size < len);
477
11296
			text = xrealloc(text, asize);
478
		}
479
11296
		op = sp = text + size;
480
263180
		for (; *p; p++) {
481
263177
			if (*p == '\\' || sawesc) {
482
				/*
483
				 * If this is a continuation from the last
484
				 * buffer, we won't have a character to
485
				 * skip over.
486
				 */
487
366
				if (sawesc)
488
					sawesc = 0;
489
				else
490
366
					p++;
491
492
366
				if (*p == '\0') {
493
					/*
494
					 * This escaped character is continued
495
					 * in the next part of the line.  Note
496
					 * this fact, then cause the loop to
497
					 * exit w/ normal EOL case and reenter
498
					 * above with the new buffer.
499
					 */
500
					sawesc = 1;
501
					p--;
502
					continue;
503
366
				} else if (strchr("123456789", *p) != NULL) {
504
177
					*sp++ = '\\';
505
177
					ref = *p - '0';
506

177
					if (s->re != NULL &&
507
					    ref > s->re->re_nsub)
508
						error(COMPILE,
509
"\\%c not defined in the RE", *p);
510
177
					if (s->maxbref < ref)
511
173
						s->maxbref = ref;
512
189
				} else if (*p == '&' || *p == '\\')
513
51
					*sp++ = '\\';
514
262811
			} else if (*p == c) {
515
11293
				p++;
516
11293
				*sp++ = '\0';
517
11293
				size += sp - op;
518
11293
				s->new = xrealloc(text, size);
519
11293
				return (p);
520
251518
			} else if (*p == '\n') {
521
				error(COMPILE,
522
"unescaped newline inside substitute pattern");
523
				/* NOTREACHED */
524
			}
525
251884
			*sp++ = *p;
526
		}
527
3
		size += sp - op;
528
3
	} while ((p = cu_fgets(&lbuf, &bufsize)));
529
	error(COMPILE, "unterminated substitute in regular expression");
530
	/* NOTREACHED */
531
}
532
533
/*
534
 * Compile the flags of the s command
535
 */
536
static char *
537
compile_flags(char *p, struct s_subst *s)
538
11293
{
539
	int gn;			/* True if we have seen g or n */
540
	long l;
541
	char wfile[PATH_MAX], *q, *eq;
542
543
11293
	s->n = 1;				/* Default */
544
11293
	s->p = 0;
545
11293
	s->wfile = NULL;
546
11293
	s->wfd = -1;
547
11293
	for (gn = 0;;) {
548

53176
		EATSPACE();			/* EXTENSION */
549

21522
		switch (*p) {
550
		case 'g':
551
10210
			if (gn)
552
				error(COMPILE, "more than one number or 'g' in"
553
				    " substitute flags");
554
10210
			gn = 1;
555
10210
			s->n = 0;
556
10210
			break;
557
		case '\0':
558
		case '\n':
559
		case ';':
560
11293
			return (p);
561
		case 'p':
562
19
			s->p = 1;
563
19
			break;
564
		case '1': case '2': case '3':
565
		case '4': case '5': case '6':
566
		case '7': case '8': case '9':
567
			if (gn)
568
				error(COMPILE, "more than one number or 'g' in"
569
				    " substitute flags");
570
			gn = 1;
571
			l = strtol(p, &p, 10);
572
			if (l <= 0 || l >= INT_MAX)
573
				error(COMPILE,
574
				    "number in substitute flags out of range");
575
			s->n = (int)l;
576
			continue;
577
		case 'w':
578
			p++;
579
#ifdef HISTORIC_PRACTICE
580
			if (*p != ' ') {
581
				error(WARNING, "space missing before w wfile");
582
				return (p);
583
			}
584
#endif
585
			EATSPACE();
586
			q = wfile;
587
			eq = wfile + sizeof(wfile) - 1;
588
			while (*p) {
589
				if (*p == '\n')
590
					break;
591
				if (q >= eq)
592
					error(COMPILE, "wfile too long");
593
				*q++ = *p++;
594
			}
595
			*q = '\0';
596
			if (q == wfile)
597
				error(COMPILE, "no wfile specified");
598
			s->wfile = strdup(wfile);
599
			if (!aflag && (s->wfd = open(wfile,
600
			    O_WRONLY|O_APPEND|O_CREAT|O_TRUNC,
601
			    DEFFILEMODE)) == -1)
602
				error(FATAL, "%s: %s", wfile, strerror(errno));
603
			return (p);
604
		default:
605
			error(COMPILE,
606
			    "bad flag in substitute command: '%c'", *p);
607
			break;
608
		}
609
10229
		p++;
610
	}
611
}
612
613
/*
614
 * Compile a translation set of strings into a lookup table.
615
 */
616
static char *
617
compile_tr(char *p, char **transtab)
618
190
{
619
	int i;
620
	char *lt, *op, *np;
621
190
	char *old = NULL, *new = NULL;
622
623
190
	if (*p == '\0' || *p == '\\')
624
		error(COMPILE,
625
"transform pattern can not be delimited by newline or backslash");
626
190
	old = xmalloc(strlen(p) + 1);
627
190
	p = compile_delimited(p, old, 1);
628
190
	if (p == NULL) {
629
		error(COMPILE, "unterminated transform source string");
630
		goto bad;
631
	}
632
190
	new = xmalloc(strlen(p) + 1);
633
190
	p = compile_delimited(--p, new, 1);
634
190
	if (p == NULL) {
635
		error(COMPILE, "unterminated transform target string");
636
		goto bad;
637
	}
638

380
	EATSPACE();
639
190
	if (strlen(new) != strlen(old)) {
640
		error(COMPILE, "transform strings are not the same length");
641
		goto bad;
642
	}
643
	/* We assume characters are 8 bits */
644
190
	lt = xmalloc(UCHAR_MAX + 1);
645
48830
	for (i = 0; i <= UCHAR_MAX; i++)
646
48640
		lt[i] = (char)i;
647
2845
	for (op = old, np = new; *op; op++, np++)
648
2655
		lt[(u_char)*op] = *np;
649
190
	*transtab = lt;
650
190
	free(old);
651
190
	free(new);
652
190
	return (p);
653
bad:
654
	free(old);
655
	free(new);
656
	return (NULL);
657
}
658
659
/*
660
 * Compile the text following an a, c, or i command.
661
 */
662
static char *
663
compile_text(void)
664
8
{
665
	int asize, esc_nl, size;
666
	char *lbuf, *text, *p, *op, *s;
667
	size_t bufsize;
668
669
8
	lbuf = text = NULL;
670
8
	asize = size = 0;
671
16
	while ((p = cu_fgets(&lbuf, &bufsize))) {
672
8
		size_t len = ROUNDLEN(strlen(p) + 1);
673
8
		if (asize - size < len) {
674
			do {
675
8
				asize += len;
676
8
			} while (asize - size < len);
677
8
			text = xrealloc(text, asize);
678
		}
679
8
		op = s = text + size;
680
312
		for (esc_nl = 0; *p != '\0'; p++) {
681

304
			if (*p == '\\' && p[1] != '\0' && *++p == '\n')
682
				esc_nl = 1;
683
304
			*s++ = *p;
684
		}
685
8
		size += s - op;
686
8
		if (!esc_nl) {
687
8
			*s = '\0';
688
8
			break;
689
		}
690
	}
691
8
	free(lbuf);
692
8
	text = xrealloc(text, size + 1);
693
8
	text[size] = '\0';
694
8
	return (text);
695
}
696
697
/*
698
 * Get an address and return a pointer to the first character after
699
 * it.  Fill the structure pointed to according to the address.
700
 */
701
static char *
702
compile_addr(char *p, struct s_addr *a)
703
1009
{
704
	char *end;
705
706

1009
	switch (*p) {
707
	case '\\':				/* Context address */
708
		++p;
709
		/* FALLTHROUGH */
710
	case '/':				/* Context address */
711
964
		p = compile_re(p, &a->u.r);
712
964
		if (p == NULL)
713
			error(COMPILE, "unterminated regular expression");
714
964
		a->type = AT_RE;
715
964
		return (p);
716
717
	case '$':				/* Last line */
718
		a->type = AT_LAST;
719
		return (p + 1);
720
						/* Line number */
721
	case '0': case '1': case '2': case '3': case '4':
722
	case '5': case '6': case '7': case '8': case '9':
723
45
		a->type = AT_LINE;
724
45
		a->u.l = strtoul(p, &end, 10);
725
45
		return (end);
726
	default:
727
		error(COMPILE, "expected context address");
728
		return (NULL);
729
	}
730
}
731
732
/*
733
 * duptoeol --
734
 *	Return a copy of all the characters up to \n or \0.
735
 */
736
static char *
737
duptoeol(char *s, char *ctype, char **semi)
738
1054
{
739
	size_t len;
740
	int ws;
741
	char *start;
742
743
1054
	ws = 0;
744
1054
	if (semi) {
745

1603
		for (start = s; *s != '\0' && *s != '\n' && *s != ';'; ++s)
746
954
			ws = isspace((unsigned char)*s);
747
	} else {
748
7830
		for (start = s; *s != '\0' && *s != '\n'; ++s)
749
7425
			ws = isspace((unsigned char)*s);
750
405
		*s = '\0';
751
	}
752
1054
	if (ws)
753
		error(WARNING, "whitespace after %s", ctype);
754
1054
	len = s - start + 1;
755
1054
	if (semi)
756
649
		*semi = s;
757
1054
	s = xmalloc(len);
758
1054
	strlcpy(s, start, len);
759
1054
	return (s);
760
}
761
762
/*
763
 * Convert goto label names to addresses, and count a and r commands, in
764
 * the given subset of the script.  Free the memory used by labels in b
765
 * and t commands (but not by :).
766
 *
767
 * TODO: Remove } nodes
768
 */
769
static void
770
fixuplabel(struct s_command *cp, struct s_command *end)
771
1279
{
772
773
15810
	for (; cp != end; cp = cp->next)
774

14531
		switch (cp->code) {
775
		case 'a':
776
		case 'r':
777
413
			appendnum++;
778
413
			break;
779
		case 'b':
780
		case 't':
781
			/* Resolve branch target. */
782
677
			if (cp->t == NULL) {
783
140
				cp->u.c = NULL;
784
140
				break;
785
			}
786
537
			if ((cp->u.c = findlabel(cp->t)) == NULL)
787
				error(COMPILE, "undefined label '%s'", cp->t);
788
537
			free(cp->t);
789
537
			break;
790
		case '{':
791
			/* Do interior commands. */
792
498
			fixuplabel(cp->u.c, cp->next);
793
			break;
794
		}
795
1279
}
796
797
/*
798
 * Associate the given command label for later lookup.
799
 */
800
static void
801
enterlabel(struct s_command *cp)
802
112
{
803
	struct labhash **lhp, *lh;
804
	u_char *p;
805
	u_int h, c;
806
807
367
	for (h = 0, p = (u_char *)cp->t; (c = *p) != 0; p++)
808
255
		h = (h << 5) + h + c;
809
112
	lhp = &labels[h & LHMASK];
810
112
	for (lh = *lhp; lh != NULL; lh = lh->lh_next)
811
		if (lh->lh_hash == h && strcmp(cp->t, lh->lh_cmd->t) == 0)
812
			error(COMPILE, "duplicate label '%s'", cp->t);
813
112
	lh = xmalloc(sizeof *lh);
814
112
	lh->lh_next = *lhp;
815
112
	lh->lh_hash = h;
816
112
	lh->lh_cmd = cp;
817
112
	lh->lh_ref = 0;
818
112
	*lhp = lh;
819
112
}
820
821
/*
822
 * Find the label contained in the command l in the command linked
823
 * list cp.  L is excluded from the search.  Return NULL if not found.
824
 */
825
static struct s_command *
826
findlabel(char *name)
827
537
{
828
	struct labhash *lh;
829
	u_char *p;
830
	u_int h, c;
831
832
1236
	for (h = 0, p = (u_char *)name; (c = *p) != 0; p++)
833
699
		h = (h << 5) + h + c;
834
537
	for (lh = labels[h & LHMASK]; lh != NULL; lh = lh->lh_next) {
835

537
		if (lh->lh_hash == h && strcmp(name, lh->lh_cmd->t) == 0) {
836
537
			lh->lh_ref = 1;
837
537
			return (lh->lh_cmd);
838
		}
839
	}
840
	return (NULL);
841
}
842
843
/*
844
 * Warn about any unused labels.  As a side effect, release the label hash
845
 * table space.
846
 */
847
static void
848
uselabel(void)
849
781
{
850
	struct labhash *lh, *next;
851
	int i;
852
853
100749
	for (i = 0; i < LHSZ; i++) {
854
100080
		for (lh = labels[i]; lh != NULL; lh = next) {
855
112
			next = lh->lh_next;
856
112
			if (!lh->lh_ref)
857
				error(WARNING, "unused label '%s'",
858
				    lh->lh_cmd->t);
859
112
			free(lh);
860
		}
861
	}
862
781
}