Annotation of src/usr.bin/sed/compile.c, Revision 1.31
1.31 ! deraadt 1: /* $OpenBSD: compile.c,v 1.30 2009/10/23 15:24:41 reyk Exp $ */
1.2 deraadt 2:
1.1 deraadt 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.
1.14 millert 19: * 3. Neither the name of the University nor the names of its contributors
1.1 deraadt 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:
1.11 millert 60: static char *compile_addr(char *, struct s_addr *);
61: static char *compile_ccl(char **, char *);
62: static char *compile_delimited(char *, char *);
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 **);
1.1 deraadt 68: static struct s_command
1.11 millert 69: **compile_stream(struct s_command **);
1.16 deraadt 70: static char *duptoeol(char *, char *, char **);
1.11 millert 71: static void enterlabel(struct s_command *);
1.1 deraadt 72: static struct s_command
1.11 millert 73: *findlabel(char *);
74: static void fixuplabel(struct s_command *, struct s_command *);
75: static void uselabel(void);
1.1 deraadt 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
1.15 deraadt 126: compile(void)
1.1 deraadt 127: {
128: *compile_stream(&prog) = NULL;
129: fixuplabel(prog, NULL);
130: uselabel();
131: appends = xmalloc(sizeof(struct s_appends) * appendnum);
132: match = xmalloc((maxnsub + 1) * sizeof(regmatch_t));
133: }
134:
135: #define EATSPACE() do { \
136: if (p) \
1.24 tedu 137: while (isascii(*p) && isspace(*p)) \
1.1 deraadt 138: p++; \
139: } while (0)
140:
141: static struct s_command **
1.15 deraadt 142: compile_stream(struct s_command **link)
1.1 deraadt 143: {
1.10 mpech 144: char *p;
1.25 millert 145: static char *lbuf; /* To avoid excessive malloc calls */
1.28 millert 146: static size_t bufsize;
1.1 deraadt 147: struct s_command *cmd, *cmd2, *stack;
148: struct s_format *fp;
149: int naddr; /* Number of addresses */
150:
151: stack = 0;
152: for (;;) {
1.28 millert 153: if ((p = cu_fgets(&lbuf, &bufsize)) == NULL) {
1.1 deraadt 154: if (stack != 0)
155: err(COMPILE, "unexpected EOF (pending }'s)");
156: return (link);
157: }
158:
159: semicolon: EATSPACE();
1.23 otto 160: if (*p == '#' || *p == '\0')
1.1 deraadt 161: continue;
1.23 otto 162: if (*p == ';') {
163: p++;
164: goto semicolon;
165: }
1.1 deraadt 166: *link = cmd = xmalloc(sizeof(struct s_command));
167: link = &cmd->next;
168: cmd->nonsel = cmd->inrange = 0;
169: /* First parse the addresses */
170: naddr = 0;
171:
172: /* Valid characters to start an address */
173: #define addrchar(c) (strchr("0123456789/\\$", (c)))
174: if (addrchar(*p)) {
175: naddr++;
176: cmd->a1 = xmalloc(sizeof(struct s_addr));
177: p = compile_addr(p, cmd->a1);
178: EATSPACE(); /* EXTENSION */
179: if (*p == ',') {
180: p++;
181: EATSPACE(); /* EXTENSION */
182: naddr++;
183: cmd->a2 = xmalloc(sizeof(struct s_addr));
184: p = compile_addr(p, cmd->a2);
185: EATSPACE();
1.22 tedu 186: } else {
1.1 deraadt 187: cmd->a2 = 0;
1.22 tedu 188: }
189: } else {
1.1 deraadt 190: cmd->a1 = cmd->a2 = 0;
1.22 tedu 191: }
1.1 deraadt 192:
193: nonsel: /* Now parse the command */
194: if (!*p)
195: err(COMPILE, "command expected");
196: cmd->code = *p;
197: for (fp = cmd_fmts; fp->code; fp++)
198: if (fp->code == *p)
199: break;
200: if (!fp->code)
201: err(COMPILE, "invalid command code %c", *p);
202: if (naddr > fp->naddr)
203: err(COMPILE,
1.22 tedu 204: "command %c expects up to %d address(es), found %d",
205: *p, fp->naddr, naddr);
1.1 deraadt 206: switch (fp->args) {
207: case NONSEL: /* ! */
208: p++;
209: EATSPACE();
210: cmd->nonsel = ! cmd->nonsel;
211: goto nonsel;
212: case GROUP: /* { */
213: p++;
214: EATSPACE();
215: cmd->next = stack;
216: stack = cmd;
217: link = &cmd->u.c;
218: if (*p)
219: goto semicolon;
220: break;
221: case ENDGROUP:
222: /*
223: * Short-circuit command processing, since end of
224: * group is really just a noop.
225: */
226: cmd->nonsel = 1;
227: if (stack == 0)
228: err(COMPILE, "unexpected }");
229: cmd2 = stack;
230: stack = cmd2->next;
231: cmd2->next = cmd;
232: /*FALLTHROUGH*/
233: case EMPTY: /* d D g G h H l n N p P q x = \0 */
234: p++;
235: EATSPACE();
236: if (*p == ';') {
237: p++;
238: link = &cmd->next;
239: goto semicolon;
240: }
241: if (*p)
242: err(COMPILE,
243: "extra characters at the end of %c command", cmd->code);
244: break;
245: case TEXT: /* a c i */
246: p++;
247: EATSPACE();
248: if (*p != '\\')
1.22 tedu 249: err(COMPILE, "command %c expects \\ followed by"
250: " text", cmd->code);
1.1 deraadt 251: p++;
252: EATSPACE();
253: if (*p)
1.22 tedu 254: err(COMPILE, "extra characters after \\ at the"
255: " end of %c command", cmd->code);
1.1 deraadt 256: cmd->t = compile_text();
257: break;
258: case COMMENT: /* \0 # */
259: break;
260: case WFILE: /* w */
261: p++;
262: EATSPACE();
263: if (*p == '\0')
264: err(COMPILE, "filename expected");
1.16 deraadt 265: cmd->t = duptoeol(p, "w command", NULL);
1.1 deraadt 266: if (aflag)
267: cmd->u.fd = -1;
1.21 deraadt 268: else if ((cmd->u.fd = open(p,
1.1 deraadt 269: O_WRONLY|O_APPEND|O_CREAT|O_TRUNC,
270: DEFFILEMODE)) == -1)
1.12 jsyn 271: err(FATAL, "%s: %s", p, strerror(errno));
1.1 deraadt 272: break;
273: case RFILE: /* r */
274: p++;
275: EATSPACE();
1.16 deraadt 276: cmd->t = duptoeol(p, "read command", NULL);
1.1 deraadt 277: break;
278: case BRANCH: /* b t */
279: p++;
280: EATSPACE();
281: if (*p == '\0')
282: cmd->t = NULL;
283: else
1.16 deraadt 284: cmd->t = duptoeol(p, "branch", &p);
285: if (*p == ';') {
286: p++;
287: goto semicolon;
288: }
1.1 deraadt 289: break;
290: case LABEL: /* : */
291: p++;
292: EATSPACE();
1.16 deraadt 293: cmd->t = duptoeol(p, "label", &p);
294: if (strlen(cmd->t) == 0)
1.1 deraadt 295: err(COMPILE, "empty label");
296: enterlabel(cmd);
1.16 deraadt 297: if (*p == ';') {
298: p++;
299: goto semicolon;
300: }
1.1 deraadt 301: break;
302: case SUBST: /* s */
303: p++;
304: if (*p == '\0' || *p == '\\')
1.22 tedu 305: err(COMPILE, "substitute pattern can not be"
306: " delimited by newline or backslash");
1.1 deraadt 307: cmd->u.s = xmalloc(sizeof(struct s_subst));
308: p = compile_re(p, &cmd->u.s->re);
309: if (p == NULL)
310: err(COMPILE, "unterminated substitute pattern");
311: --p;
312: p = compile_subst(p, cmd->u.s);
313: p = compile_flags(p, cmd->u.s);
314: EATSPACE();
315: if (*p == ';') {
316: p++;
317: link = &cmd->next;
318: goto semicolon;
319: }
320: break;
321: case TR: /* y */
322: p++;
323: p = compile_tr(p, (char **)&cmd->u.y);
324: EATSPACE();
325: if (*p == ';') {
326: p++;
327: link = &cmd->next;
328: goto semicolon;
329: }
330: if (*p)
1.22 tedu 331: err(COMPILE, "extra text at the end of a"
332: " transform command");
1.1 deraadt 333: break;
334: }
335: }
336: }
337:
338: /*
339: * Get a delimited string. P points to the delimeter of the string; d points
340: * to a buffer area. Newline and delimiter escapes are processed; other
341: * escapes are ignored.
342: *
343: * Returns a pointer to the first character after the final delimiter or NULL
344: * in the case of a non-terminated string. The character array d is filled
345: * with the processed string.
346: */
347: static char *
1.15 deraadt 348: compile_delimited(char *p, char *d)
1.1 deraadt 349: {
350: char c;
351:
352: c = *p++;
353: if (c == '\0')
354: return (NULL);
355: else if (c == '\\')
356: err(COMPILE, "\\ can not be used as a string delimiter");
357: else if (c == '\n')
358: err(COMPILE, "newline can not be used as a string delimiter");
359: while (*p) {
360: if (*p == '[') {
361: if ((d = compile_ccl(&p, d)) == NULL)
362: err(COMPILE, "unbalanced brackets ([])");
363: continue;
364: } else if (*p == '\\' && p[1] == '[') {
365: *d++ = *p++;
1.22 tedu 366: } else if (*p == '\\' && p[1] == c) {
1.1 deraadt 367: p++;
1.22 tedu 368: } else if (*p == '\\' && p[1] == 'n') {
1.1 deraadt 369: *d++ = '\n';
370: p += 2;
371: continue;
1.22 tedu 372: } else if (*p == '\\' && p[1] == '\\') {
1.1 deraadt 373: *d++ = *p++;
1.22 tedu 374: } else if (*p == c) {
1.1 deraadt 375: *d = '\0';
376: return (p + 1);
377: }
378: *d++ = *p++;
379: }
380: return (NULL);
381: }
382:
383:
384: /* compile_ccl: expand a POSIX character class */
385: static char *
1.15 deraadt 386: compile_ccl(char **sp, char *t)
1.1 deraadt 387: {
388: int c, d;
389: char *s = *sp;
390:
391: *t++ = *s++;
392: if (*s == '^')
393: *t++ = *s++;
394: if (*s == ']')
395: *t++ = *s++;
396: for (; *s && (*t = *s) != ']'; s++, t++)
397: if (*s == '[' && ((d = *(s+1)) == '.' || d == ':' || d == '=')) {
398: *++t = *++s, t++, s++;
399: for (c = *s; (*t = *s) != ']' || c != d; s++, t++)
400: if ((c = *s) == '\0')
401: return NULL;
1.22 tedu 402: } else if (*s == '\\' && s[1] == 'n') {
403: *t = '\n';
404: s++;
405: }
406: if (*s == ']') {
407: *sp = ++s;
408: return (++t);
409: } else {
410: return (NULL);
411: }
1.1 deraadt 412: }
413:
414: /*
415: * Get a regular expression. P points to the delimiter of the regular
416: * expression; repp points to the address of a regexp pointer. Newline
417: * and delimiter escapes are processed; other escapes are ignored.
418: * Returns a pointer to the first character after the final delimiter
419: * or NULL in the case of a non terminated regular expression. The regexp
420: * pointer is set to the compiled regular expression.
421: * Cflags are passed to regcomp.
422: */
423: static char *
1.15 deraadt 424: compile_re(char *p, regex_t **repp)
1.1 deraadt 425: {
426: int eval;
1.25 millert 427: char *re;
1.1 deraadt 428:
1.25 millert 429: re = xmalloc(strlen(p) + 1); /* strlen(re) <= strlen(p) */
1.1 deraadt 430: p = compile_delimited(p, re);
431: if (p && strlen(re) == 0) {
432: *repp = NULL;
1.25 millert 433: free(re);
1.1 deraadt 434: return (p);
435: }
436: *repp = xmalloc(sizeof(regex_t));
1.29 djm 437: if (p && (eval = regcomp(*repp, re, Eflag ? REG_EXTENDED : 0)) != 0)
1.1 deraadt 438: err(COMPILE, "RE error: %s", strregerror(eval, *repp));
439: if (maxnsub < (*repp)->re_nsub)
440: maxnsub = (*repp)->re_nsub;
1.25 millert 441: free(re);
1.1 deraadt 442: return (p);
443: }
444:
445: /*
446: * Compile the substitution string of a regular expression and set res to
447: * point to a saved copy of it. Nsub is the number of parenthesized regular
448: * expressions.
449: */
450: static char *
1.15 deraadt 451: compile_subst(char *p, struct s_subst *s)
1.1 deraadt 452: {
1.25 millert 453: static char *lbuf;
1.28 millert 454: static size_t bufsize;
1.1 deraadt 455: int asize, ref, size;
456: char c, *text, *op, *sp;
1.19 otto 457: int sawesc = 0;
1.1 deraadt 458:
459: c = *p++; /* Terminator character */
460: if (c == '\0')
461: return (NULL);
462:
463: s->maxbref = 0;
464: s->linenum = linenum;
1.28 millert 465: text = NULL;
466: asize = size = 0;
1.1 deraadt 467: do {
1.28 millert 468: size_t len = ROUNDLEN(strlen(p) + 1);
469: if (asize - size < len) {
1.25 millert 470: do {
1.28 millert 471: asize += len;
472: } while (asize - size < len);
1.25 millert 473: text = xrealloc(text, asize);
474: }
1.1 deraadt 475: op = sp = text + size;
476: for (; *p; p++) {
1.19 otto 477: if (*p == '\\' || sawesc) {
478: /*
479: * If this is a continuation from the last
480: * buffer, we won't have a character to
481: * skip over.
482: */
483: if (sawesc)
484: sawesc = 0;
485: else
486: p++;
487:
488: if (*p == '\0') {
489: /*
490: * This escaped character is continued
491: * in the next part of the line. Note
492: * this fact, then cause the loop to
493: * exit w/ normal EOL case and reenter
494: * above with the new buffer.
495: */
496: sawesc = 1;
497: p--;
498: continue;
499: } else if (strchr("123456789", *p) != NULL) {
1.1 deraadt 500: *sp++ = '\\';
501: ref = *p - '0';
502: if (s->re != NULL &&
503: ref > s->re->re_nsub)
504: err(COMPILE,
505: "\\%c not defined in the RE", *p);
506: if (s->maxbref < ref)
507: s->maxbref = ref;
508: } else if (*p == '&' || *p == '\\')
509: *sp++ = '\\';
510: } else if (*p == c) {
511: p++;
512: *sp++ = '\0';
513: size += sp - op;
514: s->new = xrealloc(text, size);
515: return (p);
516: } else if (*p == '\n') {
517: err(COMPILE,
518: "unescaped newline inside substitute pattern");
519: /* NOTREACHED */
520: }
521: *sp++ = *p;
522: }
523: size += sp - op;
1.28 millert 524: } while ((p = cu_fgets(&lbuf, &bufsize)));
1.1 deraadt 525: err(COMPILE, "unterminated substitute in regular expression");
526: /* NOTREACHED */
527: }
528:
529: /*
530: * Compile the flags of the s command
531: */
532: static char *
1.15 deraadt 533: compile_flags(char *p, struct s_subst *s)
1.1 deraadt 534: {
535: int gn; /* True if we have seen g or n */
1.17 otto 536: long l;
1.25 millert 537: char wfile[PATH_MAX], *q;
1.1 deraadt 538:
539: s->n = 1; /* Default */
540: s->p = 0;
541: s->wfile = NULL;
542: s->wfd = -1;
543: for (gn = 0;;) {
544: EATSPACE(); /* EXTENSION */
545: switch (*p) {
546: case 'g':
547: if (gn)
1.22 tedu 548: err(COMPILE, "more than one number or 'g' in"
549: " substitute flags");
1.1 deraadt 550: gn = 1;
551: s->n = 0;
552: break;
553: case '\0':
554: case '\n':
555: case ';':
556: return (p);
557: case 'p':
558: s->p = 1;
559: break;
560: case '1': case '2': case '3':
561: case '4': case '5': case '6':
562: case '7': case '8': case '9':
563: if (gn)
1.22 tedu 564: err(COMPILE, "more than one number or 'g' in"
565: " substitute flags");
1.1 deraadt 566: gn = 1;
1.17 otto 567: l = strtol(p, &p, 10);
568: if (l <= 0 || l >= INT_MAX)
569: err(COMPILE,
570: "number in substitute flags out of range");
571: s->n = (int)l;
572: continue;
1.1 deraadt 573: case 'w':
574: p++;
575: #ifdef HISTORIC_PRACTICE
576: if (*p != ' ') {
577: err(WARNING, "space missing before w wfile");
578: return (p);
579: }
580: #endif
581: EATSPACE();
582: q = wfile;
583: while (*p) {
584: if (*p == '\n')
585: break;
586: *q++ = *p++;
587: }
588: *q = '\0';
589: if (q == wfile)
590: err(COMPILE, "no wfile specified");
591: s->wfile = strdup(wfile);
592: if (!aflag && (s->wfd = open(wfile,
593: O_WRONLY|O_APPEND|O_CREAT|O_TRUNC,
594: DEFFILEMODE)) == -1)
1.12 jsyn 595: err(FATAL, "%s: %s", wfile, strerror(errno));
1.1 deraadt 596: return (p);
597: default:
598: err(COMPILE,
599: "bad flag in substitute command: '%c'", *p);
600: break;
601: }
602: p++;
603: }
604: }
605:
606: /*
607: * Compile a translation set of strings into a lookup table.
608: */
609: static char *
1.15 deraadt 610: compile_tr(char *p, char **transtab)
1.1 deraadt 611: {
612: int i;
613: char *lt, *op, *np;
1.25 millert 614: char *old = NULL, *new = NULL;
1.1 deraadt 615:
616: if (*p == '\0' || *p == '\\')
617: err(COMPILE,
618: "transform pattern can not be delimited by newline or backslash");
1.25 millert 619: old = xmalloc(strlen(p) + 1);
1.1 deraadt 620: p = compile_delimited(p, old);
621: if (p == NULL) {
622: err(COMPILE, "unterminated transform source string");
1.25 millert 623: goto bad;
1.1 deraadt 624: }
1.25 millert 625: new = xmalloc(strlen(p) + 1);
1.1 deraadt 626: p = compile_delimited(--p, new);
627: if (p == NULL) {
628: err(COMPILE, "unterminated transform target string");
1.25 millert 629: goto bad;
1.1 deraadt 630: }
631: EATSPACE();
632: if (strlen(new) != strlen(old)) {
633: err(COMPILE, "transform strings are not the same length");
1.25 millert 634: goto bad;
1.1 deraadt 635: }
636: /* We assume characters are 8 bits */
1.18 deraadt 637: lt = xmalloc(UCHAR_MAX + 1);
1.1 deraadt 638: for (i = 0; i <= UCHAR_MAX; i++)
639: lt[i] = (char)i;
640: for (op = old, np = new; *op; op++, np++)
641: lt[(u_char)*op] = *np;
642: *transtab = lt;
1.25 millert 643: free(old);
644: free(new);
1.1 deraadt 645: return (p);
1.25 millert 646: bad:
647: free(old);
648: free(new);
649: return (NULL);
1.1 deraadt 650: }
651:
652: /*
1.7 deraadt 653: * Compile the text following an a, c, or i command.
1.1 deraadt 654: */
655: static char *
1.15 deraadt 656: compile_text(void)
1.1 deraadt 657: {
1.4 deraadt 658: int asize, esc_nl, size;
1.25 millert 659: char *lbuf, *text, *p, *op, *s;
1.28 millert 660: size_t bufsize;
1.1 deraadt 661:
1.28 millert 662: lbuf = text = NULL;
663: asize = size = 0;
664: while ((p = cu_fgets(&lbuf, &bufsize))) {
665: size_t len = ROUNDLEN(strlen(p) + 1);
666: if (asize - size < len) {
1.25 millert 667: do {
1.28 millert 668: asize += len;
669: } while (asize - size < len);
1.25 millert 670: text = xrealloc(text, asize);
671: }
1.1 deraadt 672: op = s = text + size;
1.4 deraadt 673: for (esc_nl = 0; *p != '\0'; p++) {
674: if (*p == '\\' && p[1] != '\0' && *++p == '\n')
675: esc_nl = 1;
1.1 deraadt 676: *s++ = *p;
677: }
678: size += s - op;
1.4 deraadt 679: if (!esc_nl) {
1.1 deraadt 680: *s = '\0';
681: break;
682: }
683: }
1.25 millert 684: free(lbuf);
1.8 brian 685: text[size] = '\0';
1.1 deraadt 686: return (xrealloc(text, size + 1));
687: }
688:
689: /*
690: * Get an address and return a pointer to the first character after
691: * it. Fill the structure pointed to according to the address.
692: */
693: static char *
1.15 deraadt 694: compile_addr(char *p, struct s_addr *a)
1.1 deraadt 695: {
696: char *end;
697:
698: switch (*p) {
699: case '\\': /* Context address */
700: ++p;
701: /* FALLTHROUGH */
702: case '/': /* Context address */
703: p = compile_re(p, &a->u.r);
704: if (p == NULL)
705: err(COMPILE, "unterminated regular expression");
706: a->type = AT_RE;
707: return (p);
708:
709: case '$': /* Last line */
710: a->type = AT_LAST;
711: return (p + 1);
712: /* Line number */
1.21 deraadt 713: case '0': case '1': case '2': case '3': case '4':
1.1 deraadt 714: case '5': case '6': case '7': case '8': case '9':
715: a->type = AT_LINE;
1.3 millert 716: a->u.l = strtoul(p, &end, 10);
1.1 deraadt 717: return (end);
718: default:
719: err(COMPILE, "expected context address");
720: return (NULL);
721: }
722: }
723:
724: /*
725: * duptoeol --
726: * Return a copy of all the characters up to \n or \0.
727: */
728: static char *
1.16 deraadt 729: duptoeol(char *s, char *ctype, char **semi)
1.1 deraadt 730: {
731: size_t len;
732: int ws;
733: char *start;
734:
735: ws = 0;
1.16 deraadt 736: if (semi) {
737: for (start = s; *s != '\0' && *s != '\n' && *s != ';'; ++s)
738: ws = isspace(*s);
739: } else {
740: for (start = s; *s != '\0' && *s != '\n'; ++s)
741: ws = isspace(*s);
742: *s = '\0';
743: }
1.1 deraadt 744: if (ws)
745: err(WARNING, "whitespace after %s", ctype);
746: len = s - start + 1;
1.16 deraadt 747: if (semi)
748: *semi = s;
749: s = xmalloc(len);
750: strlcpy(s, start, len);
751: return (s);
1.1 deraadt 752: }
753:
754: /*
755: * Convert goto label names to addresses, and count a and r commands, in
756: * the given subset of the script. Free the memory used by labels in b
757: * and t commands (but not by :).
758: *
759: * TODO: Remove } nodes
760: */
761: static void
1.15 deraadt 762: fixuplabel(struct s_command *cp, struct s_command *end)
1.1 deraadt 763: {
764:
765: for (; cp != end; cp = cp->next)
766: switch (cp->code) {
767: case 'a':
768: case 'r':
769: appendnum++;
770: break;
771: case 'b':
772: case 't':
773: /* Resolve branch target. */
774: if (cp->t == NULL) {
775: cp->u.c = NULL;
776: break;
777: }
778: if ((cp->u.c = findlabel(cp->t)) == NULL)
779: err(COMPILE2, "undefined label '%s'", cp->t);
780: free(cp->t);
781: break;
782: case '{':
783: /* Do interior commands. */
784: fixuplabel(cp->u.c, cp->next);
785: break;
786: }
787: }
788:
789: /*
790: * Associate the given command label for later lookup.
791: */
792: static void
1.15 deraadt 793: enterlabel(struct s_command *cp)
1.1 deraadt 794: {
1.10 mpech 795: struct labhash **lhp, *lh;
796: u_char *p;
797: u_int h, c;
1.1 deraadt 798:
799: for (h = 0, p = (u_char *)cp->t; (c = *p) != 0; p++)
800: h = (h << 5) + h + c;
801: lhp = &labels[h & LHMASK];
802: for (lh = *lhp; lh != NULL; lh = lh->lh_next)
803: if (lh->lh_hash == h && strcmp(cp->t, lh->lh_cmd->t) == 0)
804: err(COMPILE2, "duplicate label '%s'", cp->t);
805: lh = xmalloc(sizeof *lh);
806: lh->lh_next = *lhp;
807: lh->lh_hash = h;
808: lh->lh_cmd = cp;
809: lh->lh_ref = 0;
810: *lhp = lh;
811: }
812:
813: /*
814: * Find the label contained in the command l in the command linked
815: * list cp. L is excluded from the search. Return NULL if not found.
816: */
817: static struct s_command *
1.15 deraadt 818: findlabel(char *name)
1.1 deraadt 819: {
1.10 mpech 820: struct labhash *lh;
821: u_char *p;
822: u_int h, c;
1.1 deraadt 823:
824: for (h = 0, p = (u_char *)name; (c = *p) != 0; p++)
825: h = (h << 5) + h + c;
826: for (lh = labels[h & LHMASK]; lh != NULL; lh = lh->lh_next) {
827: if (lh->lh_hash == h && strcmp(name, lh->lh_cmd->t) == 0) {
828: lh->lh_ref = 1;
829: return (lh->lh_cmd);
830: }
831: }
832: return (NULL);
833: }
834:
1.21 deraadt 835: /*
1.1 deraadt 836: * Warn about any unused labels. As a side effect, release the label hash
837: * table space.
838: */
839: static void
1.15 deraadt 840: uselabel(void)
1.1 deraadt 841: {
1.10 mpech 842: struct labhash *lh, *next;
843: int i;
1.1 deraadt 844:
845: for (i = 0; i < LHSZ; i++) {
846: for (lh = labels[i]; lh != NULL; lh = next) {
847: next = lh->lh_next;
848: if (!lh->lh_ref)
849: err(WARNING, "unused label '%s'",
850: lh->lh_cmd->t);
851: free(lh);
852: }
853: }
854: }