Annotation of src/usr.bin/sed/compile.c, Revision 1.43
1.43 ! martijn 1: /* $OpenBSD: compile.c,v 1.42 2017/08/01 18:05:53 martijn 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 *);
1.33 naddy 62: static char *compile_delimited(char *, char *, int);
1.11 millert 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();
1.36 deraadt 131: appends = xreallocarray(NULL, appendnum, sizeof(struct s_appends));
132: match = xreallocarray(NULL, maxnsub + 1, sizeof(regmatch_t));
1.1 deraadt 133: }
134:
135: #define EATSPACE() do { \
136: if (p) \
1.35 deraadt 137: while (isascii((unsigned char)*p) && \
138: isspace((unsigned char)*p)) \
1.1 deraadt 139: p++; \
140: } while (0)
141:
142: static struct s_command **
1.15 deraadt 143: compile_stream(struct s_command **link)
1.1 deraadt 144: {
1.10 mpech 145: char *p;
1.25 millert 146: static char *lbuf; /* To avoid excessive malloc calls */
1.28 millert 147: static size_t bufsize;
1.1 deraadt 148: struct s_command *cmd, *cmd2, *stack;
149: struct s_format *fp;
150: int naddr; /* Number of addresses */
151:
152: stack = 0;
153: for (;;) {
1.28 millert 154: if ((p = cu_fgets(&lbuf, &bufsize)) == NULL) {
1.1 deraadt 155: if (stack != 0)
1.39 mmcc 156: error(COMPILE, "unexpected EOF (pending }'s)");
1.1 deraadt 157: return (link);
158: }
159:
160: semicolon: EATSPACE();
1.23 otto 161: if (*p == '#' || *p == '\0')
1.1 deraadt 162: continue;
1.23 otto 163: if (*p == ';') {
164: p++;
165: goto semicolon;
166: }
1.1 deraadt 167: *link = cmd = xmalloc(sizeof(struct s_command));
168: link = &cmd->next;
169: cmd->nonsel = cmd->inrange = 0;
170: /* First parse the addresses */
171: naddr = 0;
172:
173: /* Valid characters to start an address */
174: #define addrchar(c) (strchr("0123456789/\\$", (c)))
175: if (addrchar(*p)) {
176: naddr++;
177: cmd->a1 = xmalloc(sizeof(struct s_addr));
178: p = compile_addr(p, cmd->a1);
179: EATSPACE(); /* EXTENSION */
180: if (*p == ',') {
181: p++;
182: EATSPACE(); /* EXTENSION */
183: naddr++;
184: cmd->a2 = xmalloc(sizeof(struct s_addr));
185: p = compile_addr(p, cmd->a2);
186: EATSPACE();
1.22 tedu 187: } else {
1.1 deraadt 188: cmd->a2 = 0;
1.22 tedu 189: }
190: } else {
1.1 deraadt 191: cmd->a1 = cmd->a2 = 0;
1.22 tedu 192: }
1.1 deraadt 193:
194: nonsel: /* Now parse the command */
195: if (!*p)
1.39 mmcc 196: error(COMPILE, "command expected");
1.1 deraadt 197: cmd->code = *p;
198: for (fp = cmd_fmts; fp->code; fp++)
199: if (fp->code == *p)
200: break;
201: if (!fp->code)
1.39 mmcc 202: error(COMPILE, "invalid command code %c", *p);
1.1 deraadt 203: if (naddr > fp->naddr)
1.39 mmcc 204: error(COMPILE,
1.22 tedu 205: "command %c expects up to %d address(es), found %d",
206: *p, fp->naddr, naddr);
1.1 deraadt 207: switch (fp->args) {
208: case NONSEL: /* ! */
209: p++;
210: EATSPACE();
1.38 millert 211: cmd->nonsel = 1;
1.1 deraadt 212: goto nonsel;
213: case GROUP: /* { */
214: p++;
215: EATSPACE();
216: cmd->next = stack;
217: stack = cmd;
218: link = &cmd->u.c;
219: 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: cmd->nonsel = 1;
228: if (stack == 0)
1.39 mmcc 229: error(COMPILE, "unexpected }");
1.1 deraadt 230: cmd2 = stack;
231: stack = cmd2->next;
232: cmd2->next = cmd;
233: /*FALLTHROUGH*/
234: case EMPTY: /* d D g G h H l n N p P q x = \0 */
235: p++;
236: EATSPACE();
237: if (*p == ';') {
238: p++;
239: link = &cmd->next;
240: goto semicolon;
241: }
242: if (*p)
1.39 mmcc 243: error(COMPILE,
1.1 deraadt 244: "extra characters at the end of %c command", cmd->code);
245: break;
246: case TEXT: /* a c i */
247: p++;
248: EATSPACE();
249: if (*p != '\\')
1.39 mmcc 250: error(COMPILE, "command %c expects \\ followed by"
1.22 tedu 251: " text", cmd->code);
1.1 deraadt 252: p++;
253: EATSPACE();
254: if (*p)
1.39 mmcc 255: error(COMPILE, "extra characters after \\ at the"
1.22 tedu 256: " end of %c command", cmd->code);
1.1 deraadt 257: cmd->t = compile_text();
258: break;
259: case COMMENT: /* \0 # */
260: break;
261: case WFILE: /* w */
262: p++;
263: EATSPACE();
264: if (*p == '\0')
1.39 mmcc 265: error(COMPILE, "filename expected");
1.16 deraadt 266: cmd->t = duptoeol(p, "w command", NULL);
1.42 martijn 267: if (aflag) {
1.1 deraadt 268: cmd->u.fd = -1;
1.42 martijn 269: pledge_wpath = 1;
270: }
1.21 deraadt 271: else if ((cmd->u.fd = open(p,
1.1 deraadt 272: O_WRONLY|O_APPEND|O_CREAT|O_TRUNC,
273: DEFFILEMODE)) == -1)
1.39 mmcc 274: error(FATAL, "%s: %s", p, strerror(errno));
1.1 deraadt 275: break;
276: case RFILE: /* r */
1.42 martijn 277: pledge_rpath = 1;
1.1 deraadt 278: p++;
279: EATSPACE();
1.43 ! martijn 280: if (*p == '\0')
! 281: error(COMPILE, "filename expected");
1.16 deraadt 282: cmd->t = duptoeol(p, "read command", NULL);
1.1 deraadt 283: break;
284: case BRANCH: /* b t */
285: p++;
286: EATSPACE();
287: if (*p == '\0')
288: cmd->t = NULL;
289: else
1.16 deraadt 290: cmd->t = duptoeol(p, "branch", &p);
291: if (*p == ';') {
292: p++;
293: goto semicolon;
294: }
1.1 deraadt 295: break;
296: case LABEL: /* : */
297: p++;
298: EATSPACE();
1.16 deraadt 299: cmd->t = duptoeol(p, "label", &p);
300: if (strlen(cmd->t) == 0)
1.39 mmcc 301: error(COMPILE, "empty label");
1.1 deraadt 302: enterlabel(cmd);
1.16 deraadt 303: if (*p == ';') {
304: p++;
305: goto semicolon;
306: }
1.1 deraadt 307: break;
308: case SUBST: /* s */
309: p++;
310: if (*p == '\0' || *p == '\\')
1.39 mmcc 311: error(COMPILE, "substitute pattern can not be"
1.22 tedu 312: " delimited by newline or backslash");
1.1 deraadt 313: cmd->u.s = xmalloc(sizeof(struct s_subst));
314: p = compile_re(p, &cmd->u.s->re);
315: if (p == NULL)
1.39 mmcc 316: error(COMPILE, "unterminated substitute pattern");
1.1 deraadt 317: --p;
318: p = compile_subst(p, cmd->u.s);
319: p = compile_flags(p, cmd->u.s);
320: EATSPACE();
321: if (*p == ';') {
322: p++;
323: link = &cmd->next;
324: goto semicolon;
325: }
326: break;
327: case TR: /* y */
328: p++;
329: p = compile_tr(p, (char **)&cmd->u.y);
330: EATSPACE();
331: if (*p == ';') {
332: p++;
333: link = &cmd->next;
334: goto semicolon;
335: }
336: if (*p)
1.39 mmcc 337: error(COMPILE, "extra text at the end of a"
1.22 tedu 338: " transform command");
1.1 deraadt 339: break;
340: }
341: }
342: }
343:
344: /*
345: * Get a delimited string. P points to the delimeter of the string; d points
346: * to a buffer area. Newline and delimiter escapes are processed; other
347: * escapes are ignored.
348: *
349: * Returns a pointer to the first character after the final delimiter or NULL
350: * in the case of a non-terminated string. The character array d is filled
351: * with the processed string.
352: */
353: static char *
1.33 naddy 354: compile_delimited(char *p, char *d, int is_tr)
1.1 deraadt 355: {
356: char c;
357:
358: c = *p++;
359: if (c == '\0')
360: return (NULL);
361: else if (c == '\\')
1.39 mmcc 362: error(COMPILE, "\\ can not be used as a string delimiter");
1.1 deraadt 363: else if (c == '\n')
1.39 mmcc 364: error(COMPILE, "newline can not be used as a string delimiter");
1.1 deraadt 365: while (*p) {
1.32 naddy 366: if (*p == '[' && *p != c) {
1.1 deraadt 367: if ((d = compile_ccl(&p, d)) == NULL)
1.39 mmcc 368: error(COMPILE, "unbalanced brackets ([])");
1.1 deraadt 369: continue;
370: } else if (*p == '\\' && p[1] == '[') {
371: *d++ = *p++;
1.22 tedu 372: } else if (*p == '\\' && p[1] == c) {
1.1 deraadt 373: p++;
1.22 tedu 374: } else if (*p == '\\' && p[1] == 'n') {
1.1 deraadt 375: *d++ = '\n';
376: p += 2;
377: continue;
1.22 tedu 378: } else if (*p == '\\' && p[1] == '\\') {
1.33 naddy 379: if (is_tr)
380: p++;
381: else
382: *d++ = *p++;
1.22 tedu 383: } else if (*p == c) {
1.1 deraadt 384: *d = '\0';
385: return (p + 1);
386: }
387: *d++ = *p++;
388: }
389: return (NULL);
390: }
391:
392:
393: /* compile_ccl: expand a POSIX character class */
394: static char *
1.15 deraadt 395: compile_ccl(char **sp, char *t)
1.1 deraadt 396: {
397: int c, d;
398: char *s = *sp;
399:
400: *t++ = *s++;
401: if (*s == '^')
402: *t++ = *s++;
403: if (*s == ']')
404: *t++ = *s++;
405: for (; *s && (*t = *s) != ']'; s++, t++)
406: if (*s == '[' && ((d = *(s+1)) == '.' || d == ':' || d == '=')) {
407: *++t = *++s, t++, s++;
408: for (c = *s; (*t = *s) != ']' || c != d; s++, t++)
409: if ((c = *s) == '\0')
410: return NULL;
1.22 tedu 411: } else if (*s == '\\' && s[1] == 'n') {
412: *t = '\n';
413: s++;
414: }
415: if (*s == ']') {
416: *sp = ++s;
417: return (++t);
418: } else {
419: return (NULL);
420: }
1.1 deraadt 421: }
422:
423: /*
424: * Get a regular expression. P points to the delimiter of the regular
425: * expression; repp points to the address of a regexp pointer. Newline
426: * and delimiter escapes are processed; other escapes are ignored.
427: * Returns a pointer to the first character after the final delimiter
428: * or NULL in the case of a non terminated regular expression. The regexp
429: * pointer is set to the compiled regular expression.
430: * Cflags are passed to regcomp.
431: */
432: static char *
1.15 deraadt 433: compile_re(char *p, regex_t **repp)
1.1 deraadt 434: {
435: int eval;
1.25 millert 436: char *re;
1.1 deraadt 437:
1.25 millert 438: re = xmalloc(strlen(p) + 1); /* strlen(re) <= strlen(p) */
1.33 naddy 439: p = compile_delimited(p, re, 0);
1.1 deraadt 440: if (p && strlen(re) == 0) {
441: *repp = NULL;
1.25 millert 442: free(re);
1.1 deraadt 443: return (p);
444: }
445: *repp = xmalloc(sizeof(regex_t));
1.29 djm 446: if (p && (eval = regcomp(*repp, re, Eflag ? REG_EXTENDED : 0)) != 0)
1.39 mmcc 447: error(COMPILE, "RE error: %s", strregerror(eval, *repp));
1.1 deraadt 448: if (maxnsub < (*repp)->re_nsub)
449: maxnsub = (*repp)->re_nsub;
1.25 millert 450: free(re);
1.1 deraadt 451: return (p);
452: }
453:
454: /*
455: * Compile the substitution string of a regular expression and set res to
456: * point to a saved copy of it. Nsub is the number of parenthesized regular
457: * expressions.
458: */
459: static char *
1.15 deraadt 460: compile_subst(char *p, struct s_subst *s)
1.1 deraadt 461: {
1.25 millert 462: static char *lbuf;
1.28 millert 463: static size_t bufsize;
1.1 deraadt 464: int asize, ref, size;
465: char c, *text, *op, *sp;
1.19 otto 466: int sawesc = 0;
1.1 deraadt 467:
468: c = *p++; /* Terminator character */
469: if (c == '\0')
470: return (NULL);
471:
472: s->maxbref = 0;
473: s->linenum = linenum;
1.28 millert 474: text = NULL;
475: asize = size = 0;
1.1 deraadt 476: do {
1.28 millert 477: size_t len = ROUNDLEN(strlen(p) + 1);
478: if (asize - size < len) {
1.25 millert 479: do {
1.28 millert 480: asize += len;
481: } while (asize - size < len);
1.25 millert 482: text = xrealloc(text, asize);
483: }
1.1 deraadt 484: op = sp = text + size;
485: for (; *p; p++) {
1.19 otto 486: if (*p == '\\' || sawesc) {
487: /*
488: * If this is a continuation from the last
489: * buffer, we won't have a character to
490: * skip over.
491: */
492: if (sawesc)
493: sawesc = 0;
494: else
495: p++;
496:
497: if (*p == '\0') {
498: /*
499: * This escaped character is continued
500: * in the next part of the line. Note
501: * this fact, then cause the loop to
502: * exit w/ normal EOL case and reenter
503: * above with the new buffer.
504: */
505: sawesc = 1;
506: p--;
507: continue;
508: } else if (strchr("123456789", *p) != NULL) {
1.1 deraadt 509: *sp++ = '\\';
510: ref = *p - '0';
511: if (s->re != NULL &&
512: ref > s->re->re_nsub)
1.39 mmcc 513: error(COMPILE,
1.1 deraadt 514: "\\%c not defined in the RE", *p);
515: if (s->maxbref < ref)
516: s->maxbref = ref;
517: } else if (*p == '&' || *p == '\\')
518: *sp++ = '\\';
519: } else if (*p == c) {
520: p++;
521: *sp++ = '\0';
522: size += sp - op;
523: s->new = xrealloc(text, size);
524: return (p);
525: } else if (*p == '\n') {
1.39 mmcc 526: error(COMPILE,
1.1 deraadt 527: "unescaped newline inside substitute pattern");
528: }
529: *sp++ = *p;
530: }
531: size += sp - op;
1.28 millert 532: } while ((p = cu_fgets(&lbuf, &bufsize)));
1.39 mmcc 533: error(COMPILE, "unterminated substitute in regular expression");
1.1 deraadt 534: }
535:
536: /*
537: * Compile the flags of the s command
538: */
539: static char *
1.15 deraadt 540: compile_flags(char *p, struct s_subst *s)
1.1 deraadt 541: {
542: int gn; /* True if we have seen g or n */
1.17 otto 543: long l;
1.1 deraadt 544:
545: s->n = 1; /* Default */
546: s->p = 0;
547: s->wfile = NULL;
548: s->wfd = -1;
549: for (gn = 0;;) {
550: EATSPACE(); /* EXTENSION */
551: switch (*p) {
552: case 'g':
553: if (gn)
1.39 mmcc 554: error(COMPILE, "more than one number or 'g' in"
1.22 tedu 555: " substitute flags");
1.1 deraadt 556: gn = 1;
557: s->n = 0;
558: break;
559: case '\0':
560: case '\n':
561: case ';':
562: return (p);
563: case 'p':
564: s->p = 1;
565: break;
566: case '1': case '2': case '3':
567: case '4': case '5': case '6':
568: case '7': case '8': case '9':
569: if (gn)
1.39 mmcc 570: error(COMPILE, "more than one number or 'g' in"
1.22 tedu 571: " substitute flags");
1.1 deraadt 572: gn = 1;
1.17 otto 573: l = strtol(p, &p, 10);
574: if (l <= 0 || l >= INT_MAX)
1.39 mmcc 575: error(COMPILE,
1.17 otto 576: "number in substitute flags out of range");
577: s->n = (int)l;
578: continue;
1.1 deraadt 579: case 'w':
580: p++;
581: EATSPACE();
1.43 ! martijn 582: if (*p == '\0')
! 583: error(COMPILE, "filename expected");
! 584: s->wfile = duptoeol(p, "s command w flag", NULL);
! 585: *p = '\0';
1.42 martijn 586: if (aflag)
587: pledge_wpath = 1;
1.43 ! martijn 588: else if ((s->wfd = open(s->wfile,
1.1 deraadt 589: O_WRONLY|O_APPEND|O_CREAT|O_TRUNC,
590: DEFFILEMODE)) == -1)
1.43 ! martijn 591: error(FATAL, "%s: %s", s->wfile, strerror(errno));
1.1 deraadt 592: return (p);
593: default:
1.39 mmcc 594: error(COMPILE,
1.1 deraadt 595: "bad flag in substitute command: '%c'", *p);
596: break;
597: }
598: p++;
599: }
600: }
601:
602: /*
603: * Compile a translation set of strings into a lookup table.
604: */
605: static char *
1.15 deraadt 606: compile_tr(char *p, char **transtab)
1.1 deraadt 607: {
608: int i;
609: char *lt, *op, *np;
1.25 millert 610: char *old = NULL, *new = NULL;
1.1 deraadt 611:
612: if (*p == '\0' || *p == '\\')
1.39 mmcc 613: error(COMPILE,
1.1 deraadt 614: "transform pattern can not be delimited by newline or backslash");
1.25 millert 615: old = xmalloc(strlen(p) + 1);
1.33 naddy 616: p = compile_delimited(p, old, 1);
1.1 deraadt 617: if (p == NULL) {
1.39 mmcc 618: error(COMPILE, "unterminated transform source string");
1.25 millert 619: goto bad;
1.1 deraadt 620: }
1.25 millert 621: new = xmalloc(strlen(p) + 1);
1.33 naddy 622: p = compile_delimited(--p, new, 1);
1.1 deraadt 623: if (p == NULL) {
1.39 mmcc 624: error(COMPILE, "unterminated transform target string");
1.25 millert 625: goto bad;
1.1 deraadt 626: }
627: EATSPACE();
628: if (strlen(new) != strlen(old)) {
1.39 mmcc 629: error(COMPILE, "transform strings are not the same length");
1.25 millert 630: goto bad;
1.1 deraadt 631: }
632: /* We assume characters are 8 bits */
1.18 deraadt 633: lt = xmalloc(UCHAR_MAX + 1);
1.1 deraadt 634: for (i = 0; i <= UCHAR_MAX; i++)
635: lt[i] = (char)i;
636: for (op = old, np = new; *op; op++, np++)
637: lt[(u_char)*op] = *np;
638: *transtab = lt;
1.25 millert 639: free(old);
640: free(new);
1.1 deraadt 641: return (p);
1.25 millert 642: bad:
643: free(old);
644: free(new);
645: return (NULL);
1.1 deraadt 646: }
647:
648: /*
1.7 deraadt 649: * Compile the text following an a, c, or i command.
1.1 deraadt 650: */
651: static char *
1.15 deraadt 652: compile_text(void)
1.1 deraadt 653: {
1.4 deraadt 654: int asize, esc_nl, size;
1.25 millert 655: char *lbuf, *text, *p, *op, *s;
1.28 millert 656: size_t bufsize;
1.1 deraadt 657:
1.28 millert 658: lbuf = text = NULL;
659: asize = size = 0;
660: while ((p = cu_fgets(&lbuf, &bufsize))) {
661: size_t len = ROUNDLEN(strlen(p) + 1);
662: if (asize - size < len) {
1.25 millert 663: do {
1.28 millert 664: asize += len;
665: } while (asize - size < len);
1.25 millert 666: text = xrealloc(text, asize);
667: }
1.1 deraadt 668: op = s = text + size;
1.4 deraadt 669: for (esc_nl = 0; *p != '\0'; p++) {
670: if (*p == '\\' && p[1] != '\0' && *++p == '\n')
671: esc_nl = 1;
1.1 deraadt 672: *s++ = *p;
673: }
674: size += s - op;
1.4 deraadt 675: if (!esc_nl) {
1.1 deraadt 676: *s = '\0';
677: break;
678: }
679: }
1.25 millert 680: free(lbuf);
1.34 millert 681: text = xrealloc(text, size + 1);
1.8 brian 682: text[size] = '\0';
1.34 millert 683: return (text);
1.1 deraadt 684: }
685:
686: /*
687: * Get an address and return a pointer to the first character after
688: * it. Fill the structure pointed to according to the address.
689: */
690: static char *
1.15 deraadt 691: compile_addr(char *p, struct s_addr *a)
1.1 deraadt 692: {
693: char *end;
694:
695: switch (*p) {
696: case '\\': /* Context address */
697: ++p;
698: /* FALLTHROUGH */
699: case '/': /* Context address */
700: p = compile_re(p, &a->u.r);
701: if (p == NULL)
1.39 mmcc 702: error(COMPILE, "unterminated regular expression");
1.1 deraadt 703: a->type = AT_RE;
704: return (p);
705:
706: case '$': /* Last line */
707: a->type = AT_LAST;
708: return (p + 1);
709: /* Line number */
1.21 deraadt 710: case '0': case '1': case '2': case '3': case '4':
1.1 deraadt 711: case '5': case '6': case '7': case '8': case '9':
712: a->type = AT_LINE;
1.3 millert 713: a->u.l = strtoul(p, &end, 10);
1.1 deraadt 714: return (end);
715: default:
1.39 mmcc 716: error(COMPILE, "expected context address");
1.1 deraadt 717: return (NULL);
718: }
719: }
720:
721: /*
722: * duptoeol --
723: * Return a copy of all the characters up to \n or \0.
724: */
725: static char *
1.16 deraadt 726: duptoeol(char *s, char *ctype, char **semi)
1.1 deraadt 727: {
728: size_t len;
729: int ws;
730: char *start;
731:
732: ws = 0;
1.16 deraadt 733: if (semi) {
734: for (start = s; *s != '\0' && *s != '\n' && *s != ';'; ++s)
1.35 deraadt 735: ws = isspace((unsigned char)*s);
1.16 deraadt 736: } else {
737: for (start = s; *s != '\0' && *s != '\n'; ++s)
1.35 deraadt 738: ws = isspace((unsigned char)*s);
1.16 deraadt 739: *s = '\0';
740: }
1.1 deraadt 741: if (ws)
1.41 krw 742: warning("whitespace after %s", ctype);
1.1 deraadt 743: len = s - start + 1;
1.16 deraadt 744: if (semi)
745: *semi = s;
746: s = xmalloc(len);
747: strlcpy(s, start, len);
748: return (s);
1.1 deraadt 749: }
750:
751: /*
752: * Convert goto label names to addresses, and count a and r commands, in
753: * the given subset of the script. Free the memory used by labels in b
754: * and t commands (but not by :).
755: *
756: * TODO: Remove } nodes
757: */
758: static void
1.15 deraadt 759: fixuplabel(struct s_command *cp, struct s_command *end)
1.1 deraadt 760: {
761:
762: for (; cp != end; cp = cp->next)
763: switch (cp->code) {
764: case 'a':
765: case 'r':
766: appendnum++;
767: break;
768: case 'b':
769: case 't':
770: /* Resolve branch target. */
771: if (cp->t == NULL) {
772: cp->u.c = NULL;
773: break;
774: }
775: if ((cp->u.c = findlabel(cp->t)) == NULL)
1.40 jca 776: error(COMPILE, "undefined label '%s'", cp->t);
1.1 deraadt 777: free(cp->t);
778: break;
779: case '{':
780: /* Do interior commands. */
781: fixuplabel(cp->u.c, cp->next);
782: break;
783: }
784: }
785:
786: /*
787: * Associate the given command label for later lookup.
788: */
789: static void
1.15 deraadt 790: enterlabel(struct s_command *cp)
1.1 deraadt 791: {
1.10 mpech 792: struct labhash **lhp, *lh;
793: u_char *p;
794: u_int h, c;
1.1 deraadt 795:
796: for (h = 0, p = (u_char *)cp->t; (c = *p) != 0; p++)
797: h = (h << 5) + h + c;
798: lhp = &labels[h & LHMASK];
799: for (lh = *lhp; lh != NULL; lh = lh->lh_next)
800: if (lh->lh_hash == h && strcmp(cp->t, lh->lh_cmd->t) == 0)
1.40 jca 801: error(COMPILE, "duplicate label '%s'", cp->t);
1.1 deraadt 802: lh = xmalloc(sizeof *lh);
803: lh->lh_next = *lhp;
804: lh->lh_hash = h;
805: lh->lh_cmd = cp;
806: lh->lh_ref = 0;
807: *lhp = lh;
808: }
809:
810: /*
811: * Find the label contained in the command l in the command linked
812: * list cp. L is excluded from the search. Return NULL if not found.
813: */
814: static struct s_command *
1.15 deraadt 815: findlabel(char *name)
1.1 deraadt 816: {
1.10 mpech 817: struct labhash *lh;
818: u_char *p;
819: u_int h, c;
1.1 deraadt 820:
821: for (h = 0, p = (u_char *)name; (c = *p) != 0; p++)
822: h = (h << 5) + h + c;
823: for (lh = labels[h & LHMASK]; lh != NULL; lh = lh->lh_next) {
824: if (lh->lh_hash == h && strcmp(name, lh->lh_cmd->t) == 0) {
825: lh->lh_ref = 1;
826: return (lh->lh_cmd);
827: }
828: }
829: return (NULL);
830: }
831:
1.21 deraadt 832: /*
1.1 deraadt 833: * Warn about any unused labels. As a side effect, release the label hash
834: * table space.
835: */
836: static void
1.15 deraadt 837: uselabel(void)
1.1 deraadt 838: {
1.10 mpech 839: struct labhash *lh, *next;
840: int i;
1.1 deraadt 841:
842: for (i = 0; i < LHSZ; i++) {
843: for (lh = labels[i]; lh != NULL; lh = next) {
844: next = lh->lh_next;
845: if (!lh->lh_ref)
1.41 krw 846: warning("unused label '%s'",
1.1 deraadt 847: lh->lh_cmd->t);
848: free(lh);
849: }
850: }
851: }