Annotation of src/usr.bin/sed/compile.c, Revision 1.49
1.49 ! schwarze 1: /* $OpenBSD: compile.c,v 1.48 2018/07/09 09:43:54 schwarze 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 *);
1.44 martijn 62: static char *compile_delimited(char *, char *);
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 *
1.44 martijn 354: compile_delimited(char *p, char *d)
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");
1.49 ! schwarze 365:
! 366: while (p[0]) {
! 367: /* Unescaped delimiter: We are done. */
! 368: if (p[0] == c) {
1.1 deraadt 369: *d = '\0';
1.49 ! schwarze 370: return p + 1;
1.1 deraadt 371: }
1.49 ! schwarze 372: if (p[0] == '\\') {
! 373: /* Escaped delimiter: Skip the backslash. */
! 374: if (p[1] == c) {
! 375: p++;
! 376: } else {
! 377: /* Backslash-n: Match linefeed. */
! 378: if (p[1] == 'n') {
! 379: *d++ = '\n';
! 380: p += 2;
! 381: /* Other escapes remain unchanged. */
! 382: } else {
! 383: *d++ = *p++;
! 384: *d++ = *p++;
! 385: }
! 386: continue;
! 387: }
! 388: }
! 389: if (p[0] != '[')
! 390: *d++ = *p++;
! 391: /*
! 392: * Bracket expression:
! 393: * It may contain the delimiter without escaping.
! 394: */
! 395: else if ((d = compile_ccl(&p, d)) == NULL)
! 396: error(COMPILE, "unbalanced brackets ([])");
1.1 deraadt 397: }
1.49 ! schwarze 398: return NULL;
1.1 deraadt 399: }
400:
401:
402: /* compile_ccl: expand a POSIX character class */
403: static char *
1.15 deraadt 404: compile_ccl(char **sp, char *t)
1.1 deraadt 405: {
406: int c, d;
407: char *s = *sp;
408:
409: *t++ = *s++;
410: if (*s == '^')
411: *t++ = *s++;
412: if (*s == ']')
413: *t++ = *s++;
414: for (; *s && (*t = *s) != ']'; s++, t++)
415: if (*s == '[' && ((d = *(s+1)) == '.' || d == ':' || d == '=')) {
416: *++t = *++s, t++, s++;
417: for (c = *s; (*t = *s) != ']' || c != d; s++, t++)
418: if ((c = *s) == '\0')
419: return NULL;
1.22 tedu 420: } else if (*s == '\\' && s[1] == 'n') {
421: *t = '\n';
422: s++;
423: }
424: if (*s == ']') {
425: *sp = ++s;
426: return (++t);
427: } else {
428: return (NULL);
429: }
1.1 deraadt 430: }
431:
432: /*
433: * Get a regular expression. P points to the delimiter of the regular
434: * expression; repp points to the address of a regexp pointer. Newline
435: * and delimiter escapes are processed; other escapes are ignored.
436: * Returns a pointer to the first character after the final delimiter
437: * or NULL in the case of a non terminated regular expression. The regexp
438: * pointer is set to the compiled regular expression.
439: * Cflags are passed to regcomp.
440: */
441: static char *
1.15 deraadt 442: compile_re(char *p, regex_t **repp)
1.1 deraadt 443: {
444: int eval;
1.25 millert 445: char *re;
1.1 deraadt 446:
1.25 millert 447: re = xmalloc(strlen(p) + 1); /* strlen(re) <= strlen(p) */
1.44 martijn 448: p = compile_delimited(p, re);
1.1 deraadt 449: if (p && strlen(re) == 0) {
450: *repp = NULL;
1.25 millert 451: free(re);
1.1 deraadt 452: return (p);
453: }
454: *repp = xmalloc(sizeof(regex_t));
1.29 djm 455: if (p && (eval = regcomp(*repp, re, Eflag ? REG_EXTENDED : 0)) != 0)
1.39 mmcc 456: error(COMPILE, "RE error: %s", strregerror(eval, *repp));
1.1 deraadt 457: if (maxnsub < (*repp)->re_nsub)
458: maxnsub = (*repp)->re_nsub;
1.25 millert 459: free(re);
1.1 deraadt 460: return (p);
461: }
462:
463: /*
464: * Compile the substitution string of a regular expression and set res to
465: * point to a saved copy of it. Nsub is the number of parenthesized regular
466: * expressions.
467: */
468: static char *
1.15 deraadt 469: compile_subst(char *p, struct s_subst *s)
1.1 deraadt 470: {
1.25 millert 471: static char *lbuf;
1.28 millert 472: static size_t bufsize;
1.46 millert 473: size_t asize, ref, size;
1.1 deraadt 474: char c, *text, *op, *sp;
1.19 otto 475: int sawesc = 0;
1.1 deraadt 476:
477: c = *p++; /* Terminator character */
478: if (c == '\0')
479: return (NULL);
480:
481: s->maxbref = 0;
482: s->linenum = linenum;
1.28 millert 483: text = NULL;
484: asize = size = 0;
1.1 deraadt 485: do {
1.28 millert 486: size_t len = ROUNDLEN(strlen(p) + 1);
487: if (asize - size < len) {
1.25 millert 488: do {
1.28 millert 489: asize += len;
490: } while (asize - size < len);
1.25 millert 491: text = xrealloc(text, asize);
492: }
1.1 deraadt 493: op = sp = text + size;
494: for (; *p; p++) {
1.19 otto 495: if (*p == '\\' || sawesc) {
496: /*
497: * If this is a continuation from the last
498: * buffer, we won't have a character to
499: * skip over.
500: */
501: if (sawesc)
502: sawesc = 0;
503: else
504: p++;
505:
506: if (*p == '\0') {
507: /*
508: * This escaped character is continued
509: * in the next part of the line. Note
510: * this fact, then cause the loop to
511: * exit w/ normal EOL case and reenter
512: * above with the new buffer.
513: */
514: sawesc = 1;
515: p--;
516: continue;
517: } else if (strchr("123456789", *p) != NULL) {
1.1 deraadt 518: *sp++ = '\\';
519: ref = *p - '0';
520: if (s->re != NULL &&
521: ref > s->re->re_nsub)
1.39 mmcc 522: error(COMPILE,
1.1 deraadt 523: "\\%c not defined in the RE", *p);
524: if (s->maxbref < ref)
525: s->maxbref = ref;
526: } else if (*p == '&' || *p == '\\')
527: *sp++ = '\\';
528: } else if (*p == c) {
529: p++;
530: *sp++ = '\0';
531: size += sp - op;
532: s->new = xrealloc(text, size);
533: return (p);
534: } else if (*p == '\n') {
1.39 mmcc 535: error(COMPILE,
1.1 deraadt 536: "unescaped newline inside substitute pattern");
537: }
538: *sp++ = *p;
539: }
540: size += sp - op;
1.28 millert 541: } while ((p = cu_fgets(&lbuf, &bufsize)));
1.39 mmcc 542: error(COMPILE, "unterminated substitute in regular expression");
1.1 deraadt 543: }
544:
545: /*
546: * Compile the flags of the s command
547: */
548: static char *
1.15 deraadt 549: compile_flags(char *p, struct s_subst *s)
1.1 deraadt 550: {
551: int gn; /* True if we have seen g or n */
1.17 otto 552: long l;
1.1 deraadt 553:
554: s->n = 1; /* Default */
555: s->p = 0;
556: s->wfile = NULL;
557: s->wfd = -1;
558: for (gn = 0;;) {
559: EATSPACE(); /* EXTENSION */
560: switch (*p) {
561: case 'g':
562: if (gn)
1.39 mmcc 563: error(COMPILE, "more than one number or 'g' in"
1.22 tedu 564: " substitute flags");
1.1 deraadt 565: gn = 1;
566: s->n = 0;
567: break;
568: case '\0':
569: case '\n':
570: case ';':
571: return (p);
572: case 'p':
573: s->p = 1;
574: break;
575: case '1': case '2': case '3':
576: case '4': case '5': case '6':
577: case '7': case '8': case '9':
578: if (gn)
1.39 mmcc 579: error(COMPILE, "more than one number or 'g' in"
1.22 tedu 580: " substitute flags");
1.1 deraadt 581: gn = 1;
1.17 otto 582: l = strtol(p, &p, 10);
583: if (l <= 0 || l >= INT_MAX)
1.39 mmcc 584: error(COMPILE,
1.17 otto 585: "number in substitute flags out of range");
586: s->n = (int)l;
587: continue;
1.1 deraadt 588: case 'w':
589: p++;
590: EATSPACE();
1.43 martijn 591: if (*p == '\0')
592: error(COMPILE, "filename expected");
593: s->wfile = duptoeol(p, "s command w flag", NULL);
594: *p = '\0';
1.42 martijn 595: if (aflag)
596: pledge_wpath = 1;
1.43 martijn 597: else if ((s->wfd = open(s->wfile,
1.1 deraadt 598: O_WRONLY|O_APPEND|O_CREAT|O_TRUNC,
599: DEFFILEMODE)) == -1)
1.43 martijn 600: error(FATAL, "%s: %s", s->wfile, strerror(errno));
1.1 deraadt 601: return (p);
602: default:
1.39 mmcc 603: error(COMPILE,
1.1 deraadt 604: "bad flag in substitute command: '%c'", *p);
605: break;
606: }
607: p++;
608: }
609: }
610:
611: /*
612: * Compile a translation set of strings into a lookup table.
613: */
614: static char *
1.44 martijn 615: compile_tr(char *old, char **transtab)
1.1 deraadt 616: {
617: int i;
1.45 martijn 618: char delimiter, check[UCHAR_MAX + 1];
1.44 martijn 619: char *new, *end;
620:
621: memset(check, 0, sizeof(check));
622: delimiter = *old;
623: if (delimiter == '\\')
624: error(COMPILE, "\\ can not be used as a string delimiter");
625: else if (delimiter == '\n' || delimiter == '\0')
626: error(COMPILE, "newline can not be used as a string delimiter");
627:
628: new = old++;
629: do {
630: if ((new = strchr(new + 1, delimiter)) == NULL)
631: error(COMPILE, "unterminated transform source string");
632: } while (*(new - 1) == '\\' && *(new -2) != '\\');
633: *new = '\0';
634: end = new++;
635: do {
636: if ((end = strchr(end + 1, delimiter)) == NULL)
637: error(COMPILE, "unterminated transform target string");
638: } while (*(end -1) == '\\' && *(end -2) != '\\');
639: *end = '\0';
640:
641: /* We assume characters are 8 bits */
642: *transtab = xmalloc(UCHAR_MAX + 1);
643: for (i = 0; i <= UCHAR_MAX; i++)
644: (*transtab)[i] = (char)i;
1.1 deraadt 645:
1.44 martijn 646: while (*old != '\0' && *new != '\0') {
647: if (*old == '\\') {
648: old++;
649: if (*old == 'n')
650: *old = '\n';
651: else if (*old != delimiter && *old != '\\')
652: error(COMPILE, "Unexpected character after "
653: "backslash");
654: }
655: if (*new == '\\') {
656: new++;
657: if (*new == 'n')
658: *new = '\n';
659: else if (*new != delimiter && *new != '\\')
660: error(COMPILE, "Unexpected character after "
661: "backslash");
662: }
1.47 millert 663: if (check[(u_char) *old] == 1)
1.44 martijn 664: error(COMPILE, "Repeated character in source string");
1.47 millert 665: check[(u_char) *old] = 1;
1.45 martijn 666: (*transtab)[(u_char) *old++] = *new++;
1.1 deraadt 667: }
1.44 martijn 668: if (*old != '\0' || *new != '\0')
1.39 mmcc 669: error(COMPILE, "transform strings are not the same length");
1.44 martijn 670: return end + 1;
1.1 deraadt 671: }
672:
673: /*
1.7 deraadt 674: * Compile the text following an a, c, or i command.
1.1 deraadt 675: */
676: static char *
1.15 deraadt 677: compile_text(void)
1.1 deraadt 678: {
1.46 millert 679: size_t asize, size, bufsize;
1.25 millert 680: char *lbuf, *text, *p, *op, *s;
1.46 millert 681: int esc_nl;
1.1 deraadt 682:
1.28 millert 683: lbuf = text = NULL;
684: asize = size = 0;
685: while ((p = cu_fgets(&lbuf, &bufsize))) {
686: size_t len = ROUNDLEN(strlen(p) + 1);
687: if (asize - size < len) {
1.25 millert 688: do {
1.28 millert 689: asize += len;
690: } while (asize - size < len);
1.25 millert 691: text = xrealloc(text, asize);
692: }
1.1 deraadt 693: op = s = text + size;
1.4 deraadt 694: for (esc_nl = 0; *p != '\0'; p++) {
695: if (*p == '\\' && p[1] != '\0' && *++p == '\n')
696: esc_nl = 1;
1.1 deraadt 697: *s++ = *p;
698: }
699: size += s - op;
1.4 deraadt 700: if (!esc_nl) {
1.1 deraadt 701: *s = '\0';
702: break;
703: }
704: }
1.25 millert 705: free(lbuf);
1.34 millert 706: text = xrealloc(text, size + 1);
1.8 brian 707: text[size] = '\0';
1.34 millert 708: return (text);
1.1 deraadt 709: }
710:
711: /*
712: * Get an address and return a pointer to the first character after
713: * it. Fill the structure pointed to according to the address.
714: */
715: static char *
1.15 deraadt 716: compile_addr(char *p, struct s_addr *a)
1.1 deraadt 717: {
718: char *end;
719:
720: switch (*p) {
721: case '\\': /* Context address */
722: ++p;
723: /* FALLTHROUGH */
724: case '/': /* Context address */
725: p = compile_re(p, &a->u.r);
726: if (p == NULL)
1.39 mmcc 727: error(COMPILE, "unterminated regular expression");
1.1 deraadt 728: a->type = AT_RE;
729: return (p);
730:
731: case '$': /* Last line */
732: a->type = AT_LAST;
733: return (p + 1);
734: /* Line number */
1.21 deraadt 735: case '0': case '1': case '2': case '3': case '4':
1.1 deraadt 736: case '5': case '6': case '7': case '8': case '9':
737: a->type = AT_LINE;
1.3 millert 738: a->u.l = strtoul(p, &end, 10);
1.1 deraadt 739: return (end);
740: default:
1.39 mmcc 741: error(COMPILE, "expected context address");
1.1 deraadt 742: return (NULL);
743: }
744: }
745:
746: /*
747: * duptoeol --
748: * Return a copy of all the characters up to \n or \0.
749: */
750: static char *
1.16 deraadt 751: duptoeol(char *s, char *ctype, char **semi)
1.1 deraadt 752: {
753: size_t len;
754: int ws;
755: char *start;
756:
757: ws = 0;
1.16 deraadt 758: if (semi) {
759: for (start = s; *s != '\0' && *s != '\n' && *s != ';'; ++s)
1.35 deraadt 760: ws = isspace((unsigned char)*s);
1.16 deraadt 761: } else {
762: for (start = s; *s != '\0' && *s != '\n'; ++s)
1.35 deraadt 763: ws = isspace((unsigned char)*s);
1.16 deraadt 764: *s = '\0';
765: }
1.1 deraadt 766: if (ws)
1.41 krw 767: warning("whitespace after %s", ctype);
1.1 deraadt 768: len = s - start + 1;
1.16 deraadt 769: if (semi)
770: *semi = s;
771: s = xmalloc(len);
772: strlcpy(s, start, len);
773: return (s);
1.1 deraadt 774: }
775:
776: /*
777: * Convert goto label names to addresses, and count a and r commands, in
778: * the given subset of the script. Free the memory used by labels in b
779: * and t commands (but not by :).
780: *
781: * TODO: Remove } nodes
782: */
783: static void
1.15 deraadt 784: fixuplabel(struct s_command *cp, struct s_command *end)
1.1 deraadt 785: {
786:
787: for (; cp != end; cp = cp->next)
788: switch (cp->code) {
789: case 'a':
790: case 'r':
791: appendnum++;
792: break;
793: case 'b':
794: case 't':
795: /* Resolve branch target. */
796: if (cp->t == NULL) {
797: cp->u.c = NULL;
798: break;
799: }
800: if ((cp->u.c = findlabel(cp->t)) == NULL)
1.40 jca 801: error(COMPILE, "undefined label '%s'", cp->t);
1.1 deraadt 802: free(cp->t);
803: break;
804: case '{':
805: /* Do interior commands. */
806: fixuplabel(cp->u.c, cp->next);
807: break;
808: }
809: }
810:
811: /*
812: * Associate the given command label for later lookup.
813: */
814: static void
1.15 deraadt 815: enterlabel(struct s_command *cp)
1.1 deraadt 816: {
1.10 mpech 817: struct labhash **lhp, *lh;
818: u_char *p;
819: u_int h, c;
1.1 deraadt 820:
821: for (h = 0, p = (u_char *)cp->t; (c = *p) != 0; p++)
822: h = (h << 5) + h + c;
823: lhp = &labels[h & LHMASK];
824: for (lh = *lhp; lh != NULL; lh = lh->lh_next)
825: if (lh->lh_hash == h && strcmp(cp->t, lh->lh_cmd->t) == 0)
1.40 jca 826: error(COMPILE, "duplicate label '%s'", cp->t);
1.1 deraadt 827: lh = xmalloc(sizeof *lh);
828: lh->lh_next = *lhp;
829: lh->lh_hash = h;
830: lh->lh_cmd = cp;
831: lh->lh_ref = 0;
832: *lhp = lh;
833: }
834:
835: /*
836: * Find the label contained in the command l in the command linked
837: * list cp. L is excluded from the search. Return NULL if not found.
838: */
839: static struct s_command *
1.15 deraadt 840: findlabel(char *name)
1.1 deraadt 841: {
1.10 mpech 842: struct labhash *lh;
843: u_char *p;
844: u_int h, c;
1.1 deraadt 845:
846: for (h = 0, p = (u_char *)name; (c = *p) != 0; p++)
847: h = (h << 5) + h + c;
848: for (lh = labels[h & LHMASK]; lh != NULL; lh = lh->lh_next) {
849: if (lh->lh_hash == h && strcmp(name, lh->lh_cmd->t) == 0) {
850: lh->lh_ref = 1;
851: return (lh->lh_cmd);
852: }
853: }
854: return (NULL);
855: }
856:
1.21 deraadt 857: /*
1.1 deraadt 858: * Warn about any unused labels. As a side effect, release the label hash
859: * table space.
860: */
861: static void
1.15 deraadt 862: uselabel(void)
1.1 deraadt 863: {
1.10 mpech 864: struct labhash *lh, *next;
865: int i;
1.1 deraadt 866:
867: for (i = 0; i < LHSZ; i++) {
868: for (lh = labels[i]; lh != NULL; lh = next) {
869: next = lh->lh_next;
870: if (!lh->lh_ref)
1.41 krw 871: warning("unused label '%s'",
1.1 deraadt 872: lh->lh_cmd->t);
873: free(lh);
874: }
875: }
876: }