Annotation of src/usr.bin/sed/compile.c, Revision 1.45
1.45 ! martijn 1: /* $OpenBSD: compile.c,v 1.44 2017/12/11 13:25:57 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 *);
1.44 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");
1.1 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] == '\\') {
1.44 martijn 379: *d++ = *p++;
1.22 tedu 380: } else if (*p == c) {
1.1 deraadt 381: *d = '\0';
382: return (p + 1);
383: }
384: *d++ = *p++;
385: }
386: return (NULL);
387: }
388:
389:
390: /* compile_ccl: expand a POSIX character class */
391: static char *
1.15 deraadt 392: compile_ccl(char **sp, char *t)
1.1 deraadt 393: {
394: int c, d;
395: char *s = *sp;
396:
397: *t++ = *s++;
398: if (*s == '^')
399: *t++ = *s++;
400: if (*s == ']')
401: *t++ = *s++;
402: for (; *s && (*t = *s) != ']'; s++, t++)
403: if (*s == '[' && ((d = *(s+1)) == '.' || d == ':' || d == '=')) {
404: *++t = *++s, t++, s++;
405: for (c = *s; (*t = *s) != ']' || c != d; s++, t++)
406: if ((c = *s) == '\0')
407: return NULL;
1.22 tedu 408: } else if (*s == '\\' && s[1] == 'n') {
409: *t = '\n';
410: s++;
411: }
412: if (*s == ']') {
413: *sp = ++s;
414: return (++t);
415: } else {
416: return (NULL);
417: }
1.1 deraadt 418: }
419:
420: /*
421: * Get a regular expression. P points to the delimiter of the regular
422: * expression; repp points to the address of a regexp pointer. Newline
423: * and delimiter escapes are processed; other escapes are ignored.
424: * Returns a pointer to the first character after the final delimiter
425: * or NULL in the case of a non terminated regular expression. The regexp
426: * pointer is set to the compiled regular expression.
427: * Cflags are passed to regcomp.
428: */
429: static char *
1.15 deraadt 430: compile_re(char *p, regex_t **repp)
1.1 deraadt 431: {
432: int eval;
1.25 millert 433: char *re;
1.1 deraadt 434:
1.25 millert 435: re = xmalloc(strlen(p) + 1); /* strlen(re) <= strlen(p) */
1.44 martijn 436: p = compile_delimited(p, re);
1.1 deraadt 437: if (p && strlen(re) == 0) {
438: *repp = NULL;
1.25 millert 439: free(re);
1.1 deraadt 440: return (p);
441: }
442: *repp = xmalloc(sizeof(regex_t));
1.29 djm 443: if (p && (eval = regcomp(*repp, re, Eflag ? REG_EXTENDED : 0)) != 0)
1.39 mmcc 444: error(COMPILE, "RE error: %s", strregerror(eval, *repp));
1.1 deraadt 445: if (maxnsub < (*repp)->re_nsub)
446: maxnsub = (*repp)->re_nsub;
1.25 millert 447: free(re);
1.1 deraadt 448: return (p);
449: }
450:
451: /*
452: * Compile the substitution string of a regular expression and set res to
453: * point to a saved copy of it. Nsub is the number of parenthesized regular
454: * expressions.
455: */
456: static char *
1.15 deraadt 457: compile_subst(char *p, struct s_subst *s)
1.1 deraadt 458: {
1.25 millert 459: static char *lbuf;
1.28 millert 460: static size_t bufsize;
1.1 deraadt 461: int asize, ref, size;
462: char c, *text, *op, *sp;
1.19 otto 463: int sawesc = 0;
1.1 deraadt 464:
465: c = *p++; /* Terminator character */
466: if (c == '\0')
467: return (NULL);
468:
469: s->maxbref = 0;
470: s->linenum = linenum;
1.28 millert 471: text = NULL;
472: asize = size = 0;
1.1 deraadt 473: do {
1.28 millert 474: size_t len = ROUNDLEN(strlen(p) + 1);
475: if (asize - size < len) {
1.25 millert 476: do {
1.28 millert 477: asize += len;
478: } while (asize - size < len);
1.25 millert 479: text = xrealloc(text, asize);
480: }
1.1 deraadt 481: op = sp = text + size;
482: for (; *p; p++) {
1.19 otto 483: if (*p == '\\' || sawesc) {
484: /*
485: * If this is a continuation from the last
486: * buffer, we won't have a character to
487: * skip over.
488: */
489: if (sawesc)
490: sawesc = 0;
491: else
492: p++;
493:
494: if (*p == '\0') {
495: /*
496: * This escaped character is continued
497: * in the next part of the line. Note
498: * this fact, then cause the loop to
499: * exit w/ normal EOL case and reenter
500: * above with the new buffer.
501: */
502: sawesc = 1;
503: p--;
504: continue;
505: } else if (strchr("123456789", *p) != NULL) {
1.1 deraadt 506: *sp++ = '\\';
507: ref = *p - '0';
508: if (s->re != NULL &&
509: ref > s->re->re_nsub)
1.39 mmcc 510: error(COMPILE,
1.1 deraadt 511: "\\%c not defined in the RE", *p);
512: if (s->maxbref < ref)
513: s->maxbref = ref;
514: } else if (*p == '&' || *p == '\\')
515: *sp++ = '\\';
516: } else if (*p == c) {
517: p++;
518: *sp++ = '\0';
519: size += sp - op;
520: s->new = xrealloc(text, size);
521: return (p);
522: } else if (*p == '\n') {
1.39 mmcc 523: error(COMPILE,
1.1 deraadt 524: "unescaped newline inside substitute pattern");
525: }
526: *sp++ = *p;
527: }
528: size += sp - op;
1.28 millert 529: } while ((p = cu_fgets(&lbuf, &bufsize)));
1.39 mmcc 530: error(COMPILE, "unterminated substitute in regular expression");
1.1 deraadt 531: }
532:
533: /*
534: * Compile the flags of the s command
535: */
536: static char *
1.15 deraadt 537: compile_flags(char *p, struct s_subst *s)
1.1 deraadt 538: {
539: int gn; /* True if we have seen g or n */
1.17 otto 540: long l;
1.1 deraadt 541:
542: s->n = 1; /* Default */
543: s->p = 0;
544: s->wfile = NULL;
545: s->wfd = -1;
546: for (gn = 0;;) {
547: EATSPACE(); /* EXTENSION */
548: switch (*p) {
549: case 'g':
550: if (gn)
1.39 mmcc 551: error(COMPILE, "more than one number or 'g' in"
1.22 tedu 552: " substitute flags");
1.1 deraadt 553: gn = 1;
554: s->n = 0;
555: break;
556: case '\0':
557: case '\n':
558: case ';':
559: return (p);
560: case 'p':
561: s->p = 1;
562: break;
563: case '1': case '2': case '3':
564: case '4': case '5': case '6':
565: case '7': case '8': case '9':
566: if (gn)
1.39 mmcc 567: error(COMPILE, "more than one number or 'g' in"
1.22 tedu 568: " substitute flags");
1.1 deraadt 569: gn = 1;
1.17 otto 570: l = strtol(p, &p, 10);
571: if (l <= 0 || l >= INT_MAX)
1.39 mmcc 572: error(COMPILE,
1.17 otto 573: "number in substitute flags out of range");
574: s->n = (int)l;
575: continue;
1.1 deraadt 576: case 'w':
577: p++;
578: EATSPACE();
1.43 martijn 579: if (*p == '\0')
580: error(COMPILE, "filename expected");
581: s->wfile = duptoeol(p, "s command w flag", NULL);
582: *p = '\0';
1.42 martijn 583: if (aflag)
584: pledge_wpath = 1;
1.43 martijn 585: else if ((s->wfd = open(s->wfile,
1.1 deraadt 586: O_WRONLY|O_APPEND|O_CREAT|O_TRUNC,
587: DEFFILEMODE)) == -1)
1.43 martijn 588: error(FATAL, "%s: %s", s->wfile, strerror(errno));
1.1 deraadt 589: return (p);
590: default:
1.39 mmcc 591: error(COMPILE,
1.1 deraadt 592: "bad flag in substitute command: '%c'", *p);
593: break;
594: }
595: p++;
596: }
597: }
598:
599: /*
600: * Compile a translation set of strings into a lookup table.
601: */
602: static char *
1.44 martijn 603: compile_tr(char *old, char **transtab)
1.1 deraadt 604: {
605: int i;
1.45 ! martijn 606: char delimiter, check[UCHAR_MAX + 1];
1.44 martijn 607: char *new, *end;
608:
609: memset(check, 0, sizeof(check));
610: delimiter = *old;
611: if (delimiter == '\\')
612: error(COMPILE, "\\ can not be used as a string delimiter");
613: else if (delimiter == '\n' || delimiter == '\0')
614: error(COMPILE, "newline can not be used as a string delimiter");
615:
616: new = old++;
617: do {
618: if ((new = strchr(new + 1, delimiter)) == NULL)
619: error(COMPILE, "unterminated transform source string");
620: } while (*(new - 1) == '\\' && *(new -2) != '\\');
621: *new = '\0';
622: end = new++;
623: do {
624: if ((end = strchr(end + 1, delimiter)) == NULL)
625: error(COMPILE, "unterminated transform target string");
626: } while (*(end -1) == '\\' && *(end -2) != '\\');
627: *end = '\0';
628:
629: /* We assume characters are 8 bits */
630: *transtab = xmalloc(UCHAR_MAX + 1);
631: for (i = 0; i <= UCHAR_MAX; i++)
632: (*transtab)[i] = (char)i;
1.1 deraadt 633:
1.44 martijn 634: while (*old != '\0' && *new != '\0') {
635: if (*old == '\\') {
636: old++;
637: if (*old == 'n')
638: *old = '\n';
639: else if (*old != delimiter && *old != '\\')
640: error(COMPILE, "Unexpected character after "
641: "backslash");
642:
643: }
644: if (*new == '\\') {
645: new++;
646: if (*new == 'n')
647: *new = '\n';
648: else if (*new != delimiter && *new != '\\')
649: error(COMPILE, "Unexpected character after "
650: "backslash");
651: }
652: if (check[*old] == 1)
653: error(COMPILE, "Repeated character in source string");
654: check[*old] = 1;
1.45 ! martijn 655: (*transtab)[(u_char) *old++] = *new++;
1.1 deraadt 656: }
1.44 martijn 657: if (*old != '\0' || *new != '\0')
1.39 mmcc 658: error(COMPILE, "transform strings are not the same length");
1.44 martijn 659: return end + 1;
1.1 deraadt 660: }
661:
662: /*
1.7 deraadt 663: * Compile the text following an a, c, or i command.
1.1 deraadt 664: */
665: static char *
1.15 deraadt 666: compile_text(void)
1.1 deraadt 667: {
1.4 deraadt 668: int asize, esc_nl, size;
1.25 millert 669: char *lbuf, *text, *p, *op, *s;
1.28 millert 670: size_t bufsize;
1.1 deraadt 671:
1.28 millert 672: lbuf = text = NULL;
673: asize = size = 0;
674: while ((p = cu_fgets(&lbuf, &bufsize))) {
675: size_t len = ROUNDLEN(strlen(p) + 1);
676: if (asize - size < len) {
1.25 millert 677: do {
1.28 millert 678: asize += len;
679: } while (asize - size < len);
1.25 millert 680: text = xrealloc(text, asize);
681: }
1.1 deraadt 682: op = s = text + size;
1.4 deraadt 683: for (esc_nl = 0; *p != '\0'; p++) {
684: if (*p == '\\' && p[1] != '\0' && *++p == '\n')
685: esc_nl = 1;
1.1 deraadt 686: *s++ = *p;
687: }
688: size += s - op;
1.4 deraadt 689: if (!esc_nl) {
1.1 deraadt 690: *s = '\0';
691: break;
692: }
693: }
1.25 millert 694: free(lbuf);
1.34 millert 695: text = xrealloc(text, size + 1);
1.8 brian 696: text[size] = '\0';
1.34 millert 697: return (text);
1.1 deraadt 698: }
699:
700: /*
701: * Get an address and return a pointer to the first character after
702: * it. Fill the structure pointed to according to the address.
703: */
704: static char *
1.15 deraadt 705: compile_addr(char *p, struct s_addr *a)
1.1 deraadt 706: {
707: char *end;
708:
709: switch (*p) {
710: case '\\': /* Context address */
711: ++p;
712: /* FALLTHROUGH */
713: case '/': /* Context address */
714: p = compile_re(p, &a->u.r);
715: if (p == NULL)
1.39 mmcc 716: error(COMPILE, "unterminated regular expression");
1.1 deraadt 717: a->type = AT_RE;
718: return (p);
719:
720: case '$': /* Last line */
721: a->type = AT_LAST;
722: return (p + 1);
723: /* Line number */
1.21 deraadt 724: case '0': case '1': case '2': case '3': case '4':
1.1 deraadt 725: case '5': case '6': case '7': case '8': case '9':
726: a->type = AT_LINE;
1.3 millert 727: a->u.l = strtoul(p, &end, 10);
1.1 deraadt 728: return (end);
729: default:
1.39 mmcc 730: error(COMPILE, "expected context address");
1.1 deraadt 731: return (NULL);
732: }
733: }
734:
735: /*
736: * duptoeol --
737: * Return a copy of all the characters up to \n or \0.
738: */
739: static char *
1.16 deraadt 740: duptoeol(char *s, char *ctype, char **semi)
1.1 deraadt 741: {
742: size_t len;
743: int ws;
744: char *start;
745:
746: ws = 0;
1.16 deraadt 747: if (semi) {
748: for (start = s; *s != '\0' && *s != '\n' && *s != ';'; ++s)
1.35 deraadt 749: ws = isspace((unsigned char)*s);
1.16 deraadt 750: } else {
751: for (start = s; *s != '\0' && *s != '\n'; ++s)
1.35 deraadt 752: ws = isspace((unsigned char)*s);
1.16 deraadt 753: *s = '\0';
754: }
1.1 deraadt 755: if (ws)
1.41 krw 756: warning("whitespace after %s", ctype);
1.1 deraadt 757: len = s - start + 1;
1.16 deraadt 758: if (semi)
759: *semi = s;
760: s = xmalloc(len);
761: strlcpy(s, start, len);
762: return (s);
1.1 deraadt 763: }
764:
765: /*
766: * Convert goto label names to addresses, and count a and r commands, in
767: * the given subset of the script. Free the memory used by labels in b
768: * and t commands (but not by :).
769: *
770: * TODO: Remove } nodes
771: */
772: static void
1.15 deraadt 773: fixuplabel(struct s_command *cp, struct s_command *end)
1.1 deraadt 774: {
775:
776: for (; cp != end; cp = cp->next)
777: switch (cp->code) {
778: case 'a':
779: case 'r':
780: appendnum++;
781: break;
782: case 'b':
783: case 't':
784: /* Resolve branch target. */
785: if (cp->t == NULL) {
786: cp->u.c = NULL;
787: break;
788: }
789: if ((cp->u.c = findlabel(cp->t)) == NULL)
1.40 jca 790: error(COMPILE, "undefined label '%s'", cp->t);
1.1 deraadt 791: free(cp->t);
792: break;
793: case '{':
794: /* Do interior commands. */
795: fixuplabel(cp->u.c, cp->next);
796: break;
797: }
798: }
799:
800: /*
801: * Associate the given command label for later lookup.
802: */
803: static void
1.15 deraadt 804: enterlabel(struct s_command *cp)
1.1 deraadt 805: {
1.10 mpech 806: struct labhash **lhp, *lh;
807: u_char *p;
808: u_int h, c;
1.1 deraadt 809:
810: for (h = 0, p = (u_char *)cp->t; (c = *p) != 0; p++)
811: h = (h << 5) + h + c;
812: lhp = &labels[h & LHMASK];
813: for (lh = *lhp; lh != NULL; lh = lh->lh_next)
814: if (lh->lh_hash == h && strcmp(cp->t, lh->lh_cmd->t) == 0)
1.40 jca 815: error(COMPILE, "duplicate label '%s'", cp->t);
1.1 deraadt 816: lh = xmalloc(sizeof *lh);
817: lh->lh_next = *lhp;
818: lh->lh_hash = h;
819: lh->lh_cmd = cp;
820: lh->lh_ref = 0;
821: *lhp = lh;
822: }
823:
824: /*
825: * Find the label contained in the command l in the command linked
826: * list cp. L is excluded from the search. Return NULL if not found.
827: */
828: static struct s_command *
1.15 deraadt 829: findlabel(char *name)
1.1 deraadt 830: {
1.10 mpech 831: struct labhash *lh;
832: u_char *p;
833: u_int h, c;
1.1 deraadt 834:
835: for (h = 0, p = (u_char *)name; (c = *p) != 0; p++)
836: h = (h << 5) + h + c;
837: for (lh = labels[h & LHMASK]; lh != NULL; lh = lh->lh_next) {
838: if (lh->lh_hash == h && strcmp(name, lh->lh_cmd->t) == 0) {
839: lh->lh_ref = 1;
840: return (lh->lh_cmd);
841: }
842: }
843: return (NULL);
844: }
845:
1.21 deraadt 846: /*
1.1 deraadt 847: * Warn about any unused labels. As a side effect, release the label hash
848: * table space.
849: */
850: static void
1.15 deraadt 851: uselabel(void)
1.1 deraadt 852: {
1.10 mpech 853: struct labhash *lh, *next;
854: int i;
1.1 deraadt 855:
856: for (i = 0; i < LHSZ; i++) {
857: for (lh = labels[i]; lh != NULL; lh = next) {
858: next = lh->lh_next;
859: if (!lh->lh_ref)
1.41 krw 860: warning("unused label '%s'",
1.1 deraadt 861: lh->lh_cmd->t);
862: free(lh);
863: }
864: }
865: }