2 * Copyright (c) 1992 Diomidis Spinellis.
3 * Copyright (c) 1992, 1993
4 * The Regents of the University of California. All rights reserved.
6 * This code is derived from software contributed to Berkeley by
7 * Diomidis Spinellis of Imperial College, University of London.
9 * Redistribution and use in source and binary forms, with or without
10 * modification, are permitted provided that the following conditions
12 * 1. Redistributions of source code must retain the above copyright
13 * notice, this list of conditions and the following disclaimer.
14 * 2. Redistributions in binary form must reproduce the above copyright
15 * notice, this list of conditions and the following disclaimer in the
16 * documentation and/or other materials provided with the distribution.
17 * 4. Neither the name of the University nor the names of its contributors
18 * may be used to endorse or promote products derived from this software
19 * without specific prior written permission.
21 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
22 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
23 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
24 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
25 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
26 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
27 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
28 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
29 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
30 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
34 #include <sys/cdefs.h>
35 __FBSDID("$FreeBSD$");
38 static const char sccsid[] = "@(#)compile.c 8.1 (Berkeley) 6/6/93";
41 #include <sys/types.h>
59 #define LHMASK (LHSZ - 1)
60 static struct labhash {
61 struct labhash *lh_next;
63 struct s_command *lh_cmd;
67 static char *compile_addr(char *, struct s_addr *);
68 static char *compile_ccl(char **, char *);
69 static char *compile_delimited(char *, char *, int);
70 static char *compile_flags(char *, struct s_subst *);
71 static regex_t *compile_re(char *, int);
72 static char *compile_subst(char *, struct s_subst *);
73 static char *compile_text(void);
74 static char *compile_tr(char *, struct s_tr **);
75 static struct s_command
76 **compile_stream(struct s_command **);
77 static char *duptoeol(char *, const char *);
78 static void enterlabel(struct s_command *);
79 static struct s_command
81 static void fixuplabel(struct s_command *, struct s_command *);
82 static void uselabel(void);
85 * Command specification. This is used to drive the command parser.
88 char code; /* Command code */
89 int naddr; /* Number of address args */
90 enum e_args args; /* Argument type */
93 static struct s_format cmd_fmts[] = {
125 /* The compiled program. */
126 struct s_command *prog;
129 * Compile the program into prog.
130 * Initialise appends.
135 *compile_stream(&prog) = NULL;
136 fixuplabel(prog, NULL);
140 else if ((appends = malloc(sizeof(struct s_appends) * appendnum)) ==
143 if ((match = malloc((maxnsub + 1) * sizeof(regmatch_t))) == NULL)
147 #define EATSPACE() do { \
149 while (*p && isspace((unsigned char)*p)) \
153 static struct s_command **
154 compile_stream(struct s_command **link)
157 static char lbuf[_POSIX2_LINE_MAX + 1]; /* To save stack */
158 struct s_command *cmd, *cmd2, *stack;
160 char re[_POSIX2_LINE_MAX + 1];
161 int naddr; /* Number of addresses */
165 if ((p = cu_fgets(lbuf, sizeof(lbuf), NULL)) == NULL) {
167 errx(1, "%lu: %s: unexpected EOF (pending }'s)",
172 semicolon: EATSPACE();
174 if (*p == '#' || *p == '\0')
176 else if (*p == ';') {
181 if ((*link = cmd = malloc(sizeof(struct s_command))) == NULL)
184 cmd->startline = cmd->nonsel = 0;
185 /* First parse the addresses */
188 /* Valid characters to start an address */
189 #define addrchar(c) (strchr("0123456789/\\$", (c)))
192 if ((cmd->a1 = malloc(sizeof(struct s_addr))) == NULL)
194 p = compile_addr(p, cmd->a1);
195 EATSPACE(); /* EXTENSION */
198 EATSPACE(); /* EXTENSION */
200 if ((cmd->a2 = malloc(sizeof(struct s_addr)))
203 p = compile_addr(p, cmd->a2);
208 cmd->a1 = cmd->a2 = 0;
210 nonsel: /* Now parse the command */
212 errx(1, "%lu: %s: command expected", linenum, fname);
214 for (fp = cmd_fmts; fp->code; fp++)
218 errx(1, "%lu: %s: invalid command code %c", linenum, fname, *p);
219 if (naddr > fp->naddr)
221 "%lu: %s: command %c expects up to %d address(es), found %d",
222 linenum, fname, *p, fp->naddr, naddr);
240 * Short-circuit command processing, since end of
241 * group is really just a noop.
245 errx(1, "%lu: %s: unexpected }", linenum, fname);
250 case EMPTY: /* d D g G h H l n N p P q x = \0 */
259 errx(1, "%lu: %s: extra characters at the end of %c command",
260 linenum, fname, cmd->code);
262 case TEXT: /* a c i */
267 "%lu: %s: command %c expects \\ followed by text", linenum, fname, cmd->code);
272 "%lu: %s: extra characters after \\ at the end of %c command",
273 linenum, fname, cmd->code);
274 cmd->t = compile_text();
276 case COMMENT: /* \0 # */
282 errx(1, "%lu: %s: filename expected", linenum, fname);
283 cmd->t = duptoeol(p, "w command");
286 else if ((cmd->u.fd = open(p,
287 O_WRONLY|O_APPEND|O_CREAT|O_TRUNC,
295 errx(1, "%lu: %s: filename expected", linenum, fname);
297 cmd->t = duptoeol(p, "read command");
299 case BRANCH: /* b t */
305 cmd->t = duptoeol(p, "branch");
310 cmd->t = duptoeol(p, "label");
312 errx(1, "%lu: %s: empty label", linenum, fname);
317 if (*p == '\0' || *p == '\\')
319 "%lu: %s: substitute pattern can not be delimited by newline or backslash",
321 if ((cmd->u.s = calloc(1, sizeof(struct s_subst))) == NULL)
323 p = compile_delimited(p, re, 0);
326 "%lu: %s: unterminated substitute pattern", linenum, fname);
328 /* Compile RE with no case sensitivity temporarily */
332 cmd->u.s->re = compile_re(re, 0);
334 p = compile_subst(p, cmd->u.s);
335 p = compile_flags(p, cmd->u.s);
337 /* Recompile RE with case sensitivity from "I" flag if any */
341 cmd->u.s->re = compile_re(re, cmd->u.s->icase);
351 p = compile_tr(p, &cmd->u.y);
360 "%lu: %s: extra text at the end of a transform command", linenum, fname);
367 * Get a delimited string. P points to the delimiter of the string; d points
368 * to a buffer area. Newline and delimiter escapes are processed; other
369 * escapes are ignored.
371 * Returns a pointer to the first character after the final delimiter or NULL
372 * in the case of a non-terminated string. The character array d is filled
373 * with the processed string.
376 compile_delimited(char *p, char *d, int is_tr)
384 errx(1, "%lu: %s: \\ can not be used as a string delimiter",
387 errx(1, "%lu: %s: newline can not be used as a string delimiter",
390 if (*p == '[' && *p != c) {
391 if ((d = compile_ccl(&p, d)) == NULL)
392 errx(1, "%lu: %s: unbalanced brackets ([])", linenum, fname);
394 } else if (*p == '\\' && p[1] == '[') {
396 } else if (*p == '\\' && p[1] == c)
398 else if (*p == '\\' && p[1] == 'n') {
402 } else if (*p == '\\' && p[1] == '\\') {
407 } else if (*p == c) {
417 /* compile_ccl: expand a POSIX character class */
419 compile_ccl(char **sp, char *t)
429 for (; *s && (*t = *s) != ']'; s++, t++)
430 if (*s == '[' && ((d = *(s+1)) == '.' || d == ':' || d == '=')) {
431 *++t = *++s, t++, s++;
432 for (c = *s; (*t = *s) != ']' || c != d; s++, t++)
433 if ((c = *s) == '\0')
436 return (*s == ']') ? *sp = ++s, ++t : NULL;
440 * Compiles the regular expression in RE and returns a pointer to the compiled
441 * regular expression.
442 * Cflags are passed to regcomp.
445 compile_re(char *re, int case_insensitive)
452 if (case_insensitive)
454 if ((rep = malloc(sizeof(regex_t))) == NULL)
456 if ((eval = regcomp(rep, re, flags)) != 0)
457 errx(1, "%lu: %s: RE error: %s",
458 linenum, fname, strregerror(eval, rep));
459 if (maxnsub < rep->re_nsub)
460 maxnsub = rep->re_nsub;
465 * Compile the substitution string of a regular expression and set res to
466 * point to a saved copy of it. Nsub is the number of parenthesized regular
470 compile_subst(char *p, struct s_subst *s)
472 static char lbuf[_POSIX2_LINE_MAX + 1];
475 char c, *text, *op, *sp;
476 int more = 1, sawesc = 0;
478 c = *p++; /* Terminator character */
483 s->linenum = linenum;
484 asize = 2 * _POSIX2_LINE_MAX + 1;
485 if ((text = malloc(asize)) == NULL)
489 op = sp = text + size;
491 if (*p == '\\' || sawesc) {
493 * If this is a continuation from the last
494 * buffer, we won't have a character to
504 * This escaped character is continued
505 * in the next part of the line. Note
506 * this fact, then cause the loop to
507 * exit w/ normal EOL case and reenter
508 * above with the new buffer.
513 } else if (strchr("123456789", *p) != NULL) {
517 ref > s->re->re_nsub)
518 errx(1, "%lu: %s: \\%c not defined in the RE",
520 if (s->maxbref < ref)
522 } else if (*p == '&' || *p == '\\')
524 } else if (*p == c) {
525 if (*++p == '\0' && more) {
526 if (cu_fgets(lbuf, sizeof(lbuf), &more))
531 if ((s->new = realloc(text, size)) == NULL)
534 } else if (*p == '\n') {
536 "%lu: %s: unescaped newline inside substitute pattern", linenum, fname);
542 if (asize - size < _POSIX2_LINE_MAX + 1) {
544 if ((text = realloc(text, asize)) == NULL)
547 } while (cu_fgets(p = lbuf, sizeof(lbuf), &more));
548 errx(1, "%lu: %s: unterminated substitute in regular expression",
554 * Compile the flags of the s command
557 compile_flags(char *p, struct s_subst *s)
559 int gn; /* True if we have seen g or n */
561 char wfile[_POSIX2_LINE_MAX + 1], *q, *eq;
563 s->n = 1; /* Default */
569 EATSPACE(); /* EXTENSION */
574 "%lu: %s: more than one number or 'g' in substitute flags", linenum, fname);
589 case '1': case '2': case '3':
590 case '4': case '5': case '6':
591 case '7': case '8': case '9':
594 "%lu: %s: more than one number or 'g' in substitute flags", linenum, fname);
597 nval = strtol(p, &p, 10);
598 if (errno == ERANGE || nval > INT_MAX)
600 "%lu: %s: overflow in the 'N' substitute flag", linenum, fname);
606 #ifdef HISTORIC_PRACTICE
608 warnx("%lu: %s: space missing before w wfile", linenum, fname);
614 eq = wfile + sizeof(wfile) - 1;
619 err(1, "wfile too long");
624 errx(1, "%lu: %s: no wfile specified", linenum, fname);
625 s->wfile = strdup(wfile);
626 if (!aflag && (s->wfd = open(wfile,
627 O_WRONLY|O_APPEND|O_CREAT|O_TRUNC,
632 errx(1, "%lu: %s: bad flag in substitute command: '%c'",
641 * Compile a translation set of strings into a lookup table.
644 compile_tr(char *p, struct s_tr **py)
649 char old[_POSIX2_LINE_MAX + 1];
650 char new[_POSIX2_LINE_MAX + 1];
651 size_t oclen, oldlen, nclen, newlen;
652 mbstate_t mbs1, mbs2;
654 if ((*py = y = malloc(sizeof(*y))) == NULL)
659 if (*p == '\0' || *p == '\\')
661 "%lu: %s: transform pattern can not be delimited by newline or backslash",
663 p = compile_delimited(p, old, 1);
665 errx(1, "%lu: %s: unterminated transform source string",
667 p = compile_delimited(p - 1, new, 1);
669 errx(1, "%lu: %s: unterminated transform target string",
673 oldlen = mbsrtowcs(NULL, &op, 0, NULL);
674 if (oldlen == (size_t)-1)
677 newlen = mbsrtowcs(NULL, &np, 0, NULL);
678 if (newlen == (size_t)-1)
680 if (newlen != oldlen)
681 errx(1, "%lu: %s: transform strings are not the same length",
683 if (MB_CUR_MAX == 1) {
685 * The single-byte encoding case is easy: generate a
688 for (i = 0; i <= UCHAR_MAX; i++)
689 y->bytetab[i] = (char)i;
690 for (; *op; op++, np++)
691 y->bytetab[(u_char)*op] = *np;
694 * Multi-byte encoding case: generate a lookup table as
695 * above, but only for single-byte characters. The first
696 * bytes of multi-byte characters have their lookup table
697 * entries set to 0, which causes do_tr() to search through
698 * an auxiliary vector of multi-byte mappings.
700 memset(&mbs1, 0, sizeof(mbs1));
701 memset(&mbs2, 0, sizeof(mbs2));
702 for (i = 0; i <= UCHAR_MAX; i++)
703 y->bytetab[i] = (btowc(i) != WEOF) ? i : 0;
704 while (*op != '\0') {
705 oclen = mbrlen(op, MB_LEN_MAX, &mbs1);
706 if (oclen == (size_t)-1 || oclen == (size_t)-2)
707 errc(1, EILSEQ, NULL);
708 nclen = mbrlen(np, MB_LEN_MAX, &mbs2);
709 if (nclen == (size_t)-1 || nclen == (size_t)-2)
710 errc(1, EILSEQ, NULL);
711 if (oclen == 1 && nclen == 1)
712 y->bytetab[(u_char)*op] = *np;
714 y->bytetab[(u_char)*op] = 0;
715 y->multis = realloc(y->multis,
716 (y->nmultis + 1) * sizeof(*y->multis));
717 if (y->multis == NULL)
720 y->multis[i].fromlen = oclen;
721 memcpy(y->multis[i].from, op, oclen);
722 y->multis[i].tolen = nclen;
723 memcpy(y->multis[i].to, np, nclen);
733 * Compile the text following an a or i command.
738 int asize, esc_nl, size;
739 char *text, *p, *op, *s;
740 char lbuf[_POSIX2_LINE_MAX + 1];
742 asize = 2 * _POSIX2_LINE_MAX + 1;
743 if ((text = malloc(asize)) == NULL)
746 while (cu_fgets(lbuf, sizeof(lbuf), NULL)) {
747 op = s = text + size;
750 for (esc_nl = 0; *p != '\0'; p++) {
751 if (*p == '\\' && p[1] != '\0' && *++p == '\n')
760 if (asize - size < _POSIX2_LINE_MAX + 1) {
762 if ((text = realloc(text, asize)) == NULL)
767 if ((p = realloc(text, size + 1)) == NULL)
773 * Get an address and return a pointer to the first character after
774 * it. Fill the structure pointed to according to the address.
777 compile_addr(char *p, struct s_addr *a)
779 char *end, re[_POSIX2_LINE_MAX + 1];
786 case '\\': /* Context address */
789 case '/': /* Context address */
790 p = compile_delimited(p, re, 0);
792 errx(1, "%lu: %s: unterminated regular expression", linenum, fname);
793 /* Check for case insensitive regexp flag */
801 a->u.r = compile_re(re, icase);
805 case '$': /* Last line */
809 case '+': /* Relative line number */
810 a->type = AT_RELLINE;
814 case '0': case '1': case '2': case '3': case '4':
815 case '5': case '6': case '7': case '8': case '9':
818 a->u.l = strtol(p, &end, 10);
821 errx(1, "%lu: %s: expected context address", linenum, fname);
828 * Return a copy of all the characters up to \n or \0.
831 duptoeol(char *s, const char *ctype)
838 for (start = s; *s != '\0' && *s != '\n'; ++s)
839 ws = isspace((unsigned char)*s);
842 warnx("%lu: %s: whitespace after %s", linenum, fname, ctype);
844 if ((p = malloc(len)) == NULL)
846 return (memmove(p, start, len));
850 * Convert goto label names to addresses, and count a and r commands, in
851 * the given subset of the script. Free the memory used by labels in b
852 * and t commands (but not by :).
854 * TODO: Remove } nodes
857 fixuplabel(struct s_command *cp, struct s_command *end)
860 for (; cp != end; cp = cp->next)
868 /* Resolve branch target. */
873 if ((cp->u.c = findlabel(cp->t)) == NULL)
874 errx(1, "%lu: %s: undefined label '%s'", linenum, fname, cp->t);
878 /* Do interior commands. */
879 fixuplabel(cp->u.c, cp->next);
885 * Associate the given command label for later lookup.
888 enterlabel(struct s_command *cp)
890 struct labhash **lhp, *lh;
894 for (h = 0, p = (u_char *)cp->t; (c = *p) != 0; p++)
895 h = (h << 5) + h + c;
896 lhp = &labels[h & LHMASK];
897 for (lh = *lhp; lh != NULL; lh = lh->lh_next)
898 if (lh->lh_hash == h && strcmp(cp->t, lh->lh_cmd->t) == 0)
899 errx(1, "%lu: %s: duplicate label '%s'", linenum, fname, cp->t);
900 if ((lh = malloc(sizeof *lh)) == NULL)
910 * Find the label contained in the command l in the command linked
911 * list cp. L is excluded from the search. Return NULL if not found.
913 static struct s_command *
914 findlabel(char *name)
920 for (h = 0, p = (u_char *)name; (c = *p) != 0; p++)
921 h = (h << 5) + h + c;
922 for (lh = labels[h & LHMASK]; lh != NULL; lh = lh->lh_next) {
923 if (lh->lh_hash == h && strcmp(name, lh->lh_cmd->t) == 0) {
932 * Warn about any unused labels. As a side effect, release the label hash
938 struct labhash *lh, *next;
941 for (i = 0; i < LHSZ; i++) {
942 for (lh = labels[i]; lh != NULL; lh = next) {
945 warnx("%lu: %s: unused label '%s'",
946 linenum, fname, lh->lh_cmd->t);