kernel - Cleanup, add assertions in the vnode freeing path
[dragonfly.git] / lib / libcompat / regexp / regexp.c
blob290d009895e9b8d587014c0123dfc061d33fb909
1 /* $DragonFly: src/lib/libcompat/regexp/regexp.c,v 1.3 2008/09/30 16:57:04 swildner Exp $ */
2 /*
3 * regcomp and regexec -- regsub and regerror are elsewhere
5 * Copyright (c) 1986 by University of Toronto.
6 * Written by Henry Spencer. Not derived from licensed software.
8 * Permission is granted to anyone to use this software for any
9 * purpose on any computer system, and to redistribute it freely,
10 * subject to the following restrictions:
12 * 1. The author is not responsible for the consequences of use of
13 * this software, no matter how awful, even if they arise
14 * from defects in it.
16 * 2. The origin of this software must not be misrepresented, either
17 * by explicit claim or by omission.
19 * 3. Altered versions must be plainly marked as such, and must not
20 * be misrepresented as being the original software.
21 *** THIS IS AN ALTERED VERSION. It was altered by John Gilmore,
22 *** hoptoad!gnu, on 27 Dec 1986, to add \n as an alternative to |
23 *** to assist in implementing egrep.
24 *** THIS IS AN ALTERED VERSION. It was altered by John Gilmore,
25 *** hoptoad!gnu, on 27 Dec 1986, to add \< and \> for word-matching
26 *** as in BSD grep and ex.
27 *** THIS IS AN ALTERED VERSION. It was altered by John Gilmore,
28 *** hoptoad!gnu, on 28 Dec 1986, to optimize characters quoted with \.
29 *** THIS IS AN ALTERED VERSION. It was altered by James A. Woods,
30 *** ames!jaw, on 19 June 1987, to quash a regcomp() redundancy.
32 * Beware that some of this code is subtly aware of the way operator
33 * precedence is structured in regular expressions. Serious changes in
34 * regular-expression syntax might require a total rethink.
36 #include <limits.h>
37 #include <regexp.h>
38 #include <stdio.h>
39 #include <ctype.h>
40 #include <stdlib.h>
41 #include <string.h>
42 #include "collate.h"
43 #include "regmagic.h"
46 * The "internal use only" fields in regexp.h are present to pass info from
47 * compile to execute that permits the execute phase to run lots faster on
48 * simple cases. They are:
50 * regstart char that must begin a match; '\0' if none obvious
51 * reganch is the match anchored (at beginning-of-line only)?
52 * regmust string (pointer into program) that match must include, or NULL
53 * regmlen length of regmust string
55 * Regstart and reganch permit very fast decisions on suitable starting points
56 * for a match, cutting down the work a lot. Regmust permits fast rejection
57 * of lines that cannot possibly match. The regmust tests are costly enough
58 * that regcomp() supplies a regmust only if the r.e. contains something
59 * potentially expensive (at present, the only such thing detected is * or +
60 * at the start of the r.e., which can involve a lot of backup). Regmlen is
61 * supplied because the test in regexec() needs it and regcomp() is computing
62 * it anyway.
66 * Structure for regexp "program". This is essentially a linear encoding
67 * of a nondeterministic finite-state machine (aka syntax charts or
68 * "railroad normal form" in parsing technology). Each node is an opcode
69 * plus a "next" pointer, possibly plus an operand. "Next" pointers of
70 * all nodes except BRANCH implement concatenation; a "next" pointer with
71 * a BRANCH on both ends of it is connecting two alternatives. (Here we
72 * have one of the subtle syntax dependencies: an individual BRANCH (as
73 * opposed to a collection of them) is never concatenated with anything
74 * because of operator precedence.) The operand of some types of node is
75 * a literal string; for others, it is a node leading into a sub-FSM. In
76 * particular, the operand of a BRANCH node is the first node of the branch.
77 * (NB this is *not* a tree structure: the tail of the branch connects
78 * to the thing following the set of BRANCHes.) The opcodes are:
81 /* definition number opnd? meaning */
82 #define END 0 /* no End of program. */
83 #define BOL 1 /* no Match "" at beginning of line. */
84 #define EOL 2 /* no Match "" at end of line. */
85 #define ANY 3 /* no Match any one character. */
86 #define ANYOF 4 /* str Match any character in this string. */
87 #define ANYBUT 5 /* str Match any character not in this string. */
88 #define BRANCH 6 /* node Match this alternative, or the next... */
89 #define BACK 7 /* no Match "", "next" ptr points backward. */
90 #define EXACTLY 8 /* str Match this string. */
91 #define NOTHING 9 /* no Match empty string. */
92 #define STAR 10 /* node Match this (simple) thing 0 or more times. */
93 #define PLUS 11 /* node Match this (simple) thing 1 or more times. */
94 #define WORDA 12 /* no Match "" at wordchar, where prev is nonword */
95 #define WORDZ 13 /* no Match "" at nonwordchar, where prev is word */
96 #define OPEN 20 /* no Mark this point in input as start of #n. */
97 /* OPEN+1 is number 1, etc. */
98 #define CLOSE 30 /* no Analogous to OPEN. */
101 * Opcode notes:
103 * BRANCH The set of branches constituting a single choice are hooked
104 * together with their "next" pointers, since precedence prevents
105 * anything being concatenated to any individual branch. The
106 * "next" pointer of the last BRANCH in a choice points to the
107 * thing following the whole choice. This is also where the
108 * final "next" pointer of each individual branch points; each
109 * branch starts with the operand node of a BRANCH node.
111 * BACK Normal "next" pointers all implicitly point forward; BACK
112 * exists to make loop structures possible.
114 * STAR,PLUS '?', and complex '*' and '+', are implemented as circular
115 * BRANCH structures using BACK. Simple cases (one character
116 * per match) are implemented with STAR and PLUS for speed
117 * and to minimize recursive plunges.
119 * OPEN,CLOSE ...are numbered at compile time.
123 * A node is one char of opcode followed by two chars of "next" pointer.
124 * "Next" pointers are stored as two 8-bit pieces, high order first. The
125 * value is a positive offset from the opcode of the node containing it.
126 * An operand, if any, simply follows the node. (Note that much of the
127 * code generation knows about this implicit relationship.)
129 * Using two bytes for the "next" pointer is vast overkill for most things,
130 * but allows patterns to get big without disasters.
132 #define OP(p) (*(p))
133 #define NEXT(p) (((*((p)+1)&0377)<<8) + (*((p)+2)&0377))
134 #define OPERAND(p) ((p) + 3)
137 * See regmagic.h for one further detail of program structure.
142 * Utility definitions.
144 #ifndef CHARBITS
145 #define UCHARAT(p) ((int)*(unsigned char *)(p))
146 #else
147 #define UCHARAT(p) ((int)*(p)&CHARBITS)
148 #endif
150 #define FAIL(m) { regerror(m); return(NULL); }
151 #define ISMULT(c) ((c) == '*' || (c) == '+' || (c) == '?')
154 * Flags to be passed up and down.
156 #define HASWIDTH 01 /* Known never to match null string. */
157 #define SIMPLE 02 /* Simple enough to be STAR/PLUS operand. */
158 #define SPSTART 04 /* Starts with * or +. */
159 #define WORST 0 /* Worst case. */
162 * Global work variables for regcomp().
164 static char *regparse; /* Input-scan pointer. */
165 static int regnpar; /* () count. */
166 static char regdummy;
167 static char *regcode; /* Code-emit pointer; &regdummy = don't. */
168 static long regsize; /* Code size. */
171 * Forward declarations for regcomp()'s friends.
173 #ifndef STATIC
174 #define STATIC static
175 #endif
176 STATIC char *reg(int, int *);
177 STATIC char *regbranch(int *);
178 STATIC char *regpiece(int *);
179 STATIC char *regatom(int *);
180 STATIC char *regnode(char);
181 STATIC char *regnext(char *);
182 STATIC void regc(char);
183 STATIC void reginsert(char, char *);
184 STATIC void regtail(char *, char *);
185 STATIC void regoptail(char *, char *);
186 #ifdef STRCSPN
187 STATIC int strcspn(char *, char *);
188 #endif
191 - regcomp - compile a regular expression into internal code
193 * We can't allocate space until we know how big the compiled form will be,
194 * but we can't compile it (and thus know how big it is) until we've got a
195 * place to put the code. So we cheat: we compile it twice, once with code
196 * generation turned off and size counting turned on, and once "for real".
197 * This also means that we don't allocate space until we are sure that the
198 * thing really will compile successfully, and we never have to move the
199 * code and thus invalidate pointers into it. (Note that it has to be in
200 * one piece because free() must be able to free it all.)
202 * Beware that the optimization-preparation code in here knows about some
203 * of the structure of the compiled regexp.
205 regexp *
206 regcomp(const char *exp)
208 regexp *r;
209 char *scan;
210 char *longest;
211 int len;
212 int flags;
214 if (exp == NULL)
215 FAIL("NULL argument");
217 /* First pass: determine size, legality. */
218 #ifdef notdef
219 if (exp[0] == '.' && exp[1] == '*') exp += 2; /* aid grep */
220 #endif
221 regparse = (char *)exp;
222 regnpar = 1;
223 regsize = 0L;
224 regcode = &regdummy;
225 regc(MAGIC);
226 if (reg(0, &flags) == NULL)
227 return(NULL);
229 /* Small enough for pointer-storage convention? */
230 if (regsize >= 32767L) /* Probably could be 65535L. */
231 FAIL("regexp too big");
233 /* Allocate space. */
234 r = (regexp *)malloc(sizeof(regexp) + (unsigned)regsize);
235 if (r == NULL)
236 FAIL("out of space");
238 /* Second pass: emit code. */
239 regparse = (char *)exp;
240 regnpar = 1;
241 regcode = r->program;
242 regc(MAGIC);
243 if (reg(0, &flags) == NULL)
244 return(NULL);
246 /* Dig out information for optimizations. */
247 r->regstart = '\0'; /* Worst-case defaults. */
248 r->reganch = 0;
249 r->regmust = NULL;
250 r->regmlen = 0;
251 scan = r->program+1; /* First BRANCH. */
252 if (OP(regnext(scan)) == END) { /* Only one top-level choice. */
253 scan = OPERAND(scan);
255 /* Starting-point info. */
256 if (OP(scan) == EXACTLY)
257 r->regstart = *OPERAND(scan);
258 else if (OP(scan) == BOL)
259 r->reganch++;
262 * If there's something expensive in the r.e., find the
263 * longest literal string that must appear and make it the
264 * regmust. Resolve ties in favor of later strings, since
265 * the regstart check works with the beginning of the r.e.
266 * and avoiding duplication strengthens checking. Not a
267 * strong reason, but sufficient in the absence of others.
269 if (flags&SPSTART) {
270 longest = NULL;
271 len = 0;
272 for (; scan != NULL; scan = regnext(scan))
273 if (OP(scan) == EXACTLY && strlen(OPERAND(scan)) >= len) {
274 longest = OPERAND(scan);
275 len = strlen(OPERAND(scan));
277 r->regmust = longest;
278 r->regmlen = len;
282 return(r);
286 - reg - regular expression, i.e. main body or parenthesized thing
288 * Caller must absorb opening parenthesis.
290 * Combining parenthesis handling with the base level of regular expression
291 * is a trifle forced, but the need to tie the tails of the branches to what
292 * follows makes it hard to avoid.
294 static char *
295 reg(int paren, int *flagp)
297 char *ret;
298 char *br;
299 char *ender;
300 int parno;
301 int flags;
303 *flagp = HASWIDTH; /* Tentatively. */
305 /* Make an OPEN node, if parenthesized. */
306 if (paren) {
307 if (regnpar >= NSUBEXP)
308 FAIL("too many ()");
309 parno = regnpar;
310 regnpar++;
311 ret = regnode(OPEN+parno);
312 } else
313 ret = NULL;
315 /* Pick up the branches, linking them together. */
316 br = regbranch(&flags);
317 if (br == NULL)
318 return(NULL);
319 if (ret != NULL)
320 regtail(ret, br); /* OPEN -> first. */
321 else
322 ret = br;
323 if (!(flags&HASWIDTH))
324 *flagp &= ~HASWIDTH;
325 *flagp |= flags&SPSTART;
326 while (*regparse == '|' || *regparse == '\n') {
327 regparse++;
328 br = regbranch(&flags);
329 if (br == NULL)
330 return(NULL);
331 regtail(ret, br); /* BRANCH -> BRANCH. */
332 if (!(flags&HASWIDTH))
333 *flagp &= ~HASWIDTH;
334 *flagp |= flags&SPSTART;
337 /* Make a closing node, and hook it on the end. */
338 ender = regnode((paren) ? CLOSE+parno : END);
339 regtail(ret, ender);
341 /* Hook the tails of the branches to the closing node. */
342 for (br = ret; br != NULL; br = regnext(br))
343 regoptail(br, ender);
345 /* Check for proper termination. */
346 if (paren && *regparse++ != ')') {
347 FAIL("unmatched ()");
348 } else if (!paren && *regparse != '\0') {
349 if (*regparse == ')') {
350 FAIL("unmatched ()");
351 } else
352 FAIL("junk on end"); /* "Can't happen". */
353 /* NOTREACHED */
356 return(ret);
360 - regbranch - one alternative of an | operator
362 * Implements the concatenation operator.
364 static char *
365 regbranch(int *flagp)
367 char *ret;
368 char *chain;
369 char *latest;
370 int flags;
372 *flagp = WORST; /* Tentatively. */
374 ret = regnode(BRANCH);
375 chain = NULL;
376 while (*regparse != '\0' && *regparse != ')' &&
377 *regparse != '\n' && *regparse != '|') {
378 latest = regpiece(&flags);
379 if (latest == NULL)
380 return(NULL);
381 *flagp |= flags&HASWIDTH;
382 if (chain == NULL) /* First piece. */
383 *flagp |= flags&SPSTART;
384 else
385 regtail(chain, latest);
386 chain = latest;
388 if (chain == NULL) /* Loop ran zero times. */
389 (void) regnode(NOTHING);
391 return(ret);
395 - regpiece - something followed by possible [*+?]
397 * Note that the branching code sequences used for ? and the general cases
398 * of * and + are somewhat optimized: they use the same NOTHING node as
399 * both the endmarker for their branch list and the body of the last branch.
400 * It might seem that this node could be dispensed with entirely, but the
401 * endmarker role is not redundant.
403 static char *
404 regpiece(int *flagp)
406 char *ret;
407 char op;
408 char *next;
409 int flags;
411 ret = regatom(&flags);
412 if (ret == NULL)
413 return(NULL);
415 op = *regparse;
416 if (!ISMULT(op)) {
417 *flagp = flags;
418 return(ret);
421 if (!(flags&HASWIDTH) && op != '?')
422 FAIL("*+ operand could be empty");
423 *flagp = (op != '+') ? (WORST|SPSTART) : (WORST|HASWIDTH);
425 if (op == '*' && (flags&SIMPLE))
426 reginsert(STAR, ret);
427 else if (op == '*') {
428 /* Emit x* as (x&|), where & means "self". */
429 reginsert(BRANCH, ret); /* Either x */
430 regoptail(ret, regnode(BACK)); /* and loop */
431 regoptail(ret, ret); /* back */
432 regtail(ret, regnode(BRANCH)); /* or */
433 regtail(ret, regnode(NOTHING)); /* null. */
434 } else if (op == '+' && (flags&SIMPLE))
435 reginsert(PLUS, ret);
436 else if (op == '+') {
437 /* Emit x+ as x(&|), where & means "self". */
438 next = regnode(BRANCH); /* Either */
439 regtail(ret, next);
440 regtail(regnode(BACK), ret); /* loop back */
441 regtail(next, regnode(BRANCH)); /* or */
442 regtail(ret, regnode(NOTHING)); /* null. */
443 } else if (op == '?') {
444 /* Emit x? as (x|) */
445 reginsert(BRANCH, ret); /* Either x */
446 regtail(ret, regnode(BRANCH)); /* or */
447 next = regnode(NOTHING); /* null. */
448 regtail(ret, next);
449 regoptail(ret, next);
451 regparse++;
452 if (ISMULT(*regparse))
453 FAIL("nested *?+");
455 return(ret);
459 - regatom - the lowest level
461 * Optimization: gobbles an entire sequence of ordinary characters so that
462 * it can turn them into a single node, which is smaller to store and
463 * faster to run. Backslashed characters are exceptions, each becoming a
464 * separate node; the code is simpler that way and it's not worth fixing.
466 static char *
467 regatom(int *flagp)
469 char *ret;
470 int flags;
472 *flagp = WORST; /* Tentatively. */
474 switch (*regparse++) {
475 /* FIXME: these chars only have meaning at beg/end of pat? */
476 case '^':
477 ret = regnode(BOL);
478 break;
479 case '$':
480 ret = regnode(EOL);
481 break;
482 case '.':
483 ret = regnode(ANY);
484 *flagp |= HASWIDTH|SIMPLE;
485 break;
486 case '[': {
487 int class;
488 int classend;
489 int i;
491 if (*regparse == '^') { /* Complement of range. */
492 ret = regnode(ANYBUT);
493 regparse++;
494 } else
495 ret = regnode(ANYOF);
496 if (*regparse == ']' || *regparse == '-')
497 regc(*regparse++);
498 while (*regparse != '\0' && *regparse != ']') {
499 if (*regparse == '-') {
500 regparse++;
501 if (*regparse == ']' || *regparse == '\0')
502 regc('-');
503 else {
504 class = UCHARAT(regparse-2);
505 classend = UCHARAT(regparse);
506 if (__collate_load_error) {
507 if (class > classend)
508 FAIL("invalid [] range");
509 for (class++; class <= classend; class++)
510 regc(class);
511 } else {
512 if (__collate_range_cmp(class, classend) > 0)
513 FAIL("invalid [] range");
514 for (i = 0; i <= UCHAR_MAX; i++)
515 if ( i != class
516 && __collate_range_cmp(class, i) <= 0
517 && __collate_range_cmp(i, classend) <= 0
519 regc(i);
521 regparse++;
523 } else
524 regc(*regparse++);
526 regc('\0');
527 if (*regparse != ']')
528 FAIL("unmatched []");
529 regparse++;
530 *flagp |= HASWIDTH|SIMPLE;
532 break;
533 case '(':
534 ret = reg(1, &flags);
535 if (ret == NULL)
536 return(NULL);
537 *flagp |= flags&(HASWIDTH|SPSTART);
538 break;
539 case '\0':
540 case '|':
541 case '\n':
542 case ')':
543 FAIL("internal urp"); /* Supposed to be caught earlier. */
544 break;
545 case '?':
546 case '+':
547 case '*':
548 FAIL("?+* follows nothing");
549 break;
550 case '\\':
551 switch (*regparse++) {
552 case '\0':
553 FAIL("trailing \\");
554 break;
555 case '<':
556 ret = regnode(WORDA);
557 break;
558 case '>':
559 ret = regnode(WORDZ);
560 break;
561 /* FIXME: Someday handle \1, \2, ... */
562 default:
563 /* Handle general quoted chars in exact-match routine */
564 goto de_fault;
566 break;
567 de_fault:
568 default:
570 * Encode a string of characters to be matched exactly.
572 * This is a bit tricky due to quoted chars and due to
573 * '*', '+', and '?' taking the SINGLE char previous
574 * as their operand.
576 * On entry, the char at regparse[-1] is going to go
577 * into the string, no matter what it is. (It could be
578 * following a \ if we are entered from the '\' case.)
580 * Basic idea is to pick up a good char in ch and
581 * examine the next char. If it's *+? then we twiddle.
582 * If it's \ then we frozzle. If it's other magic char
583 * we push ch and terminate the string. If none of the
584 * above, we push ch on the string and go around again.
586 * regprev is used to remember where "the current char"
587 * starts in the string, if due to a *+? we need to back
588 * up and put the current char in a separate, 1-char, string.
589 * When regprev is NULL, ch is the only char in the
590 * string; this is used in *+? handling, and in setting
591 * flags |= SIMPLE at the end.
594 char *regprev;
595 char ch;
597 regparse--; /* Look at cur char */
598 ret = regnode(EXACTLY);
599 for ( regprev = 0 ; ; ) {
600 ch = *regparse++; /* Get current char */
601 switch (*regparse) { /* look at next one */
603 default:
604 regc(ch); /* Add cur to string */
605 break;
607 case '.': case '[': case '(':
608 case ')': case '|': case '\n':
609 case '$': case '^':
610 case '\0':
611 /* FIXME, $ and ^ should not always be magic */
612 magic:
613 regc(ch); /* dump cur char */
614 goto done; /* and we are done */
616 case '?': case '+': case '*':
617 if (!regprev) /* If just ch in str, */
618 goto magic; /* use it */
619 /* End mult-char string one early */
620 regparse = regprev; /* Back up parse */
621 goto done;
623 case '\\':
624 regc(ch); /* Cur char OK */
625 switch (regparse[1]){ /* Look after \ */
626 case '\0':
627 case '<':
628 case '>':
629 /* FIXME: Someday handle \1, \2, ... */
630 goto done; /* Not quoted */
631 default:
632 /* Backup point is \, scan * point is after it. */
633 regprev = regparse;
634 regparse++;
635 continue; /* NOT break; */
638 regprev = regparse; /* Set backup point */
640 done:
641 regc('\0');
642 *flagp |= HASWIDTH;
643 if (!regprev) /* One char? */
644 *flagp |= SIMPLE;
646 break;
649 return(ret);
653 - regnode - emit a node
655 static char * /* Location. */
656 regnode(char op)
658 char *ret;
659 char *ptr;
661 ret = regcode;
662 if (ret == &regdummy) {
663 regsize += 3;
664 return(ret);
667 ptr = ret;
668 *ptr++ = op;
669 *ptr++ = '\0'; /* Null "next" pointer. */
670 *ptr++ = '\0';
671 regcode = ptr;
673 return(ret);
677 - regc - emit (if appropriate) a byte of code
679 static void
680 regc(char b)
682 if (regcode != &regdummy)
683 *regcode++ = b;
684 else
685 regsize++;
689 - reginsert - insert an operator in front of already-emitted operand
691 * Means relocating the operand.
693 static void
694 reginsert(char op, char *opnd)
696 char *src;
697 char *dst;
698 char *place;
700 if (regcode == &regdummy) {
701 regsize += 3;
702 return;
705 src = regcode;
706 regcode += 3;
707 dst = regcode;
708 while (src > opnd)
709 *--dst = *--src;
711 place = opnd; /* Op node, where operand used to be. */
712 *place++ = op;
713 *place++ = '\0';
714 *place++ = '\0';
718 - regtail - set the next-pointer at the end of a node chain
720 static void
721 regtail(char *p, char *val)
723 char *scan;
724 char *temp;
725 int offset;
727 if (p == &regdummy)
728 return;
730 /* Find last node. */
731 scan = p;
732 for (;;) {
733 temp = regnext(scan);
734 if (temp == NULL)
735 break;
736 scan = temp;
739 if (OP(scan) == BACK)
740 offset = scan - val;
741 else
742 offset = val - scan;
743 *(scan+1) = (offset>>8)&0377;
744 *(scan+2) = offset&0377;
748 - regoptail - regtail on operand of first argument; nop if operandless
750 static void
751 regoptail(char *p, char *val)
753 /* "Operandless" and "op != BRANCH" are synonymous in practice. */
754 if (p == NULL || p == &regdummy || OP(p) != BRANCH)
755 return;
756 regtail(OPERAND(p), val);
760 * regexec and friends
764 * Global work variables for regexec().
766 static char *reginput; /* String-input pointer. */
767 static char *regbol; /* Beginning of input, for ^ check. */
768 static char **regstartp; /* Pointer to startp array. */
769 static char **regendp; /* Ditto for endp. */
772 * Forwards.
774 STATIC int regtry();
775 STATIC int regmatch();
776 STATIC int regrepeat();
778 #ifdef DEBUG
779 int regnarrate = 0;
780 void regdump();
781 STATIC char *regprop();
782 #endif
785 - regexec - match a regexp against a string
788 regexec(const regexp *prog, const char *string)
790 char *s;
791 extern char *strchr();
793 /* Be paranoid... */
794 if (prog == NULL || string == NULL) {
795 regerror("NULL parameter");
796 return(0);
799 /* Check validity of program. */
800 if (UCHARAT(prog->program) != MAGIC) {
801 regerror("corrupted program");
802 return(0);
805 /* If there is a "must appear" string, look for it. */
806 if (prog->regmust != NULL) {
807 s = (char *)string;
808 while ((s = strchr(s, prog->regmust[0])) != NULL) {
809 if (strncmp(s, prog->regmust, prog->regmlen) == 0)
810 break; /* Found it. */
811 s++;
813 if (s == NULL) /* Not present. */
814 return(0);
817 /* Mark beginning of line for ^ . */
818 regbol = (char *)string;
820 /* Simplest case: anchored match need be tried only once. */
821 if (prog->reganch)
822 return(regtry(prog, string));
824 /* Messy cases: unanchored match. */
825 s = (char *)string;
826 if (prog->regstart != '\0')
827 /* We know what char it must start with. */
828 while ((s = strchr(s, prog->regstart)) != NULL) {
829 if (regtry(prog, s))
830 return(1);
831 s++;
833 else
834 /* We don't -- general case. */
835 do {
836 if (regtry(prog, s))
837 return(1);
838 } while (*s++ != '\0');
840 /* Failure. */
841 return(0);
845 - regtry - try match at specific point
847 static int /* 0 failure, 1 success */
848 regtry(regexp *prog, char *string)
850 int i;
851 char **sp;
852 char **ep;
854 reginput = string;
855 regstartp = prog->startp;
856 regendp = prog->endp;
858 sp = prog->startp;
859 ep = prog->endp;
860 for (i = NSUBEXP; i > 0; i--) {
861 *sp++ = NULL;
862 *ep++ = NULL;
864 if (regmatch(prog->program + 1)) {
865 prog->startp[0] = string;
866 prog->endp[0] = reginput;
867 return(1);
868 } else
869 return(0);
873 - regmatch - main matching routine
875 * Conceptually the strategy is simple: check to see whether the current
876 * node matches, call self recursively to see whether the rest matches,
877 * and then act accordingly. In practice we make some effort to avoid
878 * recursion, in particular by going through "ordinary" nodes (that don't
879 * need to know whether the rest of the match failed) by a loop instead of
880 * by recursion.
882 static int /* 0 failure, 1 success */
883 regmatch(char *prog)
885 char *scan; /* Current node. */
886 char *next; /* Next node. */
888 scan = prog;
889 #ifdef DEBUG
890 if (scan != NULL && regnarrate)
891 fprintf(stderr, "%s(\n", regprop(scan));
892 #endif
893 while (scan != NULL) {
894 #ifdef DEBUG
895 if (regnarrate)
896 fprintf(stderr, "%s...\n", regprop(scan));
897 #endif
898 next = regnext(scan);
900 switch (OP(scan)) {
901 case BOL:
902 if (reginput != regbol)
903 return(0);
904 break;
905 case EOL:
906 if (*reginput != '\0')
907 return(0);
908 break;
909 case WORDA:
910 /* Must be looking at a letter, digit, or _ */
911 if ((!isalnum((unsigned char)*reginput)) && *reginput != '_')
912 return(0);
913 /* Prev must be BOL or nonword */
914 if (reginput > regbol &&
915 (isalnum((unsigned char)reginput[-1]) || reginput[-1] == '_'))
916 return(0);
917 break;
918 case WORDZ:
919 /* Must be looking at non letter, digit, or _ */
920 if (isalnum((unsigned char)*reginput) || *reginput == '_')
921 return(0);
922 /* We don't care what the previous char was */
923 break;
924 case ANY:
925 if (*reginput == '\0')
926 return(0);
927 reginput++;
928 break;
929 case EXACTLY: {
930 int len;
931 char *opnd;
933 opnd = OPERAND(scan);
934 /* Inline the first character, for speed. */
935 if (*opnd != *reginput)
936 return(0);
937 len = strlen(opnd);
938 if (len > 1 && strncmp(opnd, reginput, len) != 0)
939 return(0);
940 reginput += len;
942 break;
943 case ANYOF:
944 if (*reginput == '\0' || strchr(OPERAND(scan), *reginput) == NULL)
945 return(0);
946 reginput++;
947 break;
948 case ANYBUT:
949 if (*reginput == '\0' || strchr(OPERAND(scan), *reginput) != NULL)
950 return(0);
951 reginput++;
952 break;
953 case NOTHING:
954 break;
955 case BACK:
956 break;
957 case OPEN+1:
958 case OPEN+2:
959 case OPEN+3:
960 case OPEN+4:
961 case OPEN+5:
962 case OPEN+6:
963 case OPEN+7:
964 case OPEN+8:
965 case OPEN+9: {
966 int no;
967 char *save;
969 no = OP(scan) - OPEN;
970 save = reginput;
972 if (regmatch(next)) {
974 * Don't set startp if some later
975 * invocation of the same parentheses
976 * already has.
978 if (regstartp[no] == NULL)
979 regstartp[no] = save;
980 return(1);
981 } else
982 return(0);
984 break;
985 case CLOSE+1:
986 case CLOSE+2:
987 case CLOSE+3:
988 case CLOSE+4:
989 case CLOSE+5:
990 case CLOSE+6:
991 case CLOSE+7:
992 case CLOSE+8:
993 case CLOSE+9: {
994 int no;
995 char *save;
997 no = OP(scan) - CLOSE;
998 save = reginput;
1000 if (regmatch(next)) {
1002 * Don't set endp if some later
1003 * invocation of the same parentheses
1004 * already has.
1006 if (regendp[no] == NULL)
1007 regendp[no] = save;
1008 return(1);
1009 } else
1010 return(0);
1012 break;
1013 case BRANCH: {
1014 char *save;
1016 if (OP(next) != BRANCH) /* No choice. */
1017 next = OPERAND(scan); /* Avoid recursion. */
1018 else {
1019 do {
1020 save = reginput;
1021 if (regmatch(OPERAND(scan)))
1022 return(1);
1023 reginput = save;
1024 scan = regnext(scan);
1025 } while (scan != NULL && OP(scan) == BRANCH);
1026 return(0);
1027 /* NOTREACHED */
1030 break;
1031 case STAR:
1032 case PLUS: {
1033 char nextch;
1034 int no;
1035 char *save;
1036 int min;
1039 * Lookahead to avoid useless match attempts
1040 * when we know what character comes next.
1042 nextch = '\0';
1043 if (OP(next) == EXACTLY)
1044 nextch = *OPERAND(next);
1045 min = (OP(scan) == STAR) ? 0 : 1;
1046 save = reginput;
1047 no = regrepeat(OPERAND(scan));
1048 while (no >= min) {
1049 /* If it could work, try it. */
1050 if (nextch == '\0' || *reginput == nextch)
1051 if (regmatch(next))
1052 return(1);
1053 /* Couldn't or didn't -- back up. */
1054 no--;
1055 reginput = save + no;
1057 return(0);
1059 break;
1060 case END:
1061 return(1); /* Success! */
1062 break;
1063 default:
1064 regerror("memory corruption");
1065 return(0);
1066 break;
1069 scan = next;
1073 * We get here only if there's trouble -- normally "case END" is
1074 * the terminating point.
1076 regerror("corrupted pointers");
1077 return(0);
1081 - regrepeat - repeatedly match something simple, report how many
1083 static int
1084 regrepeat(char *p)
1086 int count = 0;
1087 char *scan;
1088 char *opnd;
1090 scan = reginput;
1091 opnd = OPERAND(p);
1092 switch (OP(p)) {
1093 case ANY:
1094 count = strlen(scan);
1095 scan += count;
1096 break;
1097 case EXACTLY:
1098 while (*opnd == *scan) {
1099 count++;
1100 scan++;
1102 break;
1103 case ANYOF:
1104 while (*scan != '\0' && strchr(opnd, *scan) != NULL) {
1105 count++;
1106 scan++;
1108 break;
1109 case ANYBUT:
1110 while (*scan != '\0' && strchr(opnd, *scan) == NULL) {
1111 count++;
1112 scan++;
1114 break;
1115 default: /* Oh dear. Called inappropriately. */
1116 regerror("internal foulup");
1117 count = 0; /* Best compromise. */
1118 break;
1120 reginput = scan;
1122 return(count);
1126 - regnext - dig the "next" pointer out of a node
1128 static char *
1129 regnext(char *p)
1131 int offset;
1133 if (p == &regdummy)
1134 return(NULL);
1136 offset = NEXT(p);
1137 if (offset == 0)
1138 return(NULL);
1140 if (OP(p) == BACK)
1141 return(p-offset);
1142 else
1143 return(p+offset);
1146 #ifdef DEBUG
1148 STATIC char *regprop();
1151 - regdump - dump a regexp onto stdout in vaguely comprehensible form
1153 void
1154 regdump(regexp *r)
1156 char *s;
1157 char op = EXACTLY; /* Arbitrary non-END op. */
1158 char *next;
1159 extern char *strchr();
1162 s = r->program + 1;
1163 while (op != END) { /* While that wasn't END last time... */
1164 op = OP(s);
1165 printf("%2d%s", s-r->program, regprop(s)); /* Where, what. */
1166 next = regnext(s);
1167 if (next == NULL) /* Next ptr. */
1168 printf("(0)");
1169 else
1170 printf("(%d)", (s-r->program)+(next-s));
1171 s += 3;
1172 if (op == ANYOF || op == ANYBUT || op == EXACTLY) {
1173 /* Literal string, where present. */
1174 while (*s != '\0') {
1175 putchar(*s);
1176 s++;
1178 s++;
1180 putchar('\n');
1183 /* Header fields of interest. */
1184 if (r->regstart != '\0')
1185 printf("start `%c' ", r->regstart);
1186 if (r->reganch)
1187 printf("anchored ");
1188 if (r->regmust != NULL)
1189 printf("must have \"%s\"", r->regmust);
1190 printf("\n");
1194 - regprop - printable representation of opcode
1196 static char *
1197 regprop(char *op)
1199 char *p;
1200 static char buf[50];
1202 (void) strcpy(buf, ":");
1204 switch (OP(op)) {
1205 case BOL:
1206 p = "BOL";
1207 break;
1208 case EOL:
1209 p = "EOL";
1210 break;
1211 case ANY:
1212 p = "ANY";
1213 break;
1214 case ANYOF:
1215 p = "ANYOF";
1216 break;
1217 case ANYBUT:
1218 p = "ANYBUT";
1219 break;
1220 case BRANCH:
1221 p = "BRANCH";
1222 break;
1223 case EXACTLY:
1224 p = "EXACTLY";
1225 break;
1226 case NOTHING:
1227 p = "NOTHING";
1228 break;
1229 case BACK:
1230 p = "BACK";
1231 break;
1232 case END:
1233 p = "END";
1234 break;
1235 case OPEN+1:
1236 case OPEN+2:
1237 case OPEN+3:
1238 case OPEN+4:
1239 case OPEN+5:
1240 case OPEN+6:
1241 case OPEN+7:
1242 case OPEN+8:
1243 case OPEN+9:
1244 sprintf(buf+strlen(buf), "OPEN%d", OP(op)-OPEN);
1245 p = NULL;
1246 break;
1247 case CLOSE+1:
1248 case CLOSE+2:
1249 case CLOSE+3:
1250 case CLOSE+4:
1251 case CLOSE+5:
1252 case CLOSE+6:
1253 case CLOSE+7:
1254 case CLOSE+8:
1255 case CLOSE+9:
1256 sprintf(buf+strlen(buf), "CLOSE%d", OP(op)-CLOSE);
1257 p = NULL;
1258 break;
1259 case STAR:
1260 p = "STAR";
1261 break;
1262 case PLUS:
1263 p = "PLUS";
1264 break;
1265 case WORDA:
1266 p = "WORDA";
1267 break;
1268 case WORDZ:
1269 p = "WORDZ";
1270 break;
1271 default:
1272 regerror("corrupted opcode");
1273 break;
1275 if (p != NULL)
1276 (void) strcat(buf, p);
1277 return(buf);
1279 #endif
1282 * The following is provided for those people who do not have strcspn() in
1283 * their C libraries. They should get off their butts and do something
1284 * about it; at least one public-domain implementation of those (highly
1285 * useful) string routines has been published on Usenet.
1287 #ifdef STRCSPN
1289 * strcspn - find length of initial segment of s1 consisting entirely
1290 * of characters not from s2
1293 static int
1294 strcspn(char *s1, char *s2)
1296 char *scan1;
1297 char *scan2;
1298 int count;
1300 count = 0;
1301 for (scan1 = s1; *scan1 != '\0'; scan1++) {
1302 for (scan2 = s2; *scan2 != '\0';) /* ++ moved down. */
1303 if (*scan1 == *scan2++)
1304 return(count);
1305 count++;
1307 return(count);
1309 #endif