/usr/src/lib/libc/regex/regcomp.c

Bug Summary

File:	src/lib/libc/regex/regcomp.c
Warning:	line 1352, column 7 Assigned value is garbage or undefined
Annotated Source Code

Press '?' to see keyboard shortcuts
Show analyzer invocation
clang -cc1 -cc1 -triple amd64-unknown-openbsd7.0 -analyze -disable-free -disable-llvm-verifier -discard-value-names -main-file-name regcomp.c -analyzer-store=region -analyzer-opt-analyze-nested-blocks -analyzer-checker=core -analyzer-checker=apiModeling -analyzer-checker=unix -analyzer-checker=deadcode -analyzer-checker=security.insecureAPI.UncheckedReturn -analyzer-checker=security.insecureAPI.getpw -analyzer-checker=security.insecureAPI.gets -analyzer-checker=security.insecureAPI.mktemp -analyzer-checker=security.insecureAPI.mkstemp -analyzer-checker=security.insecureAPI.vfork -analyzer-checker=nullability.NullPassedToNonnull -analyzer-checker=nullability.NullReturnedFromNonnull -analyzer-output plist -w -setup-static-analyzer -mrelocation-model pic -pic-level 1 -fhalf-no-semantic-interposition -mframe-pointer=all -relaxed-aliasing -fno-rounding-math -mconstructor-aliases -munwind-tables -target-cpu x86-64 -target-feature +retpoline-indirect-calls -target-feature +retpoline-indirect-branches -tune-cpu generic -debugger-tuning=gdb -fcoverage-compilation-dir=/usr/src/lib/libc/obj -resource-dir /usr/local/lib/clang/13.0.0 -include namespace.h -I /usr/src/lib/libc/include -I /usr/src/lib/libc/hidden -D __LIBC__ -D APIWARN -D YP -I /usr/src/lib/libc/yp -I /usr/src/lib/libc -I /usr/src/lib/libc/gdtoa -I /usr/src/lib/libc/arch/amd64/gdtoa -D INFNAN_CHECK -D MULTIPLE_THREADS -D NO_FENV_H -D USE_LOCALE -I /usr/src/lib/libc -I /usr/src/lib/libc/citrus -D RESOLVSORT -D FLOATING_POINT -D PRINTF_WIDE_CHAR -D SCANF_WIDE_CHAR -D FUTEX -D PIC -internal-isystem /usr/local/lib/clang/13.0.0/include -internal-externc-isystem /usr/include -O2 -fdebug-compilation-dir=/usr/src/lib/libc/obj -ferror-limit 19 -fwrapv -D_RET_PROTECTOR -ret-protector -fgnuc-version=4.2.1 -vectorize-loops -vectorize-slp -fno-builtin-malloc -fno-builtin-calloc -fno-builtin-realloc -fno-builtin-valloc -fno-builtin-free -fno-builtin-strdup -fno-builtin-strndup -analyzer-output=html -faddrsig -D__GCC_HAVE_DWARF2_CFI_ASM=1 -o /home/ben/Projects/vmm/scan-build/2022-01-12-194120-40624-1 -x c /usr/src/lib/libc/regex/regcomp.c
1/*	$OpenBSD: regcomp.c,v 1.43 2021/01/03 17:07:57 tb Exp $ */
2/*-
* Copyright (c) 1992, 1993, 1994 Henry Spencer.
* Copyright (c) 1992, 1993, 1994
*	The Regents of the University of California.  All rights reserved.
*
* This code is derived from software contributed to Berkeley by
* Henry Spencer.
*
* Redistribution and use in source and binary forms, with or without
* modification, are permitted provided that the following conditions
* are met:
* 1. Redistributions of source code must retain the above copyright
*    notice, this list of conditions and the following disclaimer.
* 2. Redistributions in binary form must reproduce the above copyright
*    notice, this list of conditions and the following disclaimer in the
*    documentation and/or other materials provided with the distribution.
* 3. Neither the name of the University nor the names of its contributors
*    may be used to endorse or promote products derived from this software
*    without specific prior written permission.
*
* THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
* ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
* IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
* ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
* FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
* DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
* OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
* HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
* LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
* OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
* SUCH DAMAGE.
*
*	@(#)regcomp.c	8.5 (Berkeley) 3/20/94
*/

37#include <sys/types.h>
38#include <stdio.h>
39#include <string.h>
40#include <ctype.h>
41#include <limits.h>
42#include <stdlib.h>
43#include <regex.h>

45#include "utils.h"
46#include "regex2.h"

48#include "cclass.h"
49#include "cname.h"

51/*
* parse structure, passed up and down to avoid global variables and
* other clumsinesses
*/
55struct parse {
const char *next;	/* next character in RE */
const char *end;	/* end of string (-> NUL normally) */
int error;		/* has an error been seen? */
sop *strip;		/* malloced strip */
sopno ssize;		/* malloced strip size (allocated) */
sopno slen;		/* malloced strip length (used) */
int ncsalloc;		/* number of csets allocated */
struct re_guts *g;
64#	define	NPAREN10	10	/* we need to remember () 1-9 for back refs */
sopno pbegin[NPAREN10];	/* -> ( ([0] unused) */
sopno pend[NPAREN10];	/* -> ) ([0] unused) */
67};

69static void p_ere(struct parse *, int);
70static void p_ere_exp(struct parse *);
71static void p_str(struct parse *);
72static void p_bre(struct parse *, int, int);
73static int p_simp_re(struct parse *, int);
74static int p_count(struct parse *);
75static void p_bracket(struct parse *);
76static void p_b_term(struct parse *, cset *);
77static void p_b_cclass(struct parse *, cset *);
78static void p_b_eclass(struct parse *, cset *);
79static char p_b_symbol(struct parse *);
80static char p_b_coll_elem(struct parse *, int);
81static char othercase(int);
82static void bothcases(struct parse *, int);
83static void ordinary(struct parse *, int);
84static void backslash(struct parse *, int);
85static void nonnewline(struct parse *);
86static void repeat(struct parse *, sopno, int, int);
87static void seterr(struct parse *, int);
88static cset *allocset(struct parse *);
89static void freeset(struct parse *, cset *);
90static int freezeset(struct parse *, cset *);
91static int firstch(struct parse *, cset *);
92static int nch(struct parse *, cset *);
93static sopno dupl(struct parse *, sopno, sopno);
94static void doemit(struct parse *, sop, size_t);
95static void doinsert(struct parse *, sop, size_t, sopno);
96static void dofwd(struct parse *, sopno, sop);
97static int enlarge(struct parse *, sopno);
98static void stripsnug(struct parse *, struct re_guts *);
99static void findmust(struct parse *, struct re_guts *);
100static sopno pluscount(struct parse *, struct re_guts *);

102static char nuls[10];		/* place to point scanner in event of error */

104/*
* macros for use with parse structure
* BEWARE:  these know that the parse structure is named `p' !!!
*/
108#define	PEEK()(*p->next)	(*p->next)
109#define	PEEK2()(*(p->next+1))	(*(p->next+1))
110#define	MORE()(p->end - p->next > 0)	(p->end - p->next > 0)
111#define	MORE2()(p->end - p->next > 1)	(p->end - p->next > 1)
112#define	SEE(c)((p->end - p->next > 0) && (*p->next) == (
c))	(MORE()(p->end - p->next > 0) && PEEK()(*p->next) == (c))
113#define	SEETWO(a, b)((p->end - p->next > 1) && (*p->next) == (
a) && (*(p->next+1)) == (b))	(MORE2()(p->end - p->next > 1) && PEEK()(*p->next) == (a) && PEEK2()(*(p->next+1)) == (b))
114#define	EAT(c)((((p->end - p->next > 0) && (*p->next) ==
 (c))) ? ((p->next++), 1) : 0)	((SEE(c)((p->end - p->next > 0) && (*p->next) == (
c))) ? (NEXT()(p->next++), 1) : 0)
115#define	EATTWO(a, b)((((p->end - p->next > 1) && (*p->next) ==
 (a) && (*(p->next+1)) == (b))) ? ((p->next += 2
), 1) : 0)	((SEETWO(a, b)((p->end - p->next > 1) && (*p->next) == (
a) && (*(p->next+1)) == (b))) ? (NEXT2()(p->next += 2), 1) : 0)
116#define	NEXT()(p->next++)	(p->next++)
117#define	NEXT2()(p->next += 2)	(p->next += 2)
118#define	NEXTn(n)(p->next += (n))	(p->next += (n))
119#define	GETNEXT()(*p->next++)	(*p->next++)
120#define	SETERROR(e)seterr(p, (e))	seterr(p, (e))
121#define	REQUIRE(co, e)do { if (!(co)) seterr(p, (e)); } while (0)	do { if (!(co)) SETERROR(e)seterr(p, (e)); } while (0)
122#define	EMIT(op, sopnd)doemit(p, (sop)(op), (size_t)(sopnd))	doemit(p, (sop)(op), (size_t)(sopnd))
123#define	INSERT(op, pos)doinsert(p, (sop)(op), (p->slen)-(pos)+1, pos)	doinsert(p, (sop)(op), HERE()(p->slen)-(pos)+1, pos)
124#define	AHEAD(pos)dofwd(p, pos, (p->slen)-(pos))		dofwd(p, pos, HERE()(p->slen)-(pos))
125#define	ASTERN(sop, pos)doemit(p, (sop)(sop), (size_t)((p->slen)-pos))	EMIT(sop, HERE()-pos)doemit(p, (sop)(sop), (size_t)((p->slen)-pos))
126#define	HERE()(p->slen)		(p->slen)
127#define	THERE()(p->slen - 1)		(p->slen - 1)
128#define	THERETHERE()(p->slen - 2)	(p->slen - 2)
129#define	DROP(n)(p->slen -= (n))	(p->slen -= (n))

131#ifndef NDEBUG
132static int never0 = 0;		/* for use in asserts; shuts lint up */
133#else
134#define	never0	0		/* some <assert.h>s have bugs too */
135#endif

137/*
- regcomp - interface for parser and compilation
*/
140int				/* 0 success, otherwise REG_something */
141regcomp(regex_t *preg, const char *pattern, int cflags)
142{
struct parse pa;
struct re_guts *g;
struct parse *p = &pa;
int i;
size_t len;
148#ifdef REDEBUG
149#	define	GOODFLAGS(f)((f)&~0200)	(f)
150#else
151#	define	GOODFLAGS(f)((f)&~0200)	((f)&~REG_DUMP0200)
152#endif

cflags = GOODFLAGS(cflags)((cflags)&~0200);
if ((cflags&REG_EXTENDED0001) && (cflags&REG_NOSPEC0020))
return(REG_INVARG16);

if (cflags&REG_PEND0040) {
if (preg->re_endp < pattern)
	return(REG_INVARG16);
len = preg->re_endp - pattern;
} else
len = strlen((char *)pattern);

/* do the mallocs early so failure handling is easy */
g = malloc(sizeof(struct re_guts));
if (g == NULL((void *)0))
return(REG_ESPACE12);
p->ssize = len/(size_t)2*(size_t)3 + (size_t)1;	/* ugh */
p->strip = reallocarray(NULL((void *)0), p->ssize, sizeof(sop));
p->slen = 0;
if (p->strip == NULL((void *)0)) {
free(g);
return(REG_ESPACE12);
}

/* set things up */
p->g = g;
p->next = pattern;
p->end = p->next + len;
p->error = 0;
p->ncsalloc = 0;
for (i = 0; i < NPAREN10; i++) {
p->pbegin[i] = 0;
p->pend[i] = 0;
}
g->csetsize = NC(127 - (-127 -1) + 1);
g->sets = NULL((void *)0);
g->setbits = NULL((void *)0);
g->ncsets = 0;
g->cflags = cflags;
g->iflags = 0;
g->nbol = 0;
g->neol = 0;
g->must = NULL((void *)0);
g->mlen = 0;
g->nsub = 0;
g->backrefs = 0;

/* do it */
EMIT(OEND, 0)doemit(p, (sop)((1LU<<((unsigned)27))), (size_t)(0));
g->firststate = THERE()(p->slen - 1);
if (cflags&REG_EXTENDED0001)
p_ere(p, OUT(127 +1));
else if (cflags&REG_NOSPEC0020)
p_str(p);
else
p_bre(p, OUT(127 +1), OUT(127 +1));
EMIT(OEND, 0)doemit(p, (sop)((1LU<<((unsigned)27))), (size_t)(0));
g->laststate = THERE()(p->slen - 1);

/* tidy up loose ends and fill things in */
stripsnug(p, g);
findmust(p, g);
g->nplus = pluscount(p, g);
g->magic = MAGIC2((('R'^0200)<<8)|'E');
preg->re_nsub = g->nsub;
preg->re_g = g;
preg->re_magic = MAGIC1((('r'^0200)<<8) | 'e');
220#ifndef REDEBUG
/* not debugging, so can't rely on the assert() in regexec() */
if (g->iflags&BAD04)
SETERROR(REG_ASSERT)seterr(p, (15));
224#endif

/* win or lose, we're done */
if (p->error != 0)	/* lose */
regfree(preg);
return(p->error);
230}

232/*
- p_ere - ERE parser top level, concatenation and alternation
*/
235static void
236p_ere(struct parse *p, int stop)	/* character this ERE should end at */
237{
char c;
sopno prevback;
sopno prevfwd;
sopno conc;
int first = 1;		/* is this the first alternative? */

for (;;) {
/* do a bunch of concatenated expressions */
conc = HERE()(p->slen);
while (MORE()(p->end - p->next > 0) && (c = PEEK()(*p->next)) != '|' && c != stop)
	p_ere_exp(p);
REQUIRE(HERE() != conc, REG_EMPTY)do { if (!((p->slen) != conc)) seterr(p, (14)); } while (0
);	/* require nonempty */

if (!EAT('|')((((p->end - p->next > 0) && (*p->next) ==
 ('|'))) ? ((p->next++), 1) : 0))
	break;		/* NOTE BREAK OUT */

if (first) {
	INSERT(OCH_, conc)doinsert(p, (sop)((15LU<<((unsigned)27))), (p->slen)
-(conc)+1, conc);	/* offset is wrong */
	prevfwd = conc;
	prevback = conc;
	first = 0;
}
ASTERN(OOR1, prevback)doemit(p, (sop)((16LU<<((unsigned)27))), (size_t)((p->
slen)-prevback));
prevback = THERE()(p->slen - 1);
AHEAD(prevfwd)dofwd(p, prevfwd, (p->slen)-(prevfwd));			/* fix previous offset */
prevfwd = HERE()(p->slen);
EMIT(OOR2, 0)doemit(p, (sop)((17LU<<((unsigned)27))), (size_t)(0));			/* offset is very wrong */
}

if (!first) {		/* tail-end fixups */
AHEAD(prevfwd)dofwd(p, prevfwd, (p->slen)-(prevfwd));
ASTERN(O_CH, prevback)doemit(p, (sop)((18LU<<((unsigned)27))), (size_t)((p->
slen)-prevback));
}

assert(!MORE() || SEE(stop))((void)0);
273}

275/*
- p_ere_exp - parse one subERE, an atom possibly followed by a repetition op
*/
278static void
279p_ere_exp(struct parse *p)
280{
char c;
sopno pos;
int count;
int count2;
sopno subno;
int wascaret = 0;

assert(MORE())((void)0);		/* caller should have ensured this */
c = GETNEXT()(*p->next++);

pos = HERE()(p->slen);
switch (c) {
case '(':
REQUIRE(MORE(), REG_EPAREN)do { if (!((p->end - p->next > 0))) seterr(p, (8)); }
 while (0);
p->g->nsub++;
subno = p->g->nsub;
if (subno < NPAREN10)
	p->pbegin[subno] = HERE()(p->slen);
EMIT(OLPAREN, subno)doemit(p, (sop)((13LU<<((unsigned)27))), (size_t)(subno
));
if (!SEE(')')((p->end - p->next > 0) && (*p->next) == (
')')))
	p_ere(p, ')');
if (subno < NPAREN10) {
	p->pend[subno] = HERE()(p->slen);
	assert(p->pend[subno] != 0)((void)0);
}
EMIT(ORPAREN, subno)doemit(p, (sop)((14LU<<((unsigned)27))), (size_t)(subno
));
REQUIRE(MORE() && GETNEXT() == ')', REG_EPAREN)do { if (!((p->end - p->next > 0) && (*p->
next++) == ')')) seterr(p, (8)); } while (0);
break;
case '^':
EMIT(OBOL, 0)doemit(p, (sop)((3LU<<((unsigned)27))), (size_t)(0));
p->g->iflags |= USEBOL01;
p->g->nbol++;
wascaret = 1;
break;
case '$':
EMIT(OEOL, 0)doemit(p, (sop)((4LU<<((unsigned)27))), (size_t)(0));
p->g->iflags |= USEEOL02;
p->g->neol++;
break;
case '|':
SETERROR(REG_EMPTY)seterr(p, (14));
break;
case '*':
case '+':
case '?':
SETERROR(REG_BADRPT)seterr(p, (13));
break;
case '.':
if (p->g->cflags&REG_NEWLINE0010)
	nonnewline(p);
else
	EMIT(OANY, 0)doemit(p, (sop)((5LU<<((unsigned)27))), (size_t)(0));
break;
case '[':
p_bracket(p);
break;
case '\\':
REQUIRE(MORE(), REG_EESCAPE)do { if (!((p->end - p->next > 0))) seterr(p, (5)); }
 while (0);
c = GETNEXT()(*p->next++);
backslash(p, c);
break;
case '{':		/* okay as ordinary except if digit follows */
REQUIRE(!MORE() || !isdigit((uch)PEEK()), REG_BADRPT)do { if (!(!(p->end - p->next > 0) || !isdigit((uch)
(*p->next)))) seterr(p, (13)); } while (0);
/* FALLTHROUGH */
default:
if (p->error != 0)
	return;
ordinary(p, c);
break;
}

if (!MORE()(p->end - p->next > 0))
return;
c = PEEK()(*p->next);
/* we call { a repetition if followed by a digit */
if (!( c == '*' || c == '+' || c == '?' ||
		(c == '{' && MORE2()(p->end - p->next > 1) && isdigit((uch)PEEK2()(*(p->next+1)))) ))
return;		/* no repetition, we're done */
NEXT()(p->next++);

REQUIRE(!wascaret, REG_BADRPT)do { if (!(!wascaret)) seterr(p, (13)); } while (0);
switch (c) {
case '*':	/* implemented as +? */
/* this case does not require the (y|) trick, noKLUDGE */
INSERT(OPLUS_, pos)doinsert(p, (sop)((9LU<<((unsigned)27))), (p->slen)-
(pos)+1, pos);
ASTERN(O_PLUS, pos)doemit(p, (sop)((10LU<<((unsigned)27))), (size_t)((p->
slen)-pos));
INSERT(OQUEST_, pos)doinsert(p, (sop)((11LU<<((unsigned)27))), (p->slen)
-(pos)+1, pos);
ASTERN(O_QUEST, pos)doemit(p, (sop)((12LU<<((unsigned)27))), (size_t)((p->
slen)-pos));
break;
case '+':
INSERT(OPLUS_, pos)doinsert(p, (sop)((9LU<<((unsigned)27))), (p->slen)-
(pos)+1, pos);
ASTERN(O_PLUS, pos)doemit(p, (sop)((10LU<<((unsigned)27))), (size_t)((p->
slen)-pos));
break;
case '?':
/* KLUDGE: emit y? as (y|) until subtle bug gets fixed */
INSERT(OCH_, pos)doinsert(p, (sop)((15LU<<((unsigned)27))), (p->slen)
-(pos)+1, pos);		/* offset slightly wrong */
ASTERN(OOR1, pos)doemit(p, (sop)((16LU<<((unsigned)27))), (size_t)((p->
slen)-pos));		/* this one's right */
AHEAD(pos)dofwd(p, pos, (p->slen)-(pos));			/* fix the OCH_ */
EMIT(OOR2, 0)doemit(p, (sop)((17LU<<((unsigned)27))), (size_t)(0));			/* offset very wrong... */
AHEAD(THERE())dofwd(p, (p->slen - 1), (p->slen)-((p->slen - 1)));			/* ...so fix it */
ASTERN(O_CH, THERETHERE())doemit(p, (sop)((18LU<<((unsigned)27))), (size_t)((p->
slen)-(p->slen - 2)));
break;
case '{':
count = p_count(p);
if (EAT(',')((((p->end - p->next > 0) && (*p->next) ==
 (','))) ? ((p->next++), 1) : 0)) {
	if (isdigit((uch)PEEK()(*p->next))) {
		count2 = p_count(p);
		REQUIRE(count <= count2, REG_BADBR)do { if (!(count <= count2)) seterr(p, (10)); } while (0);
	} else		/* single number with comma */
		count2 = INFINITY(255 + 1);
} else		/* just a single number */
	count2 = count;
repeat(p, pos, count, count2);
if (!EAT('}')((((p->end - p->next > 0) && (*p->next) ==
 ('}'))) ? ((p->next++), 1) : 0)) {	/* error heuristics */
	while (MORE()(p->end - p->next > 0) && PEEK()(*p->next) != '}')
		NEXT()(p->next++);
	REQUIRE(MORE(), REG_EBRACE)do { if (!((p->end - p->next > 0))) seterr(p, (9)); }
 while (0);
	SETERROR(REG_BADBR)seterr(p, (10));
}
break;
}

if (!MORE()(p->end - p->next > 0))
return;
c = PEEK()(*p->next);
if (!( c == '*' || c == '+' || c == '?' ||
		(c == '{' && MORE2()(p->end - p->next > 1) && isdigit((uch)PEEK2()(*(p->next+1)))) ) )
return;
SETERROR(REG_BADRPT)seterr(p, (13));
410}

412/*
- p_str - string (no metacharacters) "parser"
*/
415static void
416p_str(struct parse *p)
417{
REQUIRE(MORE(), REG_EMPTY)do { if (!((p->end - p->next > 0))) seterr(p, (14));
 } while (0);
while (MORE()(p->end - p->next > 0))
ordinary(p, GETNEXT()(*p->next++));
421}

423/*
- p_bre - BRE parser top level, anchoring and concatenation
* Giving end1 as OUT essentially eliminates the end1/end2 check.
*
* This implementation is a bit of a kludge, in that a trailing $ is first
* taken as an ordinary character and then revised to be an anchor.  The
* only undesirable side effect is that '$' gets included as a character
* category in such cases.  This is fairly harmless; not worth fixing.
* The amount of lookahead needed to avoid this kludge is excessive.
*/
433static void
434p_bre(struct parse *p,
  int end1,		/* first terminating character */
  int end2)		/* second terminating character */
437{
sopno start = HERE()(p->slen);
int first = 1;			/* first subexpression? */
int wasdollar = 0;

if (EAT('^')((((p->end - p->next > 0) && (*p->next) ==
 ('^'))) ? ((p->next++), 1) : 0)) {
EMIT(OBOL, 0)doemit(p, (sop)((3LU<<((unsigned)27))), (size_t)(0));
p->g->iflags |= USEBOL01;
p->g->nbol++;
}
while (MORE()(p->end - p->next > 0) && !SEETWO(end1, end2)((p->end - p->next > 1) && (*p->next) == (
end1) && (*(p->next+1)) == (end2))) {
wasdollar = p_simp_re(p, first);
first = 0;
}
if (wasdollar) {	/* oops, that was a trailing anchor */
DROP(1)(p->slen -= (1));
EMIT(OEOL, 0)doemit(p, (sop)((4LU<<((unsigned)27))), (size_t)(0));
p->g->iflags |= USEEOL02;
p->g->neol++;
}

REQUIRE(HERE() != start, REG_EMPTY)do { if (!((p->slen) != start)) seterr(p, (14)); } while (
0);	/* require nonempty */
459}

461/*
- p_simp_re - parse a simple RE, an atom possibly followed by a repetition
*/
464static int			/* was the simple RE an unbackslashed $? */
465p_simp_re(struct parse *p,
  int starordinary)		/* is a leading * an ordinary character? */
467{
int c;
int count;
int count2;
sopno pos;
int i;
sopno subno;
474#	define	BACKSL(1<<8)	(1<<CHAR_BIT8)

pos = HERE()(p->slen);		/* repetion op, if any, covers from here */

assert(MORE())((void)0);		/* caller should have ensured this */
c = GETNEXT()(*p->next++);
if (c == '\\') {
REQUIRE(MORE(), REG_EESCAPE)do { if (!((p->end - p->next > 0))) seterr(p, (5)); }
 while (0);
c = BACKSL(1<<8) | GETNEXT()(*p->next++);
}
switch (c) {
case '.':
if (p->g->cflags&REG_NEWLINE0010)
	nonnewline(p);
else
	EMIT(OANY, 0)doemit(p, (sop)((5LU<<((unsigned)27))), (size_t)(0));
break;
case '[':
p_bracket(p);
break;
case BACKSL(1<<8)|'<':
EMIT(OBOW, 0)doemit(p, (sop)((19LU<<((unsigned)27))), (size_t)(0));
break;
case BACKSL(1<<8)|'>':
EMIT(OEOW, 0)doemit(p, (sop)((20LU<<((unsigned)27))), (size_t)(0));
break;
case BACKSL(1<<8)|'{':
SETERROR(REG_BADRPT)seterr(p, (13));
break;
case BACKSL(1<<8)|'(':
p->g->nsub++;
subno = p->g->nsub;
if (subno < NPAREN10)
	p->pbegin[subno] = HERE()(p->slen);
EMIT(OLPAREN, subno)doemit(p, (sop)((13LU<<((unsigned)27))), (size_t)(subno
));
/* the MORE here is an error heuristic */
if (MORE()(p->end - p->next > 0) && !SEETWO('\\', ')')((p->end - p->next > 1) && (*p->next) == (
'\\') && (*(p->next+1)) == (')')))
	p_bre(p, '\\', ')');
if (subno < NPAREN10) {
	p->pend[subno] = HERE()(p->slen);
	assert(p->pend[subno] != 0)((void)0);
}
EMIT(ORPAREN, subno)doemit(p, (sop)((14LU<<((unsigned)27))), (size_t)(subno
));
REQUIRE(EATTWO('\\', ')'), REG_EPAREN)do { if (!(((((p->end - p->next > 1) && (*p->
next) == ('\\') && (*(p->next+1)) == (')'))) ? ((p
->next += 2), 1) : 0))) seterr(p, (8)); } while (0);
break;
case BACKSL(1<<8)|')':	/* should not get here -- must be user */
case BACKSL(1<<8)|'}':
SETERROR(REG_EPAREN)seterr(p, (8));
break;
case BACKSL(1<<8)|'1':
case BACKSL(1<<8)|'2':
case BACKSL(1<<8)|'3':
case BACKSL(1<<8)|'4':
case BACKSL(1<<8)|'5':
case BACKSL(1<<8)|'6':
case BACKSL(1<<8)|'7':
case BACKSL(1<<8)|'8':
case BACKSL(1<<8)|'9':
i = (c&~BACKSL(1<<8)) - '0';
assert(i < NPAREN)((void)0);
if (p->pend[i] != 0) {
	assert(i <= p->g->nsub)((void)0);
	EMIT(OBACK_, i)doemit(p, (sop)((7LU<<((unsigned)27))), (size_t)(i));
	assert(p->pbegin[i] != 0)((void)0);
	assert(OP(p->strip[p->pbegin[i]]) == OLPAREN)((void)0);
	assert(OP(p->strip[p->pend[i]]) == ORPAREN)((void)0);
	(void) dupl(p, p->pbegin[i]+1, p->pend[i]);
	EMIT(O_BACK, i)doemit(p, (sop)((8LU<<((unsigned)27))), (size_t)(i));
} else
	SETERROR(REG_ESUBREG)seterr(p, (6));
p->g->backrefs = 1;
break;
case '*':
REQUIRE(starordinary, REG_BADRPT)do { if (!(starordinary)) seterr(p, (13)); } while (0);
/* FALLTHROUGH */
default:
if (p->error != 0)
	return(0);	/* Definitely not $... */
ordinary(p, (char)c);
break;
}

if (EAT('*')((((p->end - p->next > 0) && (*p->next) ==
 ('*'))) ? ((p->next++), 1) : 0)) {		/* implemented as +? */
/* this case does not require the (y|) trick, noKLUDGE */
INSERT(OPLUS_, pos)doinsert(p, (sop)((9LU<<((unsigned)27))), (p->slen)-
(pos)+1, pos);
ASTERN(O_PLUS, pos)doemit(p, (sop)((10LU<<((unsigned)27))), (size_t)((p->
slen)-pos));
INSERT(OQUEST_, pos)doinsert(p, (sop)((11LU<<((unsigned)27))), (p->slen)
-(pos)+1, pos);
ASTERN(O_QUEST, pos)doemit(p, (sop)((12LU<<((unsigned)27))), (size_t)((p->
slen)-pos));
} else if (EATTWO('\\', '{')((((p->end - p->next > 1) && (*p->next) ==
 ('\\') && (*(p->next+1)) == ('{'))) ? ((p->next
 += 2), 1) : 0)) {
count = p_count(p);
if (EAT(',')((((p->end - p->next > 0) && (*p->next) ==
 (','))) ? ((p->next++), 1) : 0)) {
	if (MORE()(p->end - p->next > 0) && isdigit((uch)PEEK()(*p->next))) {
		count2 = p_count(p);
		REQUIRE(count <= count2, REG_BADBR)do { if (!(count <= count2)) seterr(p, (10)); } while (0);
	} else		/* single number with comma */
		count2 = INFINITY(255 + 1);
} else		/* just a single number */
	count2 = count;
repeat(p, pos, count, count2);
if (!EATTWO('\\', '}')((((p->end - p->next > 1) && (*p->next) ==
 ('\\') && (*(p->next+1)) == ('}'))) ? ((p->next
 += 2), 1) : 0)) {	/* error heuristics */
	while (MORE()(p->end - p->next > 0) && !SEETWO('\\', '}')((p->end - p->next > 1) && (*p->next) == (
'\\') && (*(p->next+1)) == ('}')))
		NEXT()(p->next++);
	REQUIRE(MORE(), REG_EBRACE)do { if (!((p->end - p->next > 0))) seterr(p, (9)); }
 while (0);
	SETERROR(REG_BADBR)seterr(p, (10));
}
} else if (c == '$')	/* $ (but not \$) ends it */
return(1);

return(0);
583}

585/*
- p_count - parse a repetition count
*/
588static int			/* the value */
589p_count(struct parse *p)
590{
int count = 0;
int ndigits = 0;

while (MORE()(p->end - p->next > 0) && isdigit((uch)PEEK()(*p->next)) && count <= DUPMAX255) {
count = count*10 + (GETNEXT()(*p->next++) - '0');
ndigits++;
}

REQUIRE(ndigits > 0 && count <= DUPMAX, REG_BADBR)do { if (!(ndigits > 0 && count <= 255)) seterr
(p, (10)); } while (0);
return(count);
601}

603/*
- p_bracket - parse a bracketed character list
*
* Note a significant property of this code:  if the allocset() did SETERROR,
* no set operations are done.
*/
609static void
610p_bracket(struct parse *p)
611{
cset *cs;
int invert = 0;

/* Dept of Truly Sickening Special-Case Kludges */
if (p->end - p->next > 5) {
if (strncmp(p->next, "[:<:]]", 6) == 0) {
	EMIT(OBOW, 0)doemit(p, (sop)((19LU<<((unsigned)27))), (size_t)(0));
	NEXTn(6)(p->next += (6));
	return;
}
if (strncmp(p->next, "[:>:]]", 6) == 0) {
	EMIT(OEOW, 0)doemit(p, (sop)((20LU<<((unsigned)27))), (size_t)(0));
	NEXTn(6)(p->next += (6));
	return;
}
}

if ((cs = allocset(p)) == NULL((void *)0)) {
/* allocset did set error status in p */
return;
}

if (EAT('^')((((p->end - p->next > 0) && (*p->next) ==
 ('^'))) ? ((p->next++), 1) : 0))
invert++;	/* make note to invert set at end */
if (EAT(']')((((p->end - p->next > 0) && (*p->next) ==
 (']'))) ? ((p->next++), 1) : 0))
CHadd(cs, ']');
else if (EAT('-')((((p->end - p->next > 0) && (*p->next) ==
 ('-'))) ? ((p->next++), 1) : 0))
CHadd(cs, '-');
while (MORE()(p->end - p->next > 0) && PEEK()(*p->next) != ']' && !SEETWO('-', ']')((p->end - p->next > 1) && (*p->next) == (
'-') && (*(p->next+1)) == (']')))
p_b_term(p, cs);
if (EAT('-')((((p->end - p->next > 0) && (*p->next) ==
 ('-'))) ? ((p->next++), 1) : 0))
CHadd(cs, '-');
REQUIRE(MORE() && GETNEXT() == ']', REG_EBRACK)do { if (!((p->end - p->next > 0) && (*p->
next++) == ']')) seterr(p, (7)); } while (0);

if (p->error != 0) {	/* don't mess things up further */
freeset(p, cs);
return;
}

if (p->g->cflags&REG_ICASE0002) {
int i;
int ci;

for (i = p->g->csetsize - 1; i >= 0; i--)
	if (CHIN(cs, i) && isalpha(i)) {
		ci = othercase(i);
		if (ci != i)
			CHadd(cs, ci);
	}
}
if (invert) {
int i;

for (i = p->g->csetsize - 1; i >= 0; i--)
	if (CHIN(cs, i))
		CHsub(cs, i);
	else
		CHadd(cs, i);
if (p->g->cflags&REG_NEWLINE0010)
	CHsub(cs, '\n');
}

if (nch(p, cs) == 1) {		/* optimize singleton sets */
ordinary(p, firstch(p, cs));
freeset(p, cs);
} else
EMIT(OANYOF, freezeset(p, cs))doemit(p, (sop)((6LU<<((unsigned)27))), (size_t)(freezeset
(p, cs)));
679}

681/*
- p_b_term - parse one term of a bracketed character list
*/
684static void
685p_b_term(struct parse *p, cset *cs)
686{
char c;
char start, finish;
int i;

/* classify what we've got */
switch ((MORE()(p->end - p->next > 0)) ? PEEK()(*p->next) : '\0') {
case '[':
c = (MORE2()(p->end - p->next > 1)) ? PEEK2()(*(p->next+1)) : '\0';
break;
case '-':
SETERROR(REG_ERANGE)seterr(p, (11));
return;			/* NOTE RETURN */
break;
default:
c = '\0';
break;
}

switch (c) {
case ':':		/* character class */
NEXT2()(p->next += 2);
REQUIRE(MORE(), REG_EBRACK)do { if (!((p->end - p->next > 0))) seterr(p, (7)); }
 while (0);
c = PEEK()(*p->next);
REQUIRE(c != '-' && c != ']', REG_ECTYPE)do { if (!(c != '-' && c != ']')) seterr(p, (4)); } while
 (0);
p_b_cclass(p, cs);
REQUIRE(MORE(), REG_EBRACK)do { if (!((p->end - p->next > 0))) seterr(p, (7)); }
 while (0);
REQUIRE(EATTWO(':', ']'), REG_ECTYPE)do { if (!(((((p->end - p->next > 1) && (*p->
next) == (':') && (*(p->next+1)) == (']'))) ? ((p->
next += 2), 1) : 0))) seterr(p, (4)); } while (0);
break;
case '=':		/* equivalence class */
NEXT2()(p->next += 2);
REQUIRE(MORE(), REG_EBRACK)do { if (!((p->end - p->next > 0))) seterr(p, (7)); }
 while (0);
c = PEEK()(*p->next);
REQUIRE(c != '-' && c != ']', REG_ECOLLATE)do { if (!(c != '-' && c != ']')) seterr(p, (3)); } while
 (0);
p_b_eclass(p, cs);
REQUIRE(MORE(), REG_EBRACK)do { if (!((p->end - p->next > 0))) seterr(p, (7)); }
 while (0);
REQUIRE(EATTWO('=', ']'), REG_ECOLLATE)do { if (!(((((p->end - p->next > 1) && (*p->
next) == ('=') && (*(p->next+1)) == (']'))) ? ((p->
next += 2), 1) : 0))) seterr(p, (3)); } while (0);
break;
default:		/* symbol, ordinary character, or range */
725/* xxx revision needed for multichar stuff */
start = p_b_symbol(p);
if (SEE('-')((p->end - p->next > 0) && (*p->next) == (
'-')) && MORE2()(p->end - p->next > 1) && PEEK2()(*(p->next+1)) != ']') {
	/* range */
	NEXT()(p->next++);
	if (EAT('-')((((p->end - p->next > 0) && (*p->next) ==
 ('-'))) ? ((p->next++), 1) : 0))
		finish = '-';
	else
		finish = p_b_symbol(p);
} else
	finish = start;
736/* xxx what about signed chars here... */
REQUIRE(start <= finish, REG_ERANGE)do { if (!(start <= finish)) seterr(p, (11)); } while (0);
for (i = start; i <= finish; i++)
	CHadd(cs, i);
break;
}
742}

744/*
- p_b_cclass - parse a character-class name and deal with it
*/
747static void
748p_b_cclass(struct parse *p, cset *cs)
749{
const char *sp = p->next;
const struct cclass *cp;
size_t len;
const char *u;
char c;

while (MORE()(p->end - p->next > 0) && isalpha((uch)PEEK()(*p->next)))
NEXT()(p->next++);
len = p->next - sp;
for (cp = cclasses; cp->name != NULL((void *)0); cp++)
if (strncmp(cp->name, sp, len) == 0 && cp->name[len] == '\0')
	break;
if (cp->name == NULL((void *)0)) {
/* oops, didn't find it */
SETERROR(REG_ECTYPE)seterr(p, (4));
return;
}

u = cp->chars;
while ((c = *u++) != '\0')
CHadd(cs, c);
771}

773/*
- p_b_eclass - parse an equivalence-class name and deal with it
*
* This implementation is incomplete. xxx
*/
778static void
779p_b_eclass(struct parse *p, cset *cs)
780{
char c;

c = p_b_coll_elem(p, '=');
CHadd(cs, c);
785}

787/*
- p_b_symbol - parse a character or [..]ed multicharacter collating symbol
*/
790static char			/* value of symbol */
791p_b_symbol(struct parse *p)
792{
char value;

REQUIRE(MORE(), REG_EBRACK)do { if (!((p->end - p->next > 0))) seterr(p, (7)); }
 while (0);
if (!EATTWO('[', '.')((((p->end - p->next > 1) && (*p->next) ==
 ('[') && (*(p->next+1)) == ('.'))) ? ((p->next
 += 2), 1) : 0))
return(GETNEXT()(*p->next++));

/* collating symbol */
value = p_b_coll_elem(p, '.');
REQUIRE(EATTWO('.', ']'), REG_ECOLLATE)do { if (!(((((p->end - p->next > 1) && (*p->
next) == ('.') && (*(p->next+1)) == (']'))) ? ((p->
next += 2), 1) : 0))) seterr(p, (3)); } while (0);
return(value);
803}

805/*
- p_b_coll_elem - parse a collating-element name and look it up
*/
808static char			/* value of collating element */
809p_b_coll_elem(struct parse *p,
  int endc)			/* name ended by endc,']' */
811{
const char *sp = p->next;
const struct cname *cp;
size_t len;

while (MORE()(p->end - p->next > 0) && !SEETWO(endc, ']')((p->end - p->next > 1) && (*p->next) == (
endc) && (*(p->next+1)) == (']')))
NEXT()(p->next++);
if (!MORE()(p->end - p->next > 0)) {
SETERROR(REG_EBRACK)seterr(p, (7));
return(0);
}
len = p->next - sp;
for (cp = cnames; cp->name != NULL((void *)0); cp++)
if (strncmp(cp->name, sp, len) == 0 && strlen(cp->name) == len)
	return(cp->code);	/* known name */
if (len == 1)
return(*sp);	/* single character */
SETERROR(REG_ECOLLATE)seterr(p, (3));			/* neither */
return(0);
830}

832/*
- othercase - return the case counterpart of an alphabetic
*/
835static char			/* if no counterpart, return ch */
836othercase(int ch)
837{
ch = (uch)ch;
assert(isalpha(ch))((void)0);
if (isupper(ch))
return ((uch)tolower(ch));
else if (islower(ch))
return ((uch)toupper(ch));
else			/* peculiar, but could happen */
return(ch);
846}

848/*
- bothcases - emit a dualcase version of a two-case character
*
* Boy, is this implementation ever a kludge...
*/
853static void
854bothcases(struct parse *p, int ch)
855{
const char *oldnext = p->next;
const char *oldend = p->end;
char bracket[3];

ch = (uch)ch;
assert(othercase(ch) != ch)((void)0);	/* p_bracket() would recurse */
p->next = bracket;
p->end = bracket+2;
bracket[0] = ch;
bracket[1] = ']';
bracket[2] = '\0';
p_bracket(p);
assert(p->next == bracket+2)((void)0);
p->next = oldnext;
p->end = oldend;
871}

873/*
- ordinary - emit an ordinary character
*/
876static void
877ordinary(struct parse *p, int ch)
878{
if ((p->g->cflags&REG_ICASE0002) && isalpha((uch)ch) && othercase(ch) != ch)
bothcases(p, ch);
else
EMIT(OCHAR, (uch)ch)doemit(p, (sop)((2LU<<((unsigned)27))), (size_t)((uch)ch
));
883}

885/*
* do something magic with this character, but only if it's extra magic
*/
888static void
889backslash(struct parse *p, int ch)
890{
switch (ch) {
case '<':
EMIT(OBOW, 0)doemit(p, (sop)((19LU<<((unsigned)27))), (size_t)(0));
break;
case '>':
EMIT(OEOW, 0)doemit(p, (sop)((20LU<<((unsigned)27))), (size_t)(0));
break;
default:
ordinary(p, ch);
break;
}
902}

904/*
- nonnewline - emit REG_NEWLINE version of OANY
*
* Boy, is this implementation ever a kludge...
*/
909static void
910nonnewline(struct parse *p)
911{
const char *oldnext = p->next;
const char *oldend = p->end;
static const char bracket[4] = { '^', '\n', ']', '\0' };

p->next = bracket;
p->end = bracket+3;
p_bracket(p);
assert(p->next == bracket+3)((void)0);
p->next = oldnext;
p->end = oldend;
922}

924/*
- repeat - generate code for a bounded repetition, recursively if needed
*/
927static void
928repeat(struct parse *p,
  sopno start,		/* operand from here to end of strip */
  int from,			/* repeated from this number */
  int to)			/* to this number of times (maybe INFINITY) */
932{
sopno finish = HERE()(p->slen);
934#	define	N2	2
935#	define	INF3	3
936#	define	REP(f, t)((f)*8 + (t))	((f)*8 + (t))
937#	define	MAP(n)(((n) <= 1) ? (n) : ((n) == (255 + 1)) ? 3 : 2)	(((n) <= 1) ? (n) : ((n) == INFINITY(255 + 1)) ? INF3 : N2)
sopno copy;

if (p->error != 0)	/* head off possible runaway recursion */
return;

assert(from <= to)((void)0);

switch (REP(MAP(from), MAP(to))(((((from) <= 1) ? (from) : ((from) == (255 + 1)) ? 3 : 2)
)*8 + ((((to) <= 1) ? (to) : ((to) == (255 + 1)) ? 3 : 2))
)) {
case REP(0, 0)((0)*8 + (0)):			/* must be user doing this */
DROP(finish-start)(p->slen -= (finish-start));	/* drop the operand */
break;
case REP(0, 1)((0)*8 + (1)):			/* as x{1,1}? */
case REP(0, N)((0)*8 + (2)):			/* as x{1,n}? */
case REP(0, INF)((0)*8 + (3)):		/* as x{1,}? */
/* KLUDGE: emit y? as (y|) until subtle bug gets fixed */
INSERT(OCH_, start)doinsert(p, (sop)((15LU<<((unsigned)27))), (p->slen)
-(start)+1, start);		/* offset is wrong... */
repeat(p, start+1, 1, to);
ASTERN(OOR1, start)doemit(p, (sop)((16LU<<((unsigned)27))), (size_t)((p->
slen)-start));
AHEAD(start)dofwd(p, start, (p->slen)-(start));			/* ... fix it */
EMIT(OOR2, 0)doemit(p, (sop)((17LU<<((unsigned)27))), (size_t)(0));
AHEAD(THERE())dofwd(p, (p->slen - 1), (p->slen)-((p->slen - 1)));
ASTERN(O_CH, THERETHERE())doemit(p, (sop)((18LU<<((unsigned)27))), (size_t)((p->
slen)-(p->slen - 2)));
break;
case REP(1, 1)((1)*8 + (1)):			/* trivial case */
/* done */
break;
case REP(1, N)((1)*8 + (2)):			/* as x?x{1,n-1} */
/* KLUDGE: emit y? as (y|) until subtle bug gets fixed */
INSERT(OCH_, start)doinsert(p, (sop)((15LU<<((unsigned)27))), (p->slen)
-(start)+1, start);
ASTERN(OOR1, start)doemit(p, (sop)((16LU<<((unsigned)27))), (size_t)((p->
slen)-start));
AHEAD(start)dofwd(p, start, (p->slen)-(start));
EMIT(OOR2, 0)doemit(p, (sop)((17LU<<((unsigned)27))), (size_t)(0));			/* offset very wrong... */
AHEAD(THERE())dofwd(p, (p->slen - 1), (p->slen)-((p->slen - 1)));			/* ...so fix it */
ASTERN(O_CH, THERETHERE())doemit(p, (sop)((18LU<<((unsigned)27))), (size_t)((p->
slen)-(p->slen - 2)));
copy = dupl(p, start+1, finish+1);
assert(copy == finish+4)((void)0);
repeat(p, copy, 1, to-1);
break;
case REP(1, INF)((1)*8 + (3)):		/* as x+ */
INSERT(OPLUS_, start)doinsert(p, (sop)((9LU<<((unsigned)27))), (p->slen)-
(start)+1, start);
ASTERN(O_PLUS, start)doemit(p, (sop)((10LU<<((unsigned)27))), (size_t)((p->
slen)-start));
break;
case REP(N, N)((2)*8 + (2)):			/* as xx{m-1,n-1} */
copy = dupl(p, start, finish);
repeat(p, copy, from-1, to-1);
break;
case REP(N, INF)((2)*8 + (3)):		/* as xx{n-1,INF} */
copy = dupl(p, start, finish);
repeat(p, copy, from-1, to);
break;
default:			/* "can't happen" */
SETERROR(REG_ASSERT)seterr(p, (15));	/* just in case */
break;
}
992}

994/*
- seterr - set an error condition
*/
997static void
998seterr(struct parse *p, int e)
999{
if (p->error == 0)	/* keep earliest error condition */
p->error = e;
p->next = nuls;		/* try to bring things to a halt */
p->end = nuls;
1004}

1006/*
- allocset - allocate a set of characters for []
*/
1009static cset *
1010allocset(struct parse *p)
1011{
int no = p->g->ncsets++;
size_t nc;
size_t nbytes;
cset *cs;
size_t css = (size_t)p->g->csetsize;
int i;

if (no >= p->ncsalloc) {	/* need another column of space */
void *ptr;

p->ncsalloc += CHAR_BIT8;
nc = p->ncsalloc;
assert(nc % CHAR_BIT == 0)((void)0);

ptr = reallocarray(p->g->sets, nc, sizeof(cset));
if (ptr == NULL((void *)0))
	goto nomem;
p->g->sets = ptr;

ptr = reallocarray(p->g->setbits, nc / CHAR_BIT8, css);
if (ptr == NULL((void *)0))
	goto nomem;
nbytes = (nc / CHAR_BIT8) * css;
p->g->setbits = ptr;

for (i = 0; i < no; i++)
	p->g->sets[i].ptr = p->g->setbits + css*(i/CHAR_BIT8);

(void) memset((char *)p->g->setbits + (nbytes - css), 0, css);
}
/* XXX should not happen */
if (p->g->sets == NULL((void *)0) || p->g->setbits == NULL((void *)0))
goto nomem;

cs = &p->g->sets[no];
cs->ptr = p->g->setbits + css*((no)/CHAR_BIT8);
cs->mask = 1 << ((no) % CHAR_BIT8);
cs->hash = 0;

return(cs);
1052nomem:
free(p->g->sets);
p->g->sets = NULL((void *)0);
free(p->g->setbits);
p->g->setbits = NULL((void *)0);

SETERROR(REG_ESPACE)seterr(p, (12));
/* caller's responsibility not to do set ops */
return(NULL((void *)0));
1061}

1063/*
- freeset - free a now-unused set
*/
1066static void
1067freeset(struct parse *p, cset *cs)
1068{
int i;
cset *top = &p->g->sets[p->g->ncsets];
size_t css = (size_t)p->g->csetsize;

for (i = 0; i < css; i++)
CHsub(cs, i);
if (cs == top-1)	/* recover only the easy case */
p->g->ncsets--;
1077}

1079/*
- freezeset - final processing on a set of characters
*
* The main task here is merging identical sets.  This is usually a waste
* of time (although the hash code minimizes the overhead), but can win
* big if REG_ICASE is being used.  REG_ICASE, by the way, is why the hash
* is done using addition rather than xor -- all ASCII [aA] sets xor to
* the same value!
*/
1088static int			/* set number */
1089freezeset(struct parse *p, cset *cs)
1090{
uch h = cs->hash;
int i;
cset *top = &p->g->sets[p->g->ncsets];
cset *cs2;
size_t css = (size_t)p->g->csetsize;

/* look for an earlier one which is the same */
for (cs2 = &p->g->sets[0]; cs2 < top; cs2++)
if (cs2->hash == h && cs2 != cs) {
	/* maybe */
	for (i = 0; i < css; i++)
		if (CHIN(cs2, i) != CHIN(cs, i))
			break;		/* no */
	if (i == css)
		break;			/* yes */
}

if (cs2 < top) {	/* found one */
freeset(p, cs);
cs = cs2;
}

return((int)(cs - p->g->sets));
1114}

1116/*
- firstch - return first character in a set (which must have at least one)
*/
1119static int			/* character; there is no "none" value */
1120firstch(struct parse *p, cset *cs)
1121{
int i;
size_t css = (size_t)p->g->csetsize;

for (i = 0; i < css; i++)
if (CHIN(cs, i))
	return((char)i);
assert(never)((void)0);
return(0);		/* arbitrary */
1130}

1132/*
- nch - number of characters in a set
*/
1135static int
1136nch(struct parse *p, cset *cs)
1137{
int i;
size_t css = (size_t)p->g->csetsize;
int n = 0;

for (i = 0; i < css; i++)
if (CHIN(cs, i))
	n++;
return(n);
1146}

1148/*
- dupl - emit a duplicate of a bunch of sops
*/
1151static sopno			/* start of duplicate */
1152dupl(struct parse *p,
  sopno start,		/* from here */
  sopno finish)		/* to this less one */
1155{
sopno ret = HERE()(p->slen);
sopno len = finish - start;

assert(finish >= start)((void)0);
if (len == 0)
return(ret);
if (!enlarge(p, p->ssize + len)) /* this many unexpected additions */
return(ret);
(void) memcpy(p->strip + p->slen, p->strip + start, len * sizeof(sop));
p->slen += len;
return(ret);
1167}

1169/*
- doemit - emit a strip operator
*
* It might seem better to implement this as a macro with a function as
* hard-case backup, but it's just too big and messy unless there are
* some changes to the data structures.  Maybe later.
*/
1176static void
1177doemit(struct parse *p, sop op, size_t opnd)
1178{
/* avoid making error situations worse */
if (p->error != 0)
return;

/* deal with oversize operands ("can't happen", more or less) */
assert(opnd < 1<<OPSHIFT)((void)0);

/* deal with undersized strip */
if (p->slen >= p->ssize)
if (!enlarge(p, (p->ssize+1) / 2 * 3))	/* +50% */
	return;

/* finally, it's all reduced to the easy case */
p->strip[p->slen++] = SOP(op, opnd)((op)|(opnd));
1193}

1195/*
- doinsert - insert a sop into the strip
*/
1198static void
1199doinsert(struct parse *p, sop op, size_t opnd, sopno pos)
1200{
sopno sn;
sop s;
int i;

/* avoid making error situations worse */
if (p->error != 0)
return;

sn = HERE()(p->slen);
EMIT(op, opnd)doemit(p, (sop)(op), (size_t)(opnd));		/* do checks, ensure space */
assert(HERE() == sn+1)((void)0);
s = p->strip[sn];

/* adjust paren pointers */
assert(pos > 0)((void)0);
for (i = 1; i < NPAREN10; i++) {
if (p->pbegin[i] >= pos) {
	p->pbegin[i]++;
}
if (p->pend[i] >= pos) {
	p->pend[i]++;
}
}

memmove((char *)&p->strip[pos+1], (char *)&p->strip[pos],
				(HERE()(p->slen)-pos-1)*sizeof(sop));
p->strip[pos] = s;
1228}

1230/*
- dofwd - complete a forward reference
*/
1233static void
1234dofwd(struct parse *p, sopno pos, sop value)
1235{
/* avoid making error situations worse */
if (p->error != 0)
return;

assert(value < 1<<OPSHIFT)((void)0);
p->strip[pos] = OP(p->strip[pos])((p->strip[pos])&0xf8000000LU) | value;
1242}

1244/*
- enlarge - enlarge the strip
*/
1247static int
1248enlarge(struct parse *p, sopno size)
1249{
sop *sp;

if (p->ssize >= size)
return 1;

sp = reallocarray(p->strip, size, sizeof(sop));
if (sp == NULL((void *)0)) {
SETERROR(REG_ESPACE)seterr(p, (12));
return 0;
}
p->strip = sp;
p->ssize = size;
return 1;
1263}

1265/*
- stripsnug - compact the strip
*/
1268static void
1269stripsnug(struct parse *p, struct re_guts *g)
1270{
g->nstates = p->slen;
g->strip = reallocarray(p->strip, p->slen, sizeof(sop));
if (g->strip == NULL((void *)0)) {
SETERROR(REG_ESPACE)seterr(p, (12));
g->strip = p->strip;
}
1277}

1279/*
- findmust - fill in must and mlen with longest mandatory literal string
*
* This algorithm could do fancy things like analyzing the operands of |
* for common subsequences.  Someday.  This code is simple and finds most
* of the interesting cases.
*
* Note that must and mlen got initialized during setup.
*/
1288static void
1289findmust(struct parse *p, struct re_guts *g)
1290{
sop *scan;
sop *start;    /* start initialized in the default case, after that */
1
'start' declared without an initial value→
sop *newstart; /* newstart was initialized in the OCHAR case */
sopno newlen;
sop s;
char *cp;
sopno i;

/* avoid making error situations worse */
if (p->error != 0)
2
←
Assuming field 'error' is equal to 0→
3
←
Taking false branch→
return;

/* find the longest OCHAR sequence in strip */
newlen = 0;
scan = g->strip + 1;
do {
9
←
Loop condition is false.  Exiting loop→
s = *scan++;
switch (OP(s)((s)&0xf8000000LU)) {
4
←
Control jumps to the 'default' case at line 1332→
case OCHAR(2LU<<((unsigned)27)):		/* sequence member */
	if (newlen == 0)		/* new sequence */
		newstart = scan - 1;
	newlen++;
	break;
case OPLUS_(9LU<<((unsigned)27)):		/* things that don't break one */
case OLPAREN(13LU<<((unsigned)27)):
case ORPAREN(14LU<<((unsigned)27)):
	break;
case OQUEST_(11LU<<((unsigned)27)):		/* things that must be skipped */
case OCH_(15LU<<((unsigned)27)):
	scan--;
	do {
		scan += OPND(s)((s)&0x07ffffffLU);
		s = *scan;
		/* assert() interferes w debug printouts */
		if (OP(s)((s)&0xf8000000LU) != O_QUEST(12LU<<((unsigned)27)) && OP(s)((s)&0xf8000000LU) != O_CH(18LU<<((unsigned)27)) &&
					OP(s)((s)&0xf8000000LU) != OOR2(17LU<<((unsigned)27))) {
			g->iflags |= BAD04;
			return;
		}
	} while (OP(s)((s)&0xf8000000LU) != O_QUEST(12LU<<((unsigned)27)) && OP(s)((s)&0xf8000000LU) != O_CH(18LU<<((unsigned)27)));
	/* fallthrough */
default:		/* things that break a sequence */
	if (newlen > g->mlen) {		/* ends one */
5
←
Assuming 'newlen' is <= field 'mlen'→
6
←
Taking false branch→
		start = newstart;
		g->mlen = newlen;
	}
	newlen = 0;
	break;
7
←
 Execution continues on line 1340→
}
} while (OP(s)((s)&0xf8000000LU) != OEND(1LU<<((unsigned)27)));
8
←
Assuming the condition is false→

if (g->mlen == 0)		/* there isn't one */
10
←
Assuming field 'mlen' is not equal to 0→
11
←
Taking false branch→
return;

/* turn it into a character string */
g->must = malloc((size_t)g->mlen + 1);
if (g->must == NULL((void *)0)) {		/* argh; just forget it */
12
←
Assuming field 'must' is not equal to NULL→
13
←
Taking false branch→
g->mlen = 0;
return;
}
cp = g->must;
scan = start;
14
←
Assigned value is garbage or undefined
for (i = g->mlen; i > 0; i--) {
while (OP(s = *scan++)((s = *scan++)&0xf8000000LU) != OCHAR(2LU<<((unsigned)27)))
	continue;
assert(cp < g->must + g->mlen)((void)0);
*cp++ = (char)OPND(s)((s)&0x07ffffffLU);
}
assert(cp == g->must + g->mlen)((void)0);
*cp = '\0';		/* just on general principles */
1361}

1363/*
- pluscount - count + nesting
*/
1366static sopno			/* nesting depth */
1367pluscount(struct parse *p, struct re_guts *g)
1368{
sop *scan;
sop s;
sopno plusnest = 0;
sopno maxnest = 0;

if (p->error != 0)
return(0);	/* there may not be an OEND */

scan = g->strip + 1;
do {
s = *scan++;
switch (OP(s)((s)&0xf8000000LU)) {
case OPLUS_(9LU<<((unsigned)27)):
	plusnest++;
	break;
case O_PLUS(10LU<<((unsigned)27)):
	if (plusnest > maxnest)
		maxnest = plusnest;
	plusnest--;
	break;
}
} while (OP(s)((s)&0xf8000000LU) != OEND(1LU<<((unsigned)27)));
if (plusnest != 0)
g->iflags |= BAD04;
return(maxnest);
1394}