Bug Summary

File:src/gnu/usr.bin/clang/libLLVM/../../../llvm/llvm/lib/Support/regcomp.c
Warning:line 1662, column 10
Dereference of null pointer

Annotated Source Code

Press '?' to see keyboard shortcuts

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 -ffunction-sections -fdata-sections -fcoverage-compilation-dir=/usr/src/gnu/usr.bin/clang/libLLVM/obj -resource-dir /usr/local/lib/clang/13.0.0 -I /usr/src/gnu/usr.bin/clang/libLLVM/../../../llvm/llvm/include/llvm/Transforms -I /usr/src/gnu/usr.bin/clang/libLLVM/obj/../include/llvm/AMDGPU -I /usr/src/gnu/usr.bin/clang/libLLVM/../../../llvm/llvm/lib/Target/AMDGPU -I /usr/src/gnu/usr.bin/clang/libLLVM/obj/../include/llvm/AMDGPU -I /usr/src/gnu/usr.bin/clang/libLLVM/../../../llvm/llvm/lib/Target/AMDGPU -I /usr/src/gnu/usr.bin/clang/libLLVM/obj/../include/llvm/AMDGPU -I /usr/src/gnu/usr.bin/clang/libLLVM/../../../llvm/llvm/lib/Target/AMDGPU -I /usr/src/gnu/usr.bin/clang/libLLVM/obj/../include/llvm/AMDGPU -I /usr/src/gnu/usr.bin/clang/libLLVM/../../../llvm/llvm/lib/Target/AMDGPU -I /usr/src/gnu/usr.bin/clang/libLLVM/obj/../include/llvm/AMDGPU -I /usr/src/gnu/usr.bin/clang/libLLVM/../../../llvm/llvm/lib/Target/AMDGPU -I /usr/src/gnu/usr.bin/clang/libLLVM/obj/../include/llvm/AMDGPU -I /usr/src/gnu/usr.bin/clang/libLLVM/../../../llvm/llvm/lib/Target/AMDGPU -I /usr/src/gnu/usr.bin/clang/libLLVM/../../../llvm/llvm/include/llvm/Analysis -I /usr/src/gnu/usr.bin/clang/libLLVM/../../../llvm/llvm/include/llvm/ASMParser -I /usr/src/gnu/usr.bin/clang/libLLVM/../../../llvm/llvm/include/llvm/BinaryFormat -I /usr/src/gnu/usr.bin/clang/libLLVM/../../../llvm/llvm/include/llvm/Bitcode -I /usr/src/gnu/usr.bin/clang/libLLVM/../../../llvm/llvm/include/llvm/Bitcode -I /usr/src/gnu/usr.bin/clang/libLLVM/../../../llvm/llvm/include/llvm/Bitstream -I /usr/src/gnu/usr.bin/clang/libLLVM/../../../llvm/llvm/include/llvm/Transforms -I /include/llvm/CodeGen -I /include/llvm/CodeGen/PBQP -I /usr/src/gnu/usr.bin/clang/libLLVM/obj/../include/llvm/IR -I /usr/src/gnu/usr.bin/clang/libLLVM/../../../llvm/llvm/include/llvm/IR -I /usr/src/gnu/usr.bin/clang/libLLVM/../../../llvm/llvm/include/llvm/Transforms -I /usr/src/gnu/usr.bin/clang/libLLVM/../../../llvm/llvm/include/llvm/Transforms/Coroutines -I /usr/src/gnu/usr.bin/clang/libLLVM/../../../llvm/llvm/include/llvm/ProfileData/Coverage -I /usr/src/gnu/usr.bin/clang/libLLVM/../../../llvm/llvm/include/llvm/DebugInfo/CodeView -I /usr/src/gnu/usr.bin/clang/libLLVM/../../../llvm/llvm/include/llvm/DebugInfo/DWARF -I /usr/src/gnu/usr.bin/clang/libLLVM/../../../llvm/llvm/include/llvm/DebugInfo -I /usr/src/gnu/usr.bin/clang/libLLVM/../../../llvm/llvm/include/llvm/DebugInfo/MSF -I /usr/src/gnu/usr.bin/clang/libLLVM/../../../llvm/llvm/include/llvm/DebugInfo/PDB -I /usr/src/gnu/usr.bin/clang/libLLVM/../../../llvm/llvm/include/llvm/Demangle -I /usr/src/gnu/usr.bin/clang/libLLVM/../../../llvm/llvm/include/llvm/ExecutionEngine -I /usr/src/gnu/usr.bin/clang/libLLVM/../../../llvm/llvm/include/llvm/ExecutionEngine/JITLink -I /usr/src/gnu/usr.bin/clang/libLLVM/../../../llvm/llvm/include/llvm/ExecutionEngine/Orc -I /usr/src/gnu/usr.bin/clang/libLLVM/../../../llvm/llvm/include/llvm/Frontend -I /usr/src/gnu/usr.bin/clang/libLLVM/../../../llvm/llvm/include/llvm/Frontend/OpenACC -I /usr/src/gnu/usr.bin/clang/libLLVM/../../../llvm/llvm/include/llvm/Frontend -I /usr/src/gnu/usr.bin/clang/libLLVM/../../../llvm/llvm/include/llvm/Frontend/OpenMP -I /include/llvm/CodeGen/GlobalISel -I /usr/src/gnu/usr.bin/clang/libLLVM/../../../llvm/llvm/include/llvm/IRReader -I /usr/src/gnu/usr.bin/clang/libLLVM/../../../llvm/llvm/include/llvm/Transforms -I /usr/src/gnu/usr.bin/clang/libLLVM/../../../llvm/llvm/include/llvm/Transforms/InstCombine -I /usr/src/gnu/usr.bin/clang/libLLVM/obj/../include/llvm/Transforms/InstCombine -I /usr/src/gnu/usr.bin/clang/libLLVM/../../../llvm/llvm/include/llvm/Transforms -I /usr/src/gnu/usr.bin/clang/libLLVM/../../../llvm/llvm/include/llvm/LTO -I /usr/src/gnu/usr.bin/clang/libLLVM/../../../llvm/llvm/include/llvm/Linker -I /usr/src/gnu/usr.bin/clang/libLLVM/../../../llvm/llvm/include/llvm/MC -I /usr/src/gnu/usr.bin/clang/libLLVM/../../../llvm/llvm/include/llvm/MC/MCParser -I /include/llvm/CodeGen/MIRParser -I /usr/src/gnu/usr.bin/clang/libLLVM/../../../llvm/llvm/include/llvm/Transforms -I /usr/src/gnu/usr.bin/clang/libLLVM/../../../llvm/llvm/include/llvm/Object -I /usr/src/gnu/usr.bin/clang/libLLVM/../../../llvm/llvm/include/llvm/Option -I /usr/src/gnu/usr.bin/clang/libLLVM/../../../llvm/llvm/include/llvm/Passes -I /usr/src/gnu/usr.bin/clang/libLLVM/../../../llvm/llvm/include/llvm/ -I /usr/src/gnu/usr.bin/clang/libLLVM/../../../llvm/llvm/include/llvm/ProfileData -I /usr/src/gnu/usr.bin/clang/libLLVM/../../../llvm/llvm/include/llvm/Transforms -I /usr/src/gnu/usr.bin/clang/libLLVM/../../../llvm/llvm/include/llvm/Transforms/Scalar -I /usr/src/gnu/usr.bin/clang/libLLVM/../../../llvm/llvm/include/llvm/ADT -I /usr/src/gnu/usr.bin/clang/libLLVM/../../../llvm/llvm/include/llvm/Support -I /usr/src/gnu/usr.bin/clang/libLLVM/../../../llvm/llvm/include/llvm/DebugInfo/Symbolize -I /usr/src/gnu/usr.bin/clang/libLLVM/../../../llvm/llvm/include/llvm/Target -I /usr/src/gnu/usr.bin/clang/libLLVM/../../../llvm/llvm/include/llvm/Transforms -I /usr/src/gnu/usr.bin/clang/libLLVM/../../../llvm/llvm/include/llvm/Transforms/Utils -I /usr/src/gnu/usr.bin/clang/libLLVM/../../../llvm/llvm/include/llvm/Transforms -I /usr/src/gnu/usr.bin/clang/libLLVM/../../../llvm/llvm/include/llvm/Transforms/Vectorize -I /usr/src/gnu/usr.bin/clang/libLLVM/obj/../include/llvm/X86 -I /usr/src/gnu/usr.bin/clang/libLLVM/../../../llvm/llvm/lib/Target/X86 -I /usr/src/gnu/usr.bin/clang/libLLVM/obj/../include/llvm/X86 -I /usr/src/gnu/usr.bin/clang/libLLVM/../../../llvm/llvm/lib/Target/X86 -I /usr/src/gnu/usr.bin/clang/libLLVM/obj/../include/llvm/X86 -I /usr/src/gnu/usr.bin/clang/libLLVM/../../../llvm/llvm/lib/Target/X86 -I /usr/src/gnu/usr.bin/clang/libLLVM/obj/../include/llvm/X86 -I /usr/src/gnu/usr.bin/clang/libLLVM/../../../llvm/llvm/lib/Target/X86 -I /usr/src/gnu/usr.bin/clang/libLLVM/obj/../include/llvm/X86 -I /usr/src/gnu/usr.bin/clang/libLLVM/../../../llvm/llvm/lib/Target/X86 -I /usr/src/gnu/usr.bin/clang/libLLVM/../../../llvm/llvm/include/llvm/Transforms -I /usr/src/gnu/usr.bin/clang/libLLVM/../../../llvm/llvm/include/llvm/Transforms/IPO -I /usr/src/gnu/usr.bin/clang/libLLVM/../../../llvm/llvm/include -I /usr/src/gnu/usr.bin/clang/libLLVM/../include -I /usr/src/gnu/usr.bin/clang/libLLVM/obj -I /usr/src/gnu/usr.bin/clang/libLLVM/obj/../include -D NDEBUG -D __STDC_LIMIT_MACROS -D __STDC_CONSTANT_MACROS -D __STDC_FORMAT_MACROS -D LLVM_PREFIX="/usr" -D PIC -internal-isystem /usr/local/lib/clang/13.0.0/include -internal-externc-isystem /usr/include -O2 -fdebug-compilation-dir=/usr/src/gnu/usr.bin/clang/libLLVM/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/gnu/usr.bin/clang/libLLVM/../../../llvm/llvm/lib/Support/regcomp.c
1/*-
2 * This code is derived from OpenBSD's libc/regex, original license follows:
3 *
4 * Copyright (c) 1992, 1993, 1994 Henry Spencer.
5 * Copyright (c) 1992, 1993, 1994
6 * The Regents of the University of California. All rights reserved.
7 *
8 * This code is derived from software contributed to Berkeley by
9 * Henry Spencer.
10 *
11 * Redistribution and use in source and binary forms, with or without
12 * modification, are permitted provided that the following conditions
13 * are met:
14 * 1. Redistributions of source code must retain the above copyright
15 * notice, this list of conditions and the following disclaimer.
16 * 2. Redistributions in binary form must reproduce the above copyright
17 * notice, this list of conditions and the following disclaimer in the
18 * documentation and/or other materials provided with the distribution.
19 * 3. Neither the name of the University nor the names of its contributors
20 * may be used to endorse or promote products derived from this software
21 * without specific prior written permission.
22 *
23 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
24 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
25 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
26 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
27 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
28 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
29 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
30 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
31 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
32 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
33 * SUCH DAMAGE.
34 *
35 * @(#)regcomp.c 8.5 (Berkeley) 3/20/94
36 */
37
38#include <sys/types.h>
39#include <stdint.h>
40#include <stdio.h>
41#include <string.h>
42#include <ctype.h>
43#include <limits.h>
44#include <stdlib.h>
45#include "regex_impl.h"
46
47#include "regutils.h"
48#include "regex2.h"
49
50#include "llvm/Config/config.h"
51#include "llvm/Support/Compiler.h"
52
53/* character-class table */
54static struct cclass {
55 const char *name;
56 const char *chars;
57 const char *multis;
58} cclasses[] = {
59 { "alnum", "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz\
600123456789", ""} ,
61 { "alpha", "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz",
62 ""} ,
63 { "blank", " \t", ""} ,
64 { "cntrl", "\007\b\t\n\v\f\r\1\2\3\4\5\6\16\17\20\21\22\23\24\
65\25\26\27\30\31\32\33\34\35\36\37\177", ""} ,
66 { "digit", "0123456789", ""} ,
67 { "graph", "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz\
680123456789!\"#$%&'()*+,-./:;<=>?@[\\]^_`{|}~",
69 ""} ,
70 { "lower", "abcdefghijklmnopqrstuvwxyz",
71 ""} ,
72 { "print", "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz\
730123456789!\"#$%&'()*+,-./:;<=>?@[\\]^_`{|}~ ",
74 ""} ,
75 { "punct", "!\"#$%&'()*+,-./:;<=>?@[\\]^_`{|}~",
76 ""} ,
77 { "space", "\t\n\v\f\r ", ""} ,
78 { "upper", "ABCDEFGHIJKLMNOPQRSTUVWXYZ",
79 ""} ,
80 { "xdigit", "0123456789ABCDEFabcdef",
81 ""} ,
82 { NULL((void*)0), 0, "" }
83};
84
85/* character-name table */
86static struct cname {
87 const char *name;
88 char code;
89} cnames[] = {
90 { "NUL", '\0' },
91 { "SOH", '\001' },
92 { "STX", '\002' },
93 { "ETX", '\003' },
94 { "EOT", '\004' },
95 { "ENQ", '\005' },
96 { "ACK", '\006' },
97 { "BEL", '\007' },
98 { "alert", '\007' },
99 { "BS", '\010' },
100 { "backspace", '\b' },
101 { "HT", '\011' },
102 { "tab", '\t' },
103 { "LF", '\012' },
104 { "newline", '\n' },
105 { "VT", '\013' },
106 { "vertical-tab", '\v' },
107 { "FF", '\014' },
108 { "form-feed", '\f' },
109 { "CR", '\015' },
110 { "carriage-return", '\r' },
111 { "SO", '\016' },
112 { "SI", '\017' },
113 { "DLE", '\020' },
114 { "DC1", '\021' },
115 { "DC2", '\022' },
116 { "DC3", '\023' },
117 { "DC4", '\024' },
118 { "NAK", '\025' },
119 { "SYN", '\026' },
120 { "ETB", '\027' },
121 { "CAN", '\030' },
122 { "EM", '\031' },
123 { "SUB", '\032' },
124 { "ESC", '\033' },
125 { "IS4", '\034' },
126 { "FS", '\034' },
127 { "IS3", '\035' },
128 { "GS", '\035' },
129 { "IS2", '\036' },
130 { "RS", '\036' },
131 { "IS1", '\037' },
132 { "US", '\037' },
133 { "space", ' ' },
134 { "exclamation-mark", '!' },
135 { "quotation-mark", '"' },
136 { "number-sign", '#' },
137 { "dollar-sign", '$' },
138 { "percent-sign", '%' },
139 { "ampersand", '&' },
140 { "apostrophe", '\'' },
141 { "left-parenthesis", '(' },
142 { "right-parenthesis", ')' },
143 { "asterisk", '*' },
144 { "plus-sign", '+' },
145 { "comma", ',' },
146 { "hyphen", '-' },
147 { "hyphen-minus", '-' },
148 { "period", '.' },
149 { "full-stop", '.' },
150 { "slash", '/' },
151 { "solidus", '/' },
152 { "zero", '0' },
153 { "one", '1' },
154 { "two", '2' },
155 { "three", '3' },
156 { "four", '4' },
157 { "five", '5' },
158 { "six", '6' },
159 { "seven", '7' },
160 { "eight", '8' },
161 { "nine", '9' },
162 { "colon", ':' },
163 { "semicolon", ';' },
164 { "less-than-sign", '<' },
165 { "equals-sign", '=' },
166 { "greater-than-sign", '>' },
167 { "question-mark", '?' },
168 { "commercial-at", '@' },
169 { "left-square-bracket", '[' },
170 { "backslash", '\\' },
171 { "reverse-solidus", '\\' },
172 { "right-square-bracket", ']' },
173 { "circumflex", '^' },
174 { "circumflex-accent", '^' },
175 { "underscore", '_' },
176 { "low-line", '_' },
177 { "grave-accent", '`' },
178 { "left-brace", '{' },
179 { "left-curly-bracket", '{' },
180 { "vertical-line", '|' },
181 { "right-brace", '}' },
182 { "right-curly-bracket", '}' },
183 { "tilde", '~' },
184 { "DEL", '\177' },
185 { NULL((void*)0), 0 }
186};
187
188/*
189 * parse structure, passed up and down to avoid global variables and
190 * other clumsinesses
191 */
192struct parse {
193 char *next; /* next character in RE */
194 char *end; /* end of string (-> NUL normally) */
195 int error; /* has an error been seen? */
196 sop *strip; /* malloced strip */
197 sopno ssize; /* malloced strip size (allocated) */
198 sopno slen; /* malloced strip length (used) */
199 int ncsalloc; /* number of csets allocated */
200 struct re_guts *g;
201# define NPAREN10 10 /* we need to remember () 1-9 for back refs */
202 sopno pbegin[NPAREN10]; /* -> ( ([0] unused) */
203 sopno pend[NPAREN10]; /* -> ) ([0] unused) */
204};
205
206static void p_ere(struct parse *, int);
207static void p_ere_exp(struct parse *);
208static void p_str(struct parse *);
209static void p_bre(struct parse *, int, int);
210static int p_simp_re(struct parse *, int);
211static int p_count(struct parse *);
212static void p_bracket(struct parse *);
213static void p_b_term(struct parse *, cset *);
214static void p_b_cclass(struct parse *, cset *);
215static void p_b_eclass(struct parse *, cset *);
216static char p_b_symbol(struct parse *);
217static char p_b_coll_elem(struct parse *, int);
218static char othercase(int);
219static void bothcases(struct parse *, int);
220static void ordinary(struct parse *, int);
221static void nonnewline(struct parse *);
222static void repeat(struct parse *, sopno, int, int);
223static int seterr(struct parse *, int);
224static cset *allocset(struct parse *);
225static void freeset(struct parse *, cset *);
226static int freezeset(struct parse *, cset *);
227static int firstch(struct parse *, cset *);
228static int nch(struct parse *, cset *);
229static void mcadd(struct parse *, cset *, const char *);
230static void mcinvert(struct parse *, cset *);
231static void mccase(struct parse *, cset *);
232static int isinsets(struct re_guts *, int);
233static int samesets(struct re_guts *, int, int);
234static void categorize(struct parse *, struct re_guts *);
235static sopno dupl(struct parse *, sopno, sopno);
236static void doemit(struct parse *, sop, size_t);
237static void doinsert(struct parse *, sop, size_t, sopno);
238static void dofwd(struct parse *, sopno, sop);
239static void enlarge(struct parse *, sopno);
240static void stripsnug(struct parse *, struct re_guts *);
241static void findmust(struct parse *, struct re_guts *);
242static sopno pluscount(struct parse *, struct re_guts *);
243
244static char nuls[10]; /* place to point scanner in event of error */
245
246/*
247 * macros for use with parse structure
248 * BEWARE: these know that the parse structure is named `p' !!!
249 */
250#define PEEK()(*p->next) (*p->next)
251#define PEEK2()(*(p->next+1)) (*(p->next+1))
252#define MORE()(p->next < p->end) (p->next < p->end)
253#define MORE2()(p->next+1 < p->end) (p->next+1 < p->end)
254#define SEE(c)((p->next < p->end) && (*p->next) == (c)) (MORE()(p->next < p->end) && PEEK()(*p->next) == (c))
255#define SEETWO(a, b)((p->next < p->end) && (p->next+1 < p->
end) && (*p->next) == (a) && (*(p->next
+1)) == (b)) (MORE()(p->next < p->end) && MORE2()(p->next+1 < p->end) && PEEK()(*p->next) == (a) && PEEK2()(*(p->next+1)) == (b))
256#define EAT(c)((((p->next < p->end) && (*p->next) == (c
))) ? ((p->next++), 1) : 0) ((SEE(c)((p->next < p->end) && (*p->next) == (c))) ? (NEXT()(p->next++), 1) : 0)
257#define EATTWO(a, b)((((p->next < p->end) && (p->next+1 < p
->end) && (*p->next) == (a) && (*(p->
next+1)) == (b))) ? ((p->next += 2), 1) : 0) ((SEETWO(a, b)((p->next < p->end) && (p->next+1 < p->
end) && (*p->next) == (a) && (*(p->next
+1)) == (b))) ? (NEXT2()(p->next += 2), 1) : 0)
258#define NEXT()(p->next++) (p->next++)
259#define NEXT2()(p->next += 2) (p->next += 2)
260#define NEXTn(n)(p->next += (n)) (p->next += (n))
261#define GETNEXT()(*p->next++) (*p->next++)
262#define SETERROR(e)seterr(p, (e)) seterr(p, (e))
263#define REQUIRE(co, e)(void)((co) || seterr(p, (e))) (void)((co) || SETERROR(e)seterr(p, (e)))
264#define MUSTSEE(c, e)((void)(((p->next < p->end) && (*p->next)
== (c)) || seterr(p, (e)))) (REQUIRE(MORE() && PEEK() == (c), e)(void)(((p->next < p->end) && (*p->next) ==
(c)) || seterr(p, (e))))
265#define MUSTEAT(c, e)((void)(((p->next < p->end) && (*p->next++
) == (c)) || seterr(p, (e)))) (REQUIRE(MORE() && GETNEXT() == (c), e)(void)(((p->next < p->end) && (*p->next++
) == (c)) || seterr(p, (e))))
266#define MUSTNOTSEE(c, e)((void)((!(p->next < p->end) || (*p->next) != (c)
) || seterr(p, (e)))) (REQUIRE(!MORE() || PEEK() != (c), e)(void)((!(p->next < p->end) || (*p->next) != (c))
|| seterr(p, (e))))
267#define EMIT(op, sopnd)doemit(p, (sop)(op), (size_t)(sopnd)) doemit(p, (sop)(op), (size_t)(sopnd))
268#define INSERT(op, pos)doinsert(p, (sop)(op), (p->slen)-(pos)+1, pos) doinsert(p, (sop)(op), HERE()(p->slen)-(pos)+1, pos)
269#define AHEAD(pos)dofwd(p, pos, (p->slen)-(pos)) dofwd(p, pos, HERE()(p->slen)-(pos))
270#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))
271#define HERE()(p->slen) (p->slen)
272#define THERE()(p->slen - 1) (p->slen - 1)
273#define THERETHERE()(p->slen - 2) (p->slen - 2)
274#define DROP(n)(p->slen -= (n)) (p->slen -= (n))
275
276#ifdef _POSIX2_RE_DUP_MAX255
277#define DUPMAX255 _POSIX2_RE_DUP_MAX255
278#else
279#define DUPMAX255 255
280#endif
281#define INFINITY(255 + 1) (DUPMAX255 + 1)
282
283#ifndef NDEBUG1
284static int never0 = 0; /* for use in asserts; shuts lint up */
285#else
286#define never0 0 /* some <assert.h>s have bugs too */
287#endif
288
289/*
290 - llvm_regcomp - interface for parser and compilation
291 */
292int /* 0 success, otherwise REG_something */
293llvm_regcomp(llvm_regex_t *preg, const char *pattern, int cflags)
294{
295 struct parse pa;
296 struct re_guts *g;
297 struct parse *p = &pa;
298 int i;
299 size_t len;
300#ifdef REDEBUG
301# define GOODFLAGS(f)((f)&~0200) (f)
302#else
303# define GOODFLAGS(f)((f)&~0200) ((f)&~REG_DUMP0200)
304#endif
305
306 cflags = GOODFLAGS(cflags)((cflags)&~0200);
307 if ((cflags&REG_EXTENDED0001) && (cflags&REG_NOSPEC0020))
308 return(REG_INVARG16);
309
310 if (cflags&REG_PEND0040) {
311 if (preg->re_endp < pattern)
312 return(REG_INVARG16);
313 len = preg->re_endp - pattern;
314 } else
315 len = strlen((const char *)pattern);
316
317 /* do the mallocs early so failure handling is easy */
318 g = (struct re_guts *)malloc(sizeof(struct re_guts) +
319 (NC(127 - (-127 -1) + 1)-1)*sizeof(cat_t));
320 if (g == NULL((void*)0))
321 return(REG_ESPACE12);
322 p->ssize = len/(size_t)2*(size_t)3 + (size_t)1; /* ugh */
323 p->strip = (sop *)calloc(p->ssize, sizeof(sop));
324 p->slen = 0;
325 if (p->strip == NULL((void*)0)) {
326 free((char *)g);
327 return(REG_ESPACE12);
328 }
329
330 /* set things up */
331 p->g = g;
332 p->next = (char *)pattern; /* convenience; we do not modify it */
333 p->end = p->next + len;
334 p->error = 0;
335 p->ncsalloc = 0;
336 for (i = 0; i < NPAREN10; i++) {
337 p->pbegin[i] = 0;
338 p->pend[i] = 0;
339 }
340 g->csetsize = NC(127 - (-127 -1) + 1);
341 g->sets = NULL((void*)0);
342 g->setbits = NULL((void*)0);
343 g->ncsets = 0;
344 g->cflags = cflags;
345 g->iflags = 0;
346 g->nbol = 0;
347 g->neol = 0;
348 g->must = NULL((void*)0);
349 g->mlen = 0;
350 g->nsub = 0;
351 g->ncategories = 1; /* category 0 is "everything else" */
352 g->categories = &g->catspace[-(CHAR_MIN(-127 -1))];
353 (void) memset((char *)g->catspace, 0, NC(127 - (-127 -1) + 1)*sizeof(cat_t));
354 g->backrefs = 0;
355
356 /* do it */
357 EMIT(OEND, 0)doemit(p, (sop)((1LU<<((unsigned)27))), (size_t)(0));
358 g->firststate = THERE()(p->slen - 1);
359 if (cflags&REG_EXTENDED0001)
360 p_ere(p, OUT(127 +1));
361 else if (cflags&REG_NOSPEC0020)
362 p_str(p);
363 else
364 p_bre(p, OUT(127 +1), OUT(127 +1));
365 EMIT(OEND, 0)doemit(p, (sop)((1LU<<((unsigned)27))), (size_t)(0));
366 g->laststate = THERE()(p->slen - 1);
367
368 /* tidy up loose ends and fill things in */
369 categorize(p, g);
370 stripsnug(p, g);
371 findmust(p, g);
372 g->nplus = pluscount(p, g);
373 g->magic = MAGIC2((('R'^0200)<<8)|'E');
374 preg->re_nsub = g->nsub;
375 preg->re_g = g;
376 preg->re_magic = MAGIC1((('r'^0200)<<8) | 'e');
377#ifndef REDEBUG
378 /* not debugging, so can't rely on the assert() in llvm_regexec() */
379 if (g->iflags&REGEX_BAD04)
380 SETERROR(REG_ASSERT)seterr(p, (15));
381#endif
382
383 /* win or lose, we're done */
384 if (p->error != 0) /* lose */
385 llvm_regfree(preg);
386 return(p->error);
387}
388
389/*
390 - p_ere - ERE parser top level, concatenation and alternation
391 */
392static void
393p_ere(struct parse *p, int stop) /* character this ERE should end at */
394{
395 char c;
396 sopno prevback = 0;
397 sopno prevfwd = 0;
398 sopno conc;
399 int first = 1; /* is this the first alternative? */
400
401 for (;;) {
402 /* do a bunch of concatenated expressions */
403 conc = HERE()(p->slen);
404 while (MORE()(p->next < p->end) && (c = PEEK()(*p->next)) != '|' && c != stop)
405 p_ere_exp(p);
406 REQUIRE(HERE() != conc, REG_EMPTY)(void)(((p->slen) != conc) || seterr(p, (14))); /* require nonempty */
407
408 if (!EAT('|')((((p->next < p->end) && (*p->next) == ('|'
))) ? ((p->next++), 1) : 0))
409 break; /* NOTE BREAK OUT */
410
411 if (first) {
412 INSERT(OCH_, conc)doinsert(p, (sop)((15LU<<((unsigned)27))), (p->slen)
-(conc)+1, conc); /* offset is wrong */
413 prevfwd = conc;
414 prevback = conc;
415 first = 0;
416 }
417 ASTERN(OOR1, prevback)doemit(p, (sop)((16LU<<((unsigned)27))), (size_t)((p->
slen)-prevback));
418 prevback = THERE()(p->slen - 1);
419 AHEAD(prevfwd)dofwd(p, prevfwd, (p->slen)-(prevfwd)); /* fix previous offset */
420 prevfwd = HERE()(p->slen);
421 EMIT(OOR2, 0)doemit(p, (sop)((17LU<<((unsigned)27))), (size_t)(0)); /* offset is very wrong */
422 }
423
424 if (!first) { /* tail-end fixups */
425 AHEAD(prevfwd)dofwd(p, prevfwd, (p->slen)-(prevfwd));
426 ASTERN(O_CH, prevback)doemit(p, (sop)((18LU<<((unsigned)27))), (size_t)((p->
slen)-prevback));
427 }
428
429 assert(!MORE() || SEE(stop))((void)0);
430}
431
432/*
433 - p_ere_exp - parse one subERE, an atom possibly followed by a repetition op
434 */
435static void
436p_ere_exp(struct parse *p)
437{
438 char c;
439 sopno pos;
440 int count;
441 int count2;
442 int backrefnum;
443 sopno subno;
444 int wascaret = 0;
445
446 assert(MORE())((void)0); /* caller should have ensured this */
447 c = GETNEXT()(*p->next++);
448
449 pos = HERE()(p->slen);
450 switch (c) {
451 case '(':
452 REQUIRE(MORE(), REG_EPAREN)(void)(((p->next < p->end)) || seterr(p, (8)));
453 p->g->nsub++;
454 subno = p->g->nsub;
455 if (subno < NPAREN10)
456 p->pbegin[subno] = HERE()(p->slen);
457 EMIT(OLPAREN, subno)doemit(p, (sop)((13LU<<((unsigned)27))), (size_t)(subno
));
458 if (!SEE(')')((p->next < p->end) && (*p->next) == (')'
)))
459 p_ere(p, ')');
460 if (subno < NPAREN10) {
461 p->pend[subno] = HERE()(p->slen);
462 assert(p->pend[subno] != 0)((void)0);
463 }
464 EMIT(ORPAREN, subno)doemit(p, (sop)((14LU<<((unsigned)27))), (size_t)(subno
));
465 MUSTEAT(')', REG_EPAREN)((void)(((p->next < p->end) && (*p->next++
) == (')')) || seterr(p, (8))));
466 break;
467#ifndef POSIX_MISTAKE
468 case ')': /* happens only if no current unmatched ( */
469 /*
470 * You may ask, why the ifndef? Because I didn't notice
471 * this until slightly too late for 1003.2, and none of the
472 * other 1003.2 regular-expression reviewers noticed it at
473 * all. So an unmatched ) is legal POSIX, at least until
474 * we can get it fixed.
475 */
476 SETERROR(REG_EPAREN)seterr(p, (8));
477 break;
478#endif
479 case '^':
480 EMIT(OBOL, 0)doemit(p, (sop)((3LU<<((unsigned)27))), (size_t)(0));
481 p->g->iflags |= USEBOL01;
482 p->g->nbol++;
483 wascaret = 1;
484 break;
485 case '$':
486 EMIT(OEOL, 0)doemit(p, (sop)((4LU<<((unsigned)27))), (size_t)(0));
487 p->g->iflags |= USEEOL02;
488 p->g->neol++;
489 break;
490 case '|':
491 SETERROR(REG_EMPTY)seterr(p, (14));
492 break;
493 case '*':
494 case '+':
495 case '?':
496 SETERROR(REG_BADRPT)seterr(p, (13));
497 break;
498 case '.':
499 if (p->g->cflags&REG_NEWLINE0010)
500 nonnewline(p);
501 else
502 EMIT(OANY, 0)doemit(p, (sop)((5LU<<((unsigned)27))), (size_t)(0));
503 break;
504 case '[':
505 p_bracket(p);
506 break;
507 case '\\':
508 REQUIRE(MORE(), REG_EESCAPE)(void)(((p->next < p->end)) || seterr(p, (5)));
509 c = GETNEXT()(*p->next++);
510 if (c >= '1' && c <= '9') {
511 /* \[0-9] is taken to be a back-reference to a previously specified
512 * matching group. backrefnum will hold the number. The matching
513 * group must exist (i.e. if \4 is found there must have been at
514 * least 4 matching groups specified in the pattern previously).
515 */
516 backrefnum = c - '0';
517 if (p->pend[backrefnum] == 0) {
518 SETERROR(REG_ESUBREG)seterr(p, (6));
519 break;
520 }
521
522 /* Make sure everything checks out and emit the sequence
523 * that marks a back-reference to the parse structure.
524 */
525 assert(backrefnum <= p->g->nsub)((void)0);
526 EMIT(OBACK_, backrefnum)doemit(p, (sop)((7LU<<((unsigned)27))), (size_t)(backrefnum
));
527 assert(p->pbegin[backrefnum] != 0)((void)0);
528 assert(OP(p->strip[p->pbegin[backrefnum]]) != OLPAREN)((void)0);
529 assert(OP(p->strip[p->pend[backrefnum]]) != ORPAREN)((void)0);
530 (void) dupl(p, p->pbegin[backrefnum]+1, p->pend[backrefnum]);
531 EMIT(O_BACK, backrefnum)doemit(p, (sop)((8LU<<((unsigned)27))), (size_t)(backrefnum
));
532 p->g->backrefs = 1;
533 } else {
534 /* Other chars are simply themselves when escaped with a backslash.
535 */
536 ordinary(p, c);
537 }
538 break;
539 case '{': /* okay as ordinary except if digit follows */
540 REQUIRE(!MORE() || !isdigit((uch)PEEK()), REG_BADRPT)(void)((!(p->next < p->end) || !isdigit((uch)(*p->
next))) || seterr(p, (13)));
541 LLVM_FALLTHROUGH__attribute__((fallthrough));
542 default:
543 ordinary(p, c);
544 break;
545 }
546
547 if (!MORE()(p->next < p->end))
548 return;
549 c = PEEK()(*p->next);
550 /* we call { a repetition if followed by a digit */
551 if (!( c == '*' || c == '+' || c == '?' ||
552 (c == '{' && MORE2()(p->next+1 < p->end) && isdigit((uch)PEEK2()(*(p->next+1)))) ))
553 return; /* no repetition, we're done */
554 NEXT()(p->next++);
555
556 REQUIRE(!wascaret, REG_BADRPT)(void)((!wascaret) || seterr(p, (13)));
557 switch (c) {
558 case '*': /* implemented as +? */
559 /* this case does not require the (y|) trick, noKLUDGE */
560 INSERT(OPLUS_, pos)doinsert(p, (sop)((9LU<<((unsigned)27))), (p->slen)-
(pos)+1, pos);
561 ASTERN(O_PLUS, pos)doemit(p, (sop)((10LU<<((unsigned)27))), (size_t)((p->
slen)-pos));
562 INSERT(OQUEST_, pos)doinsert(p, (sop)((11LU<<((unsigned)27))), (p->slen)
-(pos)+1, pos);
563 ASTERN(O_QUEST, pos)doemit(p, (sop)((12LU<<((unsigned)27))), (size_t)((p->
slen)-pos));
564 break;
565 case '+':
566 INSERT(OPLUS_, pos)doinsert(p, (sop)((9LU<<((unsigned)27))), (p->slen)-
(pos)+1, pos);
567 ASTERN(O_PLUS, pos)doemit(p, (sop)((10LU<<((unsigned)27))), (size_t)((p->
slen)-pos));
568 break;
569 case '?':
570 /* KLUDGE: emit y? as (y|) until subtle bug gets fixed */
571 INSERT(OCH_, pos)doinsert(p, (sop)((15LU<<((unsigned)27))), (p->slen)
-(pos)+1, pos); /* offset slightly wrong */
572 ASTERN(OOR1, pos)doemit(p, (sop)((16LU<<((unsigned)27))), (size_t)((p->
slen)-pos)); /* this one's right */
573 AHEAD(pos)dofwd(p, pos, (p->slen)-(pos)); /* fix the OCH_ */
574 EMIT(OOR2, 0)doemit(p, (sop)((17LU<<((unsigned)27))), (size_t)(0)); /* offset very wrong... */
575 AHEAD(THERE())dofwd(p, (p->slen - 1), (p->slen)-((p->slen - 1))); /* ...so fix it */
576 ASTERN(O_CH, THERETHERE())doemit(p, (sop)((18LU<<((unsigned)27))), (size_t)((p->
slen)-(p->slen - 2)));
577 break;
578 case '{':
579 count = p_count(p);
580 if (EAT(',')((((p->next < p->end) && (*p->next) == (','
))) ? ((p->next++), 1) : 0)) {
581 if (isdigit((uch)PEEK()(*p->next))) {
582 count2 = p_count(p);
583 REQUIRE(count <= count2, REG_BADBR)(void)((count <= count2) || seterr(p, (10)));
584 } else /* single number with comma */
585 count2 = INFINITY(255 + 1);
586 } else /* just a single number */
587 count2 = count;
588 repeat(p, pos, count, count2);
589 if (!EAT('}')((((p->next < p->end) && (*p->next) == ('}'
))) ? ((p->next++), 1) : 0)) { /* error heuristics */
590 while (MORE()(p->next < p->end) && PEEK()(*p->next) != '}')
591 NEXT()(p->next++);
592 REQUIRE(MORE(), REG_EBRACE)(void)(((p->next < p->end)) || seterr(p, (9)));
593 SETERROR(REG_BADBR)seterr(p, (10));
594 }
595 break;
596 }
597
598 if (!MORE()(p->next < p->end))
599 return;
600 c = PEEK()(*p->next);
601 if (!( c == '*' || c == '+' || c == '?' ||
602 (c == '{' && MORE2()(p->next+1 < p->end) && isdigit((uch)PEEK2()(*(p->next+1)))) ) )
603 return;
604 SETERROR(REG_BADRPT)seterr(p, (13));
605}
606
607/*
608 - p_str - string (no metacharacters) "parser"
609 */
610static void
611p_str(struct parse *p)
612{
613 REQUIRE(MORE(), REG_EMPTY)(void)(((p->next < p->end)) || seterr(p, (14)));
614 while (MORE()(p->next < p->end))
615 ordinary(p, GETNEXT()(*p->next++));
616}
617
618/*
619 - p_bre - BRE parser top level, anchoring and concatenation
620 * Giving end1 as OUT essentially eliminates the end1/end2 check.
621 *
622 * This implementation is a bit of a kludge, in that a trailing $ is first
623 * taken as an ordinary character and then revised to be an anchor. The
624 * only undesirable side effect is that '$' gets included as a character
625 * category in such cases. This is fairly harmless; not worth fixing.
626 * The amount of lookahead needed to avoid this kludge is excessive.
627 */
628static void
629p_bre(struct parse *p,
630 int end1, /* first terminating character */
631 int end2) /* second terminating character */
632{
633 sopno start = HERE()(p->slen);
634 int first = 1; /* first subexpression? */
635 int wasdollar = 0;
636
637 if (EAT('^')((((p->next < p->end) && (*p->next) == ('^'
))) ? ((p->next++), 1) : 0)) {
638 EMIT(OBOL, 0)doemit(p, (sop)((3LU<<((unsigned)27))), (size_t)(0));
639 p->g->iflags |= USEBOL01;
640 p->g->nbol++;
641 }
642 while (MORE()(p->next < p->end) && !SEETWO(end1, end2)((p->next < p->end) && (p->next+1 < p->
end) && (*p->next) == (end1) && (*(p->next
+1)) == (end2))) {
643 wasdollar = p_simp_re(p, first);
644 first = 0;
645 }
646 if (wasdollar) { /* oops, that was a trailing anchor */
647 DROP(1)(p->slen -= (1));
648 EMIT(OEOL, 0)doemit(p, (sop)((4LU<<((unsigned)27))), (size_t)(0));
649 p->g->iflags |= USEEOL02;
650 p->g->neol++;
651 }
652
653 REQUIRE(HERE() != start, REG_EMPTY)(void)(((p->slen) != start) || seterr(p, (14))); /* require nonempty */
654}
655
656/*
657 - p_simp_re - parse a simple RE, an atom possibly followed by a repetition
658 */
659static int /* was the simple RE an unbackslashed $? */
660p_simp_re(struct parse *p,
661 int starordinary) /* is a leading * an ordinary character? */
662{
663 int c;
664 int count;
665 int count2;
666 sopno pos;
667 int i;
668 sopno subno;
669# define BACKSL(1<<8) (1<<CHAR_BIT8)
670
671 pos = HERE()(p->slen); /* repetition op, if any, covers from here */
672
673 assert(MORE())((void)0); /* caller should have ensured this */
674 c = GETNEXT()(*p->next++);
675 if (c == '\\') {
676 REQUIRE(MORE(), REG_EESCAPE)(void)(((p->next < p->end)) || seterr(p, (5)));
677 c = BACKSL(1<<8) | GETNEXT()(*p->next++);
678 }
679 switch (c) {
680 case '.':
681 if (p->g->cflags&REG_NEWLINE0010)
682 nonnewline(p);
683 else
684 EMIT(OANY, 0)doemit(p, (sop)((5LU<<((unsigned)27))), (size_t)(0));
685 break;
686 case '[':
687 p_bracket(p);
688 break;
689 case BACKSL(1<<8)|'{':
690 SETERROR(REG_BADRPT)seterr(p, (13));
691 break;
692 case BACKSL(1<<8)|'(':
693 p->g->nsub++;
694 subno = p->g->nsub;
695 if (subno < NPAREN10)
696 p->pbegin[subno] = HERE()(p->slen);
697 EMIT(OLPAREN, subno)doemit(p, (sop)((13LU<<((unsigned)27))), (size_t)(subno
));
698 /* the MORE here is an error heuristic */
699 if (MORE()(p->next < p->end) && !SEETWO('\\', ')')((p->next < p->end) && (p->next+1 < p->
end) && (*p->next) == ('\\') && (*(p->next
+1)) == (')')))
700 p_bre(p, '\\', ')');
701 if (subno < NPAREN10) {
702 p->pend[subno] = HERE()(p->slen);
703 assert(p->pend[subno] != 0)((void)0);
704 }
705 EMIT(ORPAREN, subno)doemit(p, (sop)((14LU<<((unsigned)27))), (size_t)(subno
));
706 REQUIRE(EATTWO('\\', ')'), REG_EPAREN)(void)((((((p->next < p->end) && (p->next
+1 < p->end) && (*p->next) == ('\\') &&
(*(p->next+1)) == (')'))) ? ((p->next += 2), 1) : 0)) ||
seterr(p, (8)));
707 break;
708 case BACKSL(1<<8)|')': /* should not get here -- must be user */
709 case BACKSL(1<<8)|'}':
710 SETERROR(REG_EPAREN)seterr(p, (8));
711 break;
712 case BACKSL(1<<8)|'1':
713 case BACKSL(1<<8)|'2':
714 case BACKSL(1<<8)|'3':
715 case BACKSL(1<<8)|'4':
716 case BACKSL(1<<8)|'5':
717 case BACKSL(1<<8)|'6':
718 case BACKSL(1<<8)|'7':
719 case BACKSL(1<<8)|'8':
720 case BACKSL(1<<8)|'9':
721 i = (c&~BACKSL(1<<8)) - '0';
722 assert(i < NPAREN)((void)0);
723 if (p->pend[i] != 0) {
724 assert(i <= p->g->nsub)((void)0);
725 EMIT(OBACK_, i)doemit(p, (sop)((7LU<<((unsigned)27))), (size_t)(i));
726 assert(p->pbegin[i] != 0)((void)0);
727 assert(OP(p->strip[p->pbegin[i]]) == OLPAREN)((void)0);
728 assert(OP(p->strip[p->pend[i]]) == ORPAREN)((void)0);
729 (void) dupl(p, p->pbegin[i]+1, p->pend[i]);
730 EMIT(O_BACK, i)doemit(p, (sop)((8LU<<((unsigned)27))), (size_t)(i));
731 } else
732 SETERROR(REG_ESUBREG)seterr(p, (6));
733 p->g->backrefs = 1;
734 break;
735 case '*':
736 REQUIRE(starordinary, REG_BADRPT)(void)((starordinary) || seterr(p, (13)));
737 LLVM_FALLTHROUGH__attribute__((fallthrough));
738 default:
739 ordinary(p, (char)c);
740 break;
741 }
742
743 if (EAT('*')((((p->next < p->end) && (*p->next) == ('*'
))) ? ((p->next++), 1) : 0)) { /* implemented as +? */
744 /* this case does not require the (y|) trick, noKLUDGE */
745 INSERT(OPLUS_, pos)doinsert(p, (sop)((9LU<<((unsigned)27))), (p->slen)-
(pos)+1, pos);
746 ASTERN(O_PLUS, pos)doemit(p, (sop)((10LU<<((unsigned)27))), (size_t)((p->
slen)-pos));
747 INSERT(OQUEST_, pos)doinsert(p, (sop)((11LU<<((unsigned)27))), (p->slen)
-(pos)+1, pos);
748 ASTERN(O_QUEST, pos)doemit(p, (sop)((12LU<<((unsigned)27))), (size_t)((p->
slen)-pos));
749 } else if (EATTWO('\\', '{')((((p->next < p->end) && (p->next+1 < p
->end) && (*p->next) == ('\\') && (*(p->
next+1)) == ('{'))) ? ((p->next += 2), 1) : 0)) {
750 count = p_count(p);
751 if (EAT(',')((((p->next < p->end) && (*p->next) == (','
))) ? ((p->next++), 1) : 0)) {
752 if (MORE()(p->next < p->end) && isdigit((uch)PEEK()(*p->next))) {
753 count2 = p_count(p);
754 REQUIRE(count <= count2, REG_BADBR)(void)((count <= count2) || seterr(p, (10)));
755 } else /* single number with comma */
756 count2 = INFINITY(255 + 1);
757 } else /* just a single number */
758 count2 = count;
759 repeat(p, pos, count, count2);
760 if (!EATTWO('\\', '}')((((p->next < p->end) && (p->next+1 < p
->end) && (*p->next) == ('\\') && (*(p->
next+1)) == ('}'))) ? ((p->next += 2), 1) : 0)) { /* error heuristics */
761 while (MORE()(p->next < p->end) && !SEETWO('\\', '}')((p->next < p->end) && (p->next+1 < p->
end) && (*p->next) == ('\\') && (*(p->next
+1)) == ('}')))
762 NEXT()(p->next++);
763 REQUIRE(MORE(), REG_EBRACE)(void)(((p->next < p->end)) || seterr(p, (9)));
764 SETERROR(REG_BADBR)seterr(p, (10));
765 }
766 } else if (c == '$') /* $ (but not \$) ends it */
767 return(1);
768
769 return(0);
770}
771
772/*
773 - p_count - parse a repetition count
774 */
775static int /* the value */
776p_count(struct parse *p)
777{
778 int count = 0;
779 int ndigits = 0;
780
781 while (MORE()(p->next < p->end) && isdigit((uch)PEEK()(*p->next)) && count <= DUPMAX255) {
782 count = count*10 + (GETNEXT()(*p->next++) - '0');
783 ndigits++;
784 }
785
786 REQUIRE(ndigits > 0 && count <= DUPMAX, REG_BADBR)(void)((ndigits > 0 && count <= 255) || seterr(
p, (10)));
787 return(count);
788}
789
790/*
791 - p_bracket - parse a bracketed character list
792 *
793 * Note a significant property of this code: if the allocset() did SETERROR,
794 * no set operations are done.
795 */
796static void
797p_bracket(struct parse *p)
798{
799 cset *cs;
800 int invert = 0;
801
802 /* Dept of Truly Sickening Special-Case Kludges */
803 if (p->next + 5 < p->end && strncmp(p->next, "[:<:]]", 6) == 0) {
804 EMIT(OBOW, 0)doemit(p, (sop)((19LU<<((unsigned)27))), (size_t)(0));
805 NEXTn(6)(p->next += (6));
806 return;
807 }
808 if (p->next + 5 < p->end && strncmp(p->next, "[:>:]]", 6) == 0) {
809 EMIT(OEOW, 0)doemit(p, (sop)((20LU<<((unsigned)27))), (size_t)(0));
810 NEXTn(6)(p->next += (6));
811 return;
812 }
813
814 if ((cs = allocset(p)) == NULL((void*)0)) {
815 /* allocset did set error status in p */
816 return;
817 }
818
819 if (EAT('^')((((p->next < p->end) && (*p->next) == ('^'
))) ? ((p->next++), 1) : 0))
820 invert++; /* make note to invert set at end */
821 if (EAT(']')((((p->next < p->end) && (*p->next) == (']'
))) ? ((p->next++), 1) : 0))
822 CHadd(cs, ']')((cs)->ptr[(uch)(']')] |= (cs)->mask, (cs)->hash += (
']'));
823 else if (EAT('-')((((p->next < p->end) && (*p->next) == ('-'
))) ? ((p->next++), 1) : 0))
824 CHadd(cs, '-')((cs)->ptr[(uch)('-')] |= (cs)->mask, (cs)->hash += (
'-'));
825 while (MORE()(p->next < p->end) && PEEK()(*p->next) != ']' && !SEETWO('-', ']')((p->next < p->end) && (p->next+1 < p->
end) && (*p->next) == ('-') && (*(p->next
+1)) == (']')))
826 p_b_term(p, cs);
827 if (EAT('-')((((p->next < p->end) && (*p->next) == ('-'
))) ? ((p->next++), 1) : 0))
828 CHadd(cs, '-')((cs)->ptr[(uch)('-')] |= (cs)->mask, (cs)->hash += (
'-'));
829 MUSTEAT(']', REG_EBRACK)((void)(((p->next < p->end) && (*p->next++
) == (']')) || seterr(p, (7))));
830
831 if (p->error != 0) { /* don't mess things up further */
832 freeset(p, cs);
833 return;
834 }
835
836 if (p->g->cflags&REG_ICASE0002) {
837 int i;
838 int ci;
839
840 for (i = p->g->csetsize - 1; i >= 0; i--)
841 if (CHIN(cs, i)((cs)->ptr[(uch)(i)] & (cs)->mask) && isalpha(i)) {
842 ci = othercase(i);
843 if (ci != i)
844 CHadd(cs, ci)((cs)->ptr[(uch)(ci)] |= (cs)->mask, (cs)->hash += (
ci));
845 }
846 if (cs->multis != NULL((void*)0))
847 mccase(p, cs);
848 }
849 if (invert) {
850 int i;
851
852 for (i = p->g->csetsize - 1; i >= 0; i--)
853 if (CHIN(cs, i)((cs)->ptr[(uch)(i)] & (cs)->mask))
854 CHsub(cs, i)((cs)->ptr[(uch)(i)] &= ~(cs)->mask, (cs)->hash -=
(i));
855 else
856 CHadd(cs, i)((cs)->ptr[(uch)(i)] |= (cs)->mask, (cs)->hash += (i
));
857 if (p->g->cflags&REG_NEWLINE0010)
858 CHsub(cs, '\n')((cs)->ptr[(uch)('\n')] &= ~(cs)->mask, (cs)->hash
-= ('\n'));
859 if (cs->multis != NULL((void*)0))
860 mcinvert(p, cs);
861 }
862
863 assert(cs->multis == NULL)((void)0); /* xxx */
864
865 if (nch(p, cs) == 1) { /* optimize singleton sets */
866 ordinary(p, firstch(p, cs));
867 freeset(p, cs);
868 } else
869 EMIT(OANYOF, freezeset(p, cs))doemit(p, (sop)((6LU<<((unsigned)27))), (size_t)(freezeset
(p, cs)));
870}
871
872/*
873 - p_b_term - parse one term of a bracketed character list
874 */
875static void
876p_b_term(struct parse *p, cset *cs)
877{
878 char c;
879 char start, finish;
880 int i;
881
882 /* classify what we've got */
883 switch ((MORE()(p->next < p->end)) ? PEEK()(*p->next) : '\0') {
884 case '[':
885 c = (MORE2()(p->next+1 < p->end)) ? PEEK2()(*(p->next+1)) : '\0';
886 break;
887 case '-':
888 SETERROR(REG_ERANGE)seterr(p, (11));
889 return; /* NOTE RETURN */
890 break;
891 default:
892 c = '\0';
893 break;
894 }
895
896 switch (c) {
897 case ':': /* character class */
898 NEXT2()(p->next += 2);
899 REQUIRE(MORE(), REG_EBRACK)(void)(((p->next < p->end)) || seterr(p, (7)));
900 c = PEEK()(*p->next);
901 REQUIRE(c != '-' && c != ']', REG_ECTYPE)(void)((c != '-' && c != ']') || seterr(p, (4)));
902 p_b_cclass(p, cs);
903 REQUIRE(MORE(), REG_EBRACK)(void)(((p->next < p->end)) || seterr(p, (7)));
904 REQUIRE(EATTWO(':', ']'), REG_ECTYPE)(void)((((((p->next < p->end) && (p->next
+1 < p->end) && (*p->next) == (':') &&
(*(p->next+1)) == (']'))) ? ((p->next += 2), 1) : 0)) ||
seterr(p, (4)));
905 break;
906 case '=': /* equivalence class */
907 NEXT2()(p->next += 2);
908 REQUIRE(MORE(), REG_EBRACK)(void)(((p->next < p->end)) || seterr(p, (7)));
909 c = PEEK()(*p->next);
910 REQUIRE(c != '-' && c != ']', REG_ECOLLATE)(void)((c != '-' && c != ']') || seterr(p, (3)));
911 p_b_eclass(p, cs);
912 REQUIRE(MORE(), REG_EBRACK)(void)(((p->next < p->end)) || seterr(p, (7)));
913 REQUIRE(EATTWO('=', ']'), REG_ECOLLATE)(void)((((((p->next < p->end) && (p->next
+1 < p->end) && (*p->next) == ('=') &&
(*(p->next+1)) == (']'))) ? ((p->next += 2), 1) : 0)) ||
seterr(p, (3)));
914 break;
915 default: /* symbol, ordinary character, or range */
916/* xxx revision needed for multichar stuff */
917 start = p_b_symbol(p);
918 if (SEE('-')((p->next < p->end) && (*p->next) == ('-'
)) && MORE2()(p->next+1 < p->end) && PEEK2()(*(p->next+1)) != ']') {
919 /* range */
920 NEXT()(p->next++);
921 if (EAT('-')((((p->next < p->end) && (*p->next) == ('-'
))) ? ((p->next++), 1) : 0))
922 finish = '-';
923 else
924 finish = p_b_symbol(p);
925 } else
926 finish = start;
927/* xxx what about signed chars here... */
928 REQUIRE(start <= finish, REG_ERANGE)(void)((start <= finish) || seterr(p, (11)));
929 for (i = start; i <= finish; i++)
930 CHadd(cs, i)((cs)->ptr[(uch)(i)] |= (cs)->mask, (cs)->hash += (i
));
931 break;
932 }
933}
934
935/*
936 - p_b_cclass - parse a character-class name and deal with it
937 */
938static void
939p_b_cclass(struct parse *p, cset *cs)
940{
941 char *sp = p->next;
942 struct cclass *cp;
943 size_t len;
944 const char *u;
945 char c;
946
947 while (MORE()(p->next < p->end) && isalpha((uch)PEEK()(*p->next)))
948 NEXT()(p->next++);
949 len = p->next - sp;
950 for (cp = cclasses; cp->name != NULL((void*)0); cp++)
951 if (strncmp(cp->name, sp, len) == 0 && cp->name[len] == '\0')
952 break;
953 if (cp->name == NULL((void*)0)) {
954 /* oops, didn't find it */
955 SETERROR(REG_ECTYPE)seterr(p, (4));
956 return;
957 }
958
959 u = cp->chars;
960 while ((c = *u++) != '\0')
961 CHadd(cs, c)((cs)->ptr[(uch)(c)] |= (cs)->mask, (cs)->hash += (c
));
962 for (u = cp->multis; *u != '\0'; u += strlen(u) + 1)
963 MCadd(p, cs, u)mcadd(p, cs, u);
964}
965
966/*
967 - p_b_eclass - parse an equivalence-class name and deal with it
968 *
969 * This implementation is incomplete. xxx
970 */
971static void
972p_b_eclass(struct parse *p, cset *cs)
973{
974 char c;
975
976 c = p_b_coll_elem(p, '=');
977 CHadd(cs, c)((cs)->ptr[(uch)(c)] |= (cs)->mask, (cs)->hash += (c
));
978}
979
980/*
981 - p_b_symbol - parse a character or [..]ed multicharacter collating symbol
982 */
983static char /* value of symbol */
984p_b_symbol(struct parse *p)
985{
986 char value;
987
988 REQUIRE(MORE(), REG_EBRACK)(void)(((p->next < p->end)) || seterr(p, (7)));
989 if (!EATTWO('[', '.')((((p->next < p->end) && (p->next+1 < p
->end) && (*p->next) == ('[') && (*(p->
next+1)) == ('.'))) ? ((p->next += 2), 1) : 0))
990 return(GETNEXT()(*p->next++));
991
992 /* collating symbol */
993 value = p_b_coll_elem(p, '.');
994 REQUIRE(EATTWO('.', ']'), REG_ECOLLATE)(void)((((((p->next < p->end) && (p->next
+1 < p->end) && (*p->next) == ('.') &&
(*(p->next+1)) == (']'))) ? ((p->next += 2), 1) : 0)) ||
seterr(p, (3)));
995 return(value);
996}
997
998/*
999 - p_b_coll_elem - parse a collating-element name and look it up
1000 */
1001static char /* value of collating element */
1002p_b_coll_elem(struct parse *p,
1003 int endc) /* name ended by endc,']' */
1004{
1005 char *sp = p->next;
1006 struct cname *cp;
1007 size_t len;
1008
1009 while (MORE()(p->next < p->end) && !SEETWO(endc, ']')((p->next < p->end) && (p->next+1 < p->
end) && (*p->next) == (endc) && (*(p->next
+1)) == (']')))
1010 NEXT()(p->next++);
1011 if (!MORE()(p->next < p->end)) {
1012 SETERROR(REG_EBRACK)seterr(p, (7));
1013 return(0);
1014 }
1015 len = p->next - sp;
1016 for (cp = cnames; cp->name != NULL((void*)0); cp++)
1017 if (strncmp(cp->name, sp, len) == 0 && strlen(cp->name) == len)
1018 return(cp->code); /* known name */
1019 if (len == 1)
1020 return(*sp); /* single character */
1021 SETERROR(REG_ECOLLATE)seterr(p, (3)); /* neither */
1022 return(0);
1023}
1024
1025/*
1026 - othercase - return the case counterpart of an alphabetic
1027 */
1028static char /* if no counterpart, return ch */
1029othercase(int ch)
1030{
1031 ch = (uch)ch;
1032 assert(isalpha(ch))((void)0);
1033 if (isupper(ch))
1034 return ((uch)tolower(ch));
1035 else if (islower(ch))
1036 return ((uch)toupper(ch));
1037 else /* peculiar, but could happen */
1038 return(ch);
1039}
1040
1041/*
1042 - bothcases - emit a dualcase version of a two-case character
1043 *
1044 * Boy, is this implementation ever a kludge...
1045 */
1046static void
1047bothcases(struct parse *p, int ch)
1048{
1049 char *oldnext = p->next;
1050 char *oldend = p->end;
1051 char bracket[3];
1052
1053 ch = (uch)ch;
1054 assert(othercase(ch) != ch)((void)0); /* p_bracket() would recurse */
1055 p->next = bracket;
1056 p->end = bracket+2;
1057 bracket[0] = ch;
1058 bracket[1] = ']';
1059 bracket[2] = '\0';
1060 p_bracket(p);
1061 assert(p->next == bracket+2)((void)0);
1062 p->next = oldnext;
1063 p->end = oldend;
1064}
1065
1066/*
1067 - ordinary - emit an ordinary character
1068 */
1069static void
1070ordinary(struct parse *p, int ch)
1071{
1072 cat_t *cap = p->g->categories;
1073
1074 if ((p->g->cflags&REG_ICASE0002) && isalpha((uch)ch) && othercase(ch) != ch)
1075 bothcases(p, ch);
1076 else {
1077 EMIT(OCHAR, (uch)ch)doemit(p, (sop)((2LU<<((unsigned)27))), (size_t)((uch)ch
));
1078 if (cap[ch] == 0)
1079 cap[ch] = p->g->ncategories++;
1080 }
1081}
1082
1083/*
1084 - nonnewline - emit REG_NEWLINE version of OANY
1085 *
1086 * Boy, is this implementation ever a kludge...
1087 */
1088static void
1089nonnewline(struct parse *p)
1090{
1091 char *oldnext = p->next;
1092 char *oldend = p->end;
1093 char bracket[4];
1094
1095 p->next = bracket;
1096 p->end = bracket+3;
1097 bracket[0] = '^';
1098 bracket[1] = '\n';
1099 bracket[2] = ']';
1100 bracket[3] = '\0';
1101 p_bracket(p);
1102 assert(p->next == bracket+3)((void)0);
1103 p->next = oldnext;
1104 p->end = oldend;
1105}
1106
1107/*
1108 - repeat - generate code for a bounded repetition, recursively if needed
1109 */
1110static void
1111repeat(struct parse *p,
1112 sopno start, /* operand from here to end of strip */
1113 int from, /* repeated from this number */
1114 int to) /* to this number of times (maybe INFINITY) */
1115{
1116 sopno finish = HERE()(p->slen);
1117# define N2 2
1118# define INF3 3
1119# define REP(f, t)((f)*8 + (t)) ((f)*8 + (t))
1120# define MAP(n)(((n) <= 1) ? (n) : ((n) == (255 + 1)) ? 3 : 2) (((n) <= 1) ? (n) : ((n) == INFINITY(255 + 1)) ? INF3 : N2)
1121 sopno copy;
1122
1123 if (p->error != 0) /* head off possible runaway recursion */
1124 return;
1125
1126 assert(from <= to)((void)0);
1127
1128 switch (REP(MAP(from), MAP(to))(((((from) <= 1) ? (from) : ((from) == (255 + 1)) ? 3 : 2)
)*8 + ((((to) <= 1) ? (to) : ((to) == (255 + 1)) ? 3 : 2))
)) {
1129 case REP(0, 0)((0)*8 + (0)): /* must be user doing this */
1130 DROP(finish-start)(p->slen -= (finish-start)); /* drop the operand */
1131 break;
1132 case REP(0, 1)((0)*8 + (1)): /* as x{1,1}? */
1133 case REP(0, N)((0)*8 + (2)): /* as x{1,n}? */
1134 case REP(0, INF)((0)*8 + (3)): /* as x{1,}? */
1135 /* KLUDGE: emit y? as (y|) until subtle bug gets fixed */
1136 INSERT(OCH_, start)doinsert(p, (sop)((15LU<<((unsigned)27))), (p->slen)
-(start)+1, start); /* offset is wrong... */
1137 repeat(p, start+1, 1, to);
1138 ASTERN(OOR1, start)doemit(p, (sop)((16LU<<((unsigned)27))), (size_t)((p->
slen)-start));
1139 AHEAD(start)dofwd(p, start, (p->slen)-(start)); /* ... fix it */
1140 EMIT(OOR2, 0)doemit(p, (sop)((17LU<<((unsigned)27))), (size_t)(0));
1141 AHEAD(THERE())dofwd(p, (p->slen - 1), (p->slen)-((p->slen - 1)));
1142 ASTERN(O_CH, THERETHERE())doemit(p, (sop)((18LU<<((unsigned)27))), (size_t)((p->
slen)-(p->slen - 2)));
1143 break;
1144 case REP(1, 1)((1)*8 + (1)): /* trivial case */
1145 /* done */
1146 break;
1147 case REP(1, N)((1)*8 + (2)): /* as x?x{1,n-1} */
1148 /* KLUDGE: emit y? as (y|) until subtle bug gets fixed */
1149 INSERT(OCH_, start)doinsert(p, (sop)((15LU<<((unsigned)27))), (p->slen)
-(start)+1, start);
1150 ASTERN(OOR1, start)doemit(p, (sop)((16LU<<((unsigned)27))), (size_t)((p->
slen)-start));
1151 AHEAD(start)dofwd(p, start, (p->slen)-(start));
1152 EMIT(OOR2, 0)doemit(p, (sop)((17LU<<((unsigned)27))), (size_t)(0)); /* offset very wrong... */
1153 AHEAD(THERE())dofwd(p, (p->slen - 1), (p->slen)-((p->slen - 1))); /* ...so fix it */
1154 ASTERN(O_CH, THERETHERE())doemit(p, (sop)((18LU<<((unsigned)27))), (size_t)((p->
slen)-(p->slen - 2)));
1155 copy = dupl(p, start+1, finish+1);
1156 assert(copy == finish+4)((void)0);
1157 repeat(p, copy, 1, to-1);
1158 break;
1159 case REP(1, INF)((1)*8 + (3)): /* as x+ */
1160 INSERT(OPLUS_, start)doinsert(p, (sop)((9LU<<((unsigned)27))), (p->slen)-
(start)+1, start);
1161 ASTERN(O_PLUS, start)doemit(p, (sop)((10LU<<((unsigned)27))), (size_t)((p->
slen)-start));
1162 break;
1163 case REP(N, N)((2)*8 + (2)): /* as xx{m-1,n-1} */
1164 copy = dupl(p, start, finish);
1165 repeat(p, copy, from-1, to-1);
1166 break;
1167 case REP(N, INF)((2)*8 + (3)): /* as xx{n-1,INF} */
1168 copy = dupl(p, start, finish);
1169 repeat(p, copy, from-1, to);
1170 break;
1171 default: /* "can't happen" */
1172 SETERROR(REG_ASSERT)seterr(p, (15)); /* just in case */
1173 break;
1174 }
1175}
1176
1177/*
1178 - seterr - set an error condition
1179 */
1180static int /* useless but makes type checking happy */
1181seterr(struct parse *p, int e)
1182{
1183 if (p->error == 0) /* keep earliest error condition */
1184 p->error = e;
1185 p->next = nuls; /* try to bring things to a halt */
1186 p->end = nuls;
1187 return(0); /* make the return value well-defined */
1188}
1189
1190/*
1191 - allocset - allocate a set of characters for []
1192 */
1193static cset *
1194allocset(struct parse *p)
1195{
1196 int no = p->g->ncsets++;
1197 size_t nc;
1198 size_t nbytes;
1199 cset *cs;
1200 size_t css = (size_t)p->g->csetsize;
1201 int i;
1202
1203 if (no >= p->ncsalloc) { /* need another column of space */
1204 void *ptr;
1205
1206 p->ncsalloc += CHAR_BIT8;
1207 nc = p->ncsalloc;
1208 if (nc > SIZE_MAX0xffffffffffffffffUL / sizeof(cset))
1209 goto nomem;
1210 assert(nc % CHAR_BIT == 0)((void)0);
1211 nbytes = nc / CHAR_BIT8 * css;
1212
1213 ptr = (cset *)realloc((char *)p->g->sets, nc * sizeof(cset));
1214 if (ptr == NULL((void*)0))
1215 goto nomem;
1216 p->g->sets = ptr;
1217
1218 ptr = (uch *)realloc((char *)p->g->setbits, nbytes);
1219 if (ptr == NULL((void*)0))
1220 goto nomem;
1221 p->g->setbits = ptr;
1222
1223 for (i = 0; i < no; i++)
1224 p->g->sets[i].ptr = p->g->setbits + css*(i/CHAR_BIT8);
1225
1226 (void) memset((char *)p->g->setbits + (nbytes - css), 0, css);
1227 }
1228 /* XXX should not happen */
1229 if (p->g->sets == NULL((void*)0) || p->g->setbits == NULL((void*)0))
1230 goto nomem;
1231
1232 cs = &p->g->sets[no];
1233 cs->ptr = p->g->setbits + css*((no)/CHAR_BIT8);
1234 cs->mask = 1 << ((no) % CHAR_BIT8);
1235 cs->hash = 0;
1236 cs->smultis = 0;
1237 cs->multis = NULL((void*)0);
1238
1239 return(cs);
1240nomem:
1241 free(p->g->sets);
1242 p->g->sets = NULL((void*)0);
1243 free(p->g->setbits);
1244 p->g->setbits = NULL((void*)0);
1245
1246 SETERROR(REG_ESPACE)seterr(p, (12));
1247 /* caller's responsibility not to do set ops */
1248 return(NULL((void*)0));
1249}
1250
1251/*
1252 - freeset - free a now-unused set
1253 */
1254static void
1255freeset(struct parse *p, cset *cs)
1256{
1257 size_t i;
1258 cset *top = &p->g->sets[p->g->ncsets];
1259 size_t css = (size_t)p->g->csetsize;
1260
1261 for (i = 0; i < css; i++)
1262 CHsub(cs, i)((cs)->ptr[(uch)(i)] &= ~(cs)->mask, (cs)->hash -=
(i));
1263 if (cs == top-1) /* recover only the easy case */
1264 p->g->ncsets--;
1265}
1266
1267/*
1268 - freezeset - final processing on a set of characters
1269 *
1270 * The main task here is merging identical sets. This is usually a waste
1271 * of time (although the hash code minimizes the overhead), but can win
1272 * big if REG_ICASE is being used. REG_ICASE, by the way, is why the hash
1273 * is done using addition rather than xor -- all ASCII [aA] sets xor to
1274 * the same value!
1275 */
1276static int /* set number */
1277freezeset(struct parse *p, cset *cs)
1278{
1279 uch h = cs->hash;
1280 size_t i;
1281 cset *top = &p->g->sets[p->g->ncsets];
1282 cset *cs2;
1283 size_t css = (size_t)p->g->csetsize;
1284
1285 /* look for an earlier one which is the same */
1286 for (cs2 = &p->g->sets[0]; cs2 < top; cs2++)
1287 if (cs2->hash == h && cs2 != cs) {
1288 /* maybe */
1289 for (i = 0; i < css; i++)
1290 if (!!CHIN(cs2, i)((cs2)->ptr[(uch)(i)] & (cs2)->mask) != !!CHIN(cs, i)((cs)->ptr[(uch)(i)] & (cs)->mask))
1291 break; /* no */
1292 if (i == css)
1293 break; /* yes */
1294 }
1295
1296 if (cs2 < top) { /* found one */
1297 freeset(p, cs);
1298 cs = cs2;
1299 }
1300
1301 return((int)(cs - p->g->sets));
1302}
1303
1304/*
1305 - firstch - return first character in a set (which must have at least one)
1306 */
1307static int /* character; there is no "none" value */
1308firstch(struct parse *p, cset *cs)
1309{
1310 size_t i;
1311 size_t css = (size_t)p->g->csetsize;
1312
1313 for (i = 0; i < css; i++)
1314 if (CHIN(cs, i)((cs)->ptr[(uch)(i)] & (cs)->mask))
1315 return((char)i);
1316 assert(never)((void)0);
1317 return(0); /* arbitrary */
1318}
1319
1320/*
1321 - nch - number of characters in a set
1322 */
1323static int
1324nch(struct parse *p, cset *cs)
1325{
1326 size_t i;
1327 size_t css = (size_t)p->g->csetsize;
1328 int n = 0;
1329
1330 for (i = 0; i < css; i++)
1331 if (CHIN(cs, i)((cs)->ptr[(uch)(i)] & (cs)->mask))
1332 n++;
1333 return(n);
1334}
1335
1336/*
1337 - mcadd - add a collating element to a cset
1338 */
1339static void
1340mcadd( struct parse *p, cset *cs, const char *cp)
1341{
1342 size_t oldend = cs->smultis;
1343 void *np;
1344
1345 cs->smultis += strlen(cp) + 1;
1346 np = realloc(cs->multis, cs->smultis);
1347 if (np == NULL((void*)0)) {
1348 if (cs->multis)
1349 free(cs->multis);
1350 cs->multis = NULL((void*)0);
1351 SETERROR(REG_ESPACE)seterr(p, (12));
1352 return;
1353 }
1354 cs->multis = np;
1355
1356 llvm_strlcpy(cs->multis + oldend - 1, cp, cs->smultis - oldend + 1);
1357}
1358
1359/*
1360 - mcinvert - invert the list of collating elements in a cset
1361 *
1362 * This would have to know the set of possibilities. Implementation
1363 * is deferred.
1364 */
1365/* ARGSUSED */
1366static void
1367mcinvert(struct parse *p, cset *cs)
1368{
1369 assert(cs->multis == NULL)((void)0); /* xxx */
1370}
1371
1372/*
1373 - mccase - add case counterparts of the list of collating elements in a cset
1374 *
1375 * This would have to know the set of possibilities. Implementation
1376 * is deferred.
1377 */
1378/* ARGSUSED */
1379static void
1380mccase(struct parse *p, cset *cs)
1381{
1382 assert(cs->multis == NULL)((void)0); /* xxx */
1383}
1384
1385/*
1386 - isinsets - is this character in any sets?
1387 */
1388static int /* predicate */
1389isinsets(struct re_guts *g, int c)
1390{
1391 uch *col;
1392 int i;
1393 int ncols = (g->ncsets+(CHAR_BIT8-1)) / CHAR_BIT8;
1394 unsigned uc = (uch)c;
1395
1396 for (i = 0, col = g->setbits; i < ncols; i++, col += g->csetsize)
1397 if (col[uc] != 0)
1398 return(1);
1399 return(0);
1400}
1401
1402/*
1403 - samesets - are these two characters in exactly the same sets?
1404 */
1405static int /* predicate */
1406samesets(struct re_guts *g, int c1, int c2)
1407{
1408 uch *col;
1409 int i;
1410 int ncols = (g->ncsets+(CHAR_BIT8-1)) / CHAR_BIT8;
1411 unsigned uc1 = (uch)c1;
1412 unsigned uc2 = (uch)c2;
1413
1414 for (i = 0, col = g->setbits; i < ncols; i++, col += g->csetsize)
1415 if (col[uc1] != col[uc2])
1416 return(0);
1417 return(1);
1418}
1419
1420/*
1421 - categorize - sort out character categories
1422 */
1423static void
1424categorize(struct parse *p, struct re_guts *g)
1425{
1426 cat_t *cats = g->categories;
1427 int c;
1428 int c2;
1429 cat_t cat;
1430
1431 /* avoid making error situations worse */
1432 if (p->error != 0)
1433 return;
1434
1435 for (c = CHAR_MIN(-127 -1); c <= CHAR_MAX127; c++)
1436 if (cats[c] == 0 && isinsets(g, c)) {
1437 cat = g->ncategories++;
1438 cats[c] = cat;
1439 for (c2 = c+1; c2 <= CHAR_MAX127; c2++)
1440 if (cats[c2] == 0 && samesets(g, c, c2))
1441 cats[c2] = cat;
1442 }
1443}
1444
1445/*
1446 - dupl - emit a duplicate of a bunch of sops
1447 */
1448static sopno /* start of duplicate */
1449dupl(struct parse *p,
1450 sopno start, /* from here */
1451 sopno finish) /* to this less one */
1452{
1453 sopno ret = HERE()(p->slen);
1454 sopno len = finish - start;
1455
1456 assert(finish >= start)((void)0);
1457 if (len == 0)
1458 return(ret);
1459 enlarge(p, p->ssize + len); /* this many unexpected additions */
1460 assert(p->ssize >= p->slen + len)((void)0);
1461 (void) memmove((char *)(p->strip + p->slen),
1462 (char *)(p->strip + start), (size_t)len*sizeof(sop));
1463 p->slen += len;
1464 return(ret);
1465}
1466
1467/*
1468 - doemit - emit a strip operator
1469 *
1470 * It might seem better to implement this as a macro with a function as
1471 * hard-case backup, but it's just too big and messy unless there are
1472 * some changes to the data structures. Maybe later.
1473 */
1474static void
1475doemit(struct parse *p, sop op, size_t opnd)
1476{
1477 /* avoid making error situations worse */
1478 if (p->error != 0)
1479 return;
1480
1481 /* deal with oversize operands ("can't happen", more or less) */
1482 assert(opnd < 1<<OPSHIFT)((void)0);
1483
1484 /* deal with undersized strip */
1485 if (p->slen >= p->ssize)
1486 enlarge(p, (p->ssize+1) / 2 * 3); /* +50% */
1487 assert(p->slen < p->ssize)((void)0);
1488
1489 /* finally, it's all reduced to the easy case */
1490 p->strip[p->slen++] = SOP(op, opnd)((op)|(opnd));
1491}
1492
1493/*
1494 - doinsert - insert a sop into the strip
1495 */
1496static void
1497doinsert(struct parse *p, sop op, size_t opnd, sopno pos)
1498{
1499 sopno sn;
1500 sop s;
1501 int i;
1502
1503 /* avoid making error situations worse */
1504 if (p->error != 0)
1505 return;
1506
1507 sn = HERE()(p->slen);
1508 EMIT(op, opnd)doemit(p, (sop)(op), (size_t)(opnd)); /* do checks, ensure space */
1509 assert(HERE() == sn+1)((void)0);
1510 s = p->strip[sn];
1511
1512 /* adjust paren pointers */
1513 assert(pos > 0)((void)0);
1514 for (i = 1; i < NPAREN10; i++) {
1515 if (p->pbegin[i] >= pos) {
1516 p->pbegin[i]++;
1517 }
1518 if (p->pend[i] >= pos) {
1519 p->pend[i]++;
1520 }
1521 }
1522
1523 memmove((char *)&p->strip[pos+1], (char *)&p->strip[pos],
1524 (HERE()(p->slen)-pos-1)*sizeof(sop));
1525 p->strip[pos] = s;
1526}
1527
1528/*
1529 - dofwd - complete a forward reference
1530 */
1531static void
1532dofwd(struct parse *p, sopno pos, sop value)
1533{
1534 /* avoid making error situations worse */
1535 if (p->error != 0)
1536 return;
1537
1538 assert(value < 1<<OPSHIFT)((void)0);
1539 p->strip[pos] = OP(p->strip[pos])((p->strip[pos])&0xf8000000LU) | value;
1540}
1541
1542/*
1543 - enlarge - enlarge the strip
1544 */
1545static void
1546enlarge(struct parse *p, sopno size)
1547{
1548 sop *sp;
1549
1550 if (p->ssize >= size)
1551 return;
1552
1553 if ((uintptr_t)size > SIZE_MAX0xffffffffffffffffUL / sizeof(sop)) {
1554 SETERROR(REG_ESPACE)seterr(p, (12));
1555 return;
1556 }
1557
1558 sp = (sop *)realloc(p->strip, size*sizeof(sop));
1559 if (sp == NULL((void*)0)) {
1560 SETERROR(REG_ESPACE)seterr(p, (12));
1561 return;
1562 }
1563 p->strip = sp;
1564 p->ssize = size;
1565}
1566
1567/*
1568 - stripsnug - compact the strip
1569 */
1570static void
1571stripsnug(struct parse *p, struct re_guts *g)
1572{
1573 g->nstates = p->slen;
1574 if ((uintptr_t)p->slen > SIZE_MAX0xffffffffffffffffUL / sizeof(sop)) {
1575 g->strip = p->strip;
1576 SETERROR(REG_ESPACE)seterr(p, (12));
1577 return;
1578 }
1579
1580 g->strip = (sop *)realloc((char *)p->strip, p->slen * sizeof(sop));
1581 if (g->strip == NULL((void*)0)) {
1582 SETERROR(REG_ESPACE)seterr(p, (12));
1583 g->strip = p->strip;
1584 }
1585}
1586
1587/*
1588 - findmust - fill in must and mlen with longest mandatory literal string
1589 *
1590 * This algorithm could do fancy things like analyzing the operands of |
1591 * for common subsequences. Someday. This code is simple and finds most
1592 * of the interesting cases.
1593 *
1594 * Note that must and mlen got initialized during setup.
1595 */
1596static void
1597findmust(struct parse *p, struct re_guts *g)
1598{
1599 sop *scan;
1600 sop *start = 0; /* start initialized in the default case, after that */
1601 sop *newstart = 0; /* newstart was initialized in the OCHAR case */
1602 sopno newlen;
1603 sop s;
1604 char *cp;
1605 sopno i;
1606
1607 /* avoid making error situations worse */
1608 if (p->error != 0)
1
Assuming field 'error' is equal to 0
→
2
←
Taking false branch
→
1609 return;
1610
1611 /* find the longest OCHAR sequence in strip */
1612 newlen = 0;
1613 scan = g->strip + 1;
1614 do {
8
←
Loop condition is false. Exiting loop
→
1615 s = *scan++;
1616 switch (OP(s)((s)&0xf8000000LU)) {
3
←
Control jumps to the 'default' case at line 1640
→
1617 case OCHAR(2LU<<((unsigned)27)): /* sequence member */
1618 if (newlen == 0) /* new sequence */
1619 newstart = scan - 1;
1620 newlen++;
1621 break;
1622 case OPLUS_(9LU<<((unsigned)27)): /* things that don't break one */
1623 case OLPAREN(13LU<<((unsigned)27)):
1624 case ORPAREN(14LU<<((unsigned)27)):
1625 break;
1626 case OQUEST_(11LU<<((unsigned)27)): /* things that must be skipped */
1627 case OCH_(15LU<<((unsigned)27)):
1628 scan--;
1629 do {
1630 scan += OPND(s)((s)&0x07ffffffLU);
1631 s = *scan;
1632 /* assert() interferes w debug printouts */
1633 if (OP(s)((s)&0xf8000000LU) != O_QUEST(12LU<<((unsigned)27)) && OP(s)((s)&0xf8000000LU) != O_CH(18LU<<((unsigned)27)) &&
1634 OP(s)((s)&0xf8000000LU) != OOR2(17LU<<((unsigned)27))) {
1635 g->iflags |= REGEX_BAD04;
1636 return;
1637 }
1638 } while (OP(s)((s)&0xf8000000LU) != O_QUEST(12LU<<((unsigned)27)) && OP(s)((s)&0xf8000000LU) != O_CH(18LU<<((unsigned)27)));
1639 LLVM_FALLTHROUGH__attribute__((fallthrough));
1640 default: /* things that break a sequence */
1641 if (newlen > g->mlen) { /* ends one */
4
←
Assuming 'newlen' is <= field 'mlen'
→
5
←
Taking false branch
→
1642 start = newstart;
1643 g->mlen = newlen;
1644 }
1645 newlen = 0;
1646 break;
6
←
 Execution continues on line 1648
→
1647 }
1648 } while (OP(s)((s)&0xf8000000LU) != OEND(1LU<<((unsigned)27)));
7
←
Assuming the condition is false
→
1649
1650 if (g->mlen == 0) /* there isn't one */
9
←
Assuming field 'mlen' is not equal to 0
→
10
←
Taking false branch
→
1651 return;
1652
1653 /* turn it into a character string */
1654 g->must = malloc((size_t)g->mlen + 1);
1655 if (g->must == NULL((void*)0)) { /* argh; just forget it */
11
←
Assuming field 'must' is not equal to NULL
→
12
←
Taking false branch
→
1656 g->mlen = 0;
1657 return;
1658 }
1659 cp = g->must;
1660 scan = start;
1661 for (i = g->mlen; i
12.1
'i' is > 0
 > 0; i--) {
13
←
Loop condition is true. Entering loop body
→
1662 while (OP(s = *scan++)((s = *scan++)&0xf8000000LU) != OCHAR(2LU<<((unsigned)27)))
14
←
Null pointer value stored to 'scan'
→
15
←
Dereference of null pointer
1663 continue;
1664 assert(cp < g->must + g->mlen)((void)0);
1665 *cp++ = (char)OPND(s)((s)&0x07ffffffLU);
1666 }
1667 assert(cp == g->must + g->mlen)((void)0);
1668 *cp++ = '\0'; /* just on general principles */
1669}
1670
1671/*
1672 - pluscount - count + nesting
1673 */
1674static sopno /* nesting depth */
1675pluscount(struct parse *p, struct re_guts *g)
1676{
1677 sop *scan;
1678 sop s;
1679 sopno plusnest = 0;
1680 sopno maxnest = 0;
1681
1682 if (p->error != 0)
1683 return(0); /* there may not be an OEND */
1684
1685 scan = g->strip + 1;
1686 do {
1687 s = *scan++;
1688 switch (OP(s)((s)&0xf8000000LU)) {
1689 case OPLUS_(9LU<<((unsigned)27)):
1690 plusnest++;
1691 break;
1692 case O_PLUS(10LU<<((unsigned)27)):
1693 if (plusnest > maxnest)
1694 maxnest = plusnest;
1695 plusnest--;
1696 break;
1697 }
1698 } while (OP(s)((s)&0xf8000000LU) != OEND(1LU<<((unsigned)27)));
1699 if (plusnest != 0)
1700 g->iflags |= REGEX_BAD04;
1701 return(maxnest);
1702}