7 files changed, 89 insertions, 52 deletions

diff --git a/src/regress/lib/libc/regex/Makefile b/src/regress/lib/libc/regex/Makefile
index 93b7bb9052..271debb5d3 100644
--- a/src/regress/lib/libc/regex/Makefile
+++ b/src/regress/lib/libc/regex/Makefile
@@ -1,16 +1,21 @@
+#       $OpenBSD: Makefile,v 1.6 2004/08/13 14:54:12 millert Exp $
 #       $NetBSD: Makefile,v 1.2 1995/02/16 19:38:45 cgd Exp $
 
 PROG=   re
-SRCS=   main.c split.c debug.c
-NOMAN=
+SRCS=   main.c split.c debug.c regcomp.c regerror.c regexec.c regfree.c
+.PATH: ${.CURDIR}/../../../../lib/libc/regex
 
-CFLAGS+= -I${.CURDIR}/../../../../lib/libc/regex
+CFLAGS+= -I${.CURDIR}/../../../../lib/libc/regex -DREDEBUG -DPOSIX_MISTAKE
 
 TESTS=  ${.CURDIR}/tests
 
-regress:
+REGRESS_TARGETS=do-reg do-reg-long do-reg-backref
+
+do-reg: ${PROG}
         ./re < ${TESTS}
+do-reg-long: ${PROG}
         ./re -el < ${TESTS}
+do-reg-backref: ${PROG}
         ./re -er < ${TESTS}
 
-.include <bsd.prog.mk>
+.include <bsd.regress.mk>
diff --git a/src/regress/lib/libc/regex/debug.c b/src/regress/lib/libc/regex/debug.c
index 861f550611..11129e7249 100644
--- a/src/regress/lib/libc/regex/debug.c
+++ b/src/regress/lib/libc/regex/debug.c
@@ -1,3 +1,4 @@
+/*      $OpenBSD: debug.c,v 1.4 2003/07/31 21:48:03 deraadt Exp $       */
 /*      $NetBSD: debug.c,v 1.2 1995/04/20 22:39:42 cgd Exp $    */
 
 #include <stdio.h>
@@ -218,7 +219,7 @@ FILE *d;
                         fprintf(d, ">");
                         break;
                 default:
-                        fprintf(d, "!%d(%d)!", OP(*s), opnd);
+                        fprintf(d, "!%ld(%ld)!", (long)OP(*s), (long)opnd);
                         break;
                 }
                 if (!done)
@@ -237,8 +238,8 @@ int ch;
         static char buf[10];
 
         if (isprint(ch) || ch == ' ')
-                sprintf(buf, "%c", ch);
+                snprintf(buf, sizeof buf, "%c", ch);
         else
-                sprintf(buf, "\\%o", ch);
+                snprintf(buf, sizeof buf, "\\%o", ch);
         return(buf);
 }
diff --git a/src/regress/lib/libc/regex/debug.ih b/src/regress/lib/libc/regex/debug.ih
index fb9bac0c75..9eb313af23 100644
--- a/src/regress/lib/libc/regex/debug.ih
+++ b/src/regress/lib/libc/regex/debug.ih
@@ -1,3 +1,4 @@
+/*      $OpenBSD: debug.ih,v 1.3 2002/02/16 21:27:32 millert Exp $      */
 /*      $NetBSD: debug.ih,v 1.2 1995/04/20 22:39:47 cgd Exp $   */
 
 /* ========= begin header generated by ./mkh ========= */
@@ -6,9 +7,9 @@ extern "C" {
 #endif
 
 /* === debug.c === */
-void regprint __P((regex_t *r, FILE *d));
-static void s_print __P((register struct re_guts *g, FILE *d));
-static char *regchar __P((int ch));
+void regprint(regex_t *r, FILE *d);
+static void s_print(register struct re_guts *g, FILE *d);
+static char *regchar(int ch);
 
 #ifdef __cplusplus
 }
diff --git a/src/regress/lib/libc/regex/main.c b/src/regress/lib/libc/regex/main.c
index 8d88a8b9b8..e0ed86f5e8 100644
--- a/src/regress/lib/libc/regex/main.c
+++ b/src/regress/lib/libc/regex/main.c
@@ -1,10 +1,13 @@
+/*      $OpenBSD: main.c,v 1.7 2007/09/12 19:32:35 otto Exp $   */
 /*      $NetBSD: main.c,v 1.2 1995/04/20 22:39:51 cgd Exp $     */
 
 #include <stdio.h>
+#include <stdlib.h>
 #include <string.h>
 #include <sys/types.h>
 #include <regex.h>
 #include <assert.h>
+#include <unistd.h>
 
 #include "main.ih"
 
@@ -19,15 +22,15 @@ regoff_t startoff = 0;
 regoff_t endoff = 0;
 
 
-extern int split();
-extern void regprint();
+extern int split(char *, char *[], int, char *);
+extern void regprint(regex_t *, FILE *);
 
 /*
  - main - do the simple case, hand off to regress() for regression
  */
-main(argc, argv)
-int argc;
-char *argv[];
+int
+main(int argc, char *argv[])
+
 {
         regex_t re;
 #       define  NS      10
@@ -43,7 +46,7 @@ char *argv[];
 
         progname = argv[0];
 
-        while ((c = getopt(argc, argv, "c:e:S:E:x")) != EOF)
+        while ((c = getopt(argc, argv, "c:e:S:E:x")) != -1)
                 switch (c) {
                 case 'c':       /* compile options */
                         copts = options('c', optarg);
@@ -79,7 +82,7 @@ char *argv[];
         err = regcomp(&re, argv[optind++], copts);
         if (err) {
                 len = regerror(err, &re, erbuf, sizeof(erbuf));
-                fprintf(stderr, "error %s, %d/%d `%s'\n",
+                fprintf(stderr, "error %s, %zu/%zu `%s'\n",
                         eprint(err), len, sizeof(erbuf), erbuf);
                 exit(status);
         }
@@ -97,15 +100,15 @@ char *argv[];
         err = regexec(&re, argv[optind], (size_t)NS, subs, eopts);
         if (err) {
                 len = regerror(err, &re, erbuf, sizeof(erbuf));
-                fprintf(stderr, "error %s, %d/%d `%s'\n",
+                fprintf(stderr, "error %s, %zu/%zu `%s'\n",
                         eprint(err), len, sizeof(erbuf), erbuf);
                 exit(status);
         }
         if (!(copts&REG_NOSUB)) {
-                len = (int)(subs[0].rm_eo - subs[0].rm_so);
+                len = (size_t)(subs[0].rm_eo - subs[0].rm_so);
                 if (subs[0].rm_so != -1) {
                         if (len != 0)
-                                printf("match `%.*s'\n", len,
+                                printf("match `%.*s'\n", (int)len,
                                         argv[optind] + subs[0].rm_so);
                         else
                                 printf("match `'@%.1s\n",
@@ -144,7 +147,7 @@ FILE *in;
                 line++;
                 if (inbuf[0] == '#' || inbuf[0] == '\n')
                         continue;                       /* NOTE CONTINUE */
-                inbuf[strlen(inbuf)-1] = '\0';  /* get rid of stupid \n */
+                inbuf[strcspn(inbuf, "\n")] = '\0';     /* get rid of stupid \n */
                 if (debug)
                         fprintf(stdout, "%d:\n", line);
                 nf = split(inbuf, f, MAXF, "\t\t");
@@ -225,14 +228,14 @@ int opts;			/* may not match f1 */
         char f0copy[1000];
         char f2copy[1000];
 
-        strcpy(f0copy, f0);
+        strlcpy(f0copy, f0, sizeof f0copy);
         re.re_endp = (opts&REG_PEND) ? f0copy + strlen(f0copy) : NULL;
         fixstr(f0copy);
         err = regcomp(&re, f0copy, opts);
         if (err != 0 && (!opt('C', f1) || err != efind(f2))) {
                 /* unexpected error or wrong error */
                 len = regerror(err, &re, erbuf, sizeof(erbuf));
-                fprintf(stderr, "%d: %s error %s, %d/%d `%s'\n",
+                fprintf(stderr, "%d: %s error %s, %d/%zu `%s'\n",
                                         line, type, eprint(err), len,
                                         sizeof(erbuf), erbuf);
                 status = 1;
@@ -249,7 +252,7 @@ int opts;			/* may not match f1 */
                 return;
         }
 
-        strcpy(f2copy, f2);
+        strlcpy(f2copy, f2, sizeof f2copy);
         fixstr(f2copy);
 
         if (options('e', f1)&REG_STARTEND) {
@@ -263,7 +266,7 @@ int opts;			/* may not match f1 */
         if (err != 0 && (f3 != NULL || err != REG_NOMATCH)) {
                 /* unexpected error or wrong error */
                 len = regerror(err, &re, erbuf, sizeof(erbuf));
-                fprintf(stderr, "%d: %s exec error %s, %d/%d `%s'\n",
+                fprintf(stderr, "%d: %s exec error %s, %d/%zu `%s'\n",
                                         line, type, eprint(err), len,
                                         sizeof(erbuf), erbuf);
                 status = 1;
@@ -427,8 +430,9 @@ char *should;
                                 (sub.rm_so != -1 && sub.rm_eo == -1) ||
                                 (sub.rm_so != -1 && sub.rm_so < 0) ||
                                 (sub.rm_eo != -1 && sub.rm_eo < 0) ) {
-                sprintf(grump, "start %ld end %ld", (long)sub.rm_so,
-                                                        (long)sub.rm_eo);
+                snprintf(grump, sizeof grump,
+                    "start %ld end %ld", (long)sub.rm_so,
+                    (long)sub.rm_eo);
                 return(grump);
         }
 
@@ -440,8 +444,9 @@ char *should;
 
         /* check for in range */
         if (sub.rm_eo > strlen(str)) {
-                sprintf(grump, "start %ld end %ld, past end of string",
-                                        (long)sub.rm_so, (long)sub.rm_eo);
+                snprintf(grump, sizeof grump,
+                        "start %ld end %ld, past end of string",
+                        (long)sub.rm_so, (long)sub.rm_eo);
                 return(grump);
         }
 
@@ -451,13 +456,13 @@ char *should;
 
         /* check for not supposed to match */
         if (should == NULL) {
-                sprintf(grump, "matched `%.*s'", len, p);
+                snprintf(grump, sizeof grump, "matched `%.*s'", len, p);
                 return(grump);
         }
 
         /* check for wrong match */
         if (len != shlen || strncmp(p, should, (size_t)shlen) != 0) {
-                sprintf(grump, "matched `%.*s' instead", len, p);
+                snprintf(grump, sizeof grump, "matched `%.*s' instead", len, p);
                 return(grump);
         }
         if (shlen > 0)
@@ -470,7 +475,7 @@ char *should;
         if (shlen == 0)
                 shlen = 1;      /* force check for end-of-string */
         if (strncmp(p, at, shlen) != 0) {
-                sprintf(grump, "matched null at `%.20s'", p);
+                snprintf(grump, sizeof grump, "matched null at `%.20s'", p);
                 return(grump);
         }
         return(NULL);
@@ -501,10 +506,9 @@ efind(name)
 char *name;
 {
         static char efbuf[100];
-        size_t n;
         regex_t re;
 
-        sprintf(efbuf, "REG_%s", name);
+        snprintf(efbuf, sizeof efbuf, "REG_%s", name);
         assert(strlen(efbuf) < sizeof(efbuf));
         re.re_endp = efbuf;
         (void) regerror(REG_ATOI, &re, efbuf, sizeof(efbuf));
diff --git a/src/regress/lib/libc/regex/main.ih b/src/regress/lib/libc/regex/main.ih
index 135e3e792d..0860e26333 100644
--- a/src/regress/lib/libc/regex/main.ih
+++ b/src/regress/lib/libc/regex/main.ih
@@ -1,3 +1,4 @@
+/*      $OpenBSD: main.ih,v 1.3 2002/02/16 21:27:32 millert Exp $       */
 /*      $NetBSD: main.ih,v 1.2 1995/04/20 22:39:55 cgd Exp $    */
 
 /* ========= begin header generated by ./mkh ========= */
@@ -6,14 +7,14 @@ extern "C" {
 #endif
 
 /* === main.c === */
-void regress __P((FILE *in));
-void try __P((char *f0, char *f1, char *f2, char *f3, char *f4, int opts));
-int options __P((int type, char *s));
-int opt __P((int c, char *s));
-void fixstr __P((register char *p));
-char *check __P((char *str, regmatch_t sub, char *should));
-static char *eprint __P((int err));
-static int efind __P((char *name));
+void regress(FILE *in);
+void try(char *f0, char *f1, char *f2, char *f3, char *f4, int opts);
+int options(int type, char *s);
+int opt(int c, char *s);
+void fixstr(register char *p);
+char *check(char *str, regmatch_t sub, char *should);
+static char *eprint(int err);
+static int efind(char *name);
 
 #ifdef __cplusplus
 }
diff --git a/src/regress/lib/libc/regex/split.c b/src/regress/lib/libc/regex/split.c
index dd1ca14480..fcd81a3503 100644
--- a/src/regress/lib/libc/regex/split.c
+++ b/src/regress/lib/libc/regex/split.c
@@ -1,18 +1,17 @@
+/*      $OpenBSD: split.c,v 1.5 2007/09/09 23:25:12 chl Exp $   */
 /*      $NetBSD: split.c,v 1.2 1995/04/20 22:39:57 cgd Exp $    */
 
 #include <stdio.h>
 #include <string.h>
 
+int split(char *string, char *fields[], int nfields, char *sep);
+
 /*
  - split - divide a string into fields, like awk split()
  = int split(char *string, char *fields[], int nfields, char *sep);
  */
 int                             /* number of fields, including overflow */
-split(string, fields, nfields, sep)
-char *string;
-char *fields[];                 /* list is not NULL-terminated */
-int nfields;                    /* number of entries available in fields[] */
-char *sep;                      /* "" white, "c" single char, "ab" [ab]+ */
+split(char *string, char *fields[], int nfields, char *sep)
 {
         register char *p = string;
         register char c;                        /* latest character */
@@ -161,18 +160,18 @@ char *argv[];
 
         if (argc > 4)
                 for (n = atoi(argv[3]); n > 0; n--) {
-                        (void) strcpy(buf, argv[1]);
+                        (void) strlcpy(buf, argv[1], sizeof buf);
                 }
         else if (argc > 3)
                 for (n = atoi(argv[3]); n > 0; n--) {
-                        (void) strcpy(buf, argv[1]);
+                        (void) strlcpy(buf, argv[1], sizeof buf);
                         (void) split(buf, fields, MNF, argv[2]);
                 }
         else if (argc > 2)
                 dosplit(argv[1], argv[2]);
         else if (argc > 1)
                 while (fgets(buf, sizeof(buf), stdin) != NULL) {
-                        buf[strlen(buf)-1] = '\0';      /* stomp newline */
+                        buf[strcspn(buf, "\n")] = '\0'; /* stomp newline */
                         dosplit(buf, argv[1]);
                 }
         else
@@ -287,7 +286,7 @@ regress()
         register char *f;
 
         for (n = 0; tests[n].str != NULL; n++) {
-                (void) strcpy(buf, tests[n].str);
+                (void) strlcpy(buf, tests[n].str, sizeof buf);
                 fields[RNF] = NULL;
                 nf = split(buf, fields, RNF, tests[n].seps);
                 printit = 0;
diff --git a/src/regress/lib/libc/regex/tests b/src/regress/lib/libc/regex/tests
index 8e89f161b1..c827c868b7 100644
--- a/src/regress/lib/libc/regex/tests
+++ b/src/regress/lib/libc/regex/tests
@@ -1,3 +1,4 @@
+#       $OpenBSD: tests,v 1.5 2004/11/29 16:50:31 otto Exp $
 #       $NetBSD: tests,v 1.5 1995/04/20 22:40:00 cgd Exp $
 
 # regular expression test set
@@ -168,6 +169,30 @@ a\(\(b\)*\2\)*d		b	abbbd	abbbd
 \(a\)\1bc*d     b       aabcccd aabcccd
 \(a\)\1bc*[ce]d b       aabcccd aabcccd
 ^\(a\)\1b\(c\)*cd$      b       aabcccd aabcccd
+\(b*\)\(a*\1\)* b       ab      a
+\([^_]*\)\(_*\1\)*      b       foo_foo_bar_bar_bar_baz foo_foo foo,_foo
+\([^_]*\)\(_*\1\)*      b       bar_bar_bar_baz bar_bar_bar     bar,_bar
+\([^_]*\)\(_*\1\)*      b       foo_bar_baz     foo     foo
+\(.*\)\1        b       ""      ""
+\(.*\)\1        b       a       ""
+\(.*\)\1        b       aa      aa
+\(.*\)\1        b       aaa     aa
+\(.*\)\1        b       aaaa    aaaa
+\([^_]*\)\1     b       ""      ""
+\([^_]*\)\1     b       a       ""
+\([^_]*\)\1     b       aa      aa
+\([^_]*\)\1     b       aaa     aa
+\([^_]*\)\1     b       aaaa    aaaa
+foo\(.*\)bar\1  b       foolbarl        foolbarl        l
+foo\(.*\)bar\1  b       foobar  foobar  ""
+\(\(.\)b\)*\1   b       aba
+\(\(.\)b\)*\1   b       abba
+\(\(.\)b\)*\1   b       abbba
+\(\(.\)b\)*\1   b       abbbba  bbbb    bb,b
+\(\(.\)b\)*\1   b       abbbbba abbbbb  bb,b
+\(\(.\)b\)*\1   b       abbbbbba        abbbbb  bb,b
+\(\(.\)b\)*\1   b       abbbbbbbbbbbbbba        abbbbbbbbbbbbb  bb,b
+\(\(.\)b\)*\1   b       abbbbbbbbbbbbbbba       abbbbbbbbbbbbbbb        bb,b
 
 # ordinary repetitions
 ab*c            &       abc     abc
@@ -475,3 +500,4 @@ CC[13]1|a{21}[23][EO][123][Es][12]a{15}aa[34][EW]aaaaaaa[X]a	-	CC11	CC11
 Char \([a-z0-9_]*\)\[.* b       Char xyz[k      Char xyz[k      xyz
 a?b     -       ab      ab
 -\{0,1\}[0-9]*$ b       -5      -5
+