file/regex.patch

diff -bBur file-3.20.1/apprentice.c zzz-file/apprentice.c
--- file-3.20.1/apprentice.c	Wed Jun 18 23:58:04 1997
+++ zzz-file/apprentice.c	Wed Mar 12 17:07:11 1997
@@ -45,7 +45,7 @@
 
 static int getvalue	__P((struct magic *, char **));
 static int hextoint	__P((int));
-static char *getstr	__P((char *, char *, int, int *));
+static char *getstr	__P((char *, char *, int, int *, char));
 static int parse	__P((char *, int *, int));
 static void eatsize	__P((char **));
 
@@ -162,6 +162,7 @@
 			v = (long) v;
 			break;
 		case STRING:
+		case REGEX:
 			break;
 		default:
 			magwarn("can't happen: m->type=%d\n",
@@ -276,6 +277,7 @@
 #define NLESHORT	7
 #define NLELONG		6
 #define NLEDATE		6
+#define NREGEX		5
 
 	if (*l == 'u') {
 		++l;
@@ -316,6 +318,10 @@
 	} else if (strncmp(l, "ledate", NLEDATE)==0) {
 		m->type = LEDATE;
 		l += NLEDATE;
+	} else if (strncmp(l, "regex", NREGEX)==0) {
+		m->type = REGEX;
+		m->reln = '=';
+		l += NREGEX;
 	} else {
 		magwarn("type %s invalid", l);
 		return -1;
@@ -329,7 +335,7 @@
 		m->mask = ~0L;
 	EATAB;
   
-	switch (*l) {
+	if (m->type!=REGEX) switch (*l) {	// regex ignores reln
 	case '>':
 	case '<':
 	/* Old-style anding: "0 byte &0x80 dynamically linked" */
@@ -402,8 +408,8 @@
 {
 	int slen;
 
-	if (m->type == STRING) {
-		*p = getstr(*p, m->value.s, sizeof(m->value.s), &slen);
+	if (m->type == STRING || m->type == REGEX) {
+		*p = getstr(*p, m->value.s, sizeof(m->value.s), &slen, m->type==REGEX);
 		m->vallen = slen;
 	} else
 		if (m->reln != 'x') {
@@ -420,18 +426,25 @@
  * Return updated scan pointer as function result.
  */
 static char *
-getstr(s, p, plen, slen)
+getstr(s, p, plen, slen, regex)
 register char	*s;
 register char	*p;
 int	plen, *slen;
+char	regex;
 {
 	char	*origs = s, *origp = p;
 	char	*pmax = p + plen - 1;
 	register int	c;
 	register int	val;
 
+	if (regex)
+	{
+		regex = *s++;
+		if (regex==0) goto out;
+	}
+
 	while ((c = *s++) != '\0') {
-		if (isspace((unsigned char) c))
+		if ((regex && (c==regex)) || (!regex && isspace((unsigned char) c)))
 			break;
 		if (p >= pmax) {
 			fprintf(stderr, "String too long: %s\n", origs);
@@ -519,6 +532,7 @@
 		} else
 			*p++ = (char)c;
 	}
+	if (regex) s++;
 out:
 	*p = '\0';
 	*slen = p - origp;
diff -bBur file-3.20.1/print.c zzz-file/print.c
--- file-3.20.1/print.c	Wed Jun 18 23:58:19 1997
+++ zzz-file/print.c	Wed Mar 12 17:09:30 1997
@@ -52,7 +52,7 @@
 	static char *typ[] = {   "invalid", "byte", "short", "invalid",
 				 "long", "string", "date", "beshort",
 				 "belong", "bedate", "leshort", "lelong",
-				 "ledate" };
+				 "ledate","regex" };
 	(void) fputc('[', stderr);
 	(void) fprintf(stderr, ">>>>>>>> %d" + 8 - (m->cont_level & 7),
 		       m->offset);
@@ -84,6 +84,7 @@
 	    case BELONG:
 		    (void) fprintf(stderr, "%ld", m->value.l);
 		    break;
+	    case REGEX:
 	    case STRING:
 		    showstr(stderr, m->value.s, -1);
 		    break;
diff -bBur file-3.20.1/softmagic.c zzz-file/softmagic.c
--- file-3.20.1/softmagic.c	Thu Jun 19 00:36:54 1997
+++ zzz-file/softmagic.c	Wed Mar 12 18:45:11 1997
@@ -30,10 +30,11 @@
 #include <string.h>
 #include <time.h>
 #include <sys/types.h>
+#include <regex.h>
 
 #include "file.h"
 
-#if 0
+#ifndef	lint
 static char *moduleid = 
 	"@(#)$Id: softmagic.c,v 1.31 1996/06/22 22:04:22 christos Exp $";
 #endif	/* lint */
@@ -233,12 +234,17 @@
 		else {
 			/* nick@feedback.com.ar   Only print things before first newline */
 			char *qqq;
-			if ((qqq=strchr(p->s,'\n'))) *qqq='\0';
+			if (qqq=strchr(p->s,'\n')) *qqq='\0';
 			(void) printf(m->desc, p->s);
 			t = m->offset + strlen(p->s);
 		}
 		break;
 
+  	case REGEX:
+		(void) fputs(m->desc,stdout);
+		t = m->offset + 32;	/* FIXME! */
+		break;
+
 	case DATE:
 	case BEDATE:
 	case LEDATE:
@@ -247,7 +253,7 @@
 			*rt = '\0';
 		(void) printf(m->desc, pp);
 		t = m->offset + sizeof(time_t);
-		break;
+		return;
 	default:
 		error("invalid m->type (%d) in mprint().\n", m->type);
 		/*NOTREACHED*/
@@ -268,6 +274,7 @@
 	case SHORT:
 	case LONG:
 	case DATE:
+	case REGEX:
 		return 1;
 	case STRING:
 		{
@@ -376,6 +383,24 @@
 	return 1;
 }
 
+static int regex_match(const char *s, const char *pat)
+{
+	regex_t reg;
+	regmatch_t *rm;
+	int err;
+
+	err=regcomp(&reg,pat,REG_NEWLINE);
+	if (err)
+	{
+		error("Bad regex pattern: %s\n",pat);
+	}
+	rm=(regmatch_t*) malloc( sizeof(regmatch_t) * (reg.re_nsub+1) );
+	err=regexec(&reg,s,reg.re_nsub+1,rm,0);
+	regfree(&reg);
+	free(rm);
+	return err;
+}
+
 static int
 mcheck(p, m)
 union VALUETYPE* p;
@@ -428,6 +453,10 @@
 				if ((v = *b++ - *a++) != 0)
 					break;
 		}
+		break;
+	case REGEX:
+		l = 0;
+		v = regex_match(p->s, m->value.s);
 		break;
 	default:
 		error("invalid type %d in mcheck().\n", m->type);