svn commit: r233289 - stable/9/bin/sh

[Jilles Tjoelker]

Author: jilles
Date: Wed Mar 21 23:10:16 2012
New Revision: 233289
URL: http://svn.freebsd.org/changeset/base/233289

Log:
  MFC r229201: sh: Make patmatch() non-recursive.

Modified:
  stable/9/bin/sh/expand.c
Directory Properties:
  stable/9/bin/sh/   (props changed)

Modified: stable/9/bin/sh/expand.c
==============================================================================
--- stable/9/bin/sh/expand.c	Wed Mar 21 20:55:21 2012	(r233288)
+++ stable/9/bin/sh/expand.c	Wed Mar 21 23:10:16 2012	(r233289)
@@ -1440,57 +1440,63 @@ static int
 patmatch(const char *pattern, const char *string, int squoted)
 {
 	const char *p, *q, *end;
+	const char *bt_p, *bt_q;
 	char c;
 	wchar_t wc, wc2;
 
 	p = pattern;
 	q = string;
+	bt_p = NULL;
+	bt_q = NULL;
 	for (;;) {
 		switch (c = *p++) {
 		case '\0':
-			goto breakloop;
+			if (*q != '\0')
+				goto backtrack;
+			return 1;
 		case CTLESC:
 			if (squoted && *q == CTLESC)
 				q++;
 			if (*q++ != *p++)
-				return 0;
+				goto backtrack;
 			break;
 		case CTLQUOTEMARK:
 			continue;
 		case '?':
 			if (squoted && *q == CTLESC)
 				q++;
-			if (localeisutf8)
+			if (*q == '\0')
+				return 0;
+			if (localeisutf8) {
 				wc = get_wc(&q);
-			else
+				/*
+				 * A '?' does not match invalid UTF-8 but a
+				 * '*' does, so backtrack.
+				 */
+				if (wc == 0)
+					goto backtrack;
+			} else
 				wc = (unsigned char)*q++;
-			if (wc == '\0')
-				return 0;
 			break;
 		case '*':
 			c = *p;
 			while (c == CTLQUOTEMARK || c == '*')
 				c = *++p;
-			if (c != CTLESC &&  c != CTLQUOTEMARK &&
-			    c != '?' && c != '*' && c != '[') {
-				while (*q != c) {
-					if (squoted && *q == CTLESC &&
-					    q[1] == c)
-						break;
-					if (*q == '\0')
-						return 0;
-					if (squoted && *q == CTLESC)
-						q++;
-					q++;
-				}
-			}
-			do {
-				if (patmatch(p, q, squoted))
-					return 1;
-				if (squoted && *q == CTLESC)
-					q++;
-			} while (*q++ != '\0');
-			return 0;
+			/*
+			 * If the pattern ends here, we know the string
+			 * matches without needing to look at the rest of it.
+			 */
+			if (c == '\0')
+				return 1;
+			/*
+			 * First try the shortest match for the '*' that
+			 * could work. We can forget any earlier '*' since
+			 * there is no way having it match more characters
+			 * can help us, given that we are already here.
+			 */
+			bt_p = p;
+			bt_q = q;
+			break;
 		case '[': {
 			const char *endp;
 			int invert, found;
@@ -1502,7 +1508,7 @@ patmatch(const char *pattern, const char
 			for (;;) {
 				while (*endp == CTLQUOTEMARK)
 					endp++;
-				if (*endp == '\0')
+				if (*endp == 0)
 					goto dft;		/* no matching ] */
 				if (*endp == CTLESC)
 					endp++;
@@ -1517,12 +1523,14 @@ patmatch(const char *pattern, const char
 			found = 0;
 			if (squoted && *q == CTLESC)
 				q++;
-			if (localeisutf8)
+			if (*q == '\0')
+				return 0;
+			if (localeisutf8) {
 				chr = get_wc(&q);
-			else
+				if (chr == 0)
+					goto backtrack;
+			} else
 				chr = (unsigned char)*q++;
-			if (chr == '\0')
-				return 0;
 			c = *p++;
 			do {
 				if (c == CTLQUOTEMARK)
@@ -1563,21 +1571,34 @@ patmatch(const char *pattern, const char
 				}
 			} while ((c = *p++) != ']');
 			if (found == invert)
-				return 0;
+				goto backtrack;
 			break;
 		}
 dft:	        default:
 			if (squoted && *q == CTLESC)
 				q++;
-			if (*q++ != c)
+			if (*q == '\0')
+				return 0;
+			if (*q++ == c)
+				break;
+backtrack:
+			/*
+			 * If we have a mismatch (other than hitting the end
+			 * of the string), go back to the last '*' seen and
+			 * have it match one additional character.
+			 */
+			if (bt_p == NULL)
+				return 0;
+			if (squoted && *bt_q == CTLESC)
+				bt_q++;
+			if (*bt_q == '\0')
 				return 0;
+			bt_q++;
+			p = bt_p;
+			q = bt_q;
 			break;
 		}
 	}
-breakloop:
-	if (*q != '\0')
-		return 0;
-	return 1;
 }