Path: utzoo!utgpu!jarvis.csri.toronto.edu!rutgers!njin!princeton!notecnirp!nfs
From: nfs@notecnirp.Princeton.EDU (Norbert Schlenker)
Newsgroups: comp.os.minix
Subject: "Bug" in strstr() and fix
Message-ID: <19847@princeton.Princeton.EDU>
Date: 4 Oct 89 11:13:51 GMT
Sender: news@princeton.Princeton.EDU
Reply-To: nfs@notecnirp.UUCP (Norbert Schlenker)
Organization: Dept. of Computer Science, Princeton University
Lines: 42

There is a bug (actually a misfeature, since the comments seem to
indicate intention) in Henry Spencer's strstr() routine.  The most
recent draft of the ANSI C standard (in 4.11.5.7) says that, if the
second argument is a zero length string, strstr() should return its
first argument.  As written, the code returns a pointer to the zero
byte ending the first string.

The fix is enclosed below as a cdiff.

------------------------------  Cut here  -----------------------------
*** strstr.c.orig	Sun Oct  1 15:18:38 1989
--- strstr.c	Wed Oct  4 06:04:48 1989
***************
*** 16,29 ****
  	extern SIZET strlen();
  
  	/*
! 	 * The odd placement of the two tests is so "" is findable.
! 	 * Also, we inline the first char for speed.
! 	 * The ++ on scan has been moved down for optimization.
! 	 */
! 	firstc = *wanted;
! 	len = strlen(wanted);
! 	for (scan = s; *scan != firstc || strncmp(scan, wanted, len) != 0; )
! 		if (*scan++ == '\0')
! 			return(NULL);
! 	return(scan);
! }
--- 16,28 ----
  	extern SIZET strlen();
  
  	/*
! 	 * We inline the first char for speed.
! 	 */
! 	if ((firstc = *wanted) == '\0')
! 		return(s);
! 	len = strlen(wanted);
! 	for (scan = s; *scan != '\0'; scan++)
! 		if (*scan == firstc && strncmp(scan, wanted, len) == 0)
! 			return(scan);
! 	return(NULL);
! }