Reporter:  Warren Jones <wjones@tc.fluke.com> Tue Mar 26 17:28:31 1996

Problem:   When using the "-l" flag, symbolic links can result in
           multiple index entries for a single document or (worse yet)
           endless looping.

Solution:  Use a hash to keep track of files that we've already
           indexed by device and inode.

Index: index.c
===================================================================
RCS file: /usr0/wjones/src/CVS.repo/swish/src/index.c,v
retrieving revision 1.1.1.1
diff -c -r1.1.1.1 index.c
*** index.c	1995/10/18 18:43:27	1.1.1.1
--- index.c	1995/10/25 21:11:16
***************
*** 7,12 ****
--- 7,60 ----
  
  #include "swish.h"
  #include "index.h"
+ #include "hash.h"
+ 
+ /* Have we already indexed a file or directory?
+ ** This function is used to avoid multiple index entries
+ ** or endless looping due to symbolic links.
+ */
+ 
+ int already_indexed( path )
+      char *path;
+ {
+ 	static struct dev_ino {
+ 	    dev_t  dev;
+ 	    ino_t  ino;
+ 	    struct dev_ino *next;
+ 	} *inode_hash[BIGHASHSIZE], *p;
+ 
+ 	struct stat buf;
+ 	char key[34];		/* Hash key -- allow for 64 bit inodes */
+ 	unsigned hashval;
+  
+ 	if ( stat( path, &buf ) )
+ 		return 0;
+ 
+ 	/* Create hash key:  string contains device and inode. */
+ 	sprintf( key, "%lx/%lx", (unsigned long)buf.st_dev,
+ 				 (unsigned long)buf.st_ino  );
+ 
+ 	hashval = bighash(key);	/* Search hash for this file. */
+ 	for ( p = inode_hash[hashval]; p != NULL; p = p->next )
+ 		if ( p->dev == buf.st_dev &&
+ 		     p->ino == buf.st_ino    )
+ 		{				/* We found it. */
+ 			if ( verbose == 3 )
+ 				printf( "Skipping %s:  %s\n",
+ 				        path, "Already indexed." );
+ 			return 1;
+ 		}
+ 
+ 	/* Not found, make new entry. */
+ 	p = (struct dev_ino*)emalloc(sizeof(struct dev_ino));
+ 	p->dev = buf.st_dev;
+ 	p->ino = buf.st_ino;
+ 	p->next = inode_hash[hashval];
+ 	inode_hash[hashval] = p;
+ 
+ 	return 0;
+ }
+ 
  
  /* Recursively goes into a directory and calls the word-indexing
  ** functions for each file that's found.
***************
*** 31,36 ****
--- 79,87 ----
  	if (islink(dir) && !followsymlinks)
  		return;
  
+ 	if ( already_indexed(dir) )
+ 		return;
+ 
          if (dir[strlen(dir) - 1] == '/')
                  dir[strlen(dir) - 1] = '\0';
  
***************
*** 103,108 ****
--- 154,162 ----
  
                  if (!isdirectory(s)) {
  
+ 			if ( already_indexed(s) )
+ 				continue;
+ 
  			if (!isoksuffix(dp->d_name, suffixlist))
  				continue;
  
***************
*** 144,149 ****
--- 198,206 ----
          struct swline *tmplist;
  
  	if (islink(path) && !followsymlinks)
+ 		return;
+ 
+ 	if ( already_indexed(path) )
  		return;
  
          if (path[strlen(path) - 1] == '/')