1 /* centipede.c - generate and install indexing information (view w/tabstop=4) */
16 #define DEFAULT_LDAPFILTER "(objectclass=*)"
18 #define CENTROID_VALUE 1
19 #define CENTROID_WORD 2
21 #define CENTROID_RELATIVE 1
22 #define CENTROID_FULL 2
24 #define WORD_BREAKS " -',.()!;:&$%*\"/\\+_<>=?[]|^~"
37 int srcldapauthmethod;
38 int destldapauthmethod;
42 static LDAP *start_ldap_search();
43 static LDAP *bind_to_destination_ldap();
44 static int create_tmp_files();
45 static int generate_new_centroids();
46 static LDAPMod **diff_centroids();
47 static LDAPMod **full_centroid();
48 static char **charray_add_dup();
50 static void usage( char *name )
52 fprintf( stderr, "usage: %s [options] -s url -d url attributes\n", name );
53 fprintf( stderr, "where:\n" );
54 fprintf( stderr, "\t-s url\t\t[[ldap://][host[:port]]/]searchbasedn\n");
55 fprintf( stderr, "\t-d url\t\t[[ldap://][host[:port]]/]centroidentrydn\n");
56 fprintf( stderr, "options:\n" );
57 fprintf( stderr, "\t-v \t\tturn on verbose mode\n" );
58 fprintf( stderr, "\t-n \t\tgenerate, but do not install index info\n" );
59 fprintf( stderr, "\t-f filter\tentry selection filter\n" );
60 fprintf( stderr, "\t-F \t\tgenerate a full centroid\n" );
61 fprintf( stderr, "\t-R \t\tgenerate a relative centroid\n" );
62 fprintf( stderr, "\t-w \t\tgenerate a word-based centroid\n" );
63 fprintf( stderr, "\t-t directory\tcentroid directory\n" );
64 fprintf( stderr, "\t-b binddn\tsource bind dn\n" );
65 fprintf( stderr, "\t-p passwd\tsource bind passwd (for simple auth)\n" );
66 fprintf( stderr, "\t-m authmethod\tsource authmethod \"simple\" or \"kerberos\"\n" );
67 fprintf( stderr, "\t-B binddn\tdestination bind dn\n" );
68 fprintf( stderr, "\t-P passwd\tdestination bind passwd (for simple auth)\n" );
69 fprintf( stderr, "\t-M authmethod\tdestination authmethod \"simple\" or \"kerberos\"\n" );
70 fprintf( stderr, "\t-c size\t\tldbm cache size\n" );
73 main( int argc, char **argv )
76 char *ldapsrcurl, *ldapdesturl;
93 srcldapauthmethod = LDAP_AUTH_SIMPLE;
94 destldapauthmethod = LDAP_AUTH_SIMPLE;
97 destldapbinddn = NULL;
98 destldappasswd = NULL;
99 ldapfilter = DEFAULT_LDAPFILTER;
100 centroidvalues = CENTROID_VALUE;
101 centroidtype = CENTROID_RELATIVE;
106 while ( (i = getopt( argc, argv, "s:d:c:b:B:f:FRWp:P:m:M:t:vwn" ))
109 case 's': /* source url [[ldap://][host[:port]]/]basedn */
110 ldapsrcurl = strdup( optarg );
113 case 'd': /* destination url [[ldap://][host[:port]]/]entrydn */
114 ldapdesturl = strdup( optarg );
117 case 'f': /* specify a filter */
118 ldapfilter = strdup( optarg );
121 case 'F': /* generate full centroid */
122 centroidtype = CENTROID_FULL;
125 case 'R': /* generate relative centroid */
126 centroidtype = CENTROID_RELATIVE;
129 case 'w': /* generate word centroid */
130 centroidvalues = CENTROID_WORD;
133 case 'W': /* generate weights */
137 case 't': /* temp file directory */
138 centdir = strdup( optarg );
141 case 'b': /* src bind dn */
142 srcldapbinddn = strdup( optarg );
145 case 'p': /* src bind password */
146 srcldappasswd = strdup( optarg );
151 case 'B': /* dest bind dn */
152 destldapbinddn = strdup( optarg );
155 case 'P': /* dest bind password */
156 destldappasswd = strdup( optarg );
161 case 'm': /* src bind method */
162 if ( strcasecmp( optarg, "simple" ) == 0 ) {
163 srcldapauthmethod = LDAP_AUTH_SIMPLE;
164 } else if ( strcasecmp( optarg, "kerberos" ) == 0 ) {
165 srcldapauthmethod = LDAP_AUTH_KRBV4;
167 fprintf( stderr, "%s: unknown auth method\n", optarg );
168 fputs( "expecting \"simple\" or \"kerberos\"\n", stderr );
173 case 'M': /* dest bind method */
174 if ( strcasecmp( optarg, "simple" ) == 0 ) {
175 destldapauthmethod = LDAP_AUTH_SIMPLE;
176 } else if ( strcasecmp( optarg, "kerberos" ) == 0 ) {
177 destldapauthmethod = LDAP_AUTH_KRBV4;
179 fprintf( stderr, "%s: unknown auth method\n", optarg );
180 fputs( "expecting \"simple\" or \"kerberos\"\n", stderr );
185 case 'c': /* ldbm cache size */
186 ldbmcachesize = atoi( optarg );
189 case 'v': /* turn on verbose mode */
193 case 'n': /* don't actually install index info */
202 if ( optind == argc || ldapsrcurl == NULL || ldapdesturl == NULL ) {
206 attrs = &argv[optind];
209 * open the ldap connection and start searching for the entries
210 * we will use to generate the centroids.
213 if ( (ld = start_ldap_search( ldapsrcurl, ldapfilter, attrs )) == NULL ) {
214 fprintf( stderr, "could not initiate ldap search\n" );
218 if ( create_tmp_files( attrs, &tmpfile, &ldbm ) != 0 ) {
219 fprintf( stderr, "could not create temp files\n" );
224 * go through the entries returned, building a centroid for each
225 * attribute as we go.
228 if ( (count = generate_new_centroids( ld, attrs, ldbm )) < 1 ) {
230 fprintf( stderr, "no entries matched\n" );
233 fprintf( stderr, "could not generate new centroid\n" );
239 * for each centroid we generated above, compare to the existing
240 * centroid, if any, and produce adds and deletes, or produce
241 * an entirely new centroid. in either case, update the "current"
242 * centroid version with the new one we just generated.
245 if ( (ld = bind_to_destination_ldap( ldapsrcurl, ldapdesturl )) == NULL ) {
247 "could not bind to index server, or could not create index entry\n" );
251 for ( i = 0; ldbm[i] != NULL; i++ ) {
252 /* generate the name of the existing centroid, if any */
253 s = strrchr( tmpfile[i], '/' );
255 sprintf( buf, "%s/cent.%s", tmpfile[i], attrs[i] );
258 /* generate the full centroid changes */
259 if ( centroidtype == CENTROID_FULL || (oldbm = ldbm_open( buf,
260 LDBM_WRITER, 0, ldbmcachesize )) == NULL ) {
261 if ( (mods = full_centroid( attrs[i], ldbm[i], count )) == NULL ) {
262 fprintf( stderr, "could not produce full centroid for %s\n",
267 /* generate the differential centroid changes */
269 if ( (mods = diff_centroids( attrs[i], oldbm, ldbm[i], count ))
271 fprintf( stderr, "could not diff centroids\n" );
279 printf("changes:\n");
280 for ( j = 0; mods[j] != NULL; j++ ) {
281 switch( mods[j]->mod_op ) {
283 printf( "\tadd: %s\n",mods[j]->mod_type );
285 case LDAP_MOD_DELETE:
286 printf( "\tdelete: %s\n",mods[j]->mod_type );
288 case LDAP_MOD_REPLACE:
289 printf( "\treplace: %s\n",mods[j]->mod_type );
292 if ( mods[j]->mod_values != NULL ) {
293 for ( k = 0; mods[j]->mod_values[k] != NULL; k++ ) {
294 printf( "\t\t%s\n", mods[j]->mod_values[k] );
298 printf("end changes:\n");
302 printf( "%sModifying centroid...", not ? "Not " : "" );
306 /* attempt to make the changes to the index server entry */
307 if ( !not && ldap_modify_s( ld, ldapbase, mods ) != LDAP_SUCCESS ) {
308 fprintf( stderr, "could not apply centroid modification for %s\n",
310 ldap_perror( ld, ldapbase );
312 ldap_mods_free( mods, 1 );
319 /* move the new centroid into the old one's place */
321 (void) unlink( buf );
322 if ( link( tmpfile[i], buf ) != 0 ) {
324 fprintf( stderr, "could not rename %s to %s\n", buf,
329 (void) unlink( tmpfile[i] );
333 for ( i = 0; attrs[i] != NULL; i++ ) {
334 ldbm_close( ldbm[i] );
344 * open an ldap connection, bind, and initiate the search
358 if ( strncmp( ldapsrcurl, "ldap://", 7 ) != 0 ) {
359 fputs( "Not an LDAP URL", stderr ); /* Should be smarter? */
363 if ( (s2 = strchr( s, '/' )) == NULL ) {
364 ldapbase = strdup( s );
368 ldaphost = strdup( s );
371 ldapbase = strdup( s2 + 1 );
375 printf( "Base: %s\n", ldapbase );
376 printf( "Attributes:" );
377 for ( i = 0; attrs[i] != NULL; i++ ) {
378 printf( " %s", attrs[i] );
381 printf( "Binding to source LDAP server..." );
385 if ( (ld = ldap_open( ldaphost, LDAP_PORT )) == NULL ) {
386 perror( "ldap_open" );
390 if ( ldap_bind_s( ld, srcldapbinddn, srcldappasswd, srcldapauthmethod )
392 ldap_perror( ld, "ldap_bind_s" );
397 printf( "\nInitiating search..." );
398 if ( ldap_search( ld, ldapbase, LDAP_SCOPE_SUBTREE, ldapfilter, attrs, 0 )
400 ldap_perror( ld, "ldap_search" );
413 * create the temporary ldbm files we will use to hold the new centroids
425 for ( i = 0; attrs[i] != NULL; i++ )
429 if ( (*tmpfile = (char **) malloc( i * sizeof(char *) )) == NULL ) {
433 if ( (*ldbm = (LDBM *) malloc( i * sizeof(LDBM) )) == NULL ) {
437 for ( i = 0; attrs[i] != NULL; i++ ) {
438 if ( ((*tmpfile)[i] = tempnam( centdir, NULL )) == NULL ) {
443 if ( ((*ldbm)[i] = ldbm_open( (*tmpfile)[i], LDBM_WRCREAT, 0600,
444 ldbmcachesize )) == NULL ) {
445 fprintf( stderr, "ldbm_open of \"%s\" failed\n", (*tmpfile)[i] );
446 perror( "ldbm_open" );
450 (*tmpfile)[i] = NULL;
457 * step through each entry returned from the search and generate
458 * the appropriate centroid values.
462 generate_new_centroids(
470 LDAPMessage *res, *e;
476 printf( "Generating new centroids for..." );
483 while ( (rc = ldap_result( ld, LDAP_RES_ANY, 0, NULL, &res ))
484 == LDAP_RES_SEARCH_ENTRY ) {
486 e = ldap_first_entry( ld, res );
487 dn = ldap_get_dn( ld, e );
489 /* for each attr we want to generate a centroid for */
490 for ( i = 0; attrs[i] != NULL; i++ ) {
491 if ( (val = ldap_get_values( ld, e, attrs[i] )) == NULL ) {
496 for ( j = 0; val[j] != NULL; j++ ) {
497 /* normalize the value */
498 for ( s = val[j]; *s; s++ ) {
499 if ( isascii( *s ) ) {
504 if ( isascii( last ) && isdigit( last ) ) {
508 /* generate a value-based centroid */
509 if ( centroidvalues == CENTROID_VALUE ) {
511 key.dsize = strlen( key.dptr ) + 1;
512 (void) ldbm_store( ldbm[i], key, data, LDBM_INSERT );
514 /* generate a word-based centroid */
516 for ( w = strtok( val[j], WORD_BREAKS ); w != NULL;
517 w = strtok( NULL, WORD_BREAKS ) ) {
519 key.dsize = strlen( key.dptr ) + 1;
520 (void) ldbm_store( ldbm[i], key, data, LDBM_INSERT );
524 ldap_value_free( val );
533 printf( "%d entries\n", count );
540 * compare the old and new centroids, generating the appropriate add
541 * and delete operations. if the underlying database is ordered, we
542 * can do this more efficiently.
559 char **avals, **dvals;
560 int amax, acur, dmax, dcur;
563 #ifdef HAVE_BERKELEY_DB2
566 #endif /* HAVE_BERKELEY_DB2 */
569 printf( "Generating mods for differential %s centroid...", attr );
573 if ( (mods = (LDAPMod **) malloc( sizeof(LDAPMod *) * 4 )) == NULL ||
574 (mods[0] = (LDAPMod *) malloc( sizeof(LDAPMod) )) == NULL ||
575 (mods[1] = (LDAPMod *) malloc( sizeof(LDAPMod) )) == NULL ||
576 (mods[2] = (LDAPMod *) malloc( sizeof(LDAPMod) )) == NULL ||
577 (vals = (char **) malloc( 2 * sizeof(char *) )) == NULL ||
578 (vals[0] = (char *) malloc( 20 )) == NULL )
583 /* add values in mods[0] */
584 mods[0]->mod_op = LDAP_MOD_ADD;
585 mods[0]->mod_type = attr;
586 mods[0]->mod_values = NULL;
589 /* delete values in mods[1] */
590 mods[1]->mod_op = LDAP_MOD_DELETE;
591 mods[1]->mod_type = attr;
592 mods[1]->mod_values = NULL;
595 /* number of entries in mods[2] */
596 sprintf( vals[0], "%d", nentries );
598 mods[2]->mod_op = LDAP_MOD_REPLACE;
599 mods[2]->mod_type = "nentries";
600 mods[2]->mod_values = vals;
601 /* null terminate list of mods */
606 * if the underlying database is ordered, we can do a more efficient
607 * dual traversal, yielding O(N) performance.
612 #ifdef HAVE_BERKELEY_DB2
613 for ( okey = ldbm_firstkey( oldbm, &ocursorp ),
614 nkey = ldbm_firstkey( nldbm, &ncursorp );
615 okey.dptr != NULL && nkey.dptr != NULL; )
617 for ( okey = ldbm_firstkey( oldbm ), nkey = ldbm_firstkey( nldbm );
618 okey.dptr != NULL && nkey.dptr != NULL; )
621 rc = strcmp( okey.dptr, nkey.dptr );
624 /* value is in both places - leave it */
625 if ( olast.dptr != NULL ) {
626 ldbm_datum_free( oldbm, olast );
629 if ( nlast.dptr != NULL ) {
630 ldbm_datum_free( nldbm, nlast );
634 #ifdef HAVE_BERKELEY_DB2
635 okey = ldbm_nextkey( oldbm, olast, ocursorp );
636 nkey = ldbm_nextkey( nldbm, nlast, ncursorp );
638 okey = ldbm_nextkey( oldbm, olast );
639 nkey = ldbm_nextkey( nldbm, nlast );
641 } else if ( rc > 0 ) {
642 /* new value is not in old centroid - add it */
643 if ( charray_add_dup( &avals, &acur, &amax, nkey.dptr ) == NULL ) {
644 ldap_mods_free( mods, 1 );
648 if ( nlast.dptr != NULL ) {
649 ldbm_datum_free( nldbm, nlast );
653 #ifdef HAVE_BERKELEY_DB2
654 nkey = ldbm_nextkey( nldbm, nlast, ncursorp );
656 nkey = ldbm_nextkey( nldbm, nlast );
659 /* old value is not in new centroid - delete it */
660 if ( charray_add_dup( &dvals, &dcur, &dmax, okey.dptr ) == NULL ) {
661 ldap_mods_free( mods, 1 );
665 if ( olast.dptr != NULL ) {
666 ldbm_datum_free( oldbm, olast );
670 #ifdef HAVE_BERKELEY_DB2
671 okey = ldbm_nextkey( oldbm, olast, ocursorp );
673 okey = ldbm_nextkey( oldbm, olast );
678 while ( okey.dptr != NULL ) {
679 if ( charray_add_dup( &dvals, &dcur, &dmax, okey.dptr ) == NULL ) {
680 ldap_mods_free( mods, 1 );
684 #ifdef HAVE_BERKELEY_DB2
685 okey = ldbm_nextkey( oldbm, olast, ocursorp );
687 okey = ldbm_nextkey( oldbm, olast );
689 if ( olast.dptr != NULL ) {
690 ldbm_datum_free( oldbm, olast );
694 if ( olast.dptr != NULL ) {
695 ldbm_datum_free( oldbm, olast );
697 while ( nkey.dptr != NULL ) {
698 if ( charray_add_dup( &avals, &acur, &amax, nkey.dptr ) == NULL ) {
699 ldap_mods_free( mods, 1 );
703 #ifdef HAVE_BERKELEY_DB2
704 nkey = ldbm_nextkey( nldbm, nlast, ncursorp );
706 nkey = ldbm_nextkey( nldbm, nlast );
708 if ( nlast.dptr != NULL ) {
709 ldbm_datum_free( nldbm, nlast );
713 if ( nlast.dptr != NULL ) {
714 ldbm_datum_free( nldbm, nlast );
718 * if the underlying database is not ordered, we have to
719 * generate list of values to add by stepping through all new
720 * values and looking them up in the old centroid (not there => add),
721 * then stepping through all old values and looking them up in the
722 * new centroid (not there => delete). this yields O(Nf(N)) performance,
723 * where f(N) is the order to retrieve a single item.
726 /* generate list of values to add */
728 #ifdef HAVE_BERKELEY_DB2
729 for ( key = ldbm_firstkey( nldbm, &ncursorp ); key.dptr != NULL;
730 key = ldbm_nextkey( nldbm, lastkey, ncursorp ) )
732 for ( key = ldbm_firstkey( nldbm ); key.dptr != NULL;
733 key = ldbm_nextkey( nldbm, lastkey ) )
736 /* see if it's in the old one */
737 data = ldbm_fetch( oldbm, key );
739 /* not there - add it */
740 if ( data.dptr == NULL ) {
741 if ( charray_add_dup( &avals, &acur, &amax, key.dptr ) == NULL ) {
742 ldap_mods_free( mods, 1 );
746 ldbm_datum_free( oldbm, data );
748 if ( lastkey.dptr != NULL ) {
749 ldbm_datum_free( nldbm, lastkey );
753 if ( lastkey.dptr != NULL ) {
754 ldbm_datum_free( nldbm, lastkey );
757 /* generate list of values to delete */
759 #ifdef HAVE_BERKELEY_DB2
760 for ( key = ldbm_firstkey( oldbm, &ocursorp ); key.dptr != NULL;
761 key = ldbm_nextkey( oldbm, lastkey, ocursorp ) )
763 for ( key = ldbm_firstkey( oldbm ); key.dptr != NULL;
764 key = ldbm_nextkey( oldbm, lastkey ) )
767 /* see if it's in the new one */
768 data = ldbm_fetch( nldbm, key );
770 /* not there - delete it */
771 if ( data.dptr == NULL ) {
772 if ( charray_add_dup( &dvals, &dcur, &dmax, key.dptr ) == NULL ) {
773 ldap_mods_free( mods, 1 );
777 ldbm_datum_free( nldbm, data );
779 if ( lastkey.dptr != NULL ) {
780 ldbm_datum_free( oldbm, lastkey );
784 if ( lastkey.dptr != NULL ) {
785 ldbm_datum_free( oldbm, lastkey );
789 mods[0]->mod_values = avals;
790 mods[1]->mod_values = dvals;
797 if ( mods[1]->mod_values == NULL ) {
798 free( (char *) mods[1] );
801 if ( mods[0]->mod_values == NULL ) {
802 free( (char *) mods[0] );
806 if ( mods[0] == NULL ) {
807 free( (char *) mods );
826 #ifdef HAVE_BERKELEY_DB2
831 printf( "Generating mods for full %s centroid...", attr );
835 if ( (mods = (LDAPMod **) malloc( sizeof(LDAPMod *) * 3 )) == NULL ||
836 (mods[0] = (LDAPMod *) malloc( sizeof(LDAPMod) )) == NULL ||
837 (mods[1] = (LDAPMod *) malloc( sizeof(LDAPMod) )) == NULL ||
838 (vals = (char **) malloc( 2 * sizeof(char *) )) == NULL ||
839 (vals[0] = (char *) malloc( 20 )) == NULL )
844 mods[0]->mod_op = LDAP_MOD_REPLACE;
845 mods[0]->mod_type = attr;
846 mods[0]->mod_values = NULL;
847 sprintf( vals[0], "%d", nentries );
849 mods[1]->mod_op = LDAP_MOD_REPLACE;
850 mods[1]->mod_type = "nentries";
851 mods[1]->mod_values = vals;
857 #ifdef HAVE_BERKELEY_DB2
858 for ( key = ldbm_firstkey( ldbm, &cursorp ); key.dptr != NULL;
859 key = ldbm_nextkey( ldbm, lastkey, cursorp ) )
861 for ( key = ldbm_firstkey( ldbm ); key.dptr != NULL;
862 key = ldbm_nextkey( ldbm, lastkey ) )
865 if ( charray_add_dup( &vals, &vcur, &vmax, key.dptr ) == NULL ) {
866 ldap_mods_free( mods, 1 );
870 if ( lastkey.dptr != NULL ) {
871 ldbm_datum_free( ldbm, lastkey );
875 if ( lastkey.dptr != NULL ) {
876 ldbm_datum_free( ldbm, lastkey );
878 mods[0]->mod_values = vals;
885 if ( mods[0]->mod_values == NULL ) {
886 free( (char *) mods[0] );
887 free( (char *) mods );
895 * extract the destination ldap host, port, and base object for the
896 * server to receive the index information. then, open a connection,
897 * bind, and see if the entry exists. if not, create it and set things
898 * up so the centroid full and diff routines can modify it to contain
899 * the new centroid information.
903 bind_to_destination_ldap(
912 char *attrs[2], *refvalues[2], *ocvalues[2];
918 printf( "Binding to destination LDAP server..." );
922 /* first, pick out the destination ldap server info */
923 if ( ldapbase != NULL ) {
927 if ( strncmp( ldapdesturl, "ldap://", 7 ) != 0 ) {
928 fputs( "Not an LDAP URL", stderr ); /* Should be smarter? */
932 if ( (s2 = strchr( s, '/' )) == NULL ) {
933 ldapbase = strdup( s );
937 if ( ldaphost != NULL )
939 ldaphost = strdup( s );
942 ldapbase = strdup( s2 + 1 );
944 strcpy( buf, "ref=" );
945 if ( strpbrk( ldapsrcurl, " ,;" ) != NULL ) {
948 for ( s = d = ldapsrcurl; *s; s++ ) {
954 strcat( buf, ldapsrcurl );
955 if ( strpbrk( ldapsrcurl, " ,;" ) != NULL ) {
959 strcat( buf, ldapbase );
961 ldapbase = strdup( buf );
963 if ( (ld = ldap_open( ldaphost, LDAP_PORT )) == NULL ) {
964 perror( "ldap_open" );
968 if ( ldap_bind_s( ld, destldapbinddn, destldappasswd, destldapauthmethod )
970 ldap_perror( ld, "ldap_bind_s" );
980 rc = ldap_search_s( ld, ldapbase, LDAP_SCOPE_BASE, "(objectclass=*)",
984 if ( rc == LDAP_NO_SUCH_OBJECT ) {
986 printf( "%sCreating centroid entry...", not ? "Not " : "" );
990 /* create the centroid index entry */
992 m[0].mod_type = "ref";
993 refvalues[0] = ldapsrcurl;
995 m[0].mod_values = refvalues;
997 m[1].mod_type = "objectclass";
998 ocvalues[0] = "indexentry";
1000 m[1].mod_values = ocvalues;
1005 if ( !not && ldap_add_s( ld, ldapbase, mp ) != LDAP_SUCCESS ) {
1006 ldap_perror( ld, ldapbase );
1015 } else if ( rc != LDAP_SUCCESS ) {
1016 ldap_perror( ld, "ldap_search_s" );
1035 *a = (char **) malloc( (BUFSIZ + 1) * sizeof(char *) );
1038 } else if ( *cur >= *max ) {
1040 *a = (char **) realloc( *a, (*max + 1) * sizeof(char *) );
1046 (*a)[(*cur)++] = strdup( s );