14 #define CACHE_SIZE 1000000
16 #define DB_FLAGS (LDBM_WRCREAT|LDBM_NEWDB)
19 extern char *first_word(char *);
20 extern char *next_word(char *);
21 extern char *word_dup(char *);
22 extern char *phonetic(char *);
24 extern int ldap_debug;
25 extern int lber_debug;
28 int ldap_syslog_level;
30 static void add(LDBM ldbm, char *s, int *count, int *size, int freeit);
33 main( int argc, char **argv )
37 int i, j, k, count, len, nentries;
38 int vcount, wcount, pcount, scount;
39 int vsize, wsize, psize, ssize;
40 struct berval **bvals;
44 LDBM wldbm, pldbm, sldbm;
45 static char *attrs[] = { "cn", "nentries", NULL };
51 if ( (ld = ldap_init( "vertigo:5555", 0 )) == NULL ) {
52 perror( "ldap_init" );
56 if ( ldap_search( ld, "cn=index", LDAP_SCOPE_ONELEVEL, "(objectclass=*)",
58 ldap_perror( ld, "ldap_search" );
62 printf( "attr\tdn\tnentries\tvcount\tvsize\twcount\twsize\tpcount\tpsize\tscount\tssize\n" );
65 while ( ldap_result( ld, LDAP_RES_ANY, 0, NULL, &res )
66 == LDAP_RES_SEARCH_ENTRY ) {
68 e = ldap_first_entry( ld, res );
69 dn = ldap_get_dn( ld, e );
70 if ( (vals = ldap_get_values( ld, e, "nentries" )) != NULL ) {
71 nentries = atoi( vals[0] );
72 ldap_value_free( vals );
74 fprintf( stderr, "no nentries attribute for (%s)\n", dn );
78 for ( i = 0; attrs[i] != NULL; i++ ) {
79 if ( strcasecmp( attrs[i], "nentries" ) == 0 ) {
82 if ( (wldbm = ldbm_open( "wcount.ldbm", DB_FLAGS, MODE,
83 CACHE_SIZE )) == NULL || (pldbm = ldbm_open( "pcount.ldbm",
84 DB_FLAGS, MODE, CACHE_SIZE )) == NULL || (sldbm = ldbm_open(
85 "scount.ldbm", DB_FLAGS, MODE, CACHE_SIZE )) == NULL ) {
86 perror( "ldbm_open" );
89 vcount = 0; vsize = 0;
90 wcount = 0; wsize = 0;
91 pcount = 0; psize = 0;
92 scount = 0; ssize = 0;
93 if ( (bvals = ldap_get_values_len( ld, e, attrs[i] )) != NULL ) {
94 for ( j = 0; bvals[j] != NULL; j++ ) {
98 ldbm_datum_init( key );
99 ldbm_datum_init( data );
101 /* update value count */
103 vsize += bvals[j]->bv_len;
105 /* update word and phoneme counts */
106 for ( w = first_word( bvals[j]->bv_val ); w != NULL;
107 w = next_word( w ) ) {
108 add( wldbm, word_dup( w ), &wcount, &wsize, 1 );
110 add( pldbm, phonetic( w ), &pcount, &psize, 1 );
113 /* update substring count */
114 len = bvals[j]->bv_len;
115 val = bvals[j]->bv_val;
116 if ( len > SUBLEN - 2 ) {
118 for ( k = 0; k < SUBLEN - 1; k++ ) {
122 add( sldbm, buf, &scount, &ssize, 0 );
124 p = val + len - SUBLEN + 1;
125 for ( k = 0; k < SUBLEN; k++ ) {
128 buf[SUBLEN - 1] = '$';
130 add( sldbm, buf, &scount, &ssize, 0 );
132 for ( p = val; p < (val + len - SUBLEN + 1); p++ ) {
133 for ( k = 0; k < SUBLEN; k++ ) {
137 add( sldbm, buf, &scount, &ssize, 0 );
140 ldap_value_free_len( bvals );
142 printf( "%s\t%s\t%d", attrs[i], dn, nentries );
143 printf( "\t%d\t%d", vcount, vsize );
144 printf( "\t%d\t%d", wcount, wsize );
145 printf( "\t%d\t%d", pcount, psize );
146 printf( "\t%d\t%d\n", scount, ssize );
157 printf( "%d entries\n", count );
160 if ( ldap_result2error( ld, res, 1 ) != LDAP_SUCCESS ) {
161 ldap_perror( ld, "ldap_result" );
165 (void) unlink( "wcount.ldbm" );
166 (void) unlink( "pcount.ldbm" );
167 (void) unlink( "scount.ldbm" );
183 ldbm_datum_init( key );
184 ldbm_datum_init( data );
187 key.dsize = strlen( key.dptr ) + 1;
190 if ( ldbm_store( ldbm, key, data, LDBM_INSERT ) == 0 ) {
192 (*size) += strlen( key.dptr );
194 if ( freeit && ( key.dptr != NULL ) )
195 ldbm_datum_free( ldbm, key );