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;
27 int ldap_syslog_level;
29 static void add(LDBM ldbm, char *s, int *count, int *size, int freeit);
32 main( int argc, char **argv )
36 int i, j, k, count, len, nentries;
37 int vcount, wcount, pcount, scount;
38 int vsize, wsize, psize, ssize;
39 struct berval **bvals;
43 LDBM wldbm, pldbm, sldbm;
44 static char *attrs[] = { "cn", "nentries", NULL };
49 if ( (ld = ldap_init( "vertigo:5555", 0 )) == NULL ) {
50 perror( "ldap_init" );
54 if ( ldap_search( ld, "cn=index", LDAP_SCOPE_ONELEVEL, "(objectclass=*)",
56 ldap_perror( ld, "ldap_search" );
60 printf( "attr\tdn\tnentries\tvcount\tvsize\twcount\twsize\tpcount\tpsize\tscount\tssize\n" );
63 while ( ldap_result( ld, LDAP_RES_ANY, 0, NULL, &res )
64 == LDAP_RES_SEARCH_ENTRY ) {
66 e = ldap_first_entry( ld, res );
67 dn = ldap_get_dn( ld, e );
68 if ( (vals = ldap_get_values( ld, e, "nentries" )) != NULL ) {
69 nentries = atoi( vals[0] );
70 ldap_value_free( vals );
72 fprintf( stderr, "no nentries attribute for (%s)\n", dn );
76 for ( i = 0; attrs[i] != NULL; i++ ) {
77 if ( strcasecmp( attrs[i], "nentries" ) == 0 ) {
80 if ( (wldbm = ldbm_open( "wcount.ldbm", DB_FLAGS, MODE,
81 CACHE_SIZE )) == NULL || (pldbm = ldbm_open( "pcount.ldbm",
82 DB_FLAGS, MODE, CACHE_SIZE )) == NULL || (sldbm = ldbm_open(
83 "scount.ldbm", DB_FLAGS, MODE, CACHE_SIZE )) == NULL ) {
84 perror( "ldbm_open" );
87 vcount = 0; vsize = 0;
88 wcount = 0; wsize = 0;
89 pcount = 0; psize = 0;
90 scount = 0; ssize = 0;
91 if ( (bvals = ldap_get_values_len( ld, e, attrs[i] )) != NULL ) {
92 for ( j = 0; bvals[j] != NULL; j++ ) {
96 ldbm_datum_init( key );
97 ldbm_datum_init( data );
99 /* update value count */
101 vsize += bvals[j]->bv_len;
103 /* update word and phoneme counts */
104 for ( w = first_word( bvals[j]->bv_val ); w != NULL;
105 w = next_word( w ) ) {
106 add( wldbm, word_dup( w ), &wcount, &wsize, 1 );
108 add( pldbm, phonetic( w ), &pcount, &psize, 1 );
111 /* update substring count */
112 len = bvals[j]->bv_len;
113 val = bvals[j]->bv_val;
114 if ( len > SUBLEN - 2 ) {
116 for ( k = 0; k < SUBLEN - 1; k++ ) {
120 add( sldbm, buf, &scount, &ssize, 0 );
122 p = val + len - SUBLEN + 1;
123 for ( k = 0; k < SUBLEN; k++ ) {
126 buf[SUBLEN - 1] = '$';
128 add( sldbm, buf, &scount, &ssize, 0 );
130 for ( p = val; p < (val + len - SUBLEN + 1); p++ ) {
131 for ( k = 0; k < SUBLEN; k++ ) {
135 add( sldbm, buf, &scount, &ssize, 0 );
138 ldap_value_free_len( bvals );
140 printf( "%s\t%s\t%d", attrs[i], dn, nentries );
141 printf( "\t%d\t%d", vcount, vsize );
142 printf( "\t%d\t%d", wcount, wsize );
143 printf( "\t%d\t%d", pcount, psize );
144 printf( "\t%d\t%d\n", scount, ssize );
155 printf( "%d entries\n", count );
158 if ( ldap_result2error( ld, res, 1 ) != LDAP_SUCCESS ) {
159 ldap_perror( ld, "ldap_result" );
163 (void) unlink( "wcount.ldbm" );
164 (void) unlink( "pcount.ldbm" );
165 (void) unlink( "scount.ldbm" );
167 exit( EXIT_SUCCESS );
181 ldbm_datum_init( key );
182 ldbm_datum_init( data );
185 key.dsize = strlen( key.dptr ) + 1;
188 if ( ldbm_store( ldbm, key, data, LDBM_INSERT ) == 0 ) {
190 (*size) += strlen( key.dptr );
192 if ( freeit && ( key.dptr != NULL ) )
193 ldbm_datum_free( ldbm, key );