static void on_msg_recv(const char *msg) {
if (msg == NULL) { /* server closed */
client_main_run = false;
return;
}
msglist_add(msg);
}
static void distribute(int mode, tunelist_t *ns_or_sp)
{
int good = mode == REG_GOOD;
int bad = 1 - good;
bool divvy = ds_flag == DS_RAM && user_robx < EPS && !msg_count_file;
mlitem_t *item;
mlhead_t *msgs = ns_or_sp->msgs;
int score_count = 0;
int train_count = 0;
static int train_good = 0;
static int train_bad = 0;
double ratio = scale(msgs->count,
LIST_COUNT + TEST_COUNT, /* small count */
LIST_COUNT + LIST_COUNT, /* large count */
LIST_COUNT / TEST_COUNT, /* small ratio */
LIST_COUNT / LIST_COUNT); /* large ratio */
for (item = msgs->head; item != NULL; item = item->next) {
wordhash_t *wh = item->wh;
/* training set */
if (divvy && train_count / ratio < score_count + 1) {
wordhash_set_counts(wh, good, bad);
wordhash_add(train, wh, &wordprop_init);
train_count += 1;
wordhash_free(wh);
train_good += good;
train_bad += bad;
}
/* scoring set */
else {
uint bin = divvy ? MOD(score_count,3) : 0;
msglist_add(ns_or_sp->u.sets[bin], wh);
score_count += 1;
}
item->wh = NULL;
}
if (divvy) {
wordhash_insert(train, w_msg_count, sizeof(wordprop_t), &wordprop_init);
set_msg_counts(train_good, train_bad);
}
if (verbose > 1)
printf("%s: train_count = %d, score_count = %d\n",
good ? "ns" : "sp",
train_count, score_count);
return;
}
static uint read_mailbox(const char *arg, mlhead_t *msgs)
{
if (verbose) {
printf("Reading %s\n", arg);
fflush(stdout);
}
init_count();
mbox_mode = true;
bogoreader_init(1, &arg);
while ((*reader_more)()) {
wordhash_t *whp = NULL;
wordhash_t *whc = wordhash_new();
collect_words(whc);
if (ds_path != NULL && (msgs_good + msgs_bad) == 0)
set_train_msg_counts(whc);
if (whc->count == 0 && !quiet) {
printf("msg #%u, count is %u\n", message_count, whc->count);
bt_trap();
}
if (bogolex_file != NULL) {
wordhash_sort(whc);
lookup_words(whc);
write_msgcount_file(whc);
}
else if (whc->count != 0) {
if (!msg_count_file)
whp = convert_wordhash_to_propslist(whc, train);
else
whp = convert_propslist_to_countlist(whc);
msglist_add(msgs, whp);
}
update_count();
if (whc != whp)
wordhash_free(whc);
}
print_final_count();
ns_and_sp->count += message_count;
bogoreader_fini();
return message_count;
}
