bool canSubmit ( unsigned long h , long now , long maxAddUrlsPerIpDomPerDay ) {
// . sometimes no limit
// . 0 means no limit because if they don't want any submission they
// can just turn off add url and we want to avoid excess
// troubleshooting for why a url can't be added
if ( maxAddUrlsPerIpDomPerDay <= 0 ) return true;
// init the table
if ( ! s_init ) {
s_htable.set ( 50000 );
s_init = true;
}
// clean out table every 24 hours
if ( now - s_lastTime > 24*60*60 ) {
s_lastTime = now;
s_htable.clear();
}
// . if table almost full clean out ALL slots
// . TODO: just clean out oldest slots
if ( s_htable.getNumSlotsUsed() > 47000 ) s_htable.clear ();
// . how many times has this IP domain submitted?
// . allow 10 times per day
long n = s_htable.getValue ( h );
// if over 24hr limit then bail
if ( n >= maxAddUrlsPerIpDomPerDay ) return false;
// otherwise, inc it
n++;
// add to table, will replace old values
s_htable.addKey ( h , n );
return true;
}
uint8_t CountryCode::getLanguageFromDMOZ(long catid) {
if(!m_init) return(0);
catcountryrec_t ccr;
ccr.lval = 0L;
if(s_catToCountry.getNumSlotsUsed() < 1) return(0);
long slot = s_catToCountry.getSlot((long)catid);
if(slot < 0) return(0);
ccr.lval = s_catToCountry.getValueFromSlot(slot);
return(ccr.sval.lang);
}
void CountryCode::debugDumpNumbers(void) {
long slot;
catcountryrec_t ccr;
for(slot = 0; slot < s_catToCountry.getNumSlotsUsed(); slot++) {
ccr.lval = 0L;
ccr.lval = s_catToCountry.getValueFromSlot(slot);
if(ccr.lval)
log( "Slot %ld has lang %d, country %d (%ld)\n",
slot, ccr.sval.lang, ccr.sval.country, ccr.lval);
}
}
// Do not call this function lightly, it takes an hour to run
int CountryCode::createHashTable(void) {
if(!fillRegexTable()) return(0);
char tmpbuf[2048];
HashTable ht;
unsigned long long entries = 0UL;
long catid;
long numcats = g_categories->m_numCats;
catcountryrec_t ccr;
SafeBuf sb(tmpbuf, 2048);
log( "cat: Creating category country/language table.\n");
if(!ht.set(2,NULL,0,"ctrycode")) {
log( "cat: Could not allocate memory for table.\n");
return(0);
}
for(long idx = 0; idx < numcats; idx++) {
catid = g_categories->m_cats[idx].m_catid;
sb.reset();
g_categories->printPathFromId(&sb, catid, true);
if(!sb.getBufStart()) continue;
if(!(numcats % 1000))
log( "init: %ld/%ld Generated %llu so far...\n",
numcats,
idx,
entries);
ccr.lval = 0L;
ccr.sval.country = lookupCountryFromDMOZTopic(sb.getBufStart(), sb.length());
ccr.sval.lang = s_getLangIdxFromDMOZ(sb.getBufStart(), sb.length());
if(!ccr.lval) continue;
if(ccr.sval.lang > 27 || ccr.sval.country > s_numCountryCodes) {
char *xx = NULL; *xx = 0;
}
if(!ht.addKey(catid, ccr.lval)) {
log( "init: Could not add %ld (%ld)\n", catid, ccr.lval);
continue;
}
entries++;
}
ht.save(g_hostdb.m_dir, "catcountry.dat");
log( "Added %llu country entries from DMOZ to %s/catcountry.dat.\n", entries,g_hostdb.m_dir);
log( "Slots %ld, Used Slots %ld.\n", ht.getNumSlots(), ht.getNumSlotsUsed());
freeRegexTable();
return(1);
}
long CountryCode::getNumEntries(void) {
if(!m_init) return(0);
return(s_catToCountry.getNumSlotsUsed());
}
