//dtree lookup
void *dtree_get(const dtree_dt_index *h,const char *str,dtree_flag_f filter)
{
const dtree_dt_node *ret;
const dtree_dt_node *fnode;
if(!str || !h->head)
return NULL;
ret=dtree_get_node(h,str,DTREE_S_FLAG_NONE,0);
if(ret)
{
if(filter)
{
fnode=dtree_get_flag(h,ret,filter,0);
if(fnode)
return fnode->payload;
}
else
return ret->payload;
}
return NULL;
}
//searches for a token with regex
static const dtree_dt_node *dtree_search_node(const dtree_dt_index *h,const dtree_dt_node *n,const char *t,dtree_pos_f pos)
{
unsigned int hash;
dtree_flag_f rf,nf;
packed_ptr pp;
const dtree_dt_node *rflag;
if(!n || !n->curr)
return NULL;
//bad match
if(n->data!=*t && n->data!=(*t|0x20) && n->data!=DTREE_PATTERN_ANY)
return NULL;
t++;
hash=dtree_hash_char(*t);
//EOS
if(!hash && *t!='0')
{
//only quit if strong with position
if(dtree_get_flag(h,n,DTREE_DT_FLAG_STRONG,pos))
return n;
else if(!n->flags)
return NULL;
}
pp=n->nodes[hash];
rflag=DTREE_DT_GETPP(h,pp);
if(pp)
rflag=dtree_search_node(h,rflag,t,pos);
else
rflag=NULL;
//wildcard
if(!rflag)
{
hash=DTREE_HASH_ANY;
pp=n->nodes[hash];
rflag=DTREE_DT_GETPP(h,pp);
if(pp)
rflag=dtree_search_node(h,rflag,t,pos);
else
rflag=NULL;
hash=dtree_hash_char(*t);
}
//n is lazy EOT or stronger match
if((n->flags & DTREE_DT_FLAG_TOKEN) && ((!hash && *t!='0') || hash>=DTREE_HASH_SEP))
{
nf=dtree_get_flags(h,n,pos);
rf=dtree_get_flags(h,rflag,pos);
if(!rflag)
rflag=n;
else if((nf & DTREE_DT_FLAG_STRONG) && !(rf & DTREE_DT_FLAG_STRONG))
rflag=n;
else if(rf & DTREE_DT_FLAG_STRONG);
else if((nf & DTREE_DT_FLAG_CHAIN) && !(nf & DTREE_DT_FLAG_BCHAIN) &&
!(rf & DTREE_DT_FLAG_CHAIN) && !(rf & DTREE_DT_FLAG_BCHAIN))
rflag=n;
else if(rf & DTREE_DT_FLAG_CHAIN);
else if((nf & DTREE_DT_FLAG_WEAK) && !(rf & DTREE_DT_FLAG_WEAK))
rflag=n;
}
return rflag;
}
//classifies a string
const dclass_keyvalue *dclass_classify(const dclass_index *di,const char *str)
{
int on=0;
int valid;
int bcvalid;
int i;
char buf[DTREE_DATA_BUFLEN];
const char *p;
const char *token="";
packed_ptr pp;
const dtree_dt_node *wnode=NULL;
const dtree_dt_node *nnode=NULL;
const dtree_dt_node *fbnode;
const dtree_dt_node *fnode;
const dtree_dt_index *h=&di->dti;
const void *cnodes[DTREE_S_MAX_CHAIN]={NULL};
if(!str || !h->head)
return dclass_get_kverror(di);
dtree_printd(DTREE_PRINT_CLASSIFY,"dtree_classify() UA: '%s'\n",str);
for(p=str;*p;p++)
{
valid=0;
if((*p>='a' && *p<='z') || (*p>='A' && *p<='Z') || (*p>='0' && *p<='9'))
{
//new token found
if(!on)
{
token=p;
on=1;
}
valid=1;
}
if((!valid || (!*(p+1))) && on)
{
//EOT found
fbnode=dtree_get_node(h,token,0);
dtree_printd(DTREE_PRINT_CLASSIFY,"dtree_classify() token:'%s' = '%s':%d\n",
token,fbnode?dtree_node_path(h,fbnode,buf):"",fbnode?(int)fbnode->flags:0);
if(fbnode && dtree_get_flag(h,fbnode,DTREE_DT_FLAG_TOKEN))
{
if((fnode=dtree_get_flag(h,fbnode,DTREE_DT_FLAG_STRONG)))
return fnode->payload;
else if((fnode=dtree_get_flag(h,fbnode,DTREE_DT_FLAG_WEAK)))
{
i=DTREE_DC_DISTANCE(h,(char*)fnode->payload);
if(!wnode || (i>=0 && i<DTREE_DC_DISTANCE(h,(char*)wnode->payload)))
wnode=fnode;
}
else if((fnode=dtree_get_flag(h,fbnode,DTREE_DT_FLAG_NONE)))
{
i=DTREE_DC_DISTANCE(h,(char*)fnode->payload);
if(!nnode || (i>=0 && i<DTREE_DC_DISTANCE(h,(char*)nnode->payload)))
nnode=fnode;
}
if((fnode=dtree_get_flag(h,fbnode,DTREE_DT_FLAG_CHAIN)))
{
dtree_printd(DTREE_PRINT_CLASSIFY,"dtree_classify() pchain detected\n");
pp=fnode->curr;
while(pp)
{
bcvalid=0;
if(fnode->flags & DTREE_DT_FLAG_CHAIN && fnode->cparam)
{
dtree_printd(DTREE_PRINT_CLASSIFY,"dtree_classify() looking for pchain %p\n",fnode->cparam);
for(i=0;i<DTREE_S_MAX_CHAIN && cnodes[i];i++)
{
//chain hit
if(cnodes[i]==fnode->cparam)
{
if(fnode->flags & DTREE_DT_FLAG_BCHAIN)
bcvalid=1;
else
return fnode->payload;
}
}
}
if(fnode->flags & DTREE_DT_FLAG_BCHAIN && (bcvalid || !fnode->cparam))
{
for(i=0;i<DTREE_S_MAX_CHAIN;i++)
{
if(!cnodes[i])
{
cnodes[i]=fnode->payload;
dtree_printd(DTREE_PRINT_CLASSIFY,"dtree_classify() pchain added: %p('%d')\n",fnode->payload,i);
break;
}
}
}
pp=fnode->dup;
fnode=DTREE_DT_GETPP(h,pp);
}
}
}
on=0;
}
}
if(wnode)
return wnode->payload;
else if(nnode)
return nnode->payload;
return dclass_get_kverror(di);
}
//searches for a token with regex
static const dtree_dt_node *dtree_search_node(const dtree_dt_index *h,const dtree_dt_node *n,const char *t)
{
int hash;
flag_f rf,nf;
packed_ptr pp;
const dtree_dt_node *rflag;
if(!n)
return NULL;
//bad match
if(n->data!=*t && n->data!=(*t|0x20) && n->data!=DTREE_PATTERN_ANY)
return NULL;
t++;
hash=dtree_hash_char(*t);
//lazy EOT
if((!hash && *t!='0') || hash>=DTREE_HASH_SEP)
{
//only quit if strong
if(dtree_get_flag(h,n,DTREE_DT_FLAG_STRONG))
return n;
else if(!t && !n->flags)
return NULL;
}
pp=n->nodes[hash];
rflag=DTREE_DT_GETPP(h,pp);
if(rflag && *t)
rflag=dtree_search_node(h,rflag,t);
//wildcard
if(!rflag)
{
hash=dtree_hash_char(DTREE_PATTERN_ANY);
pp=n->nodes[hash];
rflag=DTREE_DT_GETPP(h,pp);
if(rflag)
rflag=dtree_search_node(h,rflag,t);
hash=dtree_hash_char(*t);
}
//partial
if(!rflag && (n->flags & DTREE_DT_FLAG_TOKEN) && (h->sflags & DTREE_S_FLAG_PARTIAL))
{
if(dtree_node_depth(h,n)>=DTREE_S_PART_TLEN)
rflag=n;
}
//n is lazy EOT or stronger match
if((n->flags & DTREE_DT_FLAG_TOKEN) && ((hash==0 && *t!='0') || hash>=DTREE_HASH_SEP))
{
nf=dtree_get_flags(h,n);
rf=dtree_get_flags(h,rflag);
if(!rflag)
rflag=n;
else if((nf & DTREE_DT_FLAG_STRONG) && !(rf & DTREE_DT_FLAG_STRONG))
rflag=n;
else if((rf & DTREE_DT_FLAG_STRONG));
else if((nf & DTREE_DT_FLAG_CHAIN) && !(nf & DTREE_DT_FLAG_BCHAIN) &&
!(rf & DTREE_DT_FLAG_CHAIN) && !(rf & DTREE_DT_FLAG_BCHAIN))
rflag=n;
else if((rf & DTREE_DT_FLAG_CHAIN));
else if((nf & DTREE_DT_FLAG_WEAK))
rflag=n;
}
return rflag;
}
