//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;
}
//adds a node to the tree
static int dtree_add_node(dtree_dt_index *h,dtree_dt_node *n,char *t,const dtree_dt_add_entry *entry)
{
int hash;
char *s;
packed_ptr pp;
dtree_dt_node *next;
if(!n)
return 0;
dtree_printd(DTREE_PRINT_INITDTREE,"ADD: tree: '%c' level: %d t: '%s' p(%p) pp(%p)\n",
n->data?n->data:'^',dtree_node_depth(h,n),t,n,n->curr);
//escape
if(h->sflags & DTREE_S_FLAG_REGEX && (*t==DTREE_PATTERN_ESCAPE) && *(t-1)!=DTREE_PATTERN_ESCAPE)
t++;
//(abc)
while(h->sflags & DTREE_S_FLAG_REGEX && (*t==DTREE_PATTERN_GROUP_S || *t==DTREE_PATTERN_GROUP_E) &&
*(t-1)!=DTREE_PATTERN_ESCAPE)
{
hash=(*t==DTREE_PATTERN_GROUP_S);
s=t+1;
while(*s && hash)
{
if(*s==DTREE_PATTERN_GROUP_S && *(s-1)!=DTREE_PATTERN_ESCAPE)
hash++;
else if(*s==DTREE_PATTERN_GROUP_E && *(s-1)!=DTREE_PATTERN_ESCAPE)
hash--;
s++;
}
if(*t==DTREE_PATTERN_GROUP_S)
{
dtree_printd(DTREE_PRINT_INITDTREE,"ADD: group detected: '%c' - '%c'\n",*(t+1)?*(t+1):'#',*s?*s:'#');
//optional group
if(*s==DTREE_PATTERN_OPTIONAL && *(s-1)!=DTREE_PATTERN_ESCAPE)
{
if(dtree_add_node(h,n,s+1,entry)<0)
return -1;
}
}
t++;
}
//?
if(h->sflags & DTREE_S_FLAG_REGEX && *t==DTREE_PATTERN_OPTIONAL && *(t-1)!=DTREE_PATTERN_ESCAPE)
{
//no group
if(*(t-1)!=DTREE_PATTERN_GROUP_E || *(t-2)==DTREE_PATTERN_ESCAPE)
{
pp=n->prev;
next=DTREE_DT_GETPP(h,pp);
if(!pp)
return 0;
dtree_printd(DTREE_PRINT_INITDTREE,"ADD: optional: '%c' t: '%s'\n",next->data?next->data:'^',(t+1));
if(dtree_add_node(h,next,t+1,entry)<0)
return -1;
}
t++;
return dtree_add_node(h,n,t,entry);
}
//EOT
if(!*t && n->prev)
return dtree_set_payload(h,n,entry);
//[abc]
if(h->sflags & DTREE_S_FLAG_REGEX && *t==DTREE_PATTERN_SET_S && *(t-1)!=DTREE_PATTERN_ESCAPE)
{
for(s=t;*s;s++)
{
if(*s==DTREE_PATTERN_SET_E && *(s-1)!=DTREE_PATTERN_ESCAPE)
break;
}
if(!*s)
return 0;
t++;
while(t<s)
{
*s=*t;
dtree_printd(DTREE_PRINT_INITDTREE,"ADD: set: '%c' *t: '%c' t: '%s'\n",n->data?n->data:'^',*t,s);
if(dtree_add_node(h,n,s,entry)<0)
return -1;
t++;
}
*s=DTREE_PATTERN_SET_E;
return 1;
}
hash=dtree_hash_char(*t);
if(!hash && *t!='0')
return 0;
if(*t>='A' && *t<='Z')
*t|=0x20;
//.
if(h->sflags & DTREE_S_FLAG_REGEX && *t==DTREE_PATTERN_ANY && *(t-1)!=DTREE_PATTERN_ESCAPE)
hash=DTREE_HASH_ANY;
pp=(packed_ptr)n->nodes[hash];
next=DTREE_DT_GETPP(h,pp);
if(!pp)
{
pp=dtree_alloc_node(h);
next=DTREE_DT_GETPP(h,pp);
if(!pp)
{
fprintf(stderr,"DTREE: allocation error detected, aborting: %zu\n",h->node_count);
return -1;
}
if(hash==DTREE_HASH_ANY)
next->data=DTREE_PATTERN_ANY;
else
next->data=*t;
next->curr=pp;
next->prev=n->curr;
n->nodes[hash]=pp;
dtree_printd(DTREE_PRINT_INITDTREE,"ADD: new node '%c' created level: %d hash: %d p(%p) pp(%p)\n",
*t,dtree_node_depth(h,n),hash,next,pp);
}
//EOT
if(!*(t+1))
return dtree_set_payload(h,next,entry);
return dtree_add_node(h,next,t+1,entry);
}
//gets the flag for a token
const dtree_dt_node *dtree_get_node(const dtree_dt_index *h,const char *t,dtree_flag_f sflags,dtree_pos_f pos)
{
unsigned int hash;
char n;
const char *p;
packed_ptr pp;
const dtree_dt_node *base;
const dtree_dt_node *lflag=NULL;
if(!sflags)
sflags=h->sflags;
if(sflags & DTREE_S_FLAG_REGEX)
{
pp=h->head->nodes[dtree_hash_char(*t)];
base=DTREE_DT_GETPP(h,pp);
lflag=dtree_search_node(h,base,t,pos);
//leading wildcard
if(!lflag)
{
pp=h->head->nodes[DTREE_HASH_ANY];
base=DTREE_DT_GETPP(h,pp);
if(pp)
lflag=dtree_search_node(h,base,t,pos);
}
return lflag;
}
pp=h->head->nodes[dtree_hash_char(*t)];
base=DTREE_DT_GETPP(h,pp);
//iterative search
for(p=t;*p && pp;p++)
{
if(base->data!=*p && base->data!=(*p|0x20) && base->data!=DTREE_PATTERN_ANY)
break;
n=*(p+1);
hash=dtree_hash_char(n);
//token match
if((!hash && n!='0') || hash>=DTREE_HASH_SEP)
{
//match is valid with position
if(base->flags && (sflags & DTREE_S_FLAG_PARTIAL || !n) &&
(!base->pos || !pos || base->pos==pos))
return base;
else if(!n)
break;
}
pp=base->nodes[hash];
base=DTREE_DT_GETPP(h,pp);
}
return lflag;
}
//adds a node to the tree
static int dtree_add_node(dtree_dt_index *h,dtree_dt_node *n,char *t,void *data,flag_f flags,void *param)
{
int hash;
char *p;
packed_ptr pp;
dtree_dt_node *next;
if(!n)
return 0;
dtree_printd(DTREE_PRINT_INITDTREE,"ADD: tree: '%c' level: %d *t: '%c' p(%p) pp(%p)\n",
n->data,dtree_node_depth(h,n),*t?*t:'#',n,n->curr);
//EOT trailing wildcard
if(!*t && n->prev)
return dtree_set_payload(h,n,data,flags,param);
//?
if(h->sflags & DTREE_S_FLAG_REGEX && *t==DTREE_PATTERN_OPTIONAL)
{
pp=n->prev;
next=DTREE_DT_GETPP(h,pp);
if(!pp)
return 0;
dtree_printd(DTREE_PRINT_INITDTREE,"ADD: optional: '%c' *t: '%c'\n",next->data,*(t+1));
if(dtree_add_node(h,next,t+1,data,flags,param)<0)
return -1;
t++;
//trailing wildcard
if(!*t)
return dtree_set_payload(h,n,data,flags,param);
}
//[abc]
if(h->sflags & DTREE_S_FLAG_REGEX && *t==DTREE_PATTERN_SET_S)
{
for(p=t;*p;p++)
{
if(*p==DTREE_PATTERN_SET_E)
break;
}
if(!*p)
return 0;
t++;
while(t<p)
{
*p=*t;
dtree_printd(DTREE_PRINT_INITDTREE,"ADD: set: '%c' *t: '%c' t: '%s'\n",n->data,*t,p);
if(dtree_add_node(h,n,p,data,flags,param)<0)
return -1;
t++;
}
*p=DTREE_PATTERN_SET_E;
return 1;
}
hash=dtree_hash_char(*t);
if(!hash && *t!='0')
return 0;
if(*t>='A' && *t<='Z')
*t|=0x20;
//.
if(hash==DTREE_HASH_ANY)
*t=DTREE_PATTERN_ANY;
pp=(packed_ptr)n->nodes[hash];
next=DTREE_DT_GETPP(h,pp);
if(!pp)
{
pp=dtree_alloc_node(h);
next=DTREE_DT_GETPP(h,pp);
if(!pp)
{
fprintf(stderr,"DTREE: allocation error detected, aborting: %zu\n",h->node_count);
return -1;
}
next->data=*t;
next->curr=pp;
next->prev=n->curr;
n->nodes[hash]=pp;
dtree_printd(DTREE_PRINT_INITDTREE,"ADD: new node '%c' created level: %d hash: %d p(%p) pp(%p)\n",
*t,dtree_node_depth(h,n),hash,next,pp);
}
//EOT
if(!*(t+1))
return dtree_set_payload(h,next,data,flags,param);
return dtree_add_node(h,next,t+1,data,flags,param);
}
//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;
}
//gets the flag for a token
const dtree_dt_node *dtree_get_node(const dtree_dt_index *h,const char *t,flag_f sflags)
{
int hash;
char n;
const char *p;
packed_ptr pp;
const dtree_dt_node *base;
const dtree_dt_node *lflag=NULL;
if(!sflags)
sflags=h->sflags;
if(sflags & DTREE_S_FLAG_REGEX)
{
pp=h->head->nodes[dtree_hash_char(*t)];
base=DTREE_DT_GETPP(h,pp);
lflag=dtree_search_node(h,base,t);
//leading wildcard
if(!lflag)
{
pp=h->head->nodes[dtree_hash_char(DTREE_PATTERN_ANY)];
base=DTREE_DT_GETPP(h,pp);
if(pp)
lflag=dtree_search_node(h,base,t);
}
return lflag;
}
pp=h->head->nodes[dtree_hash_char(*t)];
base=DTREE_DT_GETPP(h,pp);
//iterative search
for(p=t;*p && pp;p++)
{
if(base->data!=*p && base->data!=(*p|0x20) && base->data!=DTREE_PATTERN_ANY)
break;
n=*(p+1);
hash=dtree_hash_char(n);
//token match
if((!hash && n!='0') || hash>=DTREE_HASH_SEP)
{
//not full string
if(base->flags && (sflags & DTREE_S_FLAG_PARTIAL) && n)
return base;
//full token
else if(base->flags && !n)
return base;
else if(!n)
break;
}
//partial token match
if((base->flags & DTREE_DT_FLAG_TOKEN) && (sflags & DTREE_S_FLAG_PARTIAL) && (p+1-t)>=DTREE_S_PART_TLEN)
lflag=base;
pp=base->nodes[hash];
base=DTREE_DT_GETPP(h,pp);
}
return lflag;
}
//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(dtree_hash_char(*p)<DTREE_HASH_SEP)
{
//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);
}