types.c

larbin是一种开源的网络爬虫/网络蜘蛛

static int dip_hostaddr(adns_state ads, const adns_rr_hostaddr *ap, const adns_rr_hostaddr *bp) {
  if (ap->astatus != bp->astatus) return ap->astatus;
  if (ap->astatus) return 0;

  assert(ap->addrs[0].addr.sa.sa_family == AF_INET);
  assert(bp->addrs[0].addr.sa.sa_family == AF_INET);
  return dip_inaddr(ads,
		    ap->addrs[0].addr.inet.sin_addr,
		    bp->addrs[0].addr.inet.sin_addr);
}

static int di_hostaddr(adns_state ads, const void *datap_a, const void *datap_b) {
  const adns_rr_hostaddr *ap= datap_a, *bp= datap_b;

  return dip_hostaddr(ads, ap,bp);
}

static void mfp_hostaddr(adns_query qu, adns_rr_hostaddr *rrp) {
  void *tablev;

  adns__makefinal_str(qu,&rrp->host);
  tablev= rrp->addrs;
  adns__makefinal_block(qu, &tablev, rrp->naddrs*sizeof(*rrp->addrs));
  rrp->addrs= tablev;
}

static void mf_hostaddr(adns_query qu, void *datap) {
  adns_rr_hostaddr *rrp= datap;

  mfp_hostaddr(qu,rrp);
}

static adns_status csp_hostaddr(vbuf *vb, const adns_rr_hostaddr *rrp) {
  const char *errstr;
  adns_status st;
  char buf[20];
  int i;

  st= csp_domain(vb,rrp->host);  if (st) return st;

  CSP_ADDSTR(" ");
  CSP_ADDSTR(adns_errtypeabbrev(rrp->astatus));

  sprintf(buf," %d ",rrp->astatus);
  CSP_ADDSTR(buf);

  CSP_ADDSTR(adns_errabbrev(rrp->astatus));
  CSP_ADDSTR(" ");

  errstr= adns_strerror(rrp->astatus);
  st= csp_qstring(vb,errstr,strlen(errstr));  if (st) return st;
  
  if (rrp->naddrs >= 0) {
    CSP_ADDSTR(" (");
    for (i=0; i<rrp->naddrs; i++) {
      CSP_ADDSTR(" ");
      st= csp_addr(vb,&rrp->addrs[i]);
    }
    CSP_ADDSTR(" )");
  } else {
    CSP_ADDSTR(" ?");
  }
  return adns_s_ok;
}

static adns_status cs_hostaddr(vbuf *vb, const void *datap) {
  const adns_rr_hostaddr *rrp= datap;

  return csp_hostaddr(vb,rrp);
}

/*
 * _mx_raw   (pa,di)
 */

static adns_status pa_mx_raw(const parseinfo *pai, int cbyte, int max, void *datap) {
  const byte *dgram= pai->dgram;
  adns_rr_intstr *rrp= datap;
  adns_status st;
  int pref;

  if (cbyte+2 > max) return adns_s_invaliddata;
  GET_W(cbyte,pref);
  rrp->i= pref;
  st= pap_domain(pai, &cbyte, max, &rrp->str,
		 pai->qu->flags & adns_qf_quoteok_anshost ? pdf_quoteok : 0);
  if (st) return st;
  
  if (cbyte != max) return adns_s_invaliddata;
  return adns_s_ok;
}

static int di_mx_raw(adns_state ads, const void *datap_a, const void *datap_b) {
  const adns_rr_intstr *ap= datap_a, *bp= datap_b;

  if (ap->i < bp->i) return 0;
  if (ap->i > bp->i) return 1;
  return 0;
}

/*
 * _mx   (pa,di)
 */

static adns_status pa_mx(const parseinfo *pai, int cbyte, int max, void *datap) {
  const byte *dgram= pai->dgram;
  adns_rr_inthostaddr *rrp= datap;
  adns_status st;
  int pref;

  if (cbyte+2 > max) return adns_s_invaliddata;
  GET_W(cbyte,pref);
  rrp->i= pref;
  st= pap_hostaddr(pai, &cbyte, max, &rrp->ha);
  if (st) return st;
  
  if (cbyte != max) return adns_s_invaliddata;
  return adns_s_ok;
}

static int di_mx(adns_state ads, const void *datap_a, const void *datap_b) {
  const adns_rr_inthostaddr *ap= datap_a, *bp= datap_b;

  if (ap->i < bp->i) return 0;
  if (ap->i > bp->i) return 1;
  return dip_hostaddr(ads, &ap->ha, &bp->ha);
}

/*
 * _inthostaddr  (mf,cs)
 */

static void mf_inthostaddr(adns_query qu, void *datap) {
  adns_rr_inthostaddr *rrp= datap;

  mfp_hostaddr(qu,&rrp->ha);
}

static adns_status cs_inthostaddr(vbuf *vb, const void *datap) {
  const adns_rr_inthostaddr *rrp= datap;
  char buf[10];

  sprintf(buf,"%u ",rrp->i);
  CSP_ADDSTR(buf);

  return csp_hostaddr(vb,&rrp->ha);
}

/*
 * _inthost  (cs)
 */

static adns_status cs_inthost(vbuf *vb, const void *datap) {
  const adns_rr_intstr *rrp= datap;
  char buf[10];

  sprintf(buf,"%u ",rrp->i);
  CSP_ADDSTR(buf);
  return csp_domain(vb,rrp->str);
}

/*
 * _ptr   (pa, +icb_ptr)
 */

static void icb_ptr(adns_query parent, adns_query child) {
  adns_answer *cans= child->answer;
  const adns_rr_addr *queried, *found;
  adns_state ads= parent->ads;
  int i;

  if (cans->status == adns_s_nxdomain || cans->status == adns_s_nodata) {
    adns__query_fail(parent,adns_s_inconsistent);
    return;
  } else if (cans->status) {
    adns__query_fail(parent,cans->status);
    return;
  }

  queried= &parent->ctx.info.ptr_parent_addr;
  for (i=0, found=cans->rrs.addr; i<cans->nrrs; i++, found++) {
    if (queried->len == found->len &&
	!memcmp(&queried->addr,&found->addr,queried->len)) {
      if (!parent->children.head) {
	adns__query_done(parent);
	return;
      } else {
	LIST_LINK_TAIL(ads->childw,parent);
	return;
      }
    }
  }

  adns__query_fail(parent,adns_s_inconsistent);
}

static adns_status pa_ptr(const parseinfo *pai, int dmstart, int max, void *datap) {
  static const char *(expectdomain[])= { DNS_INADDR_ARPA };
  
  char **rrp= datap;
  adns_status st;
  adns_rr_addr *ap;
  findlabel_state fls;
  char *ep;
  byte ipv[4];
  char labbuf[4];
  int cbyte, i, lablen, labstart, l, id;
  adns_query nqu;
  qcontext ctx;

  cbyte= dmstart;
  st= pap_domain(pai, &cbyte, max, rrp,
		 pai->qu->flags & adns_qf_quoteok_anshost ? pdf_quoteok : 0);
  if (st) return st;
  if (cbyte != max) return adns_s_invaliddata;

  ap= &pai->qu->ctx.info.ptr_parent_addr;
  if (!ap->len) {
    adns__findlabel_start(&fls, pai->ads, -1, pai->qu,
			  pai->qu->query_dgram, pai->qu->query_dglen,
			  pai->qu->query_dglen, DNS_HDRSIZE, 0);
    for (i=0; i<4; i++) {
      st= adns__findlabel_next(&fls,&lablen,&labstart); assert(!st);
      if (lablen<=0 || lablen>3) return adns_s_querydomainwrong;
      memcpy(labbuf, pai->qu->query_dgram + labstart, lablen);  labbuf[lablen]= 0;
      ipv[3-i]= strtoul(labbuf,&ep,10);  if (*ep) return adns_s_querydomainwrong;
      if (lablen>1 && pai->qu->query_dgram[labstart]=='0')
	return adns_s_querydomainwrong;
    }
    for (i=0; i<sizeof(expectdomain)/sizeof(*expectdomain); i++) {
      st= adns__findlabel_next(&fls,&lablen,&labstart); assert(!st);
      l= strlen(expectdomain[i]);
      if (lablen != l || memcmp(pai->qu->query_dgram + labstart, expectdomain[i], l))
	return adns_s_querydomainwrong;
    }
    st= adns__findlabel_next(&fls,&lablen,0); assert(!st);
    if (lablen) return adns_s_querydomainwrong;
    
    ap->len= sizeof(struct sockaddr_in);
    memset(&ap->addr,0,sizeof(ap->addr.inet));
    ap->addr.inet.sin_family= AF_INET;
    ap->addr.inet.sin_addr.s_addr=
      htonl((ipv[0]<<24) | (ipv[1]<<16) | (ipv[2]<<8) | (ipv[3]));
  }

  st= adns__mkquery_frdgram(pai->ads, &pai->qu->vb, &id,
			    pai->dgram, pai->dglen, dmstart,
			    adns_r_addr, adns_qf_quoteok_query);
  if (st) return st;

  ctx.ext= 0;
  ctx.callback= icb_ptr;
  memset(&ctx.info,0,sizeof(ctx.info));
  st= adns__internal_submit(pai->ads, &nqu, adns__findtype(adns_r_addr),
			    &pai->qu->vb, id,
			    adns_qf_quoteok_query, pai->now, &ctx);
  if (st) return st;

  nqu->parent= pai->qu;
  LIST_LINK_TAIL_PART(pai->qu->children,nqu,siblings.);
  return adns_s_ok;
}

/*
 * _strpair   (mf)
 */

static void mf_strpair(adns_query qu, void *datap) {
  adns_rr_strpair *rrp= datap;

  adns__makefinal_str(qu,&rrp->array[0]);
  adns__makefinal_str(qu,&rrp->array[1]);
}

/*
 * _intstrpair   (mf)
 */

static void mf_intstrpair(adns_query qu, void *datap) {
  adns_rr_intstrpair *rrp= datap;

  adns__makefinal_str(qu,&rrp->array[0].str);
  adns__makefinal_str(qu,&rrp->array[1].str);
}

/*
 * _hinfo   (pa)
 */

static adns_status pa_hinfo(const parseinfo *pai, int cbyte, int max, void *datap) {
  adns_rr_intstrpair *rrp= datap;
  adns_status st;
  int i;

  for (i=0; i<2; i++) {
    st= pap_qstring(pai, &cbyte, max, &rrp->array[i].i, &rrp->array[i].str);
    if (st) return st;
  }

  if (cbyte != max) return adns_s_invaliddata;
  
  return adns_s_ok;
}

/*
 * _mailbox   (pap,cs)
 */

static adns_status pap_mailbox822(const parseinfo *pai, int *cbyte_io, int max,
				  char **mb_r) {
  int lablen, labstart, i, needquote, c, r, neednorm;
  const unsigned char *p;
  char *str;
  findlabel_state fls;
  adns_status st;
  vbuf *vb;

  vb= &pai->qu->vb;
  vb->used= 0;
  adns__findlabel_start(&fls, pai->ads,
			-1, pai->qu,
			pai->dgram, pai->dglen, max,
			*cbyte_io, cbyte_io);
  st= adns__findlabel_next(&fls,&lablen,&labstart);
  if (!lablen) {
    adns__vbuf_appendstr(vb,".");
    goto x_ok;
  }

  neednorm= 1;
  for (i=0, needquote=0, p= pai->dgram+labstart; i<lablen; i++) {
    c= *p++;
    if ((c&~128) < 32 || (c&~128) == 127) return adns_s_invaliddata;
    if (c == '.' && !neednorm) neednorm= 1;
    else if (c==' ' || c>=127 || ctype_822special(c)) needquote++;
    else neednorm= 0;
  }

  if (needquote || neednorm) {
    r= adns__vbuf_ensure(vb, lablen+needquote+4); if (!r) R_NOMEM;
    adns__vbuf_appendq(vb,"\"",1);
    for (i=0, needquote=0, p= pai->dgram+labstart; i<lablen; i++, p++) {
      c= *p;
      if (c == '"' || c=='\\') adns__vbuf_appendq(vb,"\\",1);
      adns__vbuf_appendq(vb,p,1);
    }
    adns__vbuf_appendq(vb,"\"",1);
  } else {
    r= adns__vbuf_append(vb, pai->dgram+labstart, lablen); if (!r) R_NOMEM;
  }

  r= adns__vbuf_appendstr(vb,"@"); if (!r) R_NOMEM;

  st= adns__parse_domain_more(&fls,pai->ads, pai->qu,vb,0, pai->dgram);
  if (st) return st;

 x_ok:
  str= adns__alloc_interim(pai->qu, vb->used+1); if (!str) R_NOMEM;
  memcpy(str,vb->buf,vb->used);
  str[vb->used]= 0;
  *mb_r= str;
  return adns_s_ok;
}

static adns_status pap_mailbox(const parseinfo *pai, int *cbyte_io, int max,
			       char **mb_r) {
  if (pai->qu->typei->type & adns__qtf_mail822) {
    return pap_mailbox822(pai, cbyte_io, max, mb_r);
  } else {
    return pap_domain(pai, cbyte_io, max, mb_r, pdf_quoteok);
  }
}

static adns_status csp_mailbox(vbuf *vb, const char *mailbox) {
  return csp_domain(vb,mailbox);
}

/*
 * _rp   (pa,cs)
 */

static adns_status pa_rp(const parseinfo *pai, int cbyte, int max, void *datap) {
  adns_rr_strpair *rrp= datap;
  adns_status st;

  st= pap_mailbox(pai, &cbyte, max, &rrp->array[0]);
  if (st) return st;

  st= pap_domain(pai, &cbyte, max, &rrp->array[1], pdf_quoteok);
  if (st) return st;

  if (cbyte != max) return adns_s_invaliddata;
  return adns_s_ok;
}

static adns_status cs_rp(vbuf *vb, const void *datap) {
  const adns_rr_strpair *rrp= datap;
  adns_status st;

  st= csp_mailbox(vb,rrp->array[0]);  if (st) return st;
  CSP_ADDSTR(" ");
  st= csp_domain(vb,rrp->array[1]);  if (st) return st;

  return adns_s_ok;
}  

/*
 * _soa   (pa,mf,cs)
 */

static adns_status pa_soa(const parseinfo *pai, int cbyte, int max, void *datap) {
  adns_rr_soa *rrp= datap;
  const byte *dgram= pai->dgram;
  adns_status st;
  int msw, lsw, i;

  st= pap_domain(pai, &cbyte, max, &rrp->mname,
		 pai->qu->flags & adns_qf_quoteok_anshost ? pdf_quoteok : 0);
  if (st) return st;

  st= pap_mailbox(pai, &cbyte, max, &rrp->rname);
  if (st) return st;

  if (cbyte+20 != max) return adns_s_invaliddata;
  
  for (i=0; i<5; i++) {
    GET_W(cbyte,msw);
    GET_W(cbyte,lsw);
    (&rrp->serial)[i]= (msw<<16) | lsw;
  }

  return adns_s_ok;
}

static void mf_soa(adns_query qu, void *datap) {
  adns_rr_soa *rrp= datap;

  adns__makefinal_str(qu,&rrp->mname);
  adns__makefinal_str(qu,&rrp->rname);
}

static adns_status cs_soa(vbuf *vb, const void *datap) {
  const adns_rr_soa *rrp= datap;
  char buf[20];
  int i;
  adns_status st;
  
  st= csp_domain(vb,rrp->mname);  if (st) return st;
  CSP_ADDSTR(" ");
  st= csp_mailbox(vb,rrp->rname);  if (st) return st;

  for (i=0; i<5; i++) {
    sprintf(buf," %lu",(&rrp->serial)[i]);
    CSP_ADDSTR(buf);
  }

  return adns_s_ok;
}

/*
 * _flat   (mf)
 */

static void mf_flat(adns_query qu, void *data) { }

/*
 * Now the table.
 */

#define TYPESZ_M(member)           (sizeof(*((adns_answer*)0)->rrs.member))

#define DEEP_MEMB(memb) TYPESZ_M(memb), mf_##memb, cs_##memb
#define FLAT_MEMB(memb) TYPESZ_M(memb), mf_flat, cs_##memb

#define DEEP_TYPE(code,rrt,fmt,memb,parser,comparer,printer) \
 { adns_r_##code, rrt, fmt, TYPESZ_M(memb), mf_##memb, printer, parser, comparer }
#define FLAT_TYPE(code,rrt,fmt,memb,parser,comparer,printer) \
 { adns_r_##code, rrt, fmt, TYPESZ_M(memb), mf_flat, printer, parser, comparer }

static const typeinfo typeinfos[] = {
/* Must be in ascending order of rrtype ! */
/* mem-mgmt code  rrt     fmt     member      parser      comparer    printer       */
  		    		     		 	     		 	       
FLAT_TYPE(a,      "A",     0,     inaddr,     pa_inaddr,  di_inaddr,  cs_inaddr     ),
DEEP_TYPE(ns_raw, "NS",   "raw",  str,        pa_host_raw,0,          cs_domain     ),
DEEP_TYPE(cname,  "CNAME", 0,     str,        pa_dom_raw, 0,          cs_domain     ),
DEEP_TYPE(soa_raw,"SOA",  "raw",  soa,        pa_soa,     0,          cs_soa        ),
DEEP_TYPE(ptr_raw,"PTR",  "raw",  str,        pa_host_raw,0,          cs_domain     ),
DEEP_TYPE(hinfo,  "HINFO", 0,     intstrpair, pa_hinfo,   0,          cs_hinfo      ),
DEEP_TYPE(mx_raw, "MX",   "raw",  intstr,     pa_mx_raw,  di_mx_raw,  cs_inthost    ),
DEEP_TYPE(txt,    "TXT",   0,     manyistr,   pa_txt,     0,          cs_txt        ),
DEEP_TYPE(rp_raw, "RP",   "raw",  strpair,    pa_rp,      0,          cs_rp         ),
 		     	                                     		   	       
FLAT_TYPE(addr,   "A",  "addr",   addr,       pa_addr,    di_addr,    cs_addr       ),
DEEP_TYPE(ns,     "NS", "+addr",  hostaddr,   pa_hostaddr,di_hostaddr,cs_hostaddr   ),
DEEP_TYPE(ptr,    "PTR","checked",str,        pa_ptr,     0,          cs_domain     ),
DEEP_TYPE(mx,     "MX", "+addr",  inthostaddr,pa_mx,      di_mx,      cs_inthostaddr),
 		     	                                     		      
DEEP_TYPE(soa,    "SOA","822",    soa,        pa_soa,     0,          cs_soa        ),
DEEP_TYPE(rp,     "RP", "822",    strpair,    pa_rp,      0,          cs_rp         ),
};

const typeinfo *adns__findtype(adns_rrtype type) {
  const typeinfo *begin, *end, *mid;

  begin= typeinfos;  end= typeinfos+(sizeof(typeinfos)/sizeof(typeinfo));

  while (begin < end) {
    mid= begin + ((end-begin)>>1);
    if (mid->type == type) return mid;
    if (type > mid->type) begin= mid+1;
    else end= mid;
  }
  return 0;
}