setup.c

larbin是一种开源的网络爬虫/网络蜘蛛

  *p++= 0;
  return i;

 x_badline:
  saveerr(ads,EINVAL);
  while ((c= getc(file)) != EOF && c != '\n');
  return -2;
}

static int gl_text(adns_state ads, getline_ctx *src_io, const char *filename,
		   int lno, char *buf, int buflen) {
  const char *cp= src_io->text;
  int l;

  if (!cp || !*cp) return -1;

  if (*cp == ';' || *cp == '\n') cp++;
  l= strcspn(cp,";\n");
  src_io->text = cp+l;

  if (l >= buflen) {
    adns__diag(ads,-1,0,"%s:%d: line too long, ignored",filename,lno);
    saveerr(ads,EINVAL);
    return -2;
  }
    
  memcpy(buf,cp,l);
  buf[l]= 0;
  return l;
}

static void readconfiggeneric(adns_state ads, const char *filename,
			      int (*getline)(adns_state ads, getline_ctx*,
					     const char *filename, int lno,
					     char *buf, int buflen),
			      /* Returns >=0 for success, -1 for EOF or error
			       * (error will have been reported), or -2 for
			       * bad line was encountered, try again.
			       */
			      getline_ctx gl_ctx) {
  char linebuf[2000], *p, *q;
  int lno, l, dirl;
  const struct configcommandinfo *ccip;

  for (lno=1;
       (l= getline(ads,&gl_ctx, filename,lno, linebuf,sizeof(linebuf))) != -1;
       lno++) {
    if (l == -2) continue;
    while (l>0 && ctype_whitespace(linebuf[l-1])) l--;
    linebuf[l]= 0;
    p= linebuf;
    while (ctype_whitespace(*p)) p++;
    if (*p == '#' || !*p) continue;
    q= p;
    while (*q && !ctype_whitespace(*q)) q++;
    dirl= q-p;
    for (ccip=configcommandinfos;
	 ccip->name && !(strlen(ccip->name)==dirl && !memcmp(ccip->name,p,q-p));
	 ccip++);
    if (!ccip->name) {
      adns__diag(ads,-1,0,"%s:%d: unknown configuration directive `%.*s'",
		 filename,lno,q-p,p);
      continue;
    }
    while (ctype_whitespace(*q)) q++;
    ccip->fn(ads,filename,lno,q);
  }
}

static const char *instrum_getenv(adns_state ads, const char *envvar) {
  const char *value;

  value= getenv(envvar);
  if (!value) adns__debug(ads,-1,0,"environment variable %s not set",envvar);
  else adns__debug(ads,-1,0,"environment variable %s set to `%s'",envvar,value);
  return value;
}

static void readconfig(adns_state ads, const char *filename, int warnmissing) {
  getline_ctx gl_ctx;
  
  gl_ctx.file= fopen(filename,"r");
  if (!gl_ctx.file) {
    if (errno == ENOENT) {
      if (warnmissing)
	adns__debug(ads,-1,0,"configuration file `%s' does not exist",filename);
      return;
    }
    saveerr(ads,errno);
    adns__diag(ads,-1,0,"cannot open configuration file `%s': %s",
	       filename,strerror(errno));
    return;
  }

  readconfiggeneric(ads,filename,gl_file,gl_ctx);
  
  fclose(gl_ctx.file);
}

static void readconfigtext(adns_state ads, const char *text, const char *showname) {
  getline_ctx gl_ctx;
  
  gl_ctx.text= text;
  readconfiggeneric(ads,showname,gl_text,gl_ctx);
}
  
static void readconfigenv(adns_state ads, const char *envvar) {
  const char *filename;

  if (ads->iflags & adns_if_noenv) {
    adns__debug(ads,-1,0,"not checking environment variable `%s'",envvar);
    return;
  }
  filename= instrum_getenv(ads,envvar);
  if (filename) readconfig(ads,filename,1);
}

static void readconfigenvtext(adns_state ads, const char *envvar) {
  const char *textdata;

  if (ads->iflags & adns_if_noenv) {
    adns__debug(ads,-1,0,"not checking environment variable `%s'",envvar);
    return;
  }
  textdata= instrum_getenv(ads,envvar);
  if (textdata) readconfigtext(ads,textdata,envvar);
}


int adns__setnonblock(adns_state ads, int fd) {
  int r;
  
  r= fcntl(fd,F_GETFL,0); if (r<0) return errno;
  r |= O_NONBLOCK;
  r= fcntl(fd,F_SETFL,r); if (r<0) return errno;
  return 0;
}

static int init_begin(adns_state *ads_r, adns_initflags flags, FILE *diagfile) {
  adns_state ads;
  
  ads= malloc(sizeof(*ads)); if (!ads) return errno;

  ads->iflags= flags;
  ads->diagfile= diagfile;
  ads->configerrno= 0;
  LIST_INIT(ads->udpw);
  LIST_INIT(ads->tcpw);
  LIST_INIT(ads->childw);
  LIST_INIT(ads->output);
  ads->forallnext= 0;
  ads->nextid= 0x311f;
  ads->udpsocket= ads->tcpsocket= -1;
  adns__vbuf_init(&ads->tcpsend);
  adns__vbuf_init(&ads->tcprecv);
  ads->tcprecv_skip= 0;
  ads->nservers= ads->nsortlist= ads->nsearchlist= ads->tcpserver= 0;
  ads->searchndots= 1;
  ads->tcpstate= server_disconnected;
  timerclear(&ads->tcptimeout);
  ads->searchlist= 0;

  *ads_r= ads;
  return 0;
}

static int init_finish(adns_state ads) {
  struct in_addr ia;
  struct protoent *proto;
  int r;
  
  if (!ads->nservers) {
    if (ads->diagfile && ads->iflags & adns_if_debug)
      fprintf(ads->diagfile,"adns: no nameservers, using localhost\n");
    ia.s_addr= htonl(INADDR_LOOPBACK);
    addserver(ads,ia);
  }

  proto= getprotobyname("udp"); if (!proto) { r= ENOPROTOOPT; goto x_free; }
  ads->udpsocket= socket(AF_INET,SOCK_DGRAM,proto->p_proto);
  if (ads->udpsocket<0) { r= errno; goto x_free; }

  r= adns__setnonblock(ads,ads->udpsocket);
  if (r) { r= errno; goto x_closeudp; }
  
  return 0;

 x_closeudp:
  close(ads->udpsocket);
 x_free:
  free(ads);
  return r;
}

static void init_abort(adns_state ads) {
  if (ads->nsearchlist) {
    free(ads->searchlist[0]);
    free(ads->searchlist);
  }
  free(ads);
}

int adns_init(adns_state *ads_r, adns_initflags flags, FILE *diagfile) {
  adns_state ads;
  const char *res_options, *adns_res_options;
  int r;
  
  r= init_begin(&ads, flags, diagfile ? diagfile : stderr);
  if (r) return r;
  
  res_options= instrum_getenv(ads,"RES_OPTIONS");
  adns_res_options= instrum_getenv(ads,"ADNS_RES_OPTIONS");
  ccf_options(ads,"RES_OPTIONS",-1,res_options);
  ccf_options(ads,"ADNS_RES_OPTIONS",-1,adns_res_options);

  readconfig(ads,"/etc/resolv.conf",1);
  readconfig(ads,"/etc/resolv-adns.conf",0);
  readconfigenv(ads,"RES_CONF");
  readconfigenv(ads,"ADNS_RES_CONF");

  readconfigenvtext(ads,"RES_CONF_TEXT");
  readconfigenvtext(ads,"ADNS_RES_CONF_TEXT");

  ccf_options(ads,"RES_OPTIONS",-1,res_options);
  ccf_options(ads,"ADNS_RES_OPTIONS",-1,adns_res_options);

  ccf_search(ads,"LOCALDOMAIN",-1,instrum_getenv(ads,"LOCALDOMAIN"));
  ccf_search(ads,"ADNS_LOCALDOMAIN",-1,instrum_getenv(ads,"ADNS_LOCALDOMAIN"));

  if (ads->configerrno && ads->configerrno != EINVAL) {
    r= ads->configerrno;
    init_abort(ads);
    return r;
  }

  r= init_finish(ads);
  if (r) return r;

  adns__consistency(ads,0,cc_entex);
  *ads_r= ads;
  return 0;
}

int adns_init_strcfg(adns_state *ads_r, adns_initflags flags,
		     FILE *diagfile, const char *configtext) {
  adns_state ads;
  int r;

  r= init_begin(&ads, flags, diagfile);  if (r) return r;

  readconfigtext(ads,configtext,"<supplied configuration text>");
  if (ads->configerrno) {
    r= ads->configerrno;
    init_abort(ads);
    return r;
  }

  r= init_finish(ads);  if (r) return r;
  adns__consistency(ads,0,cc_entex);
  *ads_r= ads;
  return 0;
}


void adns_finish(adns_state ads) {
  adns__consistency(ads,0,cc_entex);
  for (;;) {
    if (ads->udpw.head) adns_cancel(ads->udpw.head);
    else if (ads->tcpw.head) adns_cancel(ads->tcpw.head);
    else if (ads->childw.head) adns_cancel(ads->childw.head);
    else if (ads->output.head) adns_cancel(ads->output.head);
    else break;
  }
  close(ads->udpsocket);
  if (ads->tcpsocket >= 0) close(ads->tcpsocket);
  adns__vbuf_free(&ads->tcpsend);
  adns__vbuf_free(&ads->tcprecv);
  freesearchlist(ads);
  free(ads);
}

void adns_forallqueries_begin(adns_state ads) {
  adns__consistency(ads,0,cc_entex);
  ads->forallnext=
    ads->udpw.head ? ads->udpw.head :
    ads->tcpw.head ? ads->tcpw.head :
    ads->childw.head ? ads->childw.head :
    ads->output.head;
}
  
adns_query adns_forallqueries_next(adns_state ads, void **context_r) {
  adns_query qu, nqu;

  adns__consistency(ads,0,cc_entex);
  nqu= ads->forallnext;
  for (;;) {
    qu= nqu;
    if (!qu) return 0;
    if (qu->next) {
      nqu= qu->next;
    } else if (qu == ads->udpw.tail) {
      nqu=
	ads->tcpw.head ? ads->tcpw.head :
	ads->childw.head ? ads->childw.head :
	ads->output.head;
    } else if (qu == ads->tcpw.tail) {
      nqu=
	ads->childw.head ? ads->childw.head :
	ads->output.head;
    } else if (qu == ads->childw.tail) {
      nqu= ads->output.head;
    } else {
      nqu= 0;
    }
    if (!qu->parent) break;
  }
  ads->forallnext= nqu;
  if (context_r) *context_r= qu->ctx.ext;
  return qu;
}