webalizer.c

来自「统计日志的工具」· C语言 代码 · 共 1,572 行 · 第 1/5 页

             cur_day=rec_day; cur_hour=rec_hour;
             cur_min=rec_min; cur_sec=rec_sec;
             f_day=rec_day;
         }

         /* adjust last day processed if different */
         if (rec_day > l_day) l_day = rec_day;
 
         /* update min/sec stuff */
         if (cur_sec != rec_sec) cur_sec = rec_sec;
         if (cur_min != rec_min) cur_min = rec_min;

         /* check for hour change  */
         if (cur_hour != rec_hour)
         {
            /* if yes, init hourly stuff */
            if (ht_hit > mh_hit) mh_hit = ht_hit;
            ht_hit = 0;
            cur_hour = rec_hour;
         }

         /* check for day change   */
         if (cur_day != rec_day)
         {
            /* if yes, init daily stuff */
            tm_site[cur_day-1]=dt_site; dt_site=0;
            tm_visit[cur_day-1]=tot_visit(sd_htab);
            del_hlist(sd_htab);
            cur_day = rec_day;
         }

         /* check for month change */
         if (cur_month != rec_month)
         {
            /* if yes, do monthly stuff */
            t_visit=tot_visit(sm_htab);
            month_update_exit(req_tstamp);    /* process exit pages      */
            write_month_html();               /* generate HTML for month */
            clear_month();
            cur_month = rec_month;            /* update our flags        */
            cur_year  = rec_year;
            f_day=l_day=rec_day;
         }

#ifdef USE_DNS
         /* Resolve IP address if needed */
         if (dns_db)
         {
            if (inet_addr(log_rec.hostname) != INADDR_NONE)
            resolve_dns(&log_rec);
         }
#endif

         /* lowercase hostname */
         cp1 = log_rec.hostname;
         while (*cp1 != '\0')
         {
            if ( (*cp1>='A') && (*cp1<='Z') ) *cp1 += 'a'-'A';
            if ( (isalnum((int)*cp1))||(*cp1=='.')||(*cp1=='-') ) cp1++;
            else *cp1='\0';
         }

         /* Catch blank hostnames here */
         if (log_rec.hostname[0]=='\0')
            strncpy(log_rec.hostname,"Unknown",8);

         /* Ignore/Include check */
         if ( (isinlist(include_sites,log_rec.hostname)==NULL) &&
              (isinlist(include_urls,log_rec.url)==NULL)       &&
              (isinlist(include_refs,log_rec.refer)==NULL)     &&
              (isinlist(include_agents,log_rec.agent)==NULL)   &&
              (isinlist(include_users,log_rec.ident)==NULL)    )
         {
            if (isinlist(ignored_sites,log_rec.hostname)!=NULL)
              { total_ignore++; continue; }
            if (isinlist(ignored_urls,log_rec.url)!=NULL)
              { total_ignore++; continue; }
            if (isinlist(ignored_agents,log_rec.agent)!=NULL)
              { total_ignore++; continue; }
            if (isinlist(ignored_refs,log_rec.refer)!=NULL)
              { total_ignore++; continue; }
            if (isinlist(ignored_users,log_rec.ident)!=NULL)
              { total_ignore++; continue; }
         }

         /* Bump response code totals */
         switch (log_rec.resp_code) {
          case RC_CONTINUE:         i=IDX_CONTINUE;         break;
          case RC_SWITCHPROTO:      i=IDX_SWITCHPROTO;      break;
          case RC_OK:               i=IDX_OK;               break;
          case RC_CREATED:          i=IDX_CREATED;          break;
          case RC_ACCEPTED:         i=IDX_ACCEPTED;         break;
          case RC_NONAUTHINFO:      i=IDX_NONAUTHINFO;      break;
          case RC_NOCONTENT:        i=IDX_NOCONTENT;        break;
          case RC_RESETCONTENT:     i=IDX_RESETCONTENT;     break;
          case RC_PARTIALCONTENT:   i=IDX_PARTIALCONTENT;   break;
          case RC_MULTIPLECHOICES:  i=IDX_MULTIPLECHOICES;  break;
          case RC_MOVEDPERM:        i=IDX_MOVEDPERM;        break;
          case RC_MOVEDTEMP:        i=IDX_MOVEDTEMP;        break;
          case RC_SEEOTHER:         i=IDX_SEEOTHER;         break;
          case RC_NOMOD:            i=IDX_NOMOD;            break;
          case RC_USEPROXY:         i=IDX_USEPROXY;         break;
 	  case RC_MOVEDTEMPORARILY: i=IDX_MOVEDTEMPORARILY; break;
          case RC_BAD:              i=IDX_BAD;              break;
          case RC_UNAUTH:           i=IDX_UNAUTH;           break;
          case RC_PAYMENTREQ:       i=IDX_PAYMENTREQ;       break;
          case RC_FORBIDDEN:        i=IDX_FORBIDDEN;        break;
          case RC_NOTFOUND:         i=IDX_NOTFOUND;         break;
          case RC_METHODNOTALLOWED: i=IDX_METHODNOTALLOWED; break;
          case RC_NOTACCEPTABLE:    i=IDX_NOTACCEPTABLE;    break;
          case RC_PROXYAUTHREQ:     i=IDX_PROXYAUTHREQ;     break;
          case RC_TIMEOUT:          i=IDX_TIMEOUT;          break;
          case RC_CONFLICT:         i=IDX_CONFLICT;         break;
          case RC_GONE:             i=IDX_GONE;             break;
          case RC_LENGTHREQ:        i=IDX_LENGTHREQ;        break;
          case RC_PREFAILED:        i=IDX_PREFAILED;        break;
          case RC_REQENTTOOLARGE:   i=IDX_REQENTTOOLARGE;   break;
          case RC_REQURITOOLARGE:   i=IDX_REQURITOOLARGE;   break;
          case RC_UNSUPMEDIATYPE:   i=IDX_UNSUPMEDIATYPE;   break;
	  case RC_RNGNOTSATISFIABLE:i=IDX_RNGNOTSATISFIABLE;break;
	  case RC_EXPECTATIONFAILED:i=IDX_EXPECTATIONFAILED;break;
          case RC_SERVERERR:        i=IDX_SERVERERR;        break;
          case RC_NOTIMPLEMENTED:   i=IDX_NOTIMPLEMENTED;   break;
          case RC_BADGATEWAY:       i=IDX_BADGATEWAY;       break;
          case RC_UNAVAIL:          i=IDX_UNAVAIL;          break;
          case RC_GATEWAYTIMEOUT:   i=IDX_GATEWAYTIMEOUT;   break;
          case RC_BADHTTPVER:       i=IDX_BADHTTPVER;       break;
          default:                  i=IDX_UNDEFINED;        break;
         }
         response[i].count++;

         /* now save in the various hash tables... */
         if (log_rec.resp_code==RC_OK || log_rec.resp_code==RC_PARTIALCONTENT)
            i=1; else i=0;
         
         /* URL/ident hash table (only if valid response code) */
         if ((log_rec.resp_code==RC_OK)||(log_rec.resp_code==RC_NOMOD)||
             (log_rec.resp_code==RC_PARTIALCONTENT))
         {
            /* URL hash table */
            if (put_unode(log_rec.url,OBJ_REG,(u_long)1,
                log_rec.xfer_size,&t_url,(u_long)0,(u_long)0,um_htab))
            {
               if (verbose)
               /* Error adding URL node, skipping ... */
               fprintf(stderr,"%s %s\n", msg_nomem_u, log_rec.url);
            }

            /* ident (username) hash table */
            if (put_inode(log_rec.ident,OBJ_REG,
                1,(u_long)i,log_rec.xfer_size,&t_user,
                0,rec_tstamp,im_htab))
            {
               if (verbose)
               /* Error adding ident node, skipping .... */
               fprintf(stderr,"%s %s\n", msg_nomem_i, log_rec.ident);
            }
         }

         /* referrer hash table */
         if (ntop_refs)
         {
            if (log_rec.refer[0]!='\0')
             if (put_rnode(log_rec.refer,OBJ_REG,(u_long)1,&t_ref,rm_htab))
             {
              if (verbose)
              fprintf(stderr,"%s %s\n", msg_nomem_r, log_rec.refer);
             }
         }

         /* hostname (site) hash table - daily */
         if (put_hnode(log_rec.hostname,OBJ_REG,
             1,(u_long)i,log_rec.xfer_size,&dt_site,
             0,rec_tstamp,"",sd_htab))
         {
            if (verbose)
            /* Error adding host node (daily), skipping .... */
            fprintf(stderr,"%s %s\n",msg_nomem_dh, log_rec.hostname);
         }

         /* hostname (site) hash table - monthly */
         if (put_hnode(log_rec.hostname,OBJ_REG,
             1,(u_long)i,log_rec.xfer_size,&t_site,
             0,rec_tstamp,"",sm_htab))
         {
            if (verbose)
            /* Error adding host node (monthly), skipping .... */
            fprintf(stderr,"%s %s\n", msg_nomem_mh, log_rec.hostname);
         }

         /* user agent hash table */
         if (ntop_agents)
         {
            if (log_rec.agent[0]!='\0')
             if (put_anode(log_rec.agent,OBJ_REG,(u_long)1,&t_agent,am_htab))
             {
              if (verbose)
              fprintf(stderr,"%s %s\n", msg_nomem_a, log_rec.agent);
             }
         }

         /* bump monthly/daily/hourly totals        */
         t_hit++; ht_hit++;                         /* daily/hourly hits    */
         t_xfer += log_rec.xfer_size;               /* total xfer size      */
         tm_xfer[rec_day-1] += log_rec.xfer_size;   /* daily xfer total     */
         tm_hit[rec_day-1]++;                       /* daily hits total     */
         th_xfer[rec_hour] += log_rec.xfer_size;    /* hourly xfer total    */
         th_hit[rec_hour]++;                        /* hourly hits total    */
   
         /* if RC_OK, increase file counters */
         if (log_rec.resp_code == RC_OK)
         {
            t_file++;
            tm_file[rec_day-1]++;
            th_file[rec_hour]++;
         }

         /* Pages (pageview) calculation */
         if (ispage(log_rec.url))
         {
            t_page++;
            tm_page[rec_day-1]++;
            th_page[rec_hour]++;

            /* do search string stuff if needed     */
            if (ntop_search) srch_string(log_rec.srchstr);
         }

         /*********************************************/
         /* RECORD PROCESSED - DO GROUPS HERE         */ 
         /*********************************************/

         /* URL Grouping */
         if ( (cp1=isinglist(group_urls,log_rec.url))!=NULL)
         {
            if (put_unode(cp1,OBJ_GRP,(u_long)1,log_rec.xfer_size,
                &ul_bogus,(u_long)0,(u_long)0,um_htab))
            {
               if (verbose)
               /* Error adding URL node, skipping ... */
               fprintf(stderr,"%s %s\n", msg_nomem_u, cp1);
            }
         }

         /* Site Grouping */
         if ( (cp1=isinglist(group_sites,log_rec.hostname))!=NULL)
         {
            if (put_hnode(cp1,OBJ_GRP,1,(u_long)(log_rec.resp_code==RC_OK)?1:0,
                          log_rec.xfer_size,&ul_bogus,
                          0,rec_tstamp,"",sm_htab))
            {
               if (verbose)
               /* Error adding Site node, skipping ... */
               fprintf(stderr,"%s %s\n", msg_nomem_mh, cp1);
            }
         }
         else
         {
            /* Domain Grouping */
            if (group_domains)
            {
               cp1 = get_domain(log_rec.hostname);
               if (cp1 != NULL)
               {
                  if (put_hnode(cp1,OBJ_GRP,1,
                      (u_long)(log_rec.resp_code==RC_OK)?1:0,
                      log_rec.xfer_size,&ul_bogus,
                      0,rec_tstamp,"",sm_htab))
                  {
                     if (verbose)
                     /* Error adding Site node, skipping ... */
                     fprintf(stderr,"%s %s\n", msg_nomem_mh, cp1);
                  }
               }
            }
         }

         /* Referrer Grouping */
         if ( (cp1=isinglist(group_refs,log_rec.refer))!=NULL)
         {
            if (put_rnode(cp1,OBJ_GRP,(u_long)1,&ul_bogus,rm_htab))
            {
               if (verbose)
               /* Error adding Referrer node, skipping ... */
               fprintf(stderr,"%s %s\n", msg_nomem_r, cp1);
            }
         }

         /* User Agent Grouping */
         if ( (cp1=isinglist(group_agents,log_rec.agent))!=NULL)
         {
            if (put_anode(cp1,OBJ_GRP,(u_long)1,&ul_bogus,am_htab))
            {
               if (verbose)
               /* Error adding User Agent node, skipping ... */
               fprintf(stderr,"%s %s\n", msg_nomem_a, cp1);
            }
         }

         /* Ident (username) Grouping */
         if ( (cp1=isinglist(group_users,log_rec.ident))!=NULL)
         {
            if (put_inode(cp1,OBJ_GRP,1,(u_long)(log_rec.resp_code==RC_OK)?1:0,
                          log_rec.xfer_size,&ul_bogus,
                          0,rec_tstamp,im_htab))
            {
               if (verbose)
               /* Error adding Username node, skipping ... */
               fprintf(stderr,"%s %s\n", msg_nomem_i, cp1);
            }
         }
      }

      /*********************************************/
      /* BAD RECORD                                */