Added --ignore-referer-report command line option to hide referers fr…

…om output.

src/options.c

@@ -104,6 +104,7 @@ struct option long_opts[] = {
   {"ignore-crawlers"      , no_argument       , 0 ,  0  } ,
   {"ignore-panel"         , required_argument , 0 ,  0  } ,
   {"ignore-referer"       , required_argument , 0 ,  0  } ,
+  {"ignore-referer-report", required_argument , 0 ,  0  } ,
   {"ignore-status"        , required_argument , 0 ,  0  } ,
   {"invalid-requests"     , required_argument , 0 ,  0  } ,
   {"json-pretty-print"    , no_argument       , 0 ,  0  } ,
@@ -247,6 +248,7 @@ cmd_help (void)
   "  --ignore-panel=<PANEL>          - Ignore parsing/displaying the given panel.\n"
   "  --ignore-referer=<NEEDLE>       - Ignore a referer from being counted. Wild cards\n"
   "                                    are allowed. i.e., *.bing.com\n"
+  "  --ignore-referer-report=<NEEDLE>- Ignore referer's data, e.g. ref. within same site\n"
   "  --ignore-status=<CODE>          - Ignore parsing the given status code.\n"
   "  --num-tests=<number>            - Number of lines to test. >= 0 (10 default)\n"
   "  --process-and-exit              - Parse log and exit without outputting data.\n"
@@ -522,6 +524,11 @@ parse_long_opt (const char *name, const char *oarg)
     set_array_opt (oarg, conf.ignore_referers, &conf.ignore_referer_idx,
                    MAX_IGNORE_REF);
 
+  /* ignore referer report (e.g. within same site) */
+  if (!strcmp ("ignore-referer-report", name))
+    set_array_opt (oarg, conf.ignore_referers_report,
+                   &conf.ignore_referer_report_idx, MAX_IGNORE_REF);
+
   /* ignore status code */
   if (!strcmp ("ignore-status", name))
     set_array_opt (oarg, conf.ignore_status, &conf.ignore_status_idx,

src/parser.c

@@ -1211,7 +1211,17 @@ parse_specifier (GLogItem * logitem, char **str, const char *p, const char *end)
       extract_keyphrase (tkn, &logitem->keyphrase);
       extract_referer_site (tkn, logitem->site);
     }
-    logitem->ref = tkn;
+
+
+    if (strcmp (tkn, "-") != 0) {
+      if (!ignore_referer_report (logitem->site)) {
+        logitem->ref = tkn;
+      }
+      if (ignore_referer_report (logitem->site)) {
+        logitem->site[0] = '\0';
+      }
+
+    }
     break;
     /* user agent */
   case 'u':

src/settings.h

@@ -100,6 +100,7 @@ typedef struct GConf_
   const char *ignore_ips[MAX_IGNORE_IPS];       /* array of ips to ignore */
   const char *ignore_panels[TOTAL_MODULES];     /* array of panels to ignore */
   const char *ignore_referers[MAX_IGNORE_REF];  /* referrers to ignore */
+  const char *ignore_referers_report[MAX_IGNORE_REF];  /* referrers to count as hit, but ignore referrer data */
   const char *ignore_status[MAX_IGNORE_STATUS]; /* status to ignore */
   const char *output_formats[MAX_OUTFORMATS];   /* output format, e.g. , HTML */
   const char *sort_panels[TOTAL_MODULES];       /* sorting options for each panel */
@@ -184,6 +185,7 @@ typedef struct GConf_
   int ignore_ip_idx;                /* ignored ips index */
   int ignore_panel_idx;             /* ignored panels index */
   int ignore_referer_idx;           /* ignored referrers index */
+  int ignore_referer_report_idx;     /* ignored referrers index */
   int ignore_status_idx;            /* ignore status index */
   int output_format_idx;            /* output format index */
   int sort_panel_idx;               /* sort panel index */

src/util.c

@@ -250,6 +250,34 @@ ignore_referer (const char *host)
   return ignore;
 }
 
+int
+ignore_referer_report (const char *host)
+{
+  char *needle = NULL;
+  int i, ignore = 0;
+
+  if (conf.ignore_referer_report_idx == 0)
+    return 0;
+  if (host == NULL || *host == '\0')
+    return 0;
+
+  needle = xstrdup (host);
+  for (i = 0; i < conf.ignore_referer_report_idx; ++i) {
+    if (conf.ignore_referers_report[i] == NULL ||
+        *conf.ignore_referers_report[i] == '\0')
+      continue;
+
+    if (wc_match (conf.ignore_referers_report[i], needle)) {
+      ignore = 1;
+      goto out;
+    }
+  }
+out:
+  free (needle);
+
+  return ignore;
+}
+
 /* Determine if the given ip is within a range of IPs.
  *
  * On error, or not within the range, 0 is returned

src/util.h

@@ -77,6 +77,7 @@ int convert_date (char *res, const char *data, const char *from, const char *to,
 int count_matches (const char *s1, char c);
 int find_output_type (char **filename, const char *ext, int alloc);
 int ignore_referer (const char *ref);
+int ignore_referer_report (const char *ref);
 int intlen (int num);
 int invalid_ipaddr (char *str, int *ipvx);
 int ip_in_range (const char *ip);