add knowledge about the Heritrix bot

[php-utility-classes.git] / include / classes / useragent.php-class

diff --git a/include/classes/useragent.php-class b/include/classes/useragent.php-class
index 9f4bb2b09c63118e49df86a728d379ca35675ac6..840a3987bf29af50b9b0dc8dd6c254eac3b36493 100755 (executable)
--- a/include/classes/useragent.php-class
+++ b/include/classes/useragent.php-class
@@ -402,6 +402,11 @@ class userAgent {
       $this->version = $regs[1];
       $this->bot = true;
     }
       $this->version = $regs[1];
       $this->bot = true;
     }

+    elseif (preg_match('|heritrix/([0-9a-zA-Z\.+]+)|', $this->uastring, $regs)) {
+      $this->brand = 'Heritrix';
+      $this->version = $regs[1];
+      $this->bot = true;
+    }

     elseif (preg_match('|Slurp|', $this->uastring, $regs)) {
       $this->brand = 'Slurp';
       $this->version = null;
     elseif (preg_match('|Slurp|', $this->uastring, $regs)) {
       $this->brand = 'Slurp';
       $this->version = null;


@@ -417,6 +422,11 @@ class userAgent {
       $this->version = $regs[1];
       $this->bot = true;
     }
       $this->version = $regs[1];
       $this->bot = true;
     }

+    elseif (preg_match('|Microsoft URL Control - ([0-9a-zA-Z\.+]+)|', $this->uastring, $regs)) {
+      $this->brand = 'Microsoft URL Control';
+      $this->version = $regs[1];
+      $this->bot = true;
+    }

     elseif (preg_match('|([0-9a-zA-Z\.+]+)_AC-Plug|', $this->uastring, $regs)) {
       $this->brand = 'AC-Plug';
       $this->version = $regs[1];
     elseif (preg_match('|([0-9a-zA-Z\.+]+)_AC-Plug|', $this->uastring, $regs)) {
       $this->brand = 'AC-Plug';
       $this->version = $regs[1];


@@ -556,7 +566,7 @@ class userAgent {
                       'NetResearchServer','LinkWalker','Zeus','W3C_Validator','ZyBorg','Ask Jeeves','ia_archiver',
                       'PingALink Monitoring Services','IlTrovatore-Setaccio','Nutch','Mercator','search.ch',
                       'appie','larbin','NutchCVS','ObjectsSearch','Webchat','Mediapartners-Google','Schmozilla',
                       'NetResearchServer','LinkWalker','Zeus','W3C_Validator','ZyBorg','Ask Jeeves','ia_archiver',
                       'PingALink Monitoring Services','IlTrovatore-Setaccio','Nutch','Mercator','search.ch',
                       'appie','larbin','NutchCVS','ObjectsSearch','Webchat','Mediapartners-Google','Schmozilla',

-                      'FavOrg','findlinks','DataCha0s','','','','','','','','','');
+                      'FavOrg','findlinks','DataCha0s','ichiro','Francis','','','','','','','');

 
     if (in_array($this->brand, $botArray)) {
       $this->bot = true;
 
     if (in_array($this->brand, $botArray)) {
       $this->bot = true;