Added: Begun webcrawler verifier.

CHANGELOG.md

@@ -3,6 +3,7 @@
 - Added: `Range` helper.
 - Added: `Steeler` webcrawler verifier.
 - Added: `Twitter` webcrawler verifier.
+- Added: `Begun` webcrawler verifier.
 - Chore: refactoring.
 - Chore: integrate `coveralls.io` and add badge `Coverage Status`.
 - Chore: add badge `Dependency Status`.

README.md

@@ -86,6 +86,7 @@ if (!empty($_SERVER['HTTP_USER_AGENT']) && !empty($_SERVER['REMOTE_ADDR'])) {
 
 - **Apple**: `Applebot`.
 - **Baidu**: `Baiduspider`.
+- **Begun**: `BegunAdvertising`.
 - **Bing**: `bingbot`, `msnbot`, `adidxbot`, `BingPreview`.
 - **Deuse**: `DeuSu`.
 - **Exalead**: `Exabot`.

src/DNS/HostVerifier.php

@@ -0,0 +1,14 @@
+<?php
+namespace WebcrawlerVerifier\DNS;
+
+use WebcrawlerVerifier\Helper\StringHelper as StringHelper;
+
+class HostVerifier
+{
+    public static function verify($host, $allowedHostNames)
+    {
+        return !!array_filter($allowedHostNames, function ($validHost) use ($host) {
+            return StringHelper::endsWith($validHost, $host) !== false;
+        });
+    }
+}

src/DNS/ReverseVerifier.php

@@ -1,19 +1,13 @@
 <?php
 namespace WebcrawlerVerifier\DNS;
 
-use WebcrawlerVerifier\Helper\StringHelper as StringHelper;
-
 class ReverseVerifier
 {
     public static function verify($ip, $allowedHostNames)
     {
         $host = gethostbyaddr($ip);
         $ipAfterLookup = gethostbyname($host);
 
-        $hostIsValid = !!array_filter($allowedHostNames, function ($validHost) use ($host) {
-            return StringHelper::endsWith($validHost, $host) !== false;
-        });
-
-        return $hostIsValid && $ipAfterLookup === $ip;
+        return HostVerifier::verify($host, $allowedHostNames) && $ipAfterLookup === $ip;
     }
 }

src/Webcrawler/BegunWebcrawlerVerifier.php

@@ -0,0 +1,20 @@
+<?php
+namespace WebcrawlerVerifier\Webcrawler;
+
+use WebcrawlerVerifier\DNS\HostVerifier;
+
+class BegunWebcrawlerVerifier implements WebcrawlerVerifierInterface
+{
+    protected $allowedHostNames = ['begun.ru'];
+
+    /**
+     * Checks whether the given IP address really belongs to a valid host or not
+     *
+     * @param $ip string the IP address to check
+     * @return bool true if the given IP belongs to any of the valid hosts, otherwise false
+     */
+    public function verify($ip)
+    {
+        return HostVerifier::verify(gethostbyaddr($ip), $this->allowedHostNames);
+    }
+}

src/WebcrawlerVerifier.php

@@ -12,6 +12,7 @@ class WebcrawlerVerifier
     protected $webcrawlerVerifiers = [
         'Applebot' => 'WebcrawlerVerifier\Webcrawler\AppleWebcrawlerVerifier',
         'Baiduspider' => 'WebcrawlerVerifier\Webcrawler\BaiduWebcrawlerVerifier',
+        'BegunAdvertising' => 'WebcrawlerVerifier\Webcrawler\BegunWebcrawlerVerifier',
         'bingbot' => 'WebcrawlerVerifier\Webcrawler\BingWebcrawlerVerifier',
         'msnbot' => 'WebcrawlerVerifier\Webcrawler\BingWebcrawlerVerifier',
         'adidxbot' => 'WebcrawlerVerifier\Webcrawler\BingWebcrawlerVerifier',

tests/WebcrawlerVerifierTest.php

@@ -71,6 +71,21 @@ public function testVerifyBaiduWebcrawler()
         ));
     }
 
+    public function testVerifyBegunWebcrawler()
+    {
+        $webcrawlerVerifier = new \WebcrawlerVerifier\WebcrawlerVerifier();
+        $userAgent = 'Mozilla/5.0 (compatible; BegunAdvertising/3.0; +http://begun.ru/begun/technology/indexer/)';
+
+        $this->assertEquals($webcrawlerVerifier::UNVERIFIED, $webcrawlerVerifier->verify(
+            $userAgent,
+            '192.168.0.1'
+        ));
+        $this->assertEquals($webcrawlerVerifier::VERIFIED, $webcrawlerVerifier->verify(
+            $userAgent,
+            '91.192.149.231'
+        ));
+    }
+
     public function testVerifyBingWebcrawler()
     {
         $webcrawlerVerifier = new \WebcrawlerVerifier\WebcrawlerVerifier();