Merge pull request #11 from tomverran/add-match-tests

Add match tests

Author: Tom

src/Robot/UserAgent.php

@@ -21,8 +21,8 @@ public function getMatches($userAgent)
         }
 
         $matches = array_filter($this->userAgents, function($elem) use($ua) {
-            return strpos($elem, $ua) !== false || $elem == '*';
+            return strpos($elem, $ua) !== false || strpos($ua, $elem) !== false || $elem == '*';
         });
-        return $matches;
+        return array_values($matches);
     }
 }

tests/RecordTest.php

@@ -36,4 +36,22 @@ public function givenUserAgentMatch_delegateToAccessRules()
         $this->assertFalse((new Record($alwaysMatchingUserAgent, $this->fooForbidden))->isAllowed('Googlebot', '/foo'));
         $this->assertTrue((new Record($alwaysMatchingUserAgent, $this->allUrlsAllowed))->isAllowed('Googlebot', '/foo'));
     }
+
+    /**
+     * @test
+     */
+    public function givenNoMatches_returnMatchStrengthOfZero()
+    {
+        $googleOnly = new Record(new UserAgent(['Googlebot']), new AccessRules([]));
+        $this->assertTrue($googleOnly->getMatchStrength('Bing') == 0, 'No match at all means a match strength of zero');
+    }
+
+    /**
+     * @test
+     */
+    public function givenMatch_returnLengthOfMatchedUAAsMatchStrength()
+    {
+        $googleOnly = new Record(new UserAgent(['Googlebot']), new AccessRules([]));
+        $this->assertTrue($googleOnly->getMatchStrength('G') == 9, 'Length of the matched UA is the strength');
+    }
 }

tests/UserAgentTest.php

@@ -93,4 +93,17 @@ public function givenWildcardAgent_alwaysMatch()
             $this->assertTrue($wildcard->getMatches($ua) == ['*'], 'wilcard matches all');
         }
     }
+
+    /**
+     * @test
+     * https://developers.google.com/webmasters/control-crawl-index/docs/robots_txt#order-of-precedence-for-user-agents
+     */
+    public function givenGoogleExamples_behaveAsExpected()
+    {
+        $googleUAs = new UserAgent(['googlebot-news', '*', 'googlebot']);
+        $this->assertContains('googlebot-news', $googleUAs->getMatches('Googlebot-News'));
+        $this->assertContains('googlebot', $googleUAs->getMatches('Googlebot-Images'));
+        $this->assertNotContains('googlebot-news', $googleUAs->getMatches('Googlebot-Images'));
+        $this->assertEquals(['*'], $googleUAs->getMatches('otherbot'));
+    }
 }