matomo-org#7060 feat(bot) added support for detecting bots through cl…

…ient-hints
sanchezzzhak · Dec 30, 2022 · 1bb3bc0 · 1bb3bc0
1 parent 701351c
commit 1bb3bc0
Show file tree

Hide file tree

Showing 6 changed files with 72 additions and 8 deletions.
diff --git a/ClientHints.php b/ClientHints.php
@@ -76,6 +76,12 @@ class ClientHints
      */
     protected $app = '';
 
+    /**
+     * Represents `x-client` header field: additional header from applications/bots
+     * @var string
+     */
+    protected $xClient = '';
+
     /**
      * Constructor
      *
@@ -88,8 +94,9 @@ class ClientHints
      * @param string $architecture    `Sec-CH-UA-Arch` header field
      * @param string $bitness         `Sec-CH-UA-Bitness`
      * @param string $app             `HTTP_X-REQUESTED-WITH`
+     * @param string $xClient         `HTTP_X_CLIENT`
      */
-    public function __construct(string $model = '', string $platform = '', string $platformVersion = '', string $uaFullVersion = '', array $fullVersionList = [], bool $mobile = false, string $architecture = '', string $bitness = '', string $app = '') // phpcs:ignore Generic.Files.LineLength
+    public function __construct(string $model = '', string $platform = '', string $platformVersion = '', string $uaFullVersion = '', array $fullVersionList = [], bool $mobile = false, string $architecture = '', string $bitness = '', string $app = '', string $xClient = '') // phpcs:ignore Generic.Files.LineLength
     {
         $this->model           = $model;
         $this->platform        = $platform;
@@ -100,6 +107,7 @@ public function __construct(string $model = '', string $platform = '', string $p
         $this->architecture    = $architecture;
         $this->bitness         = $bitness;
         $this->app             = $app;
+        $this->xClient         = $xClient;
     }
 
     /**
@@ -221,6 +229,16 @@ public function getApp(): string
         return $this->app;
     }
 
+    /**
+     * Returns the xClient string
+     *
+     * @return string
+     */
+    public function getXClient(): string
+    {
+        return $this->xClient;
+    }
+
     /**
      * Factory method to easily instantiate this class using an array containing all available (client hint) headers
      *
@@ -231,7 +249,7 @@ public function getApp(): string
     public static function factory(array $headers): ClientHints
     {
         $model           = $platform = $platformVersion = $uaFullVersion = $architecture = $bitness = '';
-        $app             = '';
+        $app             = $xClient = '';
         $mobile          = false;
         $fullVersionList = [];
 
@@ -294,6 +312,7 @@ public static function factory(array $headers): ClientHints
                     if (!empty($fullVersionList)) {
                         break;
                     }
+
                     // use this only if no other header already set the list
                 case 'http-sec-ch-ua-full-version-list':
                 case 'sec-ch-ua-full-version-list':
@@ -316,6 +335,11 @@ public static function factory(array $headers): ClientHints
                         $app = $value;
                     }
 
+                    break;
+                case 'http-x-client':
+                case 'x-client':
+                    $xClient = $value;
+
                     break;
             }
         }
@@ -329,7 +353,8 @@ public static function factory(array $headers): ClientHints
             $mobile,
             $architecture,
             $bitness,
-            $app
+            $app,
+            $xClient
         );
     }
 }
diff --git a/Parser/AbstractParser.php b/Parser/AbstractParser.php
@@ -53,7 +53,7 @@ abstract class AbstractParser
      * Contains a list of mappings from names we use to known client hint values
      * @var array<string, array<string>>
      */
-    protected static $clientHintMapping = [];
+    protected static $xClientMapping = [];
 
     /**
      * Holds an array with method that should be available global
@@ -269,7 +269,7 @@ protected function getRegexes(): array
      */
     protected function applyClientHintMapping(string $name): string
     {
-        foreach (static::$clientHintMapping as $mappedName => $clientHints) {
+        foreach (static::$xClientMapping as $mappedName => $clientHints) {
             foreach ($clientHints as $clientHint) {
                 if (\strtolower($name) === \strtolower($clientHint)) {
                     return $mappedName;

diff --git a/Parser/Bot.php b/Parser/Bot.php
@@ -12,6 +12,8 @@
 
 namespace DeviceDetector\Parser;
 
+use DeviceDetector\ClientHints;
+
 /**
  * Class Bot
  *
@@ -44,6 +46,19 @@ public function discardDetails(): void
         $this->discardDetails = true;
     }
 
+    /**
+     * Contains a list of mappings from xClient names we use to known x-client values of bots
+     *
+     * @var array<string, array<string>>
+     */
+    protected static $xClientMapping = [
+        'Collabim' => [
+            'name'     => 'Collabim',
+            'category' => 'Crawler',
+            'url'      => 'https://www.collabim.com/',
+        ],
+    ];
+
     /**
      * Parses the current UA and checks whether it contains bot information
      *
@@ -64,6 +79,18 @@ public function discardDetails(): void
      */
     public function parse(): ?array
     {
+        if ($this->clientHints instanceof ClientHints && $this->clientHints->getXClient()) {
+            foreach (self::$xClientMapping as $key => $result) {
+                if ($this->fuzzyCompare($key, $this->clientHints->getXClient())) {
+                    if ($this->discardDetails) {
+                        return [true];
+                    }
+
+                    return $result;
+                }
+            }
+        }
+
         $result = null;
 
         if ($this->preMatchOverall()) {

diff --git a/Parser/OperatingSystem.php b/Parser/OperatingSystem.php
@@ -219,7 +219,7 @@ class OperatingSystem extends AbstractParser
      *
      * @var array<string, array<string>>
      */
-    protected static $clientHintMapping = [
+    protected static $xClientMapping = [
         'GNU/Linux' => ['Linux'],
         'Mac'       => ['MacOS'],
     ];

diff --git a/Tests/DeviceDetectorTest.php b/Tests/DeviceDetectorTest.php
@@ -385,8 +385,12 @@ public function testVersionTruncationForClientHints(): void
      */
     public function testParseBots(array $fixtureData): void
     {
-        $ua = $fixtureData['user_agent'];
-        $dd = new DeviceDetector($ua);
+        $ua          = $fixtureData['user_agent'];
+        $headers     = $fixtureData['headers'] ?? [];
+        $clientHints = ClientHints::factory($headers);
+        $dd          = new DeviceDetector($ua);
+
+        $dd->setClientHints($clientHints);
         $dd->parse();
         $this->assertTrue($dd->isBot());
         $botData = $dd->getBot();

diff --git a/Tests/fixtures/bots.yml b/Tests/fixtures/bots.yml
@@ -5452,3 +5452,11 @@
     name: ReqBin
     category: Crawler
     url: https://reqbin.com/curl
+-
+  user_agent: Mozilla/5.0 (Macintosh; Intel Mac OS X 10_15_1) AppleWebKit/605.1.15 (KHTML, like Gecko) Version/13.1.2 Safari/605.1.15
+  headers:
+    http-x-client: Collabim
+  bot:
+    name: Collabim
+    category: Crawler
+    url: https://www.collabim.com/