Mirrors/shlink
1
0
Fork 0
mirror of https://github.com/shlinkio/shlink.git synced 2026-03-10 17:23:12 +08:00
Code Issues Packages Projects Releases Wiki Activity

Added detection of visits from potential bots

Browse Source
This commit is contained in:
Alejandro Celaya
2021-05-22 15:09:14 +02:00
parent 663ae9f6bb
commit 9fa32b5b6b
16 changed files with 123 additions and 19 deletions
Download Patch File Download Diff File Expand all files Collapse all files

6
module/Core/src/Entity/Visit.php
View File

@@ -13,6 +13,8 @@ use Shlinkio\Shlink\Core\Model\Visitor;
use Shlinkio\Shlink\Core\Visit\Model\VisitLocationInterface;
use Shlinkio\Shlink\Importer\Model\ImportedShlinkVisit;
use function Shlinkio\Shlink\Core\isCrawler;
class Visit extends AbstractEntity implements JsonSerializable
{
public const TYPE_VALID_SHORT_URL = 'valid_short_url';
@@ -29,6 +31,7 @@ class Visit extends AbstractEntity implements JsonSerializable
private string $type;
private ?ShortUrl $shortUrl;
private ?VisitLocation $visitLocation = null;
private bool $potentialBot;
private function __construct(?ShortUrl $shortUrl, string $type)
{
@@ -49,6 +52,7 @@ class Visit extends AbstractEntity implements JsonSerializable
{
$instance = new self($shortUrl, self::TYPE_IMPORTED);
$instance->userAgent = $importedVisit->userAgent();
$instance->potentialBot = isCrawler($instance->userAgent);
$instance->referer = $importedVisit->referer();
$instance->date = Chronos::instance($importedVisit->date());
@@ -88,6 +92,7 @@ class Visit extends AbstractEntity implements JsonSerializable
$this->referer = $visitor->getReferer();
$this->remoteAddr = $this->processAddress($anonymize, $visitor->getRemoteAddress());
$this->visitedUrl = $visitor->getVisitedUrl();
$this->potentialBot = $visitor->isPotentialBot();
}
private function processAddress(bool $anonymize, ?string $address): ?string
@@ -166,6 +171,7 @@ class Visit extends AbstractEntity implements JsonSerializable
'date' => $this->date->toAtomString(),
'userAgent' => $this->userAgent,
'visitLocation' => $this->visitLocation,
'potentialBot' => $this->potentialBot,
];
}

21
module/Core/src/Model/Visitor.php
View File

@@ -8,6 +8,7 @@ use Psr\Http\Message\ServerRequestInterface;
use Shlinkio\Shlink\Common\Middleware\IpAddressMiddlewareFactory;
use Shlinkio\Shlink\Core\Options\TrackingOptions;
use function Shlinkio\Shlink\Core\isCrawler;
use function substr;
final class Visitor
@@ -21,6 +22,7 @@ final class Visitor
private string $referer;
private string $visitedUrl;
private ?string $remoteAddress;
private bool $potentialBot;
public function __construct(string $userAgent, string $referer, ?string $remoteAddress, string $visitedUrl)
{
@@ -28,6 +30,7 @@ final class Visitor
$this->referer = $this->cropToLength($referer, self::REFERER_MAX_LENGTH);
$this->visitedUrl = $this->cropToLength($visitedUrl, self::VISITED_URL_MAX_LENGTH);
$this->remoteAddress = $this->cropToLength($remoteAddress, self::REMOTE_ADDRESS_MAX_LENGTH);
$this->potentialBot = isCrawler($userAgent);
}
private function cropToLength(?string $value, int $length): ?string
@@ -70,14 +73,22 @@ final class Visitor
return $this->visitedUrl;
}
public function isPotentialBot(): bool
{
return $this->potentialBot;
}
public function normalizeForTrackingOptions(TrackingOptions $options): self
{
$instance = self::emptyInstance();
$instance = new self(
$options->disableUaTracking() ? '' : $this->userAgent,
$options->disableReferrerTracking() ? '' : $this->referer,
$options->disableIpTracking() ? null : $this->remoteAddress,
$this->visitedUrl,
);
$instance->userAgent = $options->disableUaTracking() ? '' : $this->userAgent;
$instance->referer = $options->disableReferrerTracking() ? '' : $this->referer;
$instance->remoteAddress = $options->disableIpTracking() ? null : $this->remoteAddress;
$instance->visitedUrl = $this->visitedUrl;
// Keep the fact that the visit was a potential bot, even if we no longer save the user agent
$instance->potentialBot = $this->potentialBot;
return $instance;
}