Laravel Page Views

Privacy-first pageview tracking for Laravel. Records page views with anonymized IPs, cookieless visitor counting, bot filtering and optional country resolution — all without external services.

• No cookies and nothing read from the device, so no consent banner
• No third-party requests; nothing leaves your server
• IP addresses are truncated and the raw User-Agent is never stored
• Visitors are counted through a daily-rotating hash whose inputs are not retained
• Works behind full-page caching by importing access logs
• Writes happen after the response is sent, and never break the request

Requirements

• PHP 8.2+ (PHP 8.3+ for Laravel 13)
• Laravel 12 or 13

Installation

composer require mdbtq/laravel-page-views
php artisan migrate

The migration ships with the package and runs from there. You only need to publish it if you want to edit the schema:

php artisan vendor:publish --tag=page-views-migrations

Once a copy exists in database/migrations, the package stops loading its own so the migration is never registered twice.

Optionally publish the config:

php artisan vendor:publish --tag=page-views-config

Setup

Register the middleware in bootstrap/app.php:

->withMiddleware(function (Middleware $middleware) {
    $middleware->append(\Mdbtq\PageViews\Middleware\TrackPageView::class);
})

Tracking runs in the middleware's terminate() method, so it happens after the response has been sent to the browser.

What gets recorded

A page view is skipped when the request is not a GET, the response status is not trackable, the path is excluded, the User-Agent looks like a bot, or the visitor sends a Do Not Track signal.

Configuration

Visitor counting

Unique visitors are counted through a hash of the IP address, the User-Agent and a salt that includes the current date. Because the date is part of the input, the same visitor produces a different hash tomorrow, and yesterday's records cannot be correlated with today's.

That rotation only holds if the hash inputs are not recoverable, which is why the raw User-Agent is never stored and the IP is truncated before it is written. The stored row does not contain enough to recompute its own hash.

Set a dedicated salt so the hashes do not depend on a secret used for everything else:

PAGE_VIEWS_SALT=some-long-random-string
PAGE_VIEWS_SALT_ROTATE_DAYS=30

The salt itself rotates every 30 days by default, bounding how long a single secret governs the whole table. Rotation deliberately makes hashes from earlier periods unreproducible, so returning-visitor comparisons do not span a boundary. Set PAGE_VIEWS_SALT_ROTATE_DAYS=0 to keep one static salt.

When visitor_hash.enabled is false, unique-visitor counts fall back to distinct anonymized IPs.

What this does and does not claim

The package stores no cookies and reads nothing from the visitor's device, so it stays outside the consent requirement in article 5(3) of the ePrivacy directive (in the Netherlands, article 11.7a Tw). That is the basis for running it without a cookie banner.

It does not follow that the GDPR stops applying. You still need a lawful basis for the processing (typically legitimate interest), a retention period that is actually enforced, and a mention in your privacy statement. The defaults here are built to support that — anonymized IPs, no raw User-Agent, rotating salts, and a scheduled purge — but the assessment is yours to make, and this is not legal advice.

Queued writes

On busy sites, write the record from a queue worker instead of the web process:

PAGE_VIEWS_DRIVER=queue
PAGE_VIEWS_QUEUE_CONNECTION=redis
PAGE_VIEWS_QUEUE=analytics

Excluding requests yourself

Register a condition from a service provider to veto page views the config cannot express:

use Mdbtq\PageViews\PageViewRecorder;

$this->app->make(PageViewRecorder::class)->trackWhen(
    fn ($request) => ! $request->user()?->isAdmin()
);

Reacting to page views

Every recorded view dispatches an event:

use Mdbtq\PageViews\Events\PageViewRecorded;

Event::listen(PageViewRecorded::class, function (PageViewRecorded $event) {
    // $event->pageView
});

Country Resolution

Install torann/geoip for automatic country detection:

composer require torann/geoip

The package auto-detects its presence — no additional configuration needed. Addresses the resolver cannot place are stored as null rather than as a default country.

Commands

# View statistics
php artisan pageviews:stats
php artisan pageviews:stats --days=7 --top=20
php artisan pageviews:stats --json

# Aggregate into daily totals
php artisan pageviews:rollup
php artisan pageviews:rollup --days=30

# Purge old records
php artisan pageviews:purge
php artisan pageviews:purge --days=30
php artisan pageviews:purge --dry-run

# Import from an access log
php artisan pageviews:import /var/log/nginx/access.log
php artisan pageviews:import access.log --since=2024-10-01 --host=example.com
php artisan pageviews:import access.log --dry-run
cat access.log | php artisan pageviews:import -

--days sets the reporting window and --top limits the ranked breakdowns; the daily series always covers the full window.

Importing from access logs

The middleware only sees requests that reach PHP. Full-page caching at a CDN or in nginx serves a hit without ever touching Laravel, so those visits are missing from the table. pageviews:import closes that gap by reading the access log, which records the request regardless of who answered it.

Log entries run through the same recorder as live traffic, so bot filtering, excluded paths, IP anonymization and visitor hashing behave identically. The visitor hash is derived with the salt of the day the request was served, which keeps the daily rotation intact for imported rows.

Two formats are recognised: the NCSA combined format that nginx and Apache emit by default, and JSON, which covers most CDN log pipelines (field names from nginx and Cloudflare are both understood). Lines that cannot be parsed are counted and skipped rather than aborting the run.

The command has no memory of previous runs, so overlapping imports insert duplicates. Use --since to bound each run, or import whole rotated log files exactly once.

Retention and history

pageviews:purge deletes raw rows in chunks so a large backlog does not hold one long transaction open. pageviews:rollup aggregates raw views into page_view_daily (per day, path and country), which keeps long-term history even when raw rows are purged aggressively. Re-running the rollup for a day replaces that day's totals, so it is safe to run repeatedly.

Both are scheduled by default, so the retention period in purge_days is enforced rather than advisory. The rollup runs first, preserving long-term totals in page_view_daily before the purge removes the raw rows behind them.

This requires the Laravel scheduler to be running. Without it nothing is purged and the table grows indefinitely.

Disable either one if you want to run them yourself:

PAGE_VIEWS_SCHEDULE_PURGE=false
PAGE_VIEWS_SCHEDULE_ROLLUP=false

Testing

composer test

The suite runs on SQLite by default. Set DB_DRIVER to mysql or pgsql to run it against a real database; CI runs all three.

License

MIT