AbstractAdultProviderPipe::requiresAgeVerification()

Complexity

Conditions	7
Paths	11

Size

Total Lines	59
Code Lines	37

Duplication

Lines	0
Ratio	0 %

Importance

Changes	1
Bugs	0	Features	0

<?php
    protected function requiresAgeVerification(string $html): bool
    {
        // First check if this looks like a proper content page
        // Content pages have actual movie info, cast, etc.
        $contentIndicators = [
            '<title>.*?DVD.*?</title>',
            'product-info',
            'movie-details',
            'cast-list',
            'genre-list',
            '"@type":\s*"Movie"',
            '"@type":\s*"Product"',
        ];

        foreach ($contentIndicators as $pattern) {
            if (preg_match('/' . $pattern . '/is', $html)) {
                return false; // This is a content page, not an age verification page
            }
        }

        // Check for short page that might just be a redirect/age gate
        if (strlen($html) < 500) {
            return true; // Very short response likely means we got redirected
        }

        // Now check for explicit age verification indicators
        $agePatterns = [
            'age verification',
            'are you 18',
            'are you over 18',
            'confirm your age',
            'enter your age',
            'must be 18',
            'age-gate',
            'ageGate',
            'AgeConfirmation', // PopPorn specific
            'ageConfirmationButton', // ADE specific
            'age-confirmation', // Generic
            'verify your age',
            'adult content warning',
            'I am 18 or older',
            'I am over 18',
            'this site contains adult',
        ];

        // Count how many patterns match - if multiple match on a short page, it's likely age verification
        $matchCount = 0;
        foreach ($agePatterns as $pattern) {
            if (stripos($html, $pattern) !== false) {
                $matchCount++;
                // If the page is relatively short and has an age pattern, it's probably an age gate
                if (strlen($html) < 10000) {
                    return true;
                }
            }
        }

        // If multiple patterns match, it's likely an age verification page
        return $matchCount >= 2;
    }