<?php
declare(strict_types=1);

/**
 * Toy model: variable-ratio reinforcement loop (slot-machine style).
 * Not neuroscience, just a demonstrator for your paper.
 */

final class DopamineLoopSim
{
    public function __construct(
        private float $rewardChance = 0.18,    // probability that a scroll yields a "reward"
        private float $meanReward   = 1.0,     // average reward magnitude
        private float $decay        = 0.03     // baseline decay per step
    ) {}

    public function run(int $steps = 300): array
    {
        $state = [
            'arousal' => 0.2,   // "drive" / craving proxy
            'mood'    => 0.0,   // mood proxy (can drop when no reward)
            'habit'   => 0.0,   // habit strength proxy
        ];

        $timeline = [];

        for ($t = 1; $t <= $steps; $t++) {
            $gotReward = (mt_rand() / mt_getrandmax()) < $this->rewardChance;

            $reward = 0.0;
            if ($gotReward) {
                // random reward magnitude (skewed positive)
                $reward = max(0.0, $this->randomNormal($this->meanReward, 0.7));
                $state['mood'] += 0.15 * $reward;
                $state['habit'] += 0.05 * $reward;
                $state['arousal'] += 0.08 * $reward;
            } else {
                // no reward -> frustration/negative affect
                $state['mood'] -= 0.02;
                $state['arousal'] += 0.01; // craving increases when expectation isn't met
            }

            // decay back toward baseline
            $state['mood'] *= (1.0 - $this->decay);
            $state['arousal'] *= (1.0 - ($this->decay / 2));
            $state['habit'] *= (1.0 - ($this->decay / 4));

            // clamp ranges
            $state['mood'] = max(-1.0, min(1.0, $state['mood']));
            $state['arousal'] = max(0.0, min(2.0, $state['arousal']));
            $state['habit'] = max(0.0, min(5.0, $state['habit']));

            $timeline[] = [
                't' => $t,
                'reward' => $reward,
                'mood' => $state['mood'],
                'arousal' => $state['arousal'],
                'habit' => $state['habit'],
                'gotReward' => $gotReward,
            ];
        }

        return $timeline;
    }

    private function randomNormal(float $mean, float $std): float
    {
        // Box–Muller transform
        $u1 = max(1e-9, mt_rand() / mt_getrandmax());
        $u2 = max(1e-9, mt_rand() / mt_getrandmax());
        $z0 = sqrt(-2.0 * log($u1)) * cos(2.0 * M_PI * $u2);
        return $mean + $std * $z0;
    }
}

// Example:
$sim = new DopamineLoopSim(rewardChance: 0.15);
$data = $sim->run(200);

// Quick output (CSV-ish)
echo "t,gotReward,reward,mood,arousal,habit\n";
foreach ($data as $row) {
    echo implode(",", [
        $row['t'],
        $row['gotReward'] ? 1 : 0,
        number_format($row['reward'], 3, '.', ''),
        number_format($row['mood'], 3, '.', ''),
        number_format($row['arousal'], 3, '.', ''),
        number_format($row['habit'], 3, '.', ''),
    ]) . "\n";
}

<?php
declare(strict_types=1);

final class ContentItem
{
    public function __construct(
        public readonly string $id,
        public readonly float $intensity // 0..1 (higher = darker / more extreme)
    ) {}
}

final class RecommenderToy
{
    /** @param ContentItem[] $catalog */
    public function __construct(private array $catalog) {}

    /**
     * Pick top-N by predicted engagement.
     * Mood: -1..+1; lower mood => stronger pull to higher intensity.
     */
    public function recommend(float $mood, int $n = 5): array
    {
        $scored = [];

        foreach ($this->catalog as $item) {
            $engagement = $this->predictEngagement($mood, $item->intensity);
            $scored[] = ['item' => $item, 'score' => $engagement];
        }

        usort($scored, fn($a, $b) => $b['score'] <=> $a['score']);
        return array_map(fn($x) => $x['item'], array_slice($scored, 0, $n));
    }

    private function predictEngagement(float $mood, float $intensity): float
    {
        // Baseline: users generally engage more with emotionally intense stuff.
        $base = 0.2 + 0.6 * $intensity;

        // When mood is low, attraction to intense/negative content increases.
        // (This is a simplification of "mood-congruent information seeking".)
        $moodFactor = (-$mood) * 0.4 * $intensity; // mood -1 => +0.4*intensity, mood +1 => -0.4*intensity

        // Add noise
        $noise = (mt_rand() / mt_getrandmax()) * 0.08;

        return max(0.0, $base + $moodFactor + $noise);
    }
}

// Build catalog
$catalog = [];
for ($i = 1; $i <= 200; $i++) {
    $catalog[] = new ContentItem("post_$i", intensity: (mt_rand() / mt_getrandmax()));
}

$rec = new RecommenderToy($catalog);

// Simulate a user whose mood slowly drops, then see intensities recommended
$mood = 0.2;
for ($t = 1; $t <= 12; $t++) {
    $mood -= 0.12; // drifting down
    $recs = $rec->recommend($mood, 5);
    $avgIntensity = array_sum(array_map(fn($x) => $x->intensity, $recs)) / count($recs);

    echo "step=$t mood=" . number_format($mood, 2) .
        " avgRecommendedIntensity=" . number_format($avgIntensity, 2) . "\n";
}

<?php
declare(strict_types=1);

final class CrossrefClient
{
    public function fetchByDoi(string $doi): array
    {
        $doi = trim($doi);
        $url = "https://api.crossref.org/works/" . rawurlencode($doi);

        $json = $this->httpGet($url);
        $data = json_decode($json, true, flags: JSON_THROW_ON_ERROR);

        $msg = $data['message'] ?? [];

        $title = $msg['title'][0] ?? '';
        $container = $msg['container-title'][0] ?? '';
        $year = $msg['published-print']['date-parts'][0][0]
            ?? $msg['published-online']['date-parts'][0][0]
            ?? null;

        $authors = [];
        foreach (($msg['author'] ?? []) as $a) {
            $authors[] = trim(($a['given'] ?? '') . ' ' . ($a['family'] ?? ''));
        }

        return [
            'doi' => $doi,
            'title' => $title,
            'journal' => $container,
            'year' => $year,
            'url' => $msg['URL'] ?? null,
            'authors' => $authors,
        ];
    }

    private function httpGet(string $url): string
    {
        $ch = curl_init($url);
        if ($ch === false) {
            throw new RuntimeException("curl_init failed");
        }

        curl_setopt_array($ch, [
            CURLOPT_RETURNTRANSFER => true,
            CURLOPT_FOLLOWLOCATION => true,
            CURLOPT_TIMEOUT => 12,
            CURLOPT_HTTPHEADER => [
                'Accept: application/json',
                'User-Agent: rocenka-php-bib/1.0 (mailto:your@email.invalid)'
            ],
        ]);

        $out = curl_exec($ch);
        if ($out === false) {
            $err = curl_error($ch);
            curl_close($ch);
            throw new RuntimeException("HTTP request failed: $err");
        }

        $code = curl_getinfo($ch, CURLINFO_HTTP_CODE);
        curl_close($ch);

        if ($code >= 400) {
            throw new RuntimeException("HTTP $code for $url");
        }

        return $out;
    }
}

// Example usage:
$dois = [
    // sem si dáš DOI článků, co používáš
    '10.1136/bmj.o2374', // BMJ piece (example DOI format)
];

$client = new CrossrefClient();

foreach ($dois as $doi) {
    $ref = $client->fetchByDoi($doi);
    echo $ref['year'] . " | " . $ref['journal'] . " | " . $ref['title'] . "\n";
    echo "URL: " . ($ref['url'] ?? '-') . "\n";
    echo "Authors: " . implode(", ", $ref['authors']) . "\n\n";
}