nextcloud-analytics/analyticshub/lib/Service/DataProcessor.php

<?php

declare(strict_types=1);

namespace OCA\AnalyticsHub\Service;

use OCA\AnalyticsHub\Model\ClientConfig;
use OCP\Util\Logger;

/**
 * Data Processor Service
 * Handles intelligent delta calculations and data validation
 */
class DataProcessor {

    private ?Logger $logger;

    // Default thresholds
    private const DEFAULT_SIGNIFICANT_CHANGE_PCT = 20;
    private const DEFAULT_SIGNIFICANT_CHANGE_ABS = 5;

    public function __construct() {
        $this->logger = \OC::$server->getLogger();
    }

    /**
     * Process and validate GA4 data
     */
    public function process(array $rawData, ClientConfig $client): array {
        $this->logger->info("Processing data for: {$client->getName()}");

        try {
            // Validate data completeness
            $this->validateCompleteness($rawData);

            // Calculate deltas
            $processed = $this->calculateDeltas($rawData);

            // Apply client thresholds
            $processed = $this->applyThresholds($processed, $client);

            // Generate summary
            $processed['summary'] = $this->generateSummary($processed);

            return $processed;

        } catch (\Exception $e) {
            $this->logger->error("Data processing failed: {$e->getMessage()}");
            throw $e;
        }
    }

    /**
     * Validate data completeness
     */
    private function validateCompleteness(array $data): void {
        $requiredKeys = ['dates', 'metrics'];

        foreach ($requiredKeys as $key) {
            if (!isset($data[$key])) {
                throw new DataIncompleteException("Missing required key: {$key}");
            }
        }

        // Check we have at least 7 days of data
        if (!isset($data['dates']) || count($data['dates']) < 7) {
            throw new DataIncompleteException("Insufficient data - need at least 7 days");
        }

        // Check for null or empty metrics
        $metrics = $data['metrics'] ?? [];
        $requiredMetrics = ['sessions', 'totalUsers', 'conversions', 'eventCount'];

        foreach ($requiredMetrics as $metric) {
            if (!isset($metrics[$metric])) {
                throw new DataIncompleteException("Missing required metric: {$metric}");
            }

            $values = $metrics[$metric] ?? [];
            foreach ($values as $value) {
                if ($value === null || $value === '') {
                    throw new DataIncompleteException("Null or empty value in metric: {$metric}");
                }
            }
        }

        $this->logger->info('Data validation passed');
    }

    /**
     * Calculate intelligent deltas
     */
    private function calculateDeltas(array $data): array {
        $metrics = $data['metrics'] ?? [];
        $processed = [];

        foreach ($metrics as $metric => $values) {
            $processed[$metric] = $this->calculateMetricDelta($values);
        }

        $processed['dates'] = $data['dates'] ?? [];

        return $processed;
    }

    /**
     * Calculate delta for a single metric
     */
    private function calculateMetricDelta(array $values): array {
        if (count($values) < 2) {
            return [
                'change' => 0,
                'change_pct' => 0,
                'abs_change' => 0,
                'is_significant' => false,
                'label' => 'Insufficient data',
                'trend' => 'flat'
            ];
        }

        $current = $values[count($values) - 1]; // Last 7 days
        $previous = $values[0]; // First 7 days (comparison)

        // Edge case 1: Division by zero
        if ($previous == 0) {
            if ($current == 0) {
                return [
                    'change' => 0,
                    'change_pct' => 0,
                    'abs_change' => 0,
                    'is_significant' => false,
                    'label' => 'No change',
                    'trend' => 'stable'
                ];
            } else {
                return [
                    'change' => $current,
                    'change_pct' => null,
                    'abs_change' => $current,
                    'is_significant' => true,
                    'label' => "New activity (+{$current})",
                    'trend' => 'increasing'
                ];
            }
        }

        // Calculate percentage change
        $changePct = (($current - $previous) / $previous) * 100;
        $absChange = abs($current - $previous);

        // Edge case 2: Small numbers creating misleading %
        // Require both % threshold AND minimum absolute change
        $thresholds = [
            'significant_change_pct' => self::DEFAULT_SIGNIFICANT_CHANGE_PCT,
            'significant_change_abs' => self::DEFAULT_SIGNIFICANT_CHANGE_ABS
        ];

        $isSignificant = (abs($changePct) > $thresholds['significant_change_pct']
                        && $absChange > $thresholds['significant_change_abs']);

        // Determine trend
        $trend = 'stable';
        if ($absChange > $thresholds['significant_change_abs']) {
            $trend = $changePct > 0 ? 'increasing' : 'decreasing';
        }

        return [
            'current' => $current,
            'previous' => $previous,
            'change' => $current - $previous,
            'change_pct' => round($changePct, 1),
            'abs_change' => $absChange,
            'is_significant' => $isSignificant,
            'label' => $this->formatDeltaLabel($changePct, $absChange, $isSignificant),
            'trend' => $trend
        ];
    }

    /**
     * Format delta label
     */
    private function formatDeltaLabel(float $changePct, float $absChange, bool $isSignificant): string {
        if (!$isSignificant) {
            return 'No significant change';
        }

        $direction = $changePct > 0 ? '+' : '';
        $pct = abs($changePct);
        $change = (int)$absChange;

        return "{$direction}{$pct}% ({$direction}{$change})";
    }

    /**
     * Apply client-specific thresholds
     */
    private function applyThresholds(array $processed, ClientConfig $client): array {
        $thresholds = $client->getThresholds();

        if (!$thresholds) {
            // Use defaults
            $processed['significant_change_pct'] = self::DEFAULT_SIGNIFICANT_CHANGE_PCT;
            $processed['significant_change_abs'] = self::DEFAULT_SIGNIFICANT_CHANGE_ABS;
        } else {
            // Use client thresholds
            $processed['significant_change_pct'] = $thresholds['significant_change_pct'] ?? self::DEFAULT_SIGNIFICANT_CHANGE_PCT;
            $processed['significant_change_abs'] = $thresholds['significant_change_abs'] ?? self::DEFAULT_SIGNIFICANT_CHANGE_ABS;
        }

        // Re-evaluate significance based on thresholds
        foreach ($processed as $metric => $data) {
            if (is_array($data)) {
                $data['is_significant'] = (
                    abs($data['change_pct'] ?? 0) > $processed['significant_change_pct']
                    && abs($data['abs_change'] ?? 0) > $processed['significant_change_abs']
                );
            }
        }

        return $processed;
    }

    /**
     * Generate summary
     */
    private function generateSummary(array $processed): array {
        $totalMetrics = 0;
        $significantChanges = 0;
        $increasing = 0;
        $decreasing = 0;

        foreach ($processed as $metric => $data) {
            if (is_array($data)) {
                $totalMetrics++;
                if ($data['is_significant'] ?? false) {
                    $significantChanges++;
                }
                if (($data['trend'] ?? '') === 'increasing') {
                    $increasing++;
                }
                if (($data['trend'] ?? '') === 'decreasing') {
                    $decreasing++;
                }
            }
        }

        return [
            'total_metrics' => $totalMetrics,
            'significant_changes' => $significantChanges,
            'increasing' => $increasing,
            'decreasing' => $decreasing
        ];
    }
}