<?php

namespace App\Console\Commands;

use App\Models\Document;
use App\Models\DocumentAccess;
use Carbon\Carbon;
use Illuminate\Console\Command;

class ImportDocumentAccessLog extends Command
{
    protected $signature = 'documents:import-access-log
                            {file : Path to an nginx access log in NCSA combined format}
                            {--source=nginx-backfill : Value stored in the source column}
                            {--dry-run : Parse and report without writing anything}';

    protected $description = 'Backfill document_accesses from an archived nginx access log';

    /**
     * Matches e.g.
     *   172.18.0.1 - - [23/May/2026:18:50:19 +0000] "GET /api/documents/83/download HTTP/1.1" 200 12345 "-" "Mozilla/5.0 ..."
     *
     * Only status 200 counts as an access: 499 means the client aborted the
     * transfer, and 4xx/5xx never delivered the file.
     */
    private const LINE = '#^\S+ \S+ \S+ \[([^\]]+)\] "GET /api/documents/(\d+)/download [^"]*" 200 \S+ "[^"]*" "([^"]*)"#';

    public function handle(): int
    {
        $file = $this->argument('file');

        if (! is_readable($file)) {
            $this->error("Cannot read file: {$file}");

            return self::FAILURE;
        }

        $source = $this->option('source');
        $dryRun = $this->option('dry-run');

        // Only import accesses for documents that still exist; the FK would
        // reject the rest and a deleted document has no dashboard row anyway.
        $knownIds = Document::pluck('id')->flip();

        $handle = fopen($file, 'r');
        $rows = [];
        $parsed = $skippedUnknown = $unmatched = 0;

        while (($line = fgets($handle)) !== false) {
            if (! preg_match(self::LINE, $line, $m)) {
                // Most lines are ordinary page views, not downloads.
                if (str_contains($line, '/download')) {
                    $unmatched++;
                }
                continue;
            }

            [, $timestamp, $documentId, $userAgent] = $m;

            if (! $knownIds->has((int) $documentId)) {
                $skippedUnknown++;
                continue;
            }

            $rows[] = [
                'document_id' => (int) $documentId,
                'accessed_at' => Carbon::createFromFormat('d/M/Y:H:i:s O', $timestamp)->utc(),
                'ip_address' => null,   // never recorded by the old log format
                'user_agent' => $userAgent,
                'source' => $source,
                'created_at' => now(),
                'updated_at' => now(),
            ];
            $parsed++;
        }

        fclose($handle);

        $this->info("Parsed {$parsed} successful downloads from {$file}");

        if ($skippedUnknown > 0) {
            $this->warn("Skipped {$skippedUnknown} referring to documents that no longer exist");
        }

        if ($unmatched > 0) {
            $this->warn("Skipped {$unmatched} download lines that were not status 200 (aborted or failed)");
        }

        if ($dryRun) {
            $this->comment('Dry run: nothing written.');

            return self::SUCCESS;
        }

        // Idempotency: a re-run must not double-count. Every row from a given
        // log carries the same source, so clearing that source first makes the
        // import safely repeatable.
        $existing = DocumentAccess::where('source', $source)->count();

        if ($existing > 0) {
            if (! $this->confirm("{$existing} rows with source '{$source}' already exist. Replace them?", true)) {
                $this->comment('Aborted; nothing written.');

                return self::SUCCESS;
            }

            DocumentAccess::where('source', $source)->delete();
        }

        foreach (array_chunk($rows, 500) as $chunk) {
            DocumentAccess::insert($chunk);
        }

        $this->info('Imported '.count($rows)." rows with source '{$source}'.");

        return self::SUCCESS;
    }
}