<?php
/**
 * CLI-Einstiegspunkt für System-Crontab.
 *
 * Aufruf:
 *   /usr/bin/php8.4 /var/www/clients/client2/web20/web/wp-content/plugins/dgt-verkehrsmonitor/cli/cron.php <action> [--bundesland=slug] [--full]
 *
 * Aktionen:
 *   fetch      – RSS-Quellen abrufen (alle 2 Stunden)
 *   scrape     – Blaulicht-Listenseiten scrapen (alle 4 Stunden)
 *   regions    – Presseportal Regional-Seiten scrapen (alle Bundesländer)
 *   mvnet      – polizei.mvnet.de scrapen (Mecklenburg-Vorpommern)
 *   analyze    – Ausstehende Entwürfe per KI analysieren (alle 30 Min)
 *   aggregate  – Tagesstatistiken + Report generieren (täglich 06:00)
 *   search     – Presseportal-Suche nach DUI-Keywords (Deep-Import)
 *   all        – fetch + scrape + regions + mvnet + analyze + aggregate nacheinander
 *
 * Optionen:
 *   --bundesland=slug  – Nur ein Bundesland scrapen (z.B. --bundesland=mecklenburg-vorpommern)
 *   --full             – Deep-Import: alle Seiten bis 01.01.2026 durchlaufen
 *
 * @package dgt-verkehrsmonitor
 */

// Keine Ausgabe-Buffer im CLI.
if ( PHP_SAPI !== 'cli' ) {
	http_response_code( 403 );
	exit( 'Nur per CLI aufrufbar.' );
}

// WordPress laden.
$wp_load = dirname( __DIR__, 4 ) . '/wp-load.php';
if ( ! file_exists( $wp_load ) ) {
	fwrite( STDERR, "wp-load.php nicht gefunden: {$wp_load}\n" );
	exit( 1 );
}
require_once $wp_load;

// Action aus Argument.
$action = isset( $argv[1] ) ? $argv[1] : '';

// --bundesland=slug parsen.
$bundesland = '';
$full_import = false;
foreach ( $argv as $arg ) {
	if ( 0 === strpos( $arg, '--bundesland=' ) ) {
		$bundesland = substr( $arg, 13 );
	}
	if ( '--full' === $arg ) {
		$full_import = true;
	}
}

$valid_actions = array( 'fetch', 'scrape', 'regions', 'search', 'mvnet', 'analyze', 'aggregate', 'all' );
if ( ! in_array( $action, $valid_actions, true ) ) {
	fwrite( STDERR, "Verwendung: php cron.php <" . implode( '|', $valid_actions ) . "> [--bundesland=slug] [--full]\n" );
	exit( 1 );
}

$start = microtime( true );

if ( 'fetch' === $action || 'all' === $action ) {
	echo "[" . date( 'Y-m-d H:i:s' ) . "] Fetch starten ...\n";
	dgt_vm_do_fetch_sources();
	$count = get_option( 'dgt_vm_last_fetch_count', 0 );
	echo "  -> {$count} neue Meldungen\n";
}

if ( 'scrape' === $action || 'all' === $action ) {
	$scrape_pages = $full_import ? 30 : 5;
	if ( $bundesland ) {
		echo "[" . date( 'Y-m-d H:i:s' ) . "] Listing-Scrape für {$bundesland} starten ...\n";
		$count = dgt_vm_do_scrape_listings_bl( $bundesland, $scrape_pages );
		echo "  -> {$count} neue Meldungen per Listing-Scrape ({$bundesland})\n";
	} else {
		echo "[" . date( 'Y-m-d H:i:s' ) . "] Listing-Scrape starten ...\n";
		$count = dgt_vm_do_scrape_listings( $scrape_pages );
		echo "  -> {$count} neue Meldungen per Listing-Scrape\n";
	}

	echo "[" . date( 'Y-m-d H:i:s' ) . "] Kategorie-Scrape starten ...\n";
	$count2 = dgt_vm_do_scrape_categories( $full_import ? 10 : 3 );
	echo "  -> {$count2} neue Meldungen per Kategorie-Scrape\n";
}

if ( 'regions' === $action || 'all' === $action ) {
	$region_pages = $full_import ? 200 : 5;
	$min_date     = $full_import ? '2026-01-01' : '';
	echo "[" . date( 'Y-m-d H:i:s' ) . "] Regional-Scrape starten" . ( $full_import ? ' (Full-Import ab 2026-01-01)' : '' ) . " ...\n";
	$count = dgt_vm_do_scrape_regions( $region_pages, $bundesland, $min_date );
	echo "  -> {$count} neue Meldungen per Regional-Scrape\n";
}

if ( 'search' === $action || 'all' === $action ) {
	$search_pages = $full_import ? 300 : 10;
	$min_date     = $full_import ? '2026-01-01' : '';
	echo "[" . date( 'Y-m-d H:i:s' ) . "] Presseportal-Suche starten" . ( $full_import ? ' (Full-Import ab 2026-01-01)' : '' ) . " ...\n";
	$count = dgt_vm_do_search_import( $search_pages, $min_date );
	echo "  -> {$count} neue Meldungen per Suche\n";
}

if ( 'mvnet' === $action || 'all' === $action ) {
	echo "[" . date( 'Y-m-d H:i:s' ) . "] MV-Net Scrape starten ...\n";
	$count = dgt_vm_scrape_mvnet( $full_import ? 30 : 5 );
	echo "  -> {$count} neue Meldungen per MV-Net-Scrape\n";
}

if ( 'analyze' === $action || 'all' === $action ) {
	echo "[" . date( 'Y-m-d H:i:s' ) . "] Analyse starten ...\n";
	dgt_vm_do_analyze_pending();
	$count = get_option( 'dgt_vm_last_analyze_count', 0 );
	echo "  -> {$count} Meldungen analysiert\n";
}

if ( 'aggregate' === $action || 'all' === $action ) {
	echo "[" . date( 'Y-m-d H:i:s' ) . "] Aggregation starten ...\n";
	dgt_vm_do_aggregate_daily();
	echo "  -> Tagesstatistik und Report generiert\n";
}

$elapsed = round( microtime( true ) - $start, 2 );
echo "[" . date( 'Y-m-d H:i:s' ) . "] Fertig ({$elapsed}s)\n";