<?php
/**
 * Plugin Name:       Linkvane
 * Description:       Finds broken links in your posts, pages, comments and menus, and maps your internal links to show which pages receive the most link authority.
 * Version:           1.7.3
 * Requires at least: 6.2
 * Requires PHP:      8.0
 * Author:            Enrique
 * License:           GPL v2 or later
 * License URI:       https://www.gnu.org/licenses/gpl-2.0.html
 * Text Domain:       linkvane
 *
 * @package Linkvane
 */

if ( ! defined( 'ABSPATH' ) ) {
	exit;
}

// phpcs:disable WordPress.DB.DirectDatabaseQuery.DirectQuery, WordPress.DB.DirectDatabaseQuery.NoCaching -- This plugin owns custom tables, which have no core API or object cache group; every query goes through $wpdb->prepare().

define( 'NLHEALTH_VERSION', '1.7.3' );
define( 'NLHEALTH_DB_VERSION', '2.5' );
define( 'NLHEALTH_BATCH_SIZE', 5 );
define( 'NLHEALTH_PLUGIN_FILE', __FILE__ );
define( 'NLHEALTH_PLUGIN_BASENAME', plugin_basename( __FILE__ ) );
define( 'NLHEALTH_PLUGIN_DIR', plugin_dir_path( __FILE__ ) );
define( 'NLHEALTH_PLUGIN_URL', plugin_dir_url( __FILE__ ) );
define( 'NLHEALTH_TEXT_DOMAIN', 'linkvane' );

/*
 * The author's wordpress.org username. Keep it identical to the leading entry on
 * readme.txt's `Contributors:` line; wordpress.org rejects a Contributors entry
 * naming an account that does not exist. Nothing reads this constant at runtime.
 */
if ( ! defined( 'NLHEALTH_AUTHOR_WP_HANDLE' ) ) {
	define( 'NLHEALTH_AUTHOR_WP_HANDLE', 'lenrique5454' );
}

require_once NLHEALTH_PLUGIN_DIR . 'includes/class-nlhealth-activator.php';
require_once NLHEALTH_PLUGIN_DIR . 'includes/class-nlhealth-deactivator.php';
require_once NLHEALTH_PLUGIN_DIR . 'includes/class-nlhealth-scanner.php';
require_once NLHEALTH_PLUGIN_DIR . 'includes/class-nlhealth-link-graph.php';
require_once NLHEALTH_PLUGIN_DIR . 'includes/class-nlhealth-link-recommendations.php';
require_once NLHEALTH_PLUGIN_DIR . 'includes/class-nlhealth-rules-engine.php';
require_once NLHEALTH_PLUGIN_DIR . 'includes/class-nlhealth-seo-audit.php';
require_once NLHEALTH_PLUGIN_DIR . 'includes/class-nlhealth-export.php';
require_once NLHEALTH_PLUGIN_DIR . 'admin/class-nlhealth-admin.php';

/*
 * WP-CLI commands. Registered here rather than inside nlhealth_bootstrap(): that
 * function is runtime hook wiring, and registering commands is not a hook —
 * it has no business entangled with the is_admin() branch. This block sits
 * after the scanner/link-graph/seo-audit requires so the CLI class can name
 * them without load-order surprises.
 */
if ( defined( 'WP_CLI' ) && WP_CLI ) {
	require_once NLHEALTH_PLUGIN_DIR . 'includes/class-nlhealth-cli.php';
	WP_CLI::add_command( 'nlhealth', 'NLHealth_CLI' );
	WP_CLI::add_command( 'nlhealth juice', 'NLHealth_CLI_Juice' );
	WP_CLI::add_command( 'nlhealth db', 'NLHealth_CLI_DB' );
}

register_activation_hook( __FILE__, array( 'NLHealth_Activator', 'activate' ) );
register_deactivation_hook( __FILE__, array( 'NLHealth_Deactivator', 'deactivate' ) );

// A subsite created after a network activation never ran the activation hook.
add_action( 'wp_initialize_site', array( 'NLHealth_Activator', 'activate_new_site' ), 900 );

add_filter( 'cron_schedules', array( 'NLHealth_Activator', 'add_cron_schedule' ) );

// wp_schedule_event() freezes the interval at scheduling time, so a frequency
// change only takes effect once the booked event is cleared and re-created.
add_action( 'update_option_nlhealth_scan_frequency', array( 'NLHealth_Activator', 'reschedule_cron' ) );

// add_option fires instead of update_option when the value is written for
// the initial time, and the option is only seeded by set_default_options() during
// activate() — which an in-place plugin update never runs. On such an install
// the very first frequency change would otherwise save the new value while the
// scan silently kept its old cadence.
add_action( 'add_option_nlhealth_scan_frequency', array( 'NLHealth_Activator', 'reschedule_cron' ) );

/**
 * Returns the capability required to view the plugin's admin screens.
 *
 * Read access covers the dashboards (broken links, SEO audit, Link Juice) and
 * the read-only AJAX endpoints behind them. On agency sites the intended user
 * is often an Editor or a dedicated SEO role rather than an administrator, so
 * this is filterable. Destructive actions route through
 * nlhealth_get_write_capability() instead — lowering the bar for *looking* is not
 * the same as lowering it for *changing post content*.
 *
 * @since 1.5.6
 * @return string Capability slug.
 */
function nlhealth_get_capability(): string {
	/**
	 * Filters the capability required to read the plugin's admin screens.
	 *
	 * @since 1.5.6
	 * @param string $capability Default 'manage_options'.
	 */
	$capability = apply_filters( 'nlhealth_capability', 'manage_options' );

	return is_string( $capability ) && '' !== $capability ? $capability : 'manage_options';
}

/**
 * Returns the capability required for destructive/write actions.
 *
 * Applies to anything that edits post content or changes configuration:
 * single/bulk corrections, Link Juice re-linking, the ignore list, the rules
 * editor, and the Settings page. Read-only work (dashboards, scans, re-checks,
 * the SEO audit, PageRank recomputation, CSV export of data already visible on
 * the dashboard) stays on nlhealth_get_capability().
 *
 * This deliberately does NOT inherit from nlhealth_get_capability(): lowering the
 * bar to let an SEO role *look* at the dashboard must not silently also hand
 * that role bulk find-and-replace over every post on the site. Opening up reads
 * is one decision; opening up writes is a second, separate one, so a site that
 * wants a single gate has to say so by filtering both.
 *
 * @since 1.5.6
 * @return string Capability slug.
 */
function nlhealth_get_write_capability(): string {
	/**
	 * Filters the capability required for the plugin's destructive actions.
	 *
	 * @since 1.5.6
	 * @param string $capability Default 'manage_options' — independent of nlhealth_capability.
	 */
	$capability = apply_filters( 'nlhealth_write_capability', 'manage_options' );

	return is_string( $capability ) && '' !== $capability ? $capability : 'manage_options';
}

/**
 * Whether the current user may read the plugin's admin screens.
 *
 * @since 1.5.6
 * @return bool
 */
function nlhealth_current_user_can_read(): bool {
	return current_user_can( nlhealth_get_capability() );
}

/**
 * Whether the current user may perform the plugin's destructive actions.
 *
 * @since 1.5.6
 * @return bool
 */
function nlhealth_current_user_can_write(): bool {
	return current_user_can( nlhealth_get_write_capability() );
}

/**
 * Updates a link in post content with the native WordPress HTML tag processor.
 *
 * @param int    $post_id Post ID.
 * @param string $old_url URL to replace.
 * @param string $new_url Replacement URL.
 * @return bool
 */
function nlhealth_update_post_link( int $post_id, string $old_url, string $new_url ): bool {
	$scanner = new NLHealth_Scanner();
	return $scanner->update_post_link( $post_id, $old_url, $new_url );
}

/**
 * Boots plugin runtime hooks.
 *
 * @return void
 */
function nlhealth_bootstrap(): void {
	$scanner = new NLHealth_Scanner();
	add_action( 'nlhealth_run_batch', array( $scanner, 'run_batch' ) );
	add_action( 'save_post', array( $scanner, 'handle_post_saved' ), 10, 3 );
	add_action( 'deleted_post', 'nlhealth_cleanup_deleted_post' );
	add_action( 'delete_comment', 'nlhealth_cleanup_deleted_comment' );

	if ( is_admin() ) {
		// Plugin file updates don't re-run activation, so apply any pending DB
		// schema upgrades (e.g. the v2.1 link-map tables) on existing installs.
		add_action( 'admin_init', array( 'NLHealth_Activator', 'maybe_upgrade' ) );

		$admin = new NLHealth_Admin( $scanner );
		$admin->init();
	} elseif ( defined( 'WP_CLI' ) && WP_CLI ) {
		/*
		 * is_admin() is false under WP-CLI, so the admin_init above never
		 * fires. Without this, `wp nlhealth scan` on a site whose admin has not
		 * been visited since the update runs against a stale schema — exactly
		 * the headless installs the CLI exists for.
		 *
		 * elseif rather than an unconditional add_action keeps the front-end
		 * path byte-identical (no extra option read per pageview, which is why
		 * the check was gated to admin to begin with). Priority 1 on init
		 * rather than an immediate call because maybe_upgrade() → upgrade() →
		 * upgrade_to_2_1() requires wp-admin/includes/upgrade.php and dbDelta
		 * wants $wpdb->charset/collate initialised; init is comfortably before
		 * WP-CLI dispatches the command and comfortably after WP is ready, and
		 * it leaves `wp --skip-plugins` / `wp core install` bootstraps alone.
		 */
		add_action( 'init', array( 'NLHealth_Activator', 'maybe_upgrade' ), 1 );
	}
}

nlhealth_bootstrap();

/**
 * Cleans up link scan data when a post is deleted.
 *
 * @since 1.0.1
 * @param int $post_id Deleted post ID.
 */
function nlhealth_cleanup_deleted_post( int $post_id ): void {
	global $wpdb;

	$post_id = (int) $post_id;
	if ( $post_id <= 0 ) {
		return;
	}

	$errors_table = $wpdb->prefix . 'nlhealth_link_errors';
	$scanner      = new NLHealth_Scanner();
	$suffix       = $scanner->state_key_suffix( 'post', $post_id ); // "{id}".

	// Read the hashes BEFORE deleting the rows — they are the only handle on
	// this post's option/transient state.
	$hashes = $wpdb->get_col(
		$wpdb->prepare(
			'SELECT url_hash FROM %i WHERE post_id = %d AND source_type = %s',
			$errors_table,
			$post_id,
			'post'
		)
	);

	foreach ( (array) $hashes as $hash ) {
		delete_option( 'nlhealth_last_ok_' . $hash . '_' . $suffix );
		delete_option( 'nlhealth_last_soft_' . $hash . '_' . $suffix );
		delete_transient( 'nlhealth_retry_' . $hash . '_' . $suffix );
		delete_transient( 'nlhealth_fail_' . $hash . '_' . $suffix );
		// Do NOT delete 'nlhealth_ok_' . $hash — that cache is keyed by URL alone
		// and is shared site-wide; clearing it would force every other post
		// using the same URL back over the network on the next scan.
	}

	// Only the post's own link records use this post_id; comment records reuse
	// the post_id column for the comment id, so scope the errors delete by source.
	$wpdb->delete(
		$errors_table,
		array(
			'post_id'     => $post_id,
			'source_type' => 'post',
		),
		array( '%d', '%s' )
	);

	// Events are source-scoped since DB 2.4: without the source_type clause
	// this would also destroy the events of the *comment* whose id happens to
	// equal this post id.
	$wpdb->delete(
		$wpdb->prefix . 'nlhealth_link_events',
		array(
			'post_id'     => $post_id,
			'source_type' => 'post',
		),
		array( '%d', '%s' )
	);

	// nlhealth_correction_log carries post ids only — no path ever corrects a
	// comment — so it needs no source scoping.
	$wpdb->delete(
		$wpdb->prefix . 'nlhealth_correction_log',
		array( 'post_id' => $post_id ),
		array( '%d' )
	);

	// Drop the post from the internal link graph (both as a source and a
	// target) and remove its cached juice score.
	( new NLHealth_Link_Graph() )->delete_post( $post_id );
}

/**
 * Cleans up link scan data when a comment is deleted.
 *
 * Comment-sourced broken-link records key the comment id into the post_id
 * column with source_type 'comment', so removal is scoped to that comment.
 *
 * Hooked to delete_comment rather than deleted_comment because it fires while
 * the row still exists, which is what makes reading the url hashes possible.
 * trash_comment / spam_comment are deliberately NOT wired here: a trashed or
 * spammed comment can be restored, and discarding its records would lose
 * discovered_at history and could leave the forward-only comment cursor
 * (nlhealth_comment_scan_last_id) skipping it for a whole cycle.
 *
 * @param int|string $comment_id Deleted comment ID.
 */
function nlhealth_cleanup_deleted_comment( $comment_id ): void {
	global $wpdb;

	$comment_id = (int) $comment_id;
	if ( $comment_id <= 0 ) {
		return;
	}

	$errors_table = $wpdb->prefix . 'nlhealth_link_errors';
	$scanner      = new NLHealth_Scanner();
	$suffix       = $scanner->state_key_suffix( 'comment', $comment_id ); // "comment_{id}".

	// Read the hashes BEFORE deleting the rows — they are the only handle on
	// this comment's option/transient state.
	$hashes = $wpdb->get_col(
		$wpdb->prepare(
			'SELECT url_hash FROM %i WHERE post_id = %d AND source_type = %s',
			$errors_table,
			$comment_id,
			'comment'
		)
	);

	foreach ( (array) $hashes as $hash ) {
		delete_option( 'nlhealth_last_ok_' . $hash . '_' . $suffix );
		delete_option( 'nlhealth_last_soft_' . $hash . '_' . $suffix );
		delete_transient( 'nlhealth_retry_' . $hash . '_' . $suffix );
		delete_transient( 'nlhealth_fail_' . $hash . '_' . $suffix );
		// Do NOT delete 'nlhealth_ok_' . $hash — that cache is keyed by URL alone
		// and is shared site-wide; clearing it would force every post using
		// the same URL back over the network on the next scan.
	}

	$wpdb->delete(
		$errors_table,
		array(
			'post_id'     => $comment_id,
			'source_type' => 'comment',
		),
		array( '%d', '%s' )
	);

	$wpdb->delete(
		$wpdb->prefix . 'nlhealth_link_events',
		array(
			'post_id'     => $comment_id,
			'source_type' => 'comment',
		),
		array( '%d', '%s' )
	);

	// nlhealth_correction_log is left untouched on purpose: no path corrects a
	// comment (see the source_type fix in the bulk-correct query).
}