<?php

[0] Fix | Delete

/**

[1] Fix | Delete

* The Crawler Sitemap Class.

[2] Fix | Delete

*

[3] Fix | Delete

* @package LiteSpeed

[4] Fix | Delete

* @since 1.1.0

[5] Fix | Delete

*/

[6] Fix | Delete

[7] Fix | Delete

namespace LiteSpeed;

[8] Fix | Delete

[9] Fix | Delete

defined( 'WPINC' ) || exit();

[10] Fix | Delete

[11] Fix | Delete

/**

[12] Fix | Delete

* Class Crawler_Map

[13] Fix | Delete

*

[14] Fix | Delete

* Maintains and persists crawler sitemap/blacklist state, parses custom sitemaps,

[15] Fix | Delete

* and exposes helpers to query & mutate crawler results.

[16] Fix | Delete

*/

[17] Fix | Delete

class Crawler_Map extends Root {

[18] Fix | Delete

[19] Fix | Delete

const LOG_TAG = '🐞🗺️';

[20] Fix | Delete

[21] Fix | Delete

const BM_MISS = 1;

[22] Fix | Delete

const BM_HIT = 2;

[23] Fix | Delete

const BM_BLACKLIST = 4;

[24] Fix | Delete

[25] Fix | Delete

/**

[26] Fix | Delete

* Site URL used to simplify URLs.

[27] Fix | Delete

*

[28] Fix | Delete

* @var string

[29] Fix | Delete

*/

[30] Fix | Delete

private $_site_url;

[31] Fix | Delete

[32] Fix | Delete

/**

[33] Fix | Delete

* Main crawler table name.

[34] Fix | Delete

*

[35] Fix | Delete

* @var string

[36] Fix | Delete

*/

[37] Fix | Delete

private $_tb;

[38] Fix | Delete

[39] Fix | Delete

/**

[40] Fix | Delete

* Crawler blacklist table name.

[41] Fix | Delete

*

[42] Fix | Delete

* @var string

[43] Fix | Delete

*/

[44] Fix | Delete

private $_tb_blacklist;

[45] Fix | Delete

[46] Fix | Delete

/**

[47] Fix | Delete

* Data service instance.

[48] Fix | Delete

*

[49] Fix | Delete

* @var \LiteSpeed\Data

[50] Fix | Delete

*/

[51] Fix | Delete

private $__data;

[52] Fix | Delete

[53] Fix | Delete

/**

[54] Fix | Delete

* Timeout (seconds) when fetching sitemaps.

[55] Fix | Delete

*

[56] Fix | Delete

* @var int

[57] Fix | Delete

*/

[58] Fix | Delete

private $_conf_map_timeout;

[59] Fix | Delete

[60] Fix | Delete

/**

[61] Fix | Delete

* Collected URLs from parsed sitemaps.

[62] Fix | Delete

*

[63] Fix | Delete

* @var array<int,string>

[64] Fix | Delete

*/

[65] Fix | Delete

private $_urls = [];

[66] Fix | Delete

[67] Fix | Delete

/**

[68] Fix | Delete

* Instantiate the class.

[69] Fix | Delete

*

[70] Fix | Delete

* @since 1.1.0

[71] Fix | Delete

*/

[72] Fix | Delete

public function __construct() {

[73] Fix | Delete

$this->_site_url = get_site_url();

[74] Fix | Delete

$this->__data = Data::cls();

[75] Fix | Delete

$this->_tb = $this->__data->tb( 'crawler' );

[76] Fix | Delete

$this->_tb_blacklist = $this->__data->tb( 'crawler_blacklist' );

[77] Fix | Delete

// Specify the timeout while parsing the sitemap.

[78] Fix | Delete

$this->_conf_map_timeout = defined( 'LITESPEED_CRAWLER_MAP_TIMEOUT' ) ? constant( 'LITESPEED_CRAWLER_MAP_TIMEOUT' ) : 180;

[79] Fix | Delete

}

[80] Fix | Delete

[81] Fix | Delete

/**

[82] Fix | Delete

* Save URLs crawl status into DB.

[83] Fix | Delete

*

[84] Fix | Delete

* @since 3.0

[85] Fix | Delete

* @access public

[86] Fix | Delete

*

[87] Fix | Delete

* @param array<int,array<int,array{url:string,code:int}>> $items Map of bit => [ id => [url, code] ].

[88] Fix | Delete

* @param int $curr_crawler Current crawler index (0-based).

[89] Fix | Delete

* @return array<int,array>

[90] Fix | Delete

*/

[91] Fix | Delete

public function save_map_status( $items, $curr_crawler ) {

[92] Fix | Delete

global $wpdb;

[93] Fix | Delete

Utility::compatibility();

[94] Fix | Delete

[95] Fix | Delete

$total_crawler = count( Crawler::cls()->list_crawlers() );

[96] Fix | Delete

$total_crawler_pos = $total_crawler - 1;

[97] Fix | Delete

[98] Fix | Delete

// Replace current crawler's position.

[99] Fix | Delete

$curr_crawler = (int) $curr_crawler;

[100] Fix | Delete

foreach ( $items as $bit => $ids ) {

[101] Fix | Delete

// $ids = [ id => [ url, code ], ... ].

[102] Fix | Delete

if ( ! $ids ) {

[103] Fix | Delete

continue;

[104] Fix | Delete

}

[105] Fix | Delete

self::debug( 'Update map [crawler] ' . $curr_crawler . ' [bit] ' . $bit . ' [count] ' . count( $ids ) );

[106] Fix | Delete

[107] Fix | Delete

// Update res first, then reason

[108] Fix | Delete

$right_pos = $total_crawler_pos - $curr_crawler;

[109] Fix | Delete

$id_all = implode(',', array_map('intval', array_keys($ids)));

[110] Fix | Delete

[111] Fix | Delete

// phpcs:ignore WordPress.DB.DirectDatabaseQuery.DirectQuery, WordPress.DB.PreparedSQL.InterpolatedNotPrepared

[112] Fix | Delete

$wpdb->query("UPDATE `$this->_tb` SET res = CONCAT( LEFT( res, $curr_crawler ), '$bit', RIGHT( res, $right_pos ) ) WHERE id IN ( $id_all )");

[113] Fix | Delete

[114] Fix | Delete

// Add blacklist

[115] Fix | Delete

if (Crawler::STATUS_BLACKLIST === $bit || Crawler::STATUS_NOCACHE === $bit) {

[116] Fix | Delete

$q = "SELECT a.id, a.url FROM `$this->_tb_blacklist` a LEFT JOIN `$this->_tb` b ON b.url=a.url WHERE b.id IN ( $id_all )";

[117] Fix | Delete

// phpcs:ignore WordPress.DB.DirectDatabaseQuery.DirectQuery, WordPress.DB.PreparedSQL.NotPrepared

[118] Fix | Delete

$existing = $wpdb->get_results($q, ARRAY_A);

[119] Fix | Delete

// Update current crawler status tag in existing blacklist

[120] Fix | Delete

if ($existing) {

[121] Fix | Delete

// phpcs:ignore WordPress.DB.DirectDatabaseQuery.DirectQuery, WordPress.DB.PreparedSQL.InterpolatedNotPrepared, WordPress.DB.PreparedSQL.NotPrepared

[122] Fix | Delete

$count = $wpdb->query("UPDATE `$this->_tb_blacklist` SET res = CONCAT( LEFT( res, $curr_crawler ), '$bit', RIGHT( res, $right_pos ) ) WHERE id IN ( " . implode(',', array_column($existing, 'id')) . ' )');

[123] Fix | Delete

self::debug('Update blacklist [count] ' . $count);

[124] Fix | Delete

}

[125] Fix | Delete

[126] Fix | Delete

// Append new blacklist

[127] Fix | Delete

if (count($ids) > count($existing)) {

[128] Fix | Delete

$new_urls = array_diff(array_column($ids, 'url'), array_column($existing, 'url'));

[129] Fix | Delete

[130] Fix | Delete

self::debug('Insert into blacklist [count] ' . count($new_urls));

[131] Fix | Delete

[132] Fix | Delete

$q = "INSERT INTO `$this->_tb_blacklist` ( url, res, reason ) VALUES " . implode(',', array_fill(0, count($new_urls), '( %s, %s, %s )'));

[133] Fix | Delete

$data = [];

[134] Fix | Delete

$res = array_fill(0, $total_crawler, '-');

[135] Fix | Delete

$res[$curr_crawler] = $bit;

[136] Fix | Delete

$res = implode('', $res);

[137] Fix | Delete

$default_reason = $total_crawler > 1 ? str_repeat(',', $total_crawler - 1) : ''; // Pre-populate default reason value first, update later

[138] Fix | Delete

foreach ($new_urls as $url) {

[139] Fix | Delete

$data[] = $url;

[140] Fix | Delete

$data[] = $res;

[141] Fix | Delete

$data[] = $default_reason;

[142] Fix | Delete

}

[143] Fix | Delete

// phpcs:ignore WordPress.DB.DirectDatabaseQuery.DirectQuery, WordPress.DB.PreparedSQL.NotPrepared

[144] Fix | Delete

$wpdb->query($wpdb->prepare($q, $data));

[145] Fix | Delete

}

[146] Fix | Delete

}

[147] Fix | Delete

[148] Fix | Delete

// Update sitemap reason w/ HTTP code.

[149] Fix | Delete

$reason_array = [];

[150] Fix | Delete

foreach ( $ids as $row_id => $row ) {

[151] Fix | Delete

$code = (int) $row['code'];

[152] Fix | Delete

if ( empty( $reason_array[ $code ] ) ) {

[153] Fix | Delete

$reason_array[ $code ] = [];

[154] Fix | Delete

}

[155] Fix | Delete

$reason_array[ $code ][] = (int) $row_id;

[156] Fix | Delete

}

[157] Fix | Delete

[158] Fix | Delete

foreach ($reason_array as $code => $v2) {

[159] Fix | Delete

// Complement comma

[160] Fix | Delete

if ($curr_crawler) {

[161] Fix | Delete

$code = ',' . $code;

[162] Fix | Delete

}

[163] Fix | Delete

if ($curr_crawler < $total_crawler_pos) {

[164] Fix | Delete

$code .= ',';

[165] Fix | Delete

}

[166] Fix | Delete

[167] Fix | Delete

// phpcs:ignore WordPress.DB

[168] Fix | Delete

$count = $wpdb->query( "UPDATE `$this->_tb` SET reason=CONCAT(SUBSTRING_INDEX(reason, ',', $curr_crawler), '$code', SUBSTRING_INDEX(reason, ',', -$right_pos)) WHERE id IN (" . implode(',', $v2) . ')' );

[169] Fix | Delete

[170] Fix | Delete

self::debug("Update map reason [code] $code [pos] left $curr_crawler right -$right_pos [count] $count");

[171] Fix | Delete

[172] Fix | Delete

// Update blacklist reason

[173] Fix | Delete

if (Crawler::STATUS_BLACKLIST === $bit || Crawler::STATUS_NOCACHE === $bit) {

[174] Fix | Delete

// phpcs:ignore WordPress.DB

[175] Fix | Delete

$count = $wpdb->query( "UPDATE `$this->_tb_blacklist` a LEFT JOIN `$this->_tb` b ON b.url = a.url SET a.reason=CONCAT(SUBSTRING_INDEX(a.reason, ',', $curr_crawler), '$code', SUBSTRING_INDEX(a.reason, ',', -$right_pos)) WHERE b.id IN (" . implode(',', $v2) . ')' );

[176] Fix | Delete

[177] Fix | Delete

self::debug("Update blacklist [code] $code [pos] left $curr_crawler right -$right_pos [count] $count");

[178] Fix | Delete

}

[179] Fix | Delete

}

[180] Fix | Delete

[181] Fix | Delete

// Reset list.

[182] Fix | Delete

$items[ $bit ] = [];

[183] Fix | Delete

}

[184] Fix | Delete

[185] Fix | Delete

return $items;

[186] Fix | Delete

}

[187] Fix | Delete

[188] Fix | Delete

/**

[189] Fix | Delete

* Add one record to blacklist.

[190] Fix | Delete

* NOTE: $id is sitemap table ID.

[191] Fix | Delete

*

[192] Fix | Delete

* @since 3.0

[193] Fix | Delete

* @access public

[194] Fix | Delete

*

[195] Fix | Delete

* @param int $id Sitemap row ID.

[196] Fix | Delete

* @return void

[197] Fix | Delete

*/

[198] Fix | Delete

public function blacklist_add( $id ) {

[199] Fix | Delete

global $wpdb;

[200] Fix | Delete

[201] Fix | Delete

$id = (int) $id;

[202] Fix | Delete

[203] Fix | Delete

// Build res&reason.

[204] Fix | Delete

$total_crawler = count( Crawler::cls()->list_crawlers() );

[205] Fix | Delete

$res = str_repeat(Crawler::STATUS_BLACKLIST, $total_crawler);

[206] Fix | Delete

$reason = implode(',', array_fill(0, $total_crawler, 'Man'));

[207] Fix | Delete

[208] Fix | Delete

// phpcs:ignore WordPress.DB.DirectDatabaseQuery.DirectQuery, WordPress.DB.PreparedSQL.InterpolatedNotPrepared

[209] Fix | Delete

$row = $wpdb->get_row("SELECT a.url, b.id FROM `$this->_tb` a LEFT JOIN `$this->_tb_blacklist` b ON b.url = a.url WHERE a.id = '$id'", ARRAY_A);

[210] Fix | Delete

if (!$row) {

[211] Fix | Delete

self::debug('blacklist failed to add [id] ' . $id);

[212] Fix | Delete

return;

[213] Fix | Delete

}

[214] Fix | Delete

[215] Fix | Delete

self::debug('Add to blacklist [url] ' . $row['url']);

[216] Fix | Delete

[217] Fix | Delete

$q = "UPDATE `$this->_tb` SET res = %s, reason = %s WHERE id = %d";

[218] Fix | Delete

// phpcs:ignore WordPress.DB.DirectDatabaseQuery.DirectQuery, WordPress.DB.PreparedSQL.NotPrepared

[219] Fix | Delete

$wpdb->query($wpdb->prepare($q, [ $res, $reason, $id ]));

[220] Fix | Delete

[221] Fix | Delete

if ($row['id']) {

[222] Fix | Delete

$q = "UPDATE `$this->_tb_blacklist` SET res = %s, reason = %s WHERE id = %d";

[223] Fix | Delete

// phpcs:ignore WordPress.DB.DirectDatabaseQuery.DirectQuery, WordPress.DB.PreparedSQL.NotPrepared

[224] Fix | Delete

$wpdb->query($wpdb->prepare($q, [ $res, $reason, $row['id'] ]));

[225] Fix | Delete

} else {

[226] Fix | Delete

$q = "INSERT INTO `$this->_tb_blacklist` (url, res, reason) VALUES (%s, %s, %s)";

[227] Fix | Delete

// phpcs:ignore WordPress.DB.DirectDatabaseQuery.DirectQuery, WordPress.DB.PreparedSQL.NotPrepared

[228] Fix | Delete

$wpdb->query($wpdb->prepare($q, [ $row['url'], $res, $reason ]));

[229] Fix | Delete

}

[230] Fix | Delete

}

[231] Fix | Delete

[232] Fix | Delete

/**

[233] Fix | Delete

* Delete one record from blacklist.

[234] Fix | Delete

*

[235] Fix | Delete

* @since 3.0

[236] Fix | Delete

* @access public

[237] Fix | Delete

*

[238] Fix | Delete

* @param int $id Blacklist row ID.

[239] Fix | Delete

* @return void

[240] Fix | Delete

*/

[241] Fix | Delete

public function blacklist_del( $id ) {

[242] Fix | Delete

global $wpdb;

[243] Fix | Delete

if ( ! $this->__data->tb_exist( 'crawler_blacklist' ) ) {

[244] Fix | Delete

return;

[245] Fix | Delete

}

[246] Fix | Delete

[247] Fix | Delete

$id = (int) $id;

[248] Fix | Delete

self::debug('blacklist delete [id] ' . $id);

[249] Fix | Delete

[250] Fix | Delete

$sql = sprintf(

[251] Fix | Delete

"UPDATE `%s` SET res=REPLACE(REPLACE(res, '%s', '-'), '%s', '-') WHERE url=(SELECT url FROM `%s` WHERE id=%d)",

[252] Fix | Delete

$this->_tb,

[253] Fix | Delete

Crawler::STATUS_NOCACHE,

[254] Fix | Delete

Crawler::STATUS_BLACKLIST,

[255] Fix | Delete

$this->_tb_blacklist,

[256] Fix | Delete

$id

[257] Fix | Delete

);

[258] Fix | Delete

// phpcs:ignore WordPress.DB.DirectDatabaseQuery.DirectQuery, WordPress.DB.PreparedSQL.NotPrepared

[259] Fix | Delete

$wpdb->query($sql);

[260] Fix | Delete

// phpcs:ignore WordPress.DB.DirectDatabaseQuery.DirectQuery, WordPress.DB.PreparedSQL.InterpolatedNotPrepared

[261] Fix | Delete

$wpdb->query("DELETE FROM `$this->_tb_blacklist` WHERE id='$id'");

[262] Fix | Delete

}

[263] Fix | Delete

[264] Fix | Delete

/**

[265] Fix | Delete

* Empty blacklist.

[266] Fix | Delete

*

[267] Fix | Delete

* @since 3.0

[268] Fix | Delete

* @access public

[269] Fix | Delete

* @return void

[270] Fix | Delete

*/

[271] Fix | Delete

public function blacklist_empty() {

[272] Fix | Delete

global $wpdb;

[273] Fix | Delete

[274] Fix | Delete

if ( ! $this->__data->tb_exist( 'crawler_blacklist' ) ) {

[275] Fix | Delete

return;

[276] Fix | Delete

}

[277] Fix | Delete

[278] Fix | Delete

self::debug('Truncate blacklist');

[279] Fix | Delete

$sql = sprintf("UPDATE `%s` SET res=REPLACE(REPLACE(res, '%s', '-'), '%s', '-')", $this->_tb, Crawler::STATUS_NOCACHE, Crawler::STATUS_BLACKLIST);

[280] Fix | Delete

// phpcs:ignore WordPress.DB.DirectDatabaseQuery.DirectQuery, WordPress.DB.PreparedSQL.NotPrepared

[281] Fix | Delete

$wpdb->query($sql);

[282] Fix | Delete

// phpcs:ignore WordPress.DB.DirectDatabaseQuery.DirectQuery, WordPress.DB.PreparedSQL.InterpolatedNotPrepared

[283] Fix | Delete

$wpdb->query("TRUNCATE `$this->_tb_blacklist`");

[284] Fix | Delete

}

[285] Fix | Delete

[286] Fix | Delete

/**

[287] Fix | Delete

* List blacklist.

[288] Fix | Delete

*

[289] Fix | Delete

* @since 3.0

[290] Fix | Delete

* @access public

[291] Fix | Delete

*

[292] Fix | Delete

* @param int|false $limit Number of rows to fetch, or false for all.

[293] Fix | Delete

* @param int|false $offset Offset for pagination, or false to auto-calc.

[294] Fix | Delete

* @return array<int,array<string,mixed>>

[295] Fix | Delete

*/

[296] Fix | Delete

public function list_blacklist( $limit = false, $offset = false ) {

[297] Fix | Delete

global $wpdb;

[298] Fix | Delete

[299] Fix | Delete

if ( ! $this->__data->tb_exist( 'crawler_blacklist' ) ) {

[300] Fix | Delete

return [];

[301] Fix | Delete

}

[302] Fix | Delete

[303] Fix | Delete

$q = "SELECT * FROM `$this->_tb_blacklist` ORDER BY id DESC";

[304] Fix | Delete

[305] Fix | Delete

if ( false !== $limit ) {

[306] Fix | Delete

if ( false === $offset ) {

[307] Fix | Delete

$total = $this->count_blacklist();

[308] Fix | Delete

$offset = Utility::pagination($total, $limit, true);

[309] Fix | Delete

}

[310] Fix | Delete

$q .= ' LIMIT %d, %d';

[311] Fix | Delete

// phpcs:ignore WordPress.DB.PreparedSQL.NotPrepared

[312] Fix | Delete

$q = $wpdb->prepare($q, $offset, $limit);

[313] Fix | Delete

}

[314] Fix | Delete

// phpcs:ignore WordPress.DB.DirectDatabaseQuery.DirectQuery, WordPress.DB.PreparedSQL.NotPrepared

[315] Fix | Delete

return $wpdb->get_results($q, ARRAY_A);

[316] Fix | Delete

}

[317] Fix | Delete

[318] Fix | Delete

/**

[319] Fix | Delete

* Count blacklist.

[320] Fix | Delete

*

[321] Fix | Delete

* @return int|false

[322] Fix | Delete

*/

[323] Fix | Delete

public function count_blacklist() {

[324] Fix | Delete

global $wpdb;

[325] Fix | Delete

[326] Fix | Delete

if ( ! $this->__data->tb_exist( 'crawler_blacklist' ) ) {

[327] Fix | Delete

return false;

[328] Fix | Delete

}

[329] Fix | Delete

[330] Fix | Delete

$q = "SELECT COUNT(*) FROM `$this->_tb_blacklist`";

[331] Fix | Delete

// phpcs:ignore WordPress.DB.DirectDatabaseQuery.DirectQuery, WordPress.DB.PreparedSQL.NotPrepared

[332] Fix | Delete

return $wpdb->get_var($q);

[333] Fix | Delete

}

[334] Fix | Delete

[335] Fix | Delete

/**

[336] Fix | Delete

* Empty sitemap.

[337] Fix | Delete

*

[338] Fix | Delete

* @since 3.0

[339] Fix | Delete

* @access public

[340] Fix | Delete

* @return void

[341] Fix | Delete

*/

[342] Fix | Delete

public function empty_map() {

[343] Fix | Delete

Data::cls()->tb_del( 'crawler' );

[344] Fix | Delete

[345] Fix | Delete

$msg = __( 'Sitemap cleaned successfully', 'litespeed-cache' );

[346] Fix | Delete

Admin_Display::success( $msg );

[347] Fix | Delete

}

[348] Fix | Delete

[349] Fix | Delete

/**

[350] Fix | Delete

* List generated sitemap.

[351] Fix | Delete

*

[352] Fix | Delete

* @since 3.0

[353] Fix | Delete

* @access public

[354] Fix | Delete

*

[355] Fix | Delete

* @param int $limit Number of rows per page.

[356] Fix | Delete

* @param int|bool $offset Offset for pagination, or false to auto-calc.

[357] Fix | Delete

* @return array<int,array<string,mixed>>

[358] Fix | Delete

*/

[359] Fix | Delete

public function list_map( $limit, $offset = false ) {

[360] Fix | Delete

global $wpdb;

[361] Fix | Delete

[362] Fix | Delete

if ( ! $this->__data->tb_exist( 'crawler' ) ) {

[363] Fix | Delete

return [];

[364] Fix | Delete

}

[365] Fix | Delete

[366] Fix | Delete

if ( false === $offset ) {

[367] Fix | Delete

$total = $this->count_map();

[368] Fix | Delete

$offset = Utility::pagination($total, $limit, true);

[369] Fix | Delete

}

[370] Fix | Delete

[371] Fix | Delete

$type = Router::verify_type();

[372] Fix | Delete

[373] Fix | Delete

$req_uri_like = '';

[374] Fix | Delete

// phpcs:ignore WordPress.Security.NonceVerification.Missing

[375] Fix | Delete

if ( ! empty( $_POST['kw'] ) ) {

[376] Fix | Delete

// phpcs:ignore WordPress.Security.NonceVerification.Missing

[377] Fix | Delete

$kw = sanitize_text_field( wp_unslash( $_POST['kw'] ) );

[378] Fix | Delete

$q = "SELECT * FROM `$this->_tb` WHERE url LIKE %s";

[379] Fix | Delete

if ( 'hit' === $type ) {

[380] Fix | Delete

$q .= " AND res LIKE '%" . Crawler::STATUS_HIT . "%'";

[381] Fix | Delete

}

[382] Fix | Delete

if ( 'miss' === $type ) {

[383] Fix | Delete

$q .= " AND res LIKE '%" . Crawler::STATUS_MISS . "%'";

[384] Fix | Delete

}

[385] Fix | Delete

if ( 'blacklisted' === $type ) {

[386] Fix | Delete

$q .= " AND res LIKE '%" . Crawler::STATUS_BLACKLIST . "%'";

[387] Fix | Delete

}

[388] Fix | Delete

$q .= ' ORDER BY id LIMIT %d, %d';

[389] Fix | Delete

$req_uri_like = '%' . $wpdb->esc_like( $kw ) . '%';

[390] Fix | Delete

return $wpdb->get_results( $wpdb->prepare( $q, $req_uri_like, $offset, $limit ), ARRAY_A ); // phpcs:ignore WordPress.DB.DirectDatabaseQuery.DirectQuery, WordPress.DB.PreparedSQL.NotPrepared

[391] Fix | Delete

}

[392] Fix | Delete

[393] Fix | Delete

$q = "SELECT * FROM `$this->_tb`";

[394] Fix | Delete

if ( 'hit' === $type ) {

[395] Fix | Delete

$q .= " WHERE res LIKE '%" . Crawler::STATUS_HIT . "%'";

[396] Fix | Delete

}

[397] Fix | Delete

if ( 'miss' === $type ) {

[398] Fix | Delete

$q .= " WHERE res LIKE '%" . Crawler::STATUS_MISS . "%'";

[399] Fix | Delete

}

[400] Fix | Delete

if ( 'blacklisted' === $type ) {

[401] Fix | Delete

$q .= " WHERE res LIKE '%" . Crawler::STATUS_BLACKLIST . "%'";

[402] Fix | Delete

}

[403] Fix | Delete

$q .= ' ORDER BY id LIMIT %d, %d';

[404] Fix | Delete

[405] Fix | Delete

return $wpdb->get_results( $wpdb->prepare( $q, $offset, $limit ), ARRAY_A ); // phpcs:ignore WordPress.DB.DirectDatabaseQuery.DirectQuery, WordPress.DB.PreparedSQL.NotPrepared

[406] Fix | Delete

}

[407] Fix | Delete

[408] Fix | Delete

/**

[409] Fix | Delete

* Count sitemap.

[410] Fix | Delete

*

[411] Fix | Delete

* @return int|false

[412] Fix | Delete

*/

[413] Fix | Delete

public function count_map() {

[414] Fix | Delete

global $wpdb;

[415] Fix | Delete

[416] Fix | Delete

if ( ! $this->__data->tb_exist( 'crawler' ) ) {

[417] Fix | Delete

return false;

[418] Fix | Delete

}

[419] Fix | Delete

[420] Fix | Delete

$q = "SELECT COUNT(*) FROM `$this->_tb`";

[421] Fix | Delete

[422] Fix | Delete

$type = Router::verify_type();

[423] Fix | Delete

if ( 'hit' === $type ) {

[424] Fix | Delete

$q .= " WHERE res LIKE '%" . Crawler::STATUS_HIT . "%'";

[425] Fix | Delete

}

[426] Fix | Delete

if ( 'miss' === $type ) {

[427] Fix | Delete

$q .= " WHERE res LIKE '%" . Crawler::STATUS_MISS . "%'";

[428] Fix | Delete

}

[429] Fix | Delete

if ( 'blacklisted' === $type ) {

[430] Fix | Delete

$q .= " WHERE res LIKE '%" . Crawler::STATUS_BLACKLIST . "%'";

[431] Fix | Delete

}

[432] Fix | Delete

[433] Fix | Delete

return $wpdb->get_var( $q ); // phpcs:ignore WordPress.DB.DirectDatabaseQuery.DirectQuery, WordPress.DB.PreparedSQL.NotPrepared

[434] Fix | Delete

}

[435] Fix | Delete

[436] Fix | Delete

/**

[437] Fix | Delete

* Generate sitemap.

[438] Fix | Delete

*

[439] Fix | Delete

* @since 1.1.0

[440] Fix | Delete

* @access public

[441] Fix | Delete

*

[442] Fix | Delete

* @param bool $manual Whether triggered manually from UI.

[443] Fix | Delete

* @return void

[444] Fix | Delete

*/

[445] Fix | Delete

public function gen( $manual = false ) {

[446] Fix | Delete

$count = $this->_gen();

[447] Fix | Delete

[448] Fix | Delete

if ( ! $count ) {

[449] Fix | Delete

Admin_Display::error( __( 'No valid sitemap parsed for crawler.', 'litespeed-cache' ) );

[450] Fix | Delete

return;

[451] Fix | Delete

}

[452] Fix | Delete

[453] Fix | Delete

if ( ! wp_doing_cron() && $manual ) {

[454] Fix | Delete

$msg = sprintf( __( 'Sitemap created successfully: %d items', 'litespeed-cache' ), $count );

[455] Fix | Delete

Admin_Display::success( $msg );

[456] Fix | Delete

}

[457] Fix | Delete

}

[458] Fix | Delete

[459] Fix | Delete

/**

[460] Fix | Delete

* Generate the sitemap.

[461] Fix | Delete

*

[462] Fix | Delete

* @since 1.1.0

[463] Fix | Delete

* @access private

[464] Fix | Delete

* @return int|false Number of URLs generated or false on failure.

[465] Fix | Delete

*/

[466] Fix | Delete

private function _gen() {

[467] Fix | Delete

global $wpdb;

[468] Fix | Delete

[469] Fix | Delete

if ( ! $this->__data->tb_exist( 'crawler' ) ) {

[470] Fix | Delete

$this->__data->tb_create( 'crawler' );

[471] Fix | Delete

}

[472] Fix | Delete

[473] Fix | Delete

if ( ! $this->__data->tb_exist( 'crawler_blacklist' ) ) {

[474] Fix | Delete

$this->__data->tb_create( 'crawler_blacklist' );

[475] Fix | Delete

}

[476] Fix | Delete

[477] Fix | Delete

// Use custom sitemap.

[478] Fix | Delete

$sitemap = $this->conf( Base::O_CRAWLER_SITEMAP );

[479] Fix | Delete

if ( ! $sitemap ) {

[480] Fix | Delete

return false;

[481] Fix | Delete

}

[482] Fix | Delete

[483] Fix | Delete

$offset = strlen( $this->_site_url );

[484] Fix | Delete

$sitemap = Utility::sanitize_lines( $sitemap );

[485] Fix | Delete

[486] Fix | Delete

try {

[487] Fix | Delete

foreach ( $sitemap as $this_map ) {

[488] Fix | Delete

$this->_parse( $this_map );

[489] Fix | Delete

}

[490] Fix | Delete

} catch ( \Exception $e ) {

[491] Fix | Delete

self::debug( '❌ failed to parse custom sitemap: ' . $e->getMessage() );

[492] Fix | Delete

}

[493] Fix | Delete

[494] Fix | Delete

if ( is_array( $this->_urls ) && ! empty( $this->_urls ) ) {

[495] Fix | Delete

if ( defined( 'LITESPEED_CRAWLER_DROP_DOMAIN' ) && constant( 'LITESPEED_CRAWLER_DROP_DOMAIN' ) ) {

[496] Fix | Delete

foreach ( $this->_urls as $k => $v ) {

[497] Fix | Delete

if ( 0 !== stripos( $v, $this->_site_url ) ) {

[498] Fix | Delete

unset( $this->_urls[ $k ] );

[499] Fix | Delete