YGGverse
/
YGGo
mirror of https://github.com/YGGverse/YGGo.git
1
0
Fork 0
Code Issues Projects Releases Wiki Activity
YGGo! Distributed Web Search Engine
You can not select more than 25 topics Topics must start with a letter or number, can include dashes ('-') and can be up to 35 characters long.
507 Commits
2 Branches
0 Tags
4.3 MiB
 
Tree: d46c4921c5
Branches Tags
${ item.name }
Create tag ${ searchTerm }
Create branch ${ searchTerm }
from 'd46c4921c5'
${ noResults }
YGGo/library/mysql.php

688 lines
22 KiB
Raw Blame History

<?php
class MySQL {
private PDO $_db;
public function __construct(string $host, int $port, string $database, string $username, string $password) {
$this->_db = new PDO('mysql:dbname=' . $database . ';host=' . $host . ';port=' . $port . ';charset=utf8', $username, $password, [PDO::MYSQL_ATTR_INIT_COMMAND => 'SET NAMES utf8']);
$this->_db->setAttribute(PDO::ATTR_ERRMODE, PDO::ERRMODE_EXCEPTION);
$this->_db->setAttribute(PDO::ATTR_DEFAULT_FETCH_MODE, PDO::FETCH_OBJ);
$this->_db->setAttribute(PDO::ATTR_TIMEOUT, 600);
}
// System
public function beginTransaction() {
$this->_db->beginTransaction();
}
public function commit() {
$this->_db->commit();
}
public function rollBack() {
$this->_db->rollBack();
}
// Host
public function getAPIHosts(string $apiHostFields) {
$query = $this->_db->prepare('SELECT ' . $apiHostFields . ' FROM `host`');
$query->execute();
return $query->fetchAll();
}
public function getHosts() {
$query = $this->_db->query('SELECT * FROM `host`');
return $query->fetchAll();
}
public function getHost(int $hostId) {
$query = $this->_db->prepare("SELECT *,
IF (`port` IS NOT NULL,
CONCAT(`scheme`, '://', `name`, ':', `port`),
CONCAT(`scheme`, '://', `name`)
) AS `url`
FROM `host` WHERE `hostId` = ? LIMIT 1");
$query->execute([$hostId]);
return $query->fetch();
}
public function findHostByCRC32URL(int $crc32url) {
$query = $this->_db->prepare('SELECT * FROM `host` WHERE `crc32url` = ? LIMIT 1');
$query->execute([$crc32url]);
return $query->fetch();
}
public function getTotalHosts() {
$query = $this->_db->prepare('SELECT COUNT(*) AS `total` FROM `host`');
$query->execute();
return $query->fetch()->total;
}
public function addHost(string $scheme, string $name, mixed $port, int $crc32url, int $timeAdded) {
$query = $this->_db->prepare('INSERT INTO `host` (`scheme`,
`name`,
`port`,
`crc32url`,
`timeAdded`) VALUES (?, ?, ?, ?, ?)');
$query->execute([$scheme, $name, $port, $crc32url, $timeAdded]);
return $this->_db->lastInsertId();
}
// Host settings
public function findHostSettingValue(int $hostId, string $key) {
$query = $this->_db->prepare('SELECT `value` FROM `hostSetting` WHERE `hostId` = ? AND `key` = ? LIMIT 1');
$query->execute([$hostId, $key]);
return $query->rowCount() ? json_decode($query->fetch()->value) : false;
}
public function findHostSetting(int $hostId, string $key) {
$query = $this->_db->prepare('SELECT * FROM `hostSetting` WHERE `hostId` = ? AND `key` = ? LIMIT 1');
$query->execute([$hostId, $key]);
return $query->fetch();
}
public function addHostSetting(int $hostId, string $key, mixed $value, int $timeAdded) {
$query = $this->_db->prepare('INSERT INTO `hostSetting` (`hostId`, `key`, `value`, `timeAdded`) VALUES (?, ?, ?, ?)');
$value = json_encode($value);
$query->execute(
[
$hostId,
$key,
$value,
$timeAdded
]
);
return $query->rowCount();
}
public function updateHostSetting(int $hostSettingId, mixed $value, int $timeUpdated) {
$query = $this->_db->query('UPDATE `hostSetting` SET `value` = ?,
`timeUpdated` = ?
WHERE `hostSettingId` = ?
LIMIT 1');
$value = json_encode($value);
$query->execute(
[
$value,
$timeUpdated,
$hostSettingId
]
);
return $query->rowCount();
}
public function deleteHostSetting(int $hostSettingId) {
$query = $this->_db->query('DELETE FROM `hostSetting` WHERE `hostSettingId` = ?');
$query->execute([$hostSettingId]);
return $query->rowCount();
}
// Host pages
public function getTotalHostPages(int $hostId) {
$query = $this->_db->prepare('SELECT COUNT(*) AS `total` FROM `hostPage` WHERE `hostId` = ?');
$query->execute([$hostId]);
return $query->fetch()->total;
}
public function getHostPage(int $hostPageId) {
$query = $this->_db->prepare('SELECT * FROM `hostPage` WHERE `hostPageId` = ? LIMIT 1');
$query->execute([$hostPageId]);
return $query->fetch();
}
public function findHostPageByCRC32URI(int $hostId, int $crc32uri) {
$query = $this->_db->prepare('SELECT * FROM `hostPage` WHERE `hostId` = ? AND `crc32uri` = ? LIMIT 1');
$query->execute([$hostId, $crc32uri]);
return $query->fetch();
}
public function getHostPages(int $hostId) {
$query = $this->_db->prepare('SELECT * FROM `hostPage` WHERE `hostId` = ?');
$query->execute([$hostId]);
return $query->fetchAll();
}
public function getTopHostPages(int $limit = 100) {
// Get ID (to prevent memory over usage)
$query = $this->_db->query("SELECT `hostPageId` FROM `hostPage`
WHERE `httpCode` = 200
AND `rank` > 0
AND `timeBanned` IS NULL
AND `mime` IS NOT NULL
ORDER BY `rank` DESC
LIMIT " . (int) $limit);
// Get required page details
foreach ($query->fetchAll() as $top) {
$query = $this->_db->prepare("SELECT `hostPage`.`hostId`,
`hostPage`.`hostPageId`,
`hostPage`.`uri`,
`hostPage`.`rank`,
`host`.`scheme`,
`host`.`name`,
`host`.`port`,
IF (`host`.`port` IS NOT NULL,
CONCAT(`host`.`scheme`, '://', `host`.`name`, ':', `host`.`port`),
CONCAT(`host`.`scheme`, '://', `host`.`name`)
) AS `hostURL`,
IF (`host`.`port` IS NOT NULL,
CONCAT(`host`.`scheme`, '://', `host`.`name`, ':', `host`.`port`, `hostPage`.`uri`),
CONCAT(`host`.`scheme`, '://', `host`.`name`, `hostPage`.`uri`)
) AS `hostPageURL`
FROM `hostPage`
JOIN `host` ON (`hostPage`.`hostId` = `host`.`hostId`)
WHERE `hostPage`.`hostPageId` = ?
LIMIT 1");
$query->execute([$top->hostPageId]);
if ($query->rowCount()) {
$result[] = $query->fetch();
}
}
return $result;
}
public function getHostPagesByIndexed() {
$query = $this->_db->query('SELECT * FROM `hostPage` WHERE `timeUpdated` IS NOT NULL AND `timeBanned` IS NULL');
return $query->fetchAll();
}
public function getHostPagesByLimit(int $hostId, int $limit) {
$query = $this->_db->prepare('SELECT * FROM `hostPage` WHERE `hostId` = ? ORDER BY `hostPageId` DESC LIMIT ' . (int) $limit);
$query->execute([$hostId]);
return $query->fetchAll();
}
public function getLastPageDescription(int $hostPageId) {
$query = $this->_db->prepare('SELECT * FROM `hostPageDescription` WHERE `hostPageId` = ? ORDER BY `timeAdded` DESC LIMIT 1');
$query->execute([$hostPageId]);
return $query->fetch();
}
public function addHostPage(int $hostId,
int $crc32uri,
string $uri,
int $timeAdded,
mixed $timeUpdated = null,
mixed $timeBanned = null,
mixed $httpCode = null,
mixed $mime = null) {
$query = $this->_db->prepare('INSERT INTO `hostPage` (`hostId`,
`crc32uri`,
`uri`,
`timeAdded`,
`timeUpdated`,
`timeBanned`,
`httpCode`,
`mime`) VALUES (?, ?, ?, ?, ?, ?, ?, ?)');
$query->execute([$hostId, $crc32uri, $uri, $timeAdded, $timeUpdated, $timeBanned, $httpCode, $mime]);
return $this->_db->lastInsertId();
}
public function updateHostPageTimeBanned(int $hostPageId, int $timeBanned) {
$query = $this->_db->prepare('UPDATE `hostPage` SET `timeBanned` = ? WHERE `hostPageId` = ? LIMIT 1');
$query->execute([$timeBanned, $hostPageId]);
return $query->rowCount();
}
public function updateHostPageMime(int $hostPageId, string $mime) {
$query = $this->_db->prepare('UPDATE `hostPage` SET `mime` = ? WHERE `hostPageId` = ? LIMIT 1');
$query->execute([$mime, $hostPageId]);
return $query->rowCount();
}
public function updateHostPageRank(int $hostPageId, int $rank) {
$query = $this->_db->prepare('UPDATE `hostPage` SET `rank` = ? WHERE `hostPageId` = ? LIMIT 1');
$query->execute([$rank, $hostPageId]);
return $query->rowCount();
}
public function deleteHostPage(int $hostPageId) {
$query = $this->_db->prepare('DELETE FROM `hostPage` WHERE `hostPageId` = ? LIMIT 1');
$query->execute([$hostPageId]);
return $query->rowCount();
}
public function deleteHostPageDescriptions(int $hostPageId) {
$query = $this->_db->prepare('DELETE FROM `hostPageDescription` WHERE `hostPageId` = ?');
$query->execute([$hostPageId]);
return $query->rowCount();
}
public function addHostPageDescription(int $hostPageId,
mixed $title,
mixed $description,
mixed $keywords,
mixed $data,
int $timeAdded) {
$query = $this->_db->prepare('INSERT INTO `hostPageDescription` ( `hostPageId`,
`title`,
`description`,
`keywords`,
`data`,
`timeAdded`
) VALUES (?, ?, ?, ?, ?, ?)');
$query->execute([
$hostPageId,
$title,
$description,
$keywords,
$data,
$timeAdded,
]);
return $query->rowCount();
}
public function setHostPageToHostPage(int $hostPageIdSource, int $hostPageIdTarget) {
$query = $this->_db->prepare('INSERT IGNORE `hostPageToHostPage` (`hostPageIdSource`, `hostPageIdTarget`) VALUES (?, ?)');
$query->execute([$hostPageIdSource, $hostPageIdTarget]);
}
public function deleteHostPageToHostPage(int $hostPageId) {
$query = $this->_db->prepare('DELETE FROM `hostPageToHostPage` WHERE `hostPageIdSource` = ? OR `hostPageIdTarget` = ?');
$query->execute([$hostPageId, $hostPageId]);
return $query->rowCount();
}
public function getTotalHostPagesToHostPageByHostPageIdTarget(int $hostPageIdTarget) {
$query = $this->_db->prepare('SELECT COUNT(*) AS `total` FROM `hostPageToHostPage` WHERE `hostPageIdTarget` = ?');
$query->execute([$hostPageIdTarget]);
return $query->fetch()->total;
}
public function getHostPagesToHostPageByHostPageIdTarget(int $hostPageIdTarget, int $limit = 1000) {
$query = $this->_db->prepare('SELECT * FROM `hostPageToHostPage` WHERE `hostPageIdTarget` = ? LIMIT ' . (int) $limit);
$query->execute([$hostPageIdTarget]);
return $query->fetchAll();
}
public function getHostPageToHostPage(int $hostPageIdSource, int $hostPageIdTarget) {
$query = $this->_db->prepare('SELECT * FROM `hostPageToHostPage` WHERE `hostPageIdSource` = ? AND `hostPageIdTarget` = ? LIMIT 1');
$query->execute([$hostPageIdSource, $hostPageIdTarget]);
return $query->fetch();
}
public function addHostPageSnap(int $hostPageId, int $timeAdded) {
$query = $this->_db->prepare('INSERT INTO `hostPageSnap` (`hostPageId`, `timeAdded`) VALUES (?, ?)');
$query->execute([$hostPageId, $timeAdded]);
return $this->_db->lastInsertId();
}
public function deleteHostPageSnap(int $hostPageSnapId) {
$query = $this->_db->prepare('DELETE FROM `hostPageSnap` WHERE `hostPageSnapId` = ? LIMIT 1');
$query->execute([$hostPageSnapId]);
return $query->rowCount();
}
public function getTotalHostPageSnaps(int $hostPageId) {
$query = $this->_db->prepare('SELECT COUNT(*) AS `total` FROM `hostPageSnap` WHERE `hostPageId` = ?');
$query->execute([$hostPageId]);
return $query->fetch()->total;
}
public function getHostPageSnaps(int $hostPageId) {
$query = $this->_db->prepare('SELECT * FROM `hostPageSnap` WHERE `hostPageId` = ? ORDER BY `timeAdded` DESC');
$query->execute([$hostPageId]);
return $query->fetchAll();
}
public function getHostPageSnap(int $hostPageSnapId) {
$query = $this->_db->prepare('SELECT * FROM `hostPageSnap` WHERE `hostPageSnapId` = ? LIMIT 1');
$query->execute([$hostPageSnapId]);
return $query->fetch();
}
public function addHostPageSnapDownload(int $hostPageSnapStorageId, string $crc32ip, int $timeAdded) {
$query = $this->_db->prepare('INSERT INTO `hostPageSnapDownload` (`hostPageSnapStorageId`,
`crc32ip`,
`timeAdded`) VALUES (?, ?, ?)');
$query->execute([$hostPageSnapStorageId, $crc32ip, $timeAdded]);
return $this->_db->lastInsertId();
}
public function addHostPageSnapStorage(int $hostPageSnapId, int $crc32name, int $timeAdded) {
$query = $this->_db->prepare('INSERT INTO `hostPageSnapStorage` (`hostPageSnapId`,
`crc32name`,
`timeAdded`) VALUES (?, ?, ?)');
$query->execute([$hostPageSnapId, $crc32name, $timeAdded]);
return $this->_db->lastInsertId();
}
public function findHostPageSnapStorageByCRC32Name(int $hostPageSnapId, int $crc32name) {
$query = $this->_db->prepare('SELECT * FROM `hostPageSnapStorage` WHERE `hostPageSnapId` = ? AND `crc32name` = ?');
$query->execute([$hostPageSnapId, $crc32name]);
return $query->fetch();
}
public function getHostPageSnapStorages(int $hostPageSnapId) {
$query = $this->_db->prepare('SELECT * FROM `hostPageSnapStorage` WHERE `hostPageSnapId` = ?');
$query->execute([$hostPageSnapId]);
return $query->fetchAll();
}
public function deleteHostPageSnapStorages(int $hostPageSnapId) {
$query = $this->_db->prepare('DELETE FROM `hostPageSnapStorage` WHERE `hostPageSnapId` = ?');
$query->execute([$hostPageSnapId]);
return $query->rowCount();
}
public function deleteHostPageSnapDownloads(int $hostPageSnapStorageId) {
$query = $this->_db->prepare('DELETE FROM `hostPageSnapDownload` WHERE `hostPageSnapStorageId` = ?');
$query->execute([$hostPageSnapStorageId]);
return $query->rowCount();
}
public function addHostPageDom(int $hostPageId, int $timeAdded, string $selector, string $value) {
$query = $this->_db->prepare('INSERT INTO `hostPageDom` SET `hostPageId` = ?, `timeAdded` = ?, `selector` = ?, `value` = ?');
$query->execute([$hostPageId, $timeAdded, $selector, $value]);
}
public function deleteHostPageDoms(int $hostPageId) {
$query = $this->_db->prepare('DELETE FROM `hostPageDom` WHERE `hostPageId` = ?');
$query->execute([$hostPageId]);
return $query->rowCount();
}
public function deleteHostPageDomsByTimeAdded(int $timeOffset) {
$query = $this->_db->prepare('DELETE FROM `hostPageDom` WHERE `timeAdded` < ' . (int) $timeOffset);
$query->execute();
return $query->rowCount();
}
public function truncateHostPageDom() {
$query = $this->_db->query('TRUNCATE `hostPageDom`');
}
// Cleaner tools
public function resetBannedHostPages(int $timeOffset) {
$query = $this->_db->prepare('UPDATE `hostPage` SET `timeBanned` = NULL WHERE `timeBanned` IS NOT NULL AND `timeBanned` < ' . (int) $timeOffset);
$query->execute([$timeOffset]);
return $query->rowCount();
}
public function resetBannedHosts(int $timeOffset) {
$query = $this->_db->prepare('UPDATE `host` SET `timeBanned` = NULL WHERE `timeBanned` IS NOT NULL AND `timeBanned` < ' . (int) $timeOffset);
$query->execute([$timeOffset]);
return $query->rowCount();
}
// Crawler tools
public function getHostPageCrawlQueueTotal(int $timeFrom) {
$query = $this->_db->prepare("SELECT COUNT(*) AS `total` FROM `hostPage`
WHERE (`timeUpdated` IS NULL OR `timeUpdated` < ?) AND `hostPage`.`timeBanned` IS NULL");
$query->execute([$timeFrom]);
return $query->fetch()->total;
}
public function getHostPageCrawlQueue(int $limit, int $timeFrom) {
$result = [];
// Get ID (to prevent memory over usage)
$query = $this->_db->prepare("SELECT `hostPageId` FROM `hostPage`
WHERE (`timeUpdated` IS NULL OR `timeUpdated` < ?) AND `timeBanned` IS NULL
ORDER BY LENGTH(`uri`) ASC, RAND()
LIMIT " . (int) $limit);
$query->execute([$timeFrom]);
// Get required page details
foreach ($query->fetchAll() as $queue) {
$query = $this->_db->prepare("SELECT `hostPage`.`hostId`,
`hostPage`.`hostPageId`,
`hostPage`.`uri`,
`host`.`scheme`,
`host`.`name`,
`host`.`port`,
IF (`host`.`port` IS NOT NULL,
CONCAT(`host`.`scheme`, '://', `host`.`name`, ':', `host`.`port`),
CONCAT(`host`.`scheme`, '://', `host`.`name`)
) AS `hostURL`,
IF (`host`.`port` IS NOT NULL,
CONCAT(`host`.`scheme`, '://', `host`.`name`, ':', `host`.`port`, `hostPage`.`uri`),
CONCAT(`host`.`scheme`, '://', `host`.`name`, `hostPage`.`uri`)
) AS `hostPageURL`
FROM `hostPage`
JOIN `host` ON (`host`.`hostId` = `hostPage`.`hostId`)
WHERE `hostPage`.`hostPageId` = ? LIMIT 1");
$query->execute([$queue->hostPageId]);
if ($query->rowCount()) {
$result[] = $query->fetch();
}
}
return (object) $result;
}
public function updateHostPageCrawlQueue(int $hostPageId, int $timeUpdated, int $httpCode, int $size) {
$query = $this->_db->prepare('UPDATE `hostPage` SET `timeUpdated` = ?, `httpCode` = ?, `size` = ? WHERE `hostPageId` = ? LIMIT 1');
$query->execute([$timeUpdated, $httpCode, $size, $hostPageId]);
return $query->rowCount();
}
public function getHostCrawlQueue(int $limit, int $timeFrom) {
$result = [];
// Get ID (to prevent memory over usage)
$query = $this->_db->prepare("SELECT `hostId`
FROM `host`
WHERE (`timeUpdated` IS NULL OR `timeUpdated` < ?) AND `timeBanned` IS NULL
ORDER BY RAND()
LIMIT " . (int) $limit);
$query->execute([$timeFrom]);
// Get required page details
foreach ($query->fetchAll() as $host) {
$result[] = $this->getHost($host->hostId);
}
return (object) $result;
}
public function updateHostCrawlQueue(int $hostId, int $timeUpdated) {
$query = $this->_db->prepare('UPDATE `host` SET `timeUpdated` = ? WHERE `hostId` = ? LIMIT 1');
$query->execute([$timeUpdated, $hostId]);
return $query->rowCount();
}
public function optimize() {
$this->_db->query('OPTIMIZE TABLE `host`');
$this->_db->query('OPTIMIZE TABLE `hostSetting`');
$this->_db->query('OPTIMIZE TABLE `hostPage`');
$this->_db->query('OPTIMIZE TABLE `hostPageDescription`');
$this->_db->query('OPTIMIZE TABLE `hostPageDom`');
$this->_db->query('OPTIMIZE TABLE `hostPageSnap`');
$this->_db->query('OPTIMIZE TABLE `hostPageSnapStorage`');
$this->_db->query('OPTIMIZE TABLE `hostPageSnapDownload`');
$this->_db->query('OPTIMIZE TABLE `hostPageToHostPage`');
}
}