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.

401 lines
14 KiB

2 years ago
<?php
// Load system dependencies
require_once(__DIR__ . '/../config/app.php');
require_once(__DIR__ . '/../library/curl.php');
require_once(__DIR__ . '/../library/robots.php');
require_once(__DIR__ . '/../library/filter.php');
require_once(__DIR__ . '/../library/parser.php');
require_once(__DIR__ . '/../library/mysql.php');
require_once(__DIR__ . '/../library/sphinxql.php');
// Connect Sphinx search server
$sphinx = new SphinxQL(SPHINX_HOST, SPHINX_PORT);
2 years ago
// Connect database
$db = new MySQL(DB_HOST, DB_PORT, DB_NAME, DB_USERNAME, DB_PASSWORD);
2 years ago
// Filter request data
2 years ago
$t = !empty($_GET['t']) ? Filter::url($_GET['t']) : 'text';
$m = !empty($_GET['m']) ? Filter::url($_GET['m']) : 'default';
2 years ago
$q = !empty($_GET['q']) ? Filter::url($_GET['q']) : '';
$p = !empty($_GET['p']) ? (int) $_GET['p'] : 1;
2 years ago
// Search request
if (!empty($q)) {
$resultsTotal = $sphinx->searchHostPagesTotal(Filter::searchQuery($q, $m), $t);
$results = $sphinx->searchHostPages(Filter::searchQuery($q, $m), $t, $p * WEBSITE_PAGINATION_SEARCH_PAGE_RESULTS_LIMIT - WEBSITE_PAGINATION_SEARCH_PAGE_RESULTS_LIMIT, WEBSITE_PAGINATION_SEARCH_PAGE_RESULTS_LIMIT, $resultsTotal);
} else {
$resultsTotal = 0;
$results = [];
}
// Mime list
$hostPagesMime = $sphinx->searchHostPagesMime(Filter::searchQuery($q, $m));
// Define page basics
$totalPages = $sphinx->getHostPagesTotal();
$placeholder = Filter::plural($totalPages, [sprintf(_('Over %s page or enter the new one...'), $totalPages),
sprintf(_('Over %s pages or enter the new one...'), $totalPages),
sprintf(_('Over %s pages or enter the new one...'), $totalPages),
]);
2 years ago
// Crawl request
if (filter_var($q, FILTER_VALIDATE_URL) && preg_match(CRAWL_URL_REGEXP, $q)) {
$db->beginTransaction();
try {
// Parse host info
if ($hostURL = Parser::hostURL($q)) {
// Host exists
if ($host = $db->getHost(crc32($hostURL->string))) {
$hostStatus = $host->status;
$hostNsfw = $host->nsfw;
$hostPageLimit = $host->crawlPageLimit;
$hostMetaOnly = $host->crawlMetaOnly;
$hostId = $host->hostId;
$hostRobots = $host->robots;
$hostRobotsPostfix = $host->robotsPostfix;
// Register new host
} else {
// Disk quota not reached
if (CRAWL_STOP_DISK_QUOTA_MB_LEFT < disk_free_space('/') / 1000000) {
// Get robots.txt if exists
$curl = new Curl($hostURL->string . '/robots.txt', CRAWL_CURLOPT_USERAGENT);
if (200 == $curl->getCode() && false !== stripos($curl->getContent(), 'user-agent:')) {
$hostRobots = $curl->getContent();
} else {
$hostRobots = null;
}
$hostRobotsPostfix = CRAWL_ROBOTS_POSTFIX_RULES;
$hostStatus = CRAWL_HOST_DEFAULT_STATUS ? 1 : 0;
$hostNsfw = CRAWL_HOST_DEFAULT_NSFW ? 1 : 0;
$hostMetaOnly = CRAWL_HOST_DEFAULT_META_ONLY ? 1 : 0;
$hostPageLimit = CRAWL_HOST_DEFAULT_PAGES_LIMIT;
$hostId = $db->addHost( $hostURL->scheme,
$hostURL->name,
$hostURL->port,
crc32($hostURL->string),
time(),
null,
$hostPageLimit,
(string) $hostMetaOnly,
(string) $hostStatus,
(string) $hostNsfw,
$hostRobots,
$hostRobotsPostfix);
// Add web root host page to make host visible in the crawl queue
$db->addHostPage($hostId, crc32('/'), '/', time());
}
}
// Parse page URI
$hostPageURI = Parser::uri($q);
// Init robots parser
$robots = new Robots((!$hostRobots ? (string) $hostRobots : (string) CRAWL_ROBOTS_DEFAULT_RULES) . PHP_EOL . (string) $hostRobotsPostfix);
// Save page info
if ($hostStatus && // host enabled
$robots->uriAllowed($hostPageURI->string) && // page allowed by robots.txt rules
$hostPageLimit > $db->getTotalHostPages($hostId) && // pages quantity not reached host limit
!$db->getHostPage($hostId, crc32($hostPageURI->string))) { // page not exists
$db->addHostPage($hostId, crc32($hostPageURI->string), $hostPageURI->string, time());
}
}
$db->commit();
} catch(Exception $e){
var_dump($e);
$db->rollBack();
}
2 years ago
}
?>
<!DOCTYPE html>
<html lang="<?php echo _('en-US'); ?>">
<head>
<title><?php echo (empty($q) ? _('Empty request - YGGo!') : ($p > 1 ? sprintf(_('%s - #%s - YGGo!'), htmlentities($q), $p) : sprintf(_('%s - YGGo!'), htmlentities($q)))) ?></title>
2 years ago
<meta charset="utf-8" />
2 years ago
<meta name="description" content="<?php echo _('Javascript-less Open Source Web Search Engine') ?>" />
2 years ago
<meta name="keywords" content="<?php echo _('web, search, engine, crawler, php, pdo, mysql, sphinx, yggdrasil, js-less, open source') ?>" />
2 years ago
<style>
* {
border: 0;
margin: 0;
padding: 0;
font-family: Sans-serif;
}
body {
background-color: #2e3436;
}
header {
background-color: #34393b;
position: fixed;
top: 0;
left: 0;
right: 0;
}
main {
margin-top: 110px;
2 years ago
margin-bottom: 76px;
padding: 0 20px;
2 years ago
}
h1 {
position: fixed;
top: 8px;
left: 24px;
}
h1 > a,
h1 > a:visited,
h1 > a:active,
h1 > a:hover {
2 years ago
color: #fff;
font-weight: normal;
font-size: 24px;
margin: 10px 0;
text-decoration: none;
2 years ago
}
h2 {
display: block;
font-size: 16px;
font-weight: normal;
margin: 4px 0;
color: #fff;
}
form {
display: block;
max-width: 678px;
2 years ago
margin: 0 auto;
text-align: center;
}
input {
width: 100%;
margin: 12px 0;
padding: 10px 0;
2 years ago
border-radius: 32px;
background-color: #000;
color: #fff;
font-size: 16px;
text-align: center;
}
input:hover {
background-color: #111
}
input:focus {
outline: none;
background-color: #111
}
input:focus::placeholder {
color: #090808
}
label {
font-size: 14px;
color: #fff;
float: left;
margin-left: 16px;
margin-bottom: 14px;
}
label > input {
width: auto;
margin: 0 4px;
}
2 years ago
button {
padding: 8px 16px;
2 years ago
border-radius: 4px;
cursor: pointer;
background-color: #3394fb;
color: #fff;
font-size: 14px;
position: fixed;
top: 15px;
2 years ago
right: 24px;
}
button:hover {
background-color: #4b9df4;
}
a, a:visited, a:active {
color: #9ba2ac;
display: inline-block;
font-size: 12px;
margin-top: 8px;
2 years ago
}
a:hover {
color: #54a3f7;
}
img.icon {
float: left;
border-radius: 50%;
margin-right: 8px;
}
img.image {
max-width: 100%;
border-radius: 3px;
}
2 years ago
div {
max-width: 640px;
margin: 0 auto;
padding: 16px 0;
border-top: 1px #000 dashed;
font-size: 14px
}
span {
color: #ccc;
2 years ago
display: block;
margin: 8px 0;
}
p {
margin: 16px 0;
text-align: right;
font-size: 11px;
}
p > a, p > a:visited, p > a:active {
font-size: 11px;
}
2 years ago
</style>
</head>
<body>
<header>
<form name="search" method="GET" action="<?php echo WEBSITE_DOMAIN; ?>/search.php">
<h1><a href="<?php echo WEBSITE_DOMAIN; ?>"><?php echo _('YGGo!') ?></a></h1>
2 years ago
<input type="text" name="q" placeholder="<?php echo $placeholder ?>" value="<?php echo htmlentities($q) ?>" />
<?php foreach ($hostPagesMime as $hostPageMime) { ?>
<label title="<?php $searchHostPagesTotalByMime = $sphinx->searchHostPagesTotalByMime(Filter::searchQuery($q, $m), $hostPageMime->mime);
echo Filter::plural($searchHostPagesTotalByMime, [sprintf(_('%s result'), $searchHostPagesTotalByMime),
sprintf(_('%s results'), $searchHostPagesTotalByMime),
sprintf(_('%s results'), $searchHostPagesTotalByMime)]) ?>">
<input type="radio" name="t" value="<?php echo $hostPageMime->mime ?>" <?php echo ($t == $hostPageMime->mime ? 'checked="checked"' : false) ?>/> <?php echo $hostPageMime->mime ?>
</label>
<?php } ?>
2 years ago
<button type="submit"><?php echo _('Search'); ?></button>
</form>
</header>
<main>
<?php if ($results) { ?>
<div>
<span><?php echo sprintf(_('Total found: %s'), $resultsTotal) ?></span>
<?php if ($queueTotal = $db->getHostPageCrawlQueueTotal(time() - CRAWL_PAGE_SECONDS_OFFSET, time() - CRAWL_PAGE_HOME_SECONDS_OFFSET)) { ?>
<span><?php echo sprintf(_('* Please wait for all pages crawl to complete (%s in queue).'), $queueTotal) ?></span>
<?php } ?>
2 years ago
</div>
<?php foreach ($results as $result) { ?>
<?php if ($hostPage = $db->getFoundHostPage($result->id)) { ?>
<div>
<?php if ($hostPageDescription = $db->getLastPageDescription($result->id)) { ?>
<?php if (!empty($hostPageDescription->title)) { ?>
<h2><?php echo $hostPageDescription->title ?></h2>
<?php } ?>
<?php if (!empty($hostPageDescription->description)) { ?>
<span><?php echo $hostPageDescription->description ?></span>
<?php } ?>
<?php if (!empty($hostPageDescription->keywords)) { ?>
<span><?php echo $hostPageDescription->keywords ?></span>
<?php } ?>
<?php } ?>
<a href="<?php echo $hostPage->scheme . '://' . $hostPage->name . ($hostPage->port ? ':' . $hostPage->port : false) . $hostPage->uri ?>">
<img src="<?php echo WEBSITE_DOMAIN; ?>/file.php?type=identicon&query=<?php echo urlencode($hostPage->name) ?>" alt="identicon" width="16" height="16" class="icon" />
<?php echo htmlentities(urldecode($hostPage->scheme . '://' . $hostPage->name . ($hostPage->port ? ':' . $hostPage->port : false)) . (mb_strlen(urldecode($hostPage->uri)) > 32 ? '...' . mb_substr(urldecode($hostPage->uri), -32) : urldecode($hostPage->uri))) ?>
</a>
|
<a href="<?php echo WEBSITE_DOMAIN; ?>/explore.php?hp=<?php echo $result->id ?>">
<?php echo _('explore'); ?>
</a>
<?php if ($result->mime != 'text' && $totalHostPageIdSources = $db->getTotalHostPageIdSourcesByHostPageIdTarget($result->id)) { ?>
<p>
<?php echo Filter::plural($totalHostPageIdSources, [sprintf(_('%s referrer'), $totalHostPageIdSources),
sprintf(_('%s referrers'), $totalHostPageIdSources),
sprintf(_('%s referrers'), $totalHostPageIdSources),
]) ?>
</p>
<?php $i = 1 ?>
<?php foreach ($db->getHostPageIdSourcesByHostPageIdTarget($result->id, 5) as $hostPageIdSource) { ?>
<?php if ($hostPage = $db->getFoundHostPage($hostPageIdSource->hostPageIdSource)) { ?>
<?php $i++ ?>
<p>
<a href="<?php echo $hostPage->scheme . '://' . $hostPage->name . ($hostPage->port ? ':' . $hostPage->port : false) . $hostPage->uri ?>">
<img src="<?php echo WEBSITE_DOMAIN; ?>/file.php?type=identicon&query=<?php echo urlencode($hostPage->name) ?>" alt="identicon" width="16" height="16" class="icon" />
<?php echo htmlentities(urldecode($hostPage->scheme . '://' . $hostPage->name . ($hostPage->port ? ':' . $hostPage->port : false)) . (mb_strlen(urldecode($hostPage->uri)) > 32 ? '...' . mb_substr(urldecode($hostPage->uri), -32) : urldecode($hostPage->uri))) ?>
</a>
<!--
|
<a href="<?php echo WEBSITE_DOMAIN; ?>/explore.php?hp=<?php echo $hostPage->hostPageId ?>">
<?php echo _('explore'); ?>
</a>
-->
</p>
<?php } ?>
<?php } ?>
<?php if ($i < $totalHostPageIdSources) { ?>
<p>
<a href="<?php echo WEBSITE_DOMAIN; ?>/explore.php?hp=<?php echo $result->id ?>#referrers">
<?php echo _('view all'); ?>
</a>
</p>
<?php } ?>
<?php } ?>
</div>
<?php } ?>
2 years ago
<?php } ?>
<?php if ($p * WEBSITE_PAGINATION_SEARCH_PAGE_RESULTS_LIMIT <= $resultsTotal) { ?>
<div>
<a href="<?php echo WEBSITE_DOMAIN; ?>/search.php?q=<?php echo urlencode(htmlentities($q)) ?>&t=<?php echo $t ?>&m=<?php echo $m ?>&p=<?php echo $p + 1 ?>"><?php echo _('Next page') ?></a>
</div>
<?php } ?>
2 years ago
<?php } else { ?>
<div style="text-align:center">
<span><?php echo sprintf(_('Total found: %s'), $resultsTotal) ?></span>
<?php if ($q && $queueTotal = $db->getHostPageCrawlQueueTotal(time() - CRAWL_PAGE_SECONDS_OFFSET, time() - CRAWL_PAGE_HOME_SECONDS_OFFSET)) { ?>
<span><?php echo sprintf(_('* Please wait for all pages crawl to complete (%s in queue).'), $queueTotal) ?></span>
<?php } ?>
2 years ago
</div>
<?php } ?>
</main>
</body>
</html>