Browse Source

implement index explorer

main
ghost 2 years ago
parent
commit
eeeb3dceac
  1. 7
      README.md
  2. 8
      library/mysql.php
  3. 9
      library/sphinxql.php
  4. 274
      public/explore.php
  5. 34
      public/search.php

7
README.md

@ -143,9 +143,11 @@ GET m=SphinxQL
* [x] Web pages full text ranking search * [x] Web pages full text ranking search
* [x] Unlimited content type groups * [x] Unlimited content type groups
* [x] Safe proxy images preview
* [x] Extended syntax support
* [x] Flexible settings compatible with IPv4/IPv6 networks * [x] Flexible settings compatible with IPv4/IPv6 networks
* [x] Index explorer
* [x] Safe images preview
* [x] Extended search syntax support
* [ ] Page history snaps
##### UI ##### UI
@ -163,7 +165,6 @@ GET m=SphinxQL
+ [x] Search + [x] Search
+ [x] Hosts + [x] Hosts
+ [ ] MIME list + [ ] MIME list
* [ ] Remote content DB API
* [ ] Context advertising API * [ ] Context advertising API
##### Crawler ##### Crawler

8
library/mysql.php

@ -218,7 +218,11 @@ class MySQL {
public function getFoundHostPage(int $hostPageId) { public function getFoundHostPage(int $hostPageId) {
$query = $this->_db->prepare('SELECT `hostPage`.`uri`, $query = $this->_db->prepare('SELECT `hostPage`.`hostPageId`,
`hostPage`.`uri`,
`hostPage`.`timeAdded`,
`hostPage`.`timeUpdated`,
`hostPage`.`mime`,
`host`.`scheme`, `host`.`scheme`,
`host`.`name`, `host`.`name`,
`host`.`port` `host`.`port`
@ -349,7 +353,7 @@ class MySQL {
return $query->fetch()->total; return $query->fetch()->total;
} }
public function getHostPageIdSourcesByHostPageIdTarget(int $hostPageIdTarget, int $limit) { public function getHostPageIdSourcesByHostPageIdTarget(int $hostPageIdTarget, int $limit = 1000) {
$query = $this->_db->prepare('SELECT * FROM `hostPageToHostPage` WHERE `hostPageIdTarget` = ? ORDER BY `quantity` DESC LIMIT ' . (int) $limit); $query = $this->_db->prepare('SELECT * FROM `hostPageToHostPage` WHERE `hostPageIdTarget` = ? ORDER BY `quantity` DESC LIMIT ' . (int) $limit);

9
library/sphinxql.php

@ -39,6 +39,15 @@ class SphinxQL {
return $query->fetch()->total; return $query->fetch()->total;
} }
public function getHostPagesMime() {
$query = $this->_sphinx->prepare('SELECT `mime` FROM `hostPage` GROUP BY `mime` ORDER BY `mime` ASC');
$query->execute();
return $query->fetchAll();
}
public function searchHostPagesTotalByMime(string $keyword, string $mime) { public function searchHostPagesTotalByMime(string $keyword, string $mime) {
$query = $this->_sphinx->prepare('SELECT COUNT(*) AS `total` FROM `hostPage` WHERE MATCH(?) AND `mime` = ?'); $query = $this->_sphinx->prepare('SELECT COUNT(*) AS `total` FROM `hostPage` WHERE MATCH(?) AND `mime` = ?');

274
public/explore.php

@ -0,0 +1,274 @@
<?php
// Load system dependencies
require_once('../config/app.php');
require_once('../library/curl.php');
require_once('../library/robots.php');
require_once('../library/filter.php');
require_once('../library/parser.php');
require_once('../library/mysql.php');
require_once('../library/sphinxql.php');
// Connect Sphinx search server
$sphinx = new SphinxQL(SPHINX_HOST, SPHINX_PORT);
// Connect database
$db = new MySQL(DB_HOST, DB_PORT, DB_NAME, DB_USERNAME, DB_PASSWORD);
// Filter request data
$hp = !empty($_GET['hp']) ? Filter::url($_GET['hp']) : 0;
// 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),
]);
?>
<!DOCTYPE html>
<html lang="<?php echo _('en-US'); ?>">
<head>
<title><?php echo sprintf(_('#%s info - YGGo!'), (int) $hp) ?></title>
<meta charset="utf-8" />
<meta name="description" content="<?php echo _('Javascript-less Open Source Web Search Engine') ?>" />
<meta name="keywords" content="<?php echo _('web, search, engine, crawler, php, pdo, mysql, sphinx, yggdrasil, js-less, open source') ?>" />
<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;
margin-bottom: 76px;
padding: 0 20px;
}
h1 {
position: fixed;
top: 8px;
left: 24px;
}
h1 > a,
h1 > a:visited,
h1 > a:active,
h1 > a:hover {
color: #fff;
font-weight: normal;
font-size: 24px;
margin: 10px 0;
text-decoration: none;
}
h2 {
display: block;
font-size: 16px;
font-weight: normal;
margin: 4px 0;
color: #fff;
}
form {
display: block;
max-width: 678px;
margin: 0 auto;
text-align: center;
}
input {
width: 100%;
margin: 12px 0;
padding: 10px 0;
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;
}
button {
padding: 8px 16px;
border-radius: 4px;
cursor: pointer;
background-color: #3394fb;
color: #fff;
font-size: 14px;
position: fixed;
top: 15px;
right: 24px;
}
button:hover {
background-color: #4b9df4;
}
a, a:visited, a:active {
color: #9ba2ac;
display: inline-block;
font-size: 12px;
margin-top: 8px;
}
a:hover {
color: #54a3f7;
}
img.icon {
float: left;
border-radius: 50%;
margin-right: 8px;
}
img.image {
max-width: 100%;
border-radius: 3px;
}
div {
max-width: 640px;
margin: 0 auto;
padding: 16px 0;
border-top: 1px #000 dashed;
font-size: 14px
}
span {
color: #ccc;
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;
}
</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>
<input type="text" name="q" placeholder="<?php echo $placeholder ?>" value="" />
<?php foreach ($sphinx->getHostPagesMime() as $mime) { ?>
<label><input type="radio" name="t" value="<?php echo $mime->mime ?>" /><?php echo $mime->mime ?></label>
<?php } ?>
<button type="submit"><?php echo _('Search'); ?></button>
</form>
</header>
<main>
<?php if ($hostPage = $db->getFoundHostPage($hp)) { ?>
<div>
<?php if ($hostPageDescription = $db->getLastPageDescription($hp)) { ?>
<?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) . $hostPage->uri ?>">
<img src="<?php echo WEBSITE_DOMAIN; ?>/image.php?q=<?php echo urlencode($hostPage->name) ?>" alt="favicon" width="16" height="16" class="icon" />
<?php echo htmlentities(urldecode($hostPage->scheme . '://' . $hostPage->name . ($hostPage->port ? ':' . $hostPage->port : false)) . (mb_strlen(urldecode($hostPage->uri)) > 48 ? '...' . mb_substr(urldecode($hostPage->uri), -48) : urldecode($hostPage->uri))) ?>
</a>
</div>
<div>
<p><?php echo _('MIME') ?></p>
<p><?php echo $hostPage->mime ?></p>
<p><?php echo _('Time added') ?></p>
<p><?php echo date('c', $hostPage->timeAdded) ?></p>
<p><?php echo _('Time updated') ?></p>
<p><?php echo date('c', $hostPage->timeUpdated) ?></p>
<?php if ($totalHostPageIdSources = $db->getTotalHostPageIdSourcesByHostPageIdTarget($hp)) { ?>
<p>
<?php echo Filter::plural($totalHostPageIdSources, [sprintf(_('%s referrer'), $totalHostPageIdSources),
sprintf(_('%s referrers'), $totalHostPageIdSources),
sprintf(_('%s referrers'), $totalHostPageIdSources),
]) ?>
</p>
<?php foreach ($db->getHostPageIdSourcesByHostPageIdTarget($hp) as $hostPageIdSource) { ?>
<?php if ($hostPage = $db->getFoundHostPage($hostPageIdSource->hostPageIdSource)) { ?>
<p>
<?php echo Filter::plural($hostPageIdSource->quantity, [sprintf(_('%s ref'), $hostPageIdSource->quantity),
sprintf(_('%s refs'), $hostPageIdSource->quantity),
sprintf(_('%s refs'), $hostPageIdSource->quantity),
]) ?>
<a href="<?php echo $hostPage->scheme . '://' . $hostPage->name . ($hostPage->port ? ':' . $hostPage->port : false) . $hostPage->uri ?>">
<img src="<?php echo WEBSITE_DOMAIN; ?>/image.php?q=<?php echo urlencode($hostPage->name) ?>" alt="favicon" width="16" height="16" class="icon" />
<?php echo htmlentities(urldecode($hostPage->scheme . '://' . $hostPage->name . ($hostPage->port ? ':' . $hostPage->port : false)) . (mb_strlen(urldecode($hostPage->uri)) > 48 ? '...' . mb_substr(urldecode($hostPage->uri), -48) : urldecode($hostPage->uri))) ?>
</a>
|
<a href="<?php echo WEBSITE_DOMAIN; ?>/explore.php?hp=<?php echo $hostPage->hostPageId ?>">
<?php echo _('explore'); ?>
</a>
</p>
<?php } ?>
<?php } ?>
<?php } ?>
</div>
<?php } else { ?>
<div style="text-align:center">
<span><?php echo _('Not found') ?></span>
<?php if ($queueTotal = $db->getTotalPagesByHttpCode(null)) { ?>
<span><?php echo sprintf(_('* Please wait for all pages crawl to complete (%s in queue).'), $queueTotal) ?></span>
<?php } ?>
</div>
<?php } ?>
</main>
</body>
</html>

34
public/search.php

@ -20,13 +20,12 @@ $t = !empty($_GET['t']) ? Filter::url($_GET['t']) : 'text';
$m = !empty($_GET['m']) ? Filter::url($_GET['m']) : 'default'; $m = !empty($_GET['m']) ? Filter::url($_GET['m']) : 'default';
$q = !empty($_GET['q']) ? Filter::url($_GET['q']) : ''; $q = !empty($_GET['q']) ? Filter::url($_GET['q']) : '';
$p = !empty($_GET['p']) ? (int) $_GET['p'] : 1; $p = !empty($_GET['p']) ? (int) $_GET['p'] : 1;
$i = !empty($_GET['i']) ? (int) $_GET['i'] : 0;
// Search request // Search request
if (!empty($q)) { if (!empty($q)) {
$resultsTotal = $sphinx->searchHostPagesTotal(Filter::searchQuery($q, $m), $t); $resultsTotal = $sphinx->searchHostPagesTotal(Filter::searchQuery($q, $m), $t);
$results = $sphinx->searchHostPages(Filter::searchQuery($q, $m), $t, ($i ? 1 : $p * WEBSITE_PAGINATION_SEARCH_PAGE_RESULTS_LIMIT - WEBSITE_PAGINATION_SEARCH_PAGE_RESULTS_LIMIT), ($i ? 1 : WEBSITE_PAGINATION_SEARCH_PAGE_RESULTS_LIMIT), ($i ? 1 : $resultsTotal)); $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, $resultsTotal);
} else { } else {
@ -296,7 +295,7 @@ if (filter_var($q, FILTER_VALIDATE_URL) && preg_match(CRAWL_URL_REGEXP, $q)) {
font-size: 11px; font-size: 11px;
} }
p > a { p > a, p > a:visited, p > a:active {
font-size: 11px; font-size: 11px;
} }
@ -337,19 +336,23 @@ if (filter_var($q, FILTER_VALIDATE_URL) && preg_match(CRAWL_URL_REGEXP, $q)) {
<?php } ?> <?php } ?>
<a href="<?php echo $hostPage->scheme . '://' . $hostPage->name . ($hostPage->port ? ':' . $hostPage->port : false) . $hostPage->uri ?>"> <a href="<?php echo $hostPage->scheme . '://' . $hostPage->name . ($hostPage->port ? ':' . $hostPage->port : false) . $hostPage->uri ?>">
<img src="<?php echo WEBSITE_DOMAIN; ?>/image.php?q=<?php echo urlencode($hostPage->name) ?>" alt="favicon" width="16" height="16" class="icon" /> <img src="<?php echo WEBSITE_DOMAIN; ?>/image.php?q=<?php echo urlencode($hostPage->name) ?>" alt="favicon" width="16" height="16" class="icon" />
<?php echo htmlentities(urldecode($hostPage->scheme . '://' . $hostPage->name . ($hostPage->port ? ':' . $hostPage->port : false)) . (mb_strlen(urldecode($hostPage->uri)) > 48 ? '...' . mb_substr(urldecode($hostPage->uri), -48) : urldecode($hostPage->uri))) ?> <?php echo htmlentities(urldecode($hostPage->scheme . '://' . $hostPage->name . ($hostPage->port ? ':' . $hostPage->port : false)) . (mb_strlen(urldecode($hostPage->uri)) > 36 ? '...' . mb_substr(urldecode($hostPage->uri), -36) : urldecode($hostPage->uri))) ?>
</a>
|
<a href="<?php echo WEBSITE_DOMAIN; ?>/explore.php?hp=<?php echo $result->id ?>">
<?php echo _('explore'); ?>
</a> </a>
<?php if ($result->mime != 'text' && $totalHostPageIdSources = $db->getTotalHostPageIdSourcesByHostPageIdTarget($result->id)) { ?> <?php if ($result->mime != 'text' && $totalHostPageIdSources = $db->getTotalHostPageIdSourcesByHostPageIdTarget($result->id)) { ?>
<p> <p>
<a href="search.php?q=<?php echo urlencode($q) ?>&t=<?php echo $t ?>&m=<?php echo $m ?>&i=<?php echo $result->id ?>&p=<?php echo $p ?>">
<?php echo Filter::plural($totalHostPageIdSources, [sprintf(_('%s referrer'), $totalHostPageIdSources), <?php echo Filter::plural($totalHostPageIdSources, [sprintf(_('%s referrer'), $totalHostPageIdSources),
sprintf(_('%s referrers'), $totalHostPageIdSources), sprintf(_('%s referrers'), $totalHostPageIdSources),
sprintf(_('%s referrers'), $totalHostPageIdSources), sprintf(_('%s referrers'), $totalHostPageIdSources),
]) ?> ]) ?>
</a>
</p> </p>
<?php foreach ($db->getHostPageIdSourcesByHostPageIdTarget($result->id, ($i ? 1000 : 5)) as $j => $hostPageIdSource) { ?> <?php $i = 1 ?>
<?php foreach ($db->getHostPageIdSourcesByHostPageIdTarget($result->id, 5) as $hostPageIdSource) { ?>
<?php if ($hostPage = $db->getFoundHostPage($hostPageIdSource->hostPageIdSource)) { ?> <?php if ($hostPage = $db->getFoundHostPage($hostPageIdSource->hostPageIdSource)) { ?>
<?php $i++ ?>
<p> <p>
<?php echo Filter::plural($hostPageIdSource->quantity, [sprintf(_('%s ref'), $hostPageIdSource->quantity), <?php echo Filter::plural($hostPageIdSource->quantity, [sprintf(_('%s ref'), $hostPageIdSource->quantity),
sprintf(_('%s refs'), $hostPageIdSource->quantity), sprintf(_('%s refs'), $hostPageIdSource->quantity),
@ -357,16 +360,29 @@ if (filter_var($q, FILTER_VALIDATE_URL) && preg_match(CRAWL_URL_REGEXP, $q)) {
]) ?> ]) ?>
<a href="<?php echo $hostPage->scheme . '://' . $hostPage->name . ($hostPage->port ? ':' . $hostPage->port : false) . $hostPage->uri ?>"> <a href="<?php echo $hostPage->scheme . '://' . $hostPage->name . ($hostPage->port ? ':' . $hostPage->port : false) . $hostPage->uri ?>">
<img src="<?php echo WEBSITE_DOMAIN; ?>/image.php?q=<?php echo urlencode($hostPage->name) ?>" alt="favicon" width="16" height="16" class="icon" /> <img src="<?php echo WEBSITE_DOMAIN; ?>/image.php?q=<?php echo urlencode($hostPage->name) ?>" alt="favicon" width="16" height="16" class="icon" />
<?php echo htmlentities(urldecode($hostPage->scheme . '://' . $hostPage->name . ($hostPage->port ? ':' . $hostPage->port : false)) . (mb_strlen(urldecode($hostPage->uri)) > 48 ? '...' . mb_substr(urldecode($hostPage->uri), -48) : urldecode($hostPage->uri))) ?> <?php echo htmlentities(urldecode($hostPage->scheme . '://' . $hostPage->name . ($hostPage->port ? ':' . $hostPage->port : false)) . (mb_strlen(urldecode($hostPage->uri)) > 36 ? '...' . mb_substr(urldecode($hostPage->uri), -36) : urldecode($hostPage->uri))) ?>
</a>
<!--
|
<a href="<?php echo WEBSITE_DOMAIN; ?>/explore.php?hp=<?php echo $hostPage->hostPageId ?>">
<?php echo _('explore'); ?>
</a> </a>
-->
</p> </p>
<?php } ?> <?php } ?>
<?php } ?> <?php } ?>
<?php if ($i < $totalHostPageIdSources) { ?>
<p>
<a href="<?php echo WEBSITE_DOMAIN; ?>/explore.php?hp=<?php echo $result->id ?>#referrers">
<?php echo _('view all'); ?>
</a>
</p>
<?php } ?>
<?php } ?> <?php } ?>
</div> </div>
<?php } ?> <?php } ?>
<?php } ?> <?php } ?>
<?php if (!$i && $p * WEBSITE_PAGINATION_SEARCH_PAGE_RESULTS_LIMIT <= $resultsTotal) { ?> <?php if ($p * WEBSITE_PAGINATION_SEARCH_PAGE_RESULTS_LIMIT <= $resultsTotal) { ?>
<div> <div>
<a href="<?php echo WEBSITE_DOMAIN; ?>/search.php?q=<?php echo urlencode(htmlentities($q)) ?>&t=<?php echo $t ?>&m=<?php echo $m ?>&p=<?php echo $p + 1 ?>"><?php echo _('Next page') ?></a> <a href="<?php echo WEBSITE_DOMAIN; ?>/search.php?q=<?php echo urlencode(htmlentities($q)) ?>&t=<?php echo $t ?>&m=<?php echo $m ?>&p=<?php echo $p + 1 ?>"><?php echo _('Next page') ?></a>
</div> </div>

Loading…
Cancel
Save