Browse Source

add metaYggo field index

main
ghost 2 years ago
parent
commit
11aa404807
  1. 12
      crontab/crawler.php
  2. BIN
      database/yggo.mwb
  3. 4
      library/mysql.php

12
crontab/crawler.php

@ -71,9 +71,10 @@ foreach ($db->getCrawlQueue(CRAWL_PAGE_LIMIT, time() - CRAWL_PAGE_SECONDS_OFFSET @@ -71,9 +71,10 @@ foreach ($db->getCrawlQueue(CRAWL_PAGE_LIMIT, time() - CRAWL_PAGE_SECONDS_OFFSET
}
// Get optional page meta data
$metaDescription = '';
$metaKeywords = '';
$metaRobots = '';
$metaDescription = null;
$metaKeywords = null;
$metaRobots = null;
$metaYggo = null;
foreach (@$dom->getElementsByTagName('meta') as $meta) {
@ -88,6 +89,10 @@ foreach ($db->getCrawlQueue(CRAWL_PAGE_LIMIT, time() - CRAWL_PAGE_SECONDS_OFFSET @@ -88,6 +89,10 @@ foreach ($db->getCrawlQueue(CRAWL_PAGE_LIMIT, time() - CRAWL_PAGE_SECONDS_OFFSET
if (@$meta->getAttribute('name') == 'robots') {
$metaRobots = @$meta->getAttribute('content');
}
if (@$meta->getAttribute('name') == 'yggo') {
$metaYggo = @$meta->getAttribute('content');
}
}
// Update queued page data
@ -95,6 +100,7 @@ foreach ($db->getCrawlQueue(CRAWL_PAGE_LIMIT, time() - CRAWL_PAGE_SECONDS_OFFSET @@ -95,6 +100,7 @@ foreach ($db->getCrawlQueue(CRAWL_PAGE_LIMIT, time() - CRAWL_PAGE_SECONDS_OFFSET
Filter::pageTitle($title->item(0)->nodeValue),
Filter::pageDescription($metaDescription),
Filter::pageKeywords($metaKeywords),
Filter::url($metaYggo),
CRAWL_HOST_DEFAULT_META_ONLY ? null : Filter::pageData($content));
// Append page with meta robots:noindex value to the robotsPostfix disallow list

BIN
database/yggo.mwb

Binary file not shown.

4
library/mysql.php

@ -193,14 +193,16 @@ class MySQL { @@ -193,14 +193,16 @@ class MySQL {
mixed $metaTitle,
mixed $metaDescription,
mixed $metaKeywords,
mixed $metaYggo,
mixed $data) {
$query = $this->_db->prepare('UPDATE `hostPage` SET `metaTitle` = ?,
`metaDescription` = ?,
`metaKeywords` = ?,
`metaYggo` = ?,
`data` = ? WHERE `hostPageId` = ? LIMIT 1');
$query->execute([$metaTitle, $metaDescription, $metaKeywords, $data, $hostPageId]);
$query->execute([$metaTitle, $metaDescription, $metaKeywords, $metaYggo, $data, $hostPageId]);
return $query->rowCount();
}

Loading…
Cancel
Save