YGGverse/YGGo
1
0
Fork 0
mirror of https://github.com/YGGverse/YGGo.git synced 2025-01-24 21:44:59 +00:00
Code Issues Projects Releases Wiki Activity

crawl host page DOM selectors on meta robots:index/follow condition enabled only

Browse Source
This commit is contained in:
ghost 2023-07-12 12:16:26 +03:00
parent 5c0cee7561
commit b7c415a8b0
1 changed files with 27 additions and 27 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

54
crontab/crawler.php
View File

@ -492,32 +492,6 @@ foreach ($db->getHostPageCrawlQueue(CRAWL_PAGE_LIMIT, time() - CRAWL_PAGE_SECOND
$metaKeywords = null; $metaKeywords = null;
$metaYggoManifest = null; $metaYggoManifest = null;
// Collect page DOM elements data
if (CRAWL_HOST_PAGE_DOM_SELECTORS) {
// Begin selectors extraction
$html = str_get_html($content);
foreach ((array) explode(',', CRAWL_HOST_PAGE_DOM_SELECTORS) as $selector) {
foreach($html->find($selector) as $element) {
if (!empty($element->innertext)) {
$db->addHostPageDom($queueHostPage->hostPageId,
time(),
$selector,
trim(CRAWL_HOST_PAGE_DOM_STRIP_TAGS ? strip_tags(
preg_replace('/[\s]+/',
' ',
str_replace(['<br />', '<br/>', '<br>', '</'],
[' ', ' ', ' ', ' </'],
$element->innertext))) : $element->innertext));
}
}
}
}
// Parse page content // Parse page content
$dom = new DomDocument(); $dom = new DomDocument();
@ -594,13 +568,39 @@ foreach ($db->getHostPageCrawlQueue(CRAWL_PAGE_LIMIT, time() - CRAWL_PAGE_SECOND
$content ? ($queueHostPage->crawlMetaOnly ? null : base64_encode($content)) : null, $content ? ($queueHostPage->crawlMetaOnly ? null : base64_encode($content)) : null,
time()); time());
// Collect page DOM elements data on enabled
if (CRAWL_HOST_PAGE_DOM_SELECTORS) {
// Begin selectors extraction
$html = str_get_html($content);
foreach ((array) explode(',', CRAWL_HOST_PAGE_DOM_SELECTORS) as $selector) {
foreach($html->find($selector) as $element) {
if (!empty($element->innertext)) {
$db->addHostPageDom($queueHostPage->hostPageId,
time(),
$selector,
trim(CRAWL_HOST_PAGE_DOM_STRIP_TAGS ? strip_tags(
preg_replace('/[\s]+/',
' ',
str_replace(['<br />', '<br/>', '<br>', '</'],
[' ', ' ', ' ', ' </'],
$element->innertext))) : $element->innertext));
}
}
}
}
// Update manifest registry // Update manifest registry
if (CRAWL_MANIFEST && !empty($metaYggoManifest) && filter_var($metaYggoManifest, FILTER_VALIDATE_URL) && preg_match(CRAWL_URL_REGEXP, $metaYggoManifest)) { if (CRAWL_MANIFEST && !empty($metaYggoManifest) && filter_var($metaYggoManifest, FILTER_VALIDATE_URL) && preg_match(CRAWL_URL_REGEXP, $metaYggoManifest)) {
$metaYggoManifestCRC32 = crc32($metaYggoManifest); $metaYggoManifestCRC32 = crc32($metaYggoManifest);
if (!$db->getManifest($metaYggoManifestCRC32)) { if (!$db->getManifest($metaYggoManifestCRC32)) {
$db->addManifest($metaYggoManifestCRC32, $db->addManifest($metaYggoManifestCRC32,
$metaYggoManifest, $metaYggoManifest,
(string) CRAWL_MANIFEST_DEFAULT_STATUS, (string) CRAWL_MANIFEST_DEFAULT_STATUS,
time()); time());