summaryrefslogtreecommitdiff
path: root/src/Readability.php
diff options
context:
space:
mode:
authorAndrew Dolgov <[email protected]>2024-04-03 17:38:00 +0000
committerAndrew Dolgov <[email protected]>2024-04-03 17:38:00 +0000
commit37c0c2ce76aa90e8adafbd5cb0b0332a54df1523 (patch)
treee6b44c6309edadf6e933caf81321740a0ace79a1 /src/Readability.php
parent8ac5abdd497b37d2be4833bcf18d6819bba4d9c9 (diff)
parent29c099cb9c1a9e2f6871068513fc321828da31d2 (diff)
Merge branch 'feature/ignore-more-elements' into 'master'HEADmaster
Add config support for ignoring additional elements. See merge request main/libraries/readability-php!1
Diffstat (limited to 'src/Readability.php')
-rw-r--r--src/Readability.php8
1 files changed, 4 insertions, 4 deletions
diff --git a/src/Readability.php b/src/Readability.php
index 6407a92..c5318a4 100644
--- a/src/Readability.php
+++ b/src/Readability.php
@@ -342,7 +342,7 @@ class Readability
// Extract JSON-LD metadata before removing scripts
$this->jsonld = $this->configuration->getDisableJSONLD() ? [] : $this->getJSONLD($dom);
- $this->removeScripts($dom);
+ $this->removeIgnoredElements($dom);
$this->prepDocument($dom);
@@ -1188,13 +1188,13 @@ class Readability
}
/**
- * Removes all the scripts of the html.
+ * Removes elements that should be ignored.
*
* @param DOMDocument $dom
*/
- private function removeScripts(DOMDocument $dom)
+ private function removeIgnoredElements(DOMDocument $dom)
{
- foreach (['script', 'noscript'] as $tag) {
+ foreach (['noscript', 'script', ...$this->configuration->getExtraIgnoredElements()] as $tag) {
$nodes = $dom->getElementsByTagName($tag);
foreach (iterator_to_array($nodes) as $node) {
NodeUtility::removeNode($node);