Merge pull request #128 from internetarchive/schema-rewrtie

Raise flag on malformed URLs
internetarchive · Feb 9, 2024 · 857e4e0 · 857e4e0
2 parents d61f73d + fbc3584
commit 857e4e0
Showing 1 changed file with 6 additions and 2 deletions.
diff --git a/app/src/Core/parse.php b/app/src/Core/parse.php
@@ -316,7 +316,7 @@ public function analyzePage( &$modifiedLinks = [], $webRequest = false, &$editEr
 					) {
 						//Populate URLs to submit for archiving.
 						if( $i == 1 ) {
-							$toArchive["$tid:$id"] = $link['url'];
+							if( !isset( $link['malformed_url'] ) ) $toArchive["$tid:$id"] = $link['url'];
 						} else {
 							//If it archived, then tally the success, otherwise, note it.
 							if( $archiveResponse["$tid:$id"] === true ) {
@@ -331,7 +331,7 @@ public function analyzePage( &$modifiedLinks = [], $webRequest = false, &$editEr
 					) {
 						//Populate URLs to submit for archiving.
 						if( $i == 1 ) {
-							$toArchive[$tid] = $link['url'];
+							if( !isset( $link['malformed_url'] ) ) $toArchive[$tid] = $link['url'];
 						} else {
 							//If it archived, then tally the success, otherwise, note it.
 							if( $archiveResponse[$tid] === true ) {
@@ -2390,6 +2390,10 @@ public function getLinkDetails( $linkString, $remainder ) {
 			unset( $returnArray['original_url'] );
 		}
 
+		if( empty( strtolower( parse_url( $this->deadCheck->sanitizeURL( $returnArray['url'] ), PHP_URL_SCHEME ) ) ) ) {
+			$returnArray['malformed_url'] = true;
+		}
+
 		if( isset( $returnArray['archive_template'] ) ) {
 			if( isset( $returnArray['archive_template']['parameters']['__FORMAT__'] ) ) {
 				$returnArray['archive_template']['format'] =