From 8b90e0c91fcb08b27fca29e95822fb5e9d7ecb6b Mon Sep 17 00:00:00 2001
From: LSmyrnaios <lsmyrnaios@gmail.com>
Date: Wed, 29 Nov 2023 13:33:23 +0200
Subject: [PATCH] Improve MetaDocUrl detection.

---
 .../publications_retriever/crawler/MetaDocUrlsHandler.java    | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/src/main/java/eu/openaire/publications_retriever/crawler/MetaDocUrlsHandler.java b/src/main/java/eu/openaire/publications_retriever/crawler/MetaDocUrlsHandler.java
index e8f7e03..7057352 100644
--- a/src/main/java/eu/openaire/publications_retriever/crawler/MetaDocUrlsHandler.java
+++ b/src/main/java/eu/openaire/publications_retriever/crawler/MetaDocUrlsHandler.java
@@ -19,9 +19,9 @@ public class MetaDocUrlsHandler {
 
     // Order-independent META_DOC_URL-regex.
     // <meta(?:[^<]*name=\"(?:[^<]*(?:citation|wkhealth)_pdf|eprints.document)_url\"[^<]*content=\"(http[^\"]+)\"|[^<]*content=\"(http[^\"]+)\"[^<]*name=\"(?:[^<]*(?:citation|wkhealth)_pdf|eprints.document)_url\")[^>]*[/]?>
-    private static final String metaName = "name=\"(?:[^<]*(?:citation|wkhealth)_pdf|eprints.document)_url\"";
+    private static final String metaName = "name=\"(?:[^<]*(?:(?:citation|wkhealth)(?:_fulltext)?_)?pdf|eprints.document)_url\"";
     private static final String metaContent = "content=\"(http[^\"]+)\"";
-    public static final Pattern META_DOC_URL = Pattern.compile("<meta(?:[^<]*" + metaName + "[^<]*" + metaContent + "|[^<]*" + metaContent + "[^<]*" + metaName + ")[^>]*[/]?>");
+    public static final Pattern META_DOC_URL = Pattern.compile("<meta(?:[^<]*" + metaName + "[^<]*" + metaContent + "|[^<]*" + metaContent + "[^<]*" + metaName + ")[^>]*[/]?>", Pattern.CASE_INSENSITIVE);
 
     public static Pattern COMMON_UNSUPPORTED_META_DOC_OR_DATASET_URL_EXTENSIONS;    // Its pattern gets compiled at runtime, only one time, depending on the Datatype.
     static {