From 20b432587d586279f47bc7952b1502d7139a1907 Mon Sep 17 00:00:00 2001
From: PascalEgn <pascal.egner@googlemail.com>
Date: Tue, 27 Aug 2024 17:07:20 +0200
Subject: [PATCH] editor: improve author xml detection

---
 .../editor/test_editor_views/author_list.xml        |  6 ++++--
 .../author-extract-actions.component.ts             | 13 +++++++------
 2 files changed, 11 insertions(+), 8 deletions(-)

diff --git a/backend/tests/integration/editor/test_editor_views/author_list.xml b/backend/tests/integration/editor/test_editor_views/author_list.xml
index 6dda41236..22b9b65ce 100644
--- a/backend/tests/integration/editor/test_editor_views/author_list.xml
+++ b/backend/tests/integration/editor/test_editor_views/author_list.xml
@@ -1,4 +1,6 @@
-    <collaborationauthorlist xmlns:foaf="http://xmlns.com/foaf/0.1/" xmlns:cal="http://inspirehep.net/info/HepNames/tools/authors_xml/">
+<?xml version="1.0" encoding="UTF-8"?>
+<!DOCTYPE collaborationauthorlist SYSTEM "authors.dtd">
+<collaborationauthorlist xmlns:foaf="http://xmlns.com/foaf/0.1/" xmlns:cal="http://inspirehep.net/info/HepNames/tools/authors_xml/">
     <cal:creationDate>2022-01-25</cal:creationDate>
 
     <cal:publicationReference>Fermilab-PUB-2022-01-25</cal:publicationReference>
@@ -34,4 +36,4 @@
             </cal:authorIDs>
         </foaf:Person>
     </cal:authors>
-    </collaborationauthorlist>
+</collaborationauthorlist>
diff --git a/record-editor/src/app/shared/components/extract-actions/author-extract-actions.component.ts b/record-editor/src/app/shared/components/extract-actions/author-extract-actions.component.ts
index 3b0a8b8ef..399c2cc80 100644
--- a/record-editor/src/app/shared/components/extract-actions/author-extract-actions.component.ts
+++ b/record-editor/src/app/shared/components/extract-actions/author-extract-actions.component.ts
@@ -113,11 +113,12 @@ export class AuthorExtractActionsComponent {
     if (isUrl) {
       return 'url';
     }
-
-    const isXml = this.source.startsWith('<collaborationauthorlist');
-    if (isXml) {
-      return 'xml';
-    }
-    return 'text';
+    const parser = new DOMParser();
+    const doc = parser.parseFromString(this.source, 'application/xml');
+    const errorNode = doc.querySelector('parsererror');
+    return !errorNode &&
+      doc.documentElement.nodeName === 'collaborationauthorlist'
+      ? 'xml'
+      : 'text';
   }
 }