From 805378baa6620383aa847a553079a218364ca006 Mon Sep 17 00:00:00 2001
From: Sam Leeflang <sam.leeflang@naturalis.nl>
Date: Mon, 2 Dec 2024 14:58:56 +0100
Subject: [PATCH] Improve DOI identifier check

---
 pom.xml                                                    | 2 +-
 .../core/translator/terms/utils/IdentifierUtils.java       | 4 ++--
 .../core/translator/terms/utils/IdentifierUtilsTest.java   | 7 +++++++
 3 files changed, 10 insertions(+), 3 deletions(-)

diff --git a/pom.xml b/pom.xml
index 052b1bd..5709dc9 100644
--- a/pom.xml
+++ b/pom.xml
@@ -6,7 +6,7 @@
   <parent>
     <groupId>org.springframework.boot</groupId>
     <artifactId>spring-boot-starter-parent</artifactId>
-    <version>3.3.5</version>
+    <version>3.4.0</version>
     <relativePath/> <!-- lookup parent from repository -->
   </parent>
   <groupId>eu.dissco.core</groupId>
diff --git a/src/main/java/eu/dissco/core/translator/terms/utils/IdentifierUtils.java b/src/main/java/eu/dissco/core/translator/terms/utils/IdentifierUtils.java
index db97dcd..1ab5e31 100644
--- a/src/main/java/eu/dissco/core/translator/terms/utils/IdentifierUtils.java
+++ b/src/main/java/eu/dissco/core/translator/terms/utils/IdentifierUtils.java
@@ -33,9 +33,9 @@ private IdentifierUtils() {
 
   private static Map<List<Pattern>, Triple<DctermsType, String, OdsGupriLevel>> patternMap() {
     var linkedMap = new LinkedHashMap<List<Pattern>, Triple<DctermsType, String, OdsGupriLevel>>();
-    linkedMap.put(List.of(compile("^https?://doi.org")),
+    linkedMap.put(List.of(compile("^https?://doi.org/10.")),
         Triple.of(DOI, "DOI", GLOBALLY_UNIQUE_STABLE_PERSISTENT_RESOLVABLE_FDO_COMPLIANT));
-    linkedMap.put(List.of(compile("^https?://hdl.handle.net")),
+    linkedMap.put(List.of(compile("^https?://hdl.handle.net"), compile("^https?://doi.org/")),
         Triple.of(HANDLE, "Handle", GLOBALLY_UNIQUE_STABLE_PERSISTENT_RESOLVABLE_FDO_COMPLIANT));
     linkedMap.put(List.of(compile("^https?://www.wikidata.org")),
         Triple.of(URL, "Wikidata", GLOBALLY_UNIQUE_STABLE_PERSISTENT_RESOLVABLE));
diff --git a/src/test/java/eu/dissco/core/translator/terms/utils/IdentifierUtilsTest.java b/src/test/java/eu/dissco/core/translator/terms/utils/IdentifierUtilsTest.java
index ce4b3f8..f75578a 100644
--- a/src/test/java/eu/dissco/core/translator/terms/utils/IdentifierUtilsTest.java
+++ b/src/test/java/eu/dissco/core/translator/terms/utils/IdentifierUtilsTest.java
@@ -1,6 +1,7 @@
 package eu.dissco.core.translator.terms.utils;
 
 import static eu.dissco.core.translator.schema.Identifier.DctermsType.ARK;
+import static eu.dissco.core.translator.schema.Identifier.DctermsType.DOI;
 import static eu.dissco.core.translator.schema.Identifier.DctermsType.HANDLE;
 import static eu.dissco.core.translator.schema.Identifier.DctermsType.LOCALLY_UNIQUE_IDENTIFIER;
 import static eu.dissco.core.translator.schema.Identifier.DctermsType.PURL;
@@ -31,6 +32,12 @@ class IdentifierUtilsTest {
 
   public static Stream<Arguments> identifierProvider() {
     return Stream.of(
+        Arguments.of("https://doi.org/10.3535/M42-Z4P-DRD", null, PREFERRED,
+            createIdentifier("https://doi.org/10.3535/M42-Z4P-DRD", DOI, "DOI",
+                GLOBALLY_UNIQUE_STABLE_PERSISTENT_RESOLVABLE_FDO_COMPLIANT, PREFERRED)),
+        Arguments.of("https://doi.org/21.T11148/894b1e6cad57e921764e", null, PREFERRED,
+            createIdentifier("https://doi.org/21.T11148/894b1e6cad57e921764e", HANDLE, "Handle",
+                GLOBALLY_UNIQUE_STABLE_PERSISTENT_RESOLVABLE_FDO_COMPLIANT, PREFERRED)),
         Arguments.of("https://www.wikidata.org/wiki/Q66581882", null, null,
             createIdentifier("https://www.wikidata.org/wiki/Q66581882", URL, "Wikidata",
                 GLOBALLY_UNIQUE_STABLE_PERSISTENT_RESOLVABLE, null)),