From ce78752aa33629f82c401443de88a53ba8ea1c69 Mon Sep 17 00:00:00 2001
From: Claudio Atzori <claudio.atzori@isti.cnr.it>
Date: Wed, 7 Aug 2024 15:46:44 +0200
Subject: [PATCH] BIPAffiliations to include also input data from publisher
 websites

---
 .../input_actionset_parameter.json            | 25 ++++++++++---------
 .../bipaffiliations/oozie_app/workflow.xml    |  5 ++++
 .../PrepareAffiliationRelationsTest.java      |  9 ++++---
 3 files changed, 23 insertions(+), 16 deletions(-)

diff --git a/dhp-workflows/dhp-aggregation/src/main/resources/eu/dnetlib/dhp/actionmanager/bipaffiliations/input_actionset_parameter.json b/dhp-workflows/dhp-aggregation/src/main/resources/eu/dnetlib/dhp/actionmanager/bipaffiliations/input_actionset_parameter.json
index 941f84525..f80d9e446 100644
--- a/dhp-workflows/dhp-aggregation/src/main/resources/eu/dnetlib/dhp/actionmanager/bipaffiliations/input_actionset_parameter.json
+++ b/dhp-workflows/dhp-aggregation/src/main/resources/eu/dnetlib/dhp/actionmanager/bipaffiliations/input_actionset_parameter.json
@@ -28,18 +28,19 @@
     "paramLongName": "dataciteInputPath",
     "paramDescription": "the path to get the input data from Datacite",
     "paramRequired": true
-  },{
-  "paramName": "wip",
-  "paramLongName": "webCrawlInputPath",
-  "paramDescription": "the path to get the input data from Web Crawl",
-  "paramRequired": true
-},{
-  "paramName": "pip",
-  "paramLongName": "publisherInputPath",
-  "paramDescription": "the path to get the input data from publishers",
-  "paramRequired": true
-}
-,
+  },
+  {
+    "paramName": "wip",
+    "paramLongName": "webCrawlInputPath",
+    "paramDescription": "the path to get the input data from Web Crawl",
+    "paramRequired": true
+  },
+  {
+    "paramName": "pip",
+    "paramLongName": "publisherInputPath",
+    "paramDescription": "the path to get the input data from publishers",
+    "paramRequired": true
+  },
   {
     "paramName": "o",
     "paramLongName": "outputPath",
diff --git a/dhp-workflows/dhp-aggregation/src/main/resources/eu/dnetlib/dhp/actionmanager/bipaffiliations/oozie_app/workflow.xml b/dhp-workflows/dhp-aggregation/src/main/resources/eu/dnetlib/dhp/actionmanager/bipaffiliations/oozie_app/workflow.xml
index 2e89c07fd..77683b6b3 100644
--- a/dhp-workflows/dhp-aggregation/src/main/resources/eu/dnetlib/dhp/actionmanager/bipaffiliations/oozie_app/workflow.xml
+++ b/dhp-workflows/dhp-aggregation/src/main/resources/eu/dnetlib/dhp/actionmanager/bipaffiliations/oozie_app/workflow.xml
@@ -21,6 +21,10 @@
             <name>webCrawlInputPath</name>
             <description>the path where to find the inferred affiliation relations from webCrawl</description>
         </property>
+        <property>
+            <name>publisherInputPath</name>
+            <description>the path where to find the inferred affiliation relations from publisher websites</description>
+        </property>
         <property>
             <name>outputPath</name>
             <description>the path where to store the actionset</description>
@@ -117,6 +121,7 @@
             <arg>--openapcInputPath</arg><arg>${openapcInputPath}</arg>
             <arg>--dataciteInputPath</arg><arg>${dataciteInputPath}</arg>
             <arg>--webCrawlInputPath</arg><arg>${webCrawlInputPath}</arg>
+            <arg>--publisherInputPath</arg><arg>${publisherInputPath}</arg>
             <arg>--outputPath</arg><arg>${outputPath}</arg>
         </spark>
         <ok to="End"/>
diff --git a/dhp-workflows/dhp-aggregation/src/test/java/eu/dnetlib/dhp/actionmanager/bipaffiliations/PrepareAffiliationRelationsTest.java b/dhp-workflows/dhp-aggregation/src/test/java/eu/dnetlib/dhp/actionmanager/bipaffiliations/PrepareAffiliationRelationsTest.java
index c90288f59..ac9977a7e 100644
--- a/dhp-workflows/dhp-aggregation/src/test/java/eu/dnetlib/dhp/actionmanager/bipaffiliations/PrepareAffiliationRelationsTest.java
+++ b/dhp-workflows/dhp-aggregation/src/test/java/eu/dnetlib/dhp/actionmanager/bipaffiliations/PrepareAffiliationRelationsTest.java
@@ -150,11 +150,12 @@ public class PrepareAffiliationRelationsTest {
 					.get(0)
 					.getString(4));
 
-
-		final String publisherid = ID_PREFIX + IdentifierFactory.md5(CleaningFunctions.normalizePidValue("doi", "10.1007/s00217-010-1268-9"));
+		final String publisherid = ID_PREFIX
+			+ IdentifierFactory.md5(CleaningFunctions.normalizePidValue("doi", "10.1007/s00217-010-1268-9"));
 		final String rorId = "20|ror_________::" + IdentifierFactory.md5("https://ror.org/03265fv13");
 
-		Assertions.assertEquals(1, execVerification.filter("source = '" + publisherid + "' and target = '" + rorId +"'").count()
-				);
+		Assertions
+			.assertEquals(
+				1, execVerification.filter("source = '" + publisherid + "' and target = '" + rorId + "'").count());
 	}
 }