From 8461dc62cc19f18ad0acc702673a9665749d43b3 Mon Sep 17 00:00:00 2001
From: Giambattista Bloisi <giambattista.bloisi@openaire.eu>
Date: Fri, 5 Apr 2024 19:09:36 +0200
Subject: [PATCH] initial stage

---
 airflow/dags/EOSC_entity_trasform.py | 35 ++++++++++++++--------------
 1 file changed, 17 insertions(+), 18 deletions(-)

diff --git a/airflow/dags/EOSC_entity_trasform.py b/airflow/dags/EOSC_entity_trasform.py
index fbe1a8d..cb3ef0b 100644
--- a/airflow/dags/EOSC_entity_trasform.py
+++ b/airflow/dags/EOSC_entity_trasform.py
@@ -25,33 +25,33 @@ def trasform_catalog_entity(p: dict) -> dict:
 
 
 def map_fos_topic_to_domain(fos: str):
-    res = dict()
     if fos.startswith('01'):
-        res['domain'] = 'Natural Sciences'
+        return 'Natural Sciences'
     elif fos.startswith('02'):
-        res['domain'] = 'Engineering & Technology'
+        return 'Engineering & Technology'
     elif fos.startswith('03'):
-        res['domain'] = 'Medical & Health Sciences'
+        return 'Medical & Health Sciences'
     elif fos.startswith('04'):
-        res['domain'] = 'Agricultural Sciences'
+        return 'Agricultural Sciences'
     elif fos.startswith('05'):
-        res['domain'] = 'Social Sciences'
+        return 'Social Sciences'
     elif fos.startswith('06'):
-        res['domain'] = 'Humanities'
+        return 'Humanities'
 
-    return res
+    return None
 
 def trasform_product(p: dict) -> dict:
     p = trasform_graph_entity(p)
-    p['accessRights'] = list(
-        filter(lambda ar: ar != '', map(lambda m: map_access_right(m.get('access_right')), p.get('manifestations'))))
-    p['keyword'] = list(
-        map(lambda topic: topic.get('value'),
-            filter(lambda topic: topic.get('topic').get('scheme') == 'keyword', p.get('topics'))))
-    p['domain'] = list(filter(lambda fos: fos is not None,
-                              map(lambda topic: map_fos_topic_to_domain(topic.get('topic').get('value')),
-                                  filter(lambda topic: topic.get('topic').get('scheme') == 'FOS', p.get('topics')))))
-
+    p['accessRights'] = list(set(
+        filter(lambda ar: ar != '', map(lambda m: map_access_right(m.get('access_right')), p.get('manifestations')))))
+    p['keyword'] = list(set(
+        map(lambda topic: topic.get('topic').get('value'),
+            filter(lambda topic: topic.get('topic').get('scheme') == 'keyword', p.get('topics')))))
+    p['domain'] = list(
+        map(lambda fos: {"domain": fos},
+            set(filter(lambda fos: fos is not None,
+                       map(lambda topic: map_fos_topic_to_domain(topic.get('topic').get('value')),
+                           filter(lambda topic: topic.get('topic').get('scheme') == 'FOS', p.get('topics')))))))
     p['firstPublishDate'] = next(
         iter(
             sorted(
@@ -59,7 +59,6 @@ def trasform_product(p: dict) -> dict:
                     filter(lambda date: date.get('type') == 'publishing',
                            [date for m in (p.get('manifestations') or []) for date in (m.get('dates') or [])])))),
         None)
-
     return p