Import Belga biographies into Superdesk [SDBELGA-623] (#386)

* Import Belga biographies into Superdesk [SDBELGA-623] * Minor change * Added testcase
superdesk · Mar 22, 2022 · 6a25cd0 · 6a25cd0
1 parent 2200d17
commit 6a25cd0
Show file tree

Hide file tree

Showing 3 changed files with 18 additions and 7 deletions.
diff --git a/server/belga/io/feed_parsers/belga_newsml_1_2.py b/server/belga/io/feed_parsers/belga_newsml_1_2.py
@@ -625,11 +625,11 @@ def parse_administrativemetadata(self, item, admin_el):
         signoff_list = []
         for element in admin_el.findall('Creator/Party'):
             if element is not None and element.get('FormalName'):
-                _sign_off = author_name = element.get('FormalName', '').replace(' ', '')
+                _sign_off = author_name = element.get('FormalName', '').replace(' ', '').strip('()')
                 _topic = element.get('Topic', '')
                 author = {
-                    '_id': [_topic], 'name': _topic, 'role': _topic,
-                    'sub_label': author_name
+                    '_id': [author_name, _topic], 'role': _topic,
+                    'name': _topic, 'sub_label': author_name
                 }
                 # try to find an author in DB
                 user = get_resource_service('users').find_one(req=None, username=author_name)
@@ -647,8 +647,12 @@ def parse_administrativemetadata(self, item, admin_el):
                 signoff_list.append(_sign_off)
                 item.setdefault('authors', []).append(author)
 
+        # Check and remove duplicates authors if any
+        if item.get('authors'):
+            item['authors'] = [dict(i) for i, _ in itertools.groupby(sorted(item['authors'], key=lambda k: k['_id']))]
+
         if signoff_list:
-            item["sign_off"] = "/".join(signoff_list)
+            item['sign_off'] = "/".join(signoff_list)
 
         element = admin_el.find('Contributor/Party')
         if element is not None and element.get('FormalName'):

diff --git a/server/tests/io/feed_parsers/belga_newsml_1_2_test.py b/server/tests/io/feed_parsers/belga_newsml_1_2_test.py
@@ -46,7 +46,6 @@ def test_can_parse(self):
 
     def test_content(self):
         item = self.item[0]
-
         self.assertEqual(
             item['administrative']['foreign_id'],
             'BIN118'
@@ -67,7 +66,14 @@ def test_content(self):
                 'role': 'AUTHOR',
                 'sub_label': 'DWM',
             },
-        ])
+            {
+                '_id': ['TIP', 'EDITOR'],
+                'name': 'EDITOR',
+                'role': 'EDITOR',
+                'sub_label': 'TIP',
+            }
+        ]
+        )
         self.assertEqual(
             item['body_html'],
             '<p>Steven &lt;b&gt;Van Geel&lt;/b&gt; gaf zich op 31 mei 2014 aan bij de politie van zijn '
@@ -133,7 +139,7 @@ def test_content(self):
         self.assertEqual(item['subject'], expected_subjects)
 
         self.assertEqual(item['type'], 'text')
-        self.assertEqual(item['sign_off'], 'DWM')
+        self.assertEqual(item['sign_off'], 'DWM/TIP')
         self.assertEqual(item['version'], 4)
         self.assertEqual(item['versioncreated'], datetime.datetime(2019, 1, 29, 12, 34, tzinfo=pytz.utc))
         self.assertEqual(item["genre"], [{'name': 'CURRENT', 'qcode': 'CURRENT'}])

diff --git a/server/tests/io/fixtures/belga_newsml_1_2.xml b/server/tests/io/fixtures/belga_newsml_1_2.xml
@@ -52,6 +52,7 @@
                     </Provider>
                     <Creator>
                         <Party FormalName="DWM" Topic="AUTHOR"/>
+                        <Party FormalName="(TIP)" Topic="EDITOR"/>
                     </Creator>
                     <Source>
                         <Party FormalName="BELGA"/>