normailize journal titles: use hep request

inspirehep · Nov 9, 2023 · b9f5bfb · b9f5bfb
1 parent 4604f3a
commit b9f5bfb
Showing 1 changed file with 16 additions and 8 deletions.
diff --git a/inspirehep/modules/workflows/tasks/actions.py b/inspirehep/modules/workflows/tasks/actions.py
@@ -638,21 +638,29 @@ def normalize_journal_titles(obj, eng):
         normalize_journal_title_entry(obj, publication, add_inspire_categories=True)
 
     references = obj.data.get("references", [])
+    reference_journal_titles = get_value(references, 'reference.publication_info.journal_title', [])
+    response = requests.post(
+        "{inspirehep_url}/curation/literature/normalize-journal-titles".format(
+            inspirehep_url=current_app.config["INSPIREHEP_URL"]
+        ),
+        headers=_get_headers_for_hep_root_table_request(),
+        data=json.dumps({'journal_titles_list': reference_journal_titles})
+    )
+    response.raise_for_status()
+    normalized_journal_titles_mapping = response.json()['normalized_journal_titles']
     for reference in references:
-        publication_info = get_value(reference, 'reference.publication_info')
-        if not publication_info:
+        publication_info = get_value(reference, 'reference.publication_info', {})
+        journal_title = publication_info.get('journal_title')
+        if not journal_title:
             continue
-        normalize_journal_title_entry(obj, publication_info)
+        normalized_joutnal_title = normalized_journal_titles_mapping['journal_title']
+        normalize_journal_title_entry(obj, publication_info, normalized_joutnal_title)
 
     if obj.extra_data.get('journal_inspire_categories'):
         obj.extra_data['journal_inspire_categories'] = dedupe_list(obj.extra_data['journal_inspire_categories'])
 
 
-def normalize_journal_title_entry(obj, publication_info, add_inspire_categories=False):
-    if 'journal_title' not in publication_info:
-        return
-
-    normalized_title = normalize_journal_title(publication_info['journal_title'])
+def normalize_journal_title_entry(obj, publication_info, normalized_title, add_inspire_categories=False):
     publication_info['journal_title'] = normalized_title
 
     ref_query = RecordMetadata.query.filter(