Merge pull request #2943 from opensafely-core/Jongmassey/fix-bnf-dmd-mapping-scraper

Fix link html parsing for Sept 2025 release

Author: lucyb

mappings/bnfdmd/data_downloader.py

@@ -47,7 +47,10 @@ def get_releases(self):
                     valid_from := datepat.search(inner_text.replace("\xa0", " "))
                 ):
                     # sometimes "valid from" inner text is across adjacent elements
-                    inner_text += a_tag.next_sibling()[0].text
+                    if a_tag.next_sibling:
+                        inner_text += a_tag.next_sibling.text
+                    elif a_tag.previous_sibling:
+                        inner_text = a_tag.previous_sibling.text + inner_text
                 valid_from = datetime.strptime(valid_from.group(), datefmt)
                 matches.append((match.group("date"), url, filename, valid_from))
 

mappings/bnfdmd/tests/conftest.py

@@ -39,7 +39,8 @@ def mocked_responses_homepage():
             "https://www.nhsbsa.nhs.uk/prescription-data/understanding-our-data/bnf-snomed-mapping",
             body=f"""
             <p><a href="/sites/default/files/{MOCK_FILEPATH}">January 2024 (ZIP file: 16.76MB)</a></p>
-            <p><a href="/sites/default/files/2023-01/BNF%20Snomed%20Mapping%20data%2020230116.zip">January 2023 (ZIP file: 16.76MB)</a></p>
+            <p><a href="/sites/default/files/2023-01/BNF%20Snomed%20Mapping%20data%2020230116.zip">January</a> 2023 (ZIP file: 16.76MB)</p>
+            <p>September<a href="/sites/default/files/2023-11/BNF%20Snomed%20Mapping%20data%2020231120.zip"> 2023 (ZIP file: 18.77MB)</a></p>
         """,
             status=200,
         )

mappings/bnfdmd/tests/test_data_downloader.py

@@ -17,3 +17,10 @@ def test_get_latest_release_with_existing(mocked_responses_homepage, tmp_path):
     downloader = Downloader(tmp_path)
     with pytest.raises(ValueError, match="Latest release already exists"):
         downloader.download_latest_release()
+
+
+def test_historical_release_dates_parsed(mocked_responses_homepage, tmp_path):
+    downloader = Downloader(tmp_path)
+    releases = downloader.get_releases()
+
+    assert len(releases) == 3