Filter procedures to ngtube and chest x-ray

cli/src/pixl_cli/_io.py

@@ -126,8 +126,11 @@ def _check_and_parse_parquet(private_dir: Path, public_dir: Path) -> pd.DataFram
     accessions = pd.read_parquet(private_dir / "PROCEDURE_OCCURRENCE_LINKS.parquet")
     # study_date is in procedure.procedure_date
     procedure = pd.read_parquet(public_dir / "PROCEDURE_OCCURRENCE.parquet")
+    # TODO: move from hardcoded ng tube and chest x-rays to being configurable  # noqa: FIX002
+    # https://github.com/UCLH-Foundry/PIXL/issues/212
+    imaging_procedures = procedure.loc[procedure.procedure_concept_id.isin([4163872, 42538241])]
     # joining data together
-    people_procedures = people.merge(procedure, on="person_id")
+    people_procedures = people.merge(imaging_procedures, on="person_id")
     return people_procedures.merge(accessions, on="procedure_occurrence_id")
 
 

cli/tests/test_messages_from_parquet.py

@@ -26,9 +26,9 @@
 
 def test_messages_from_parquet(resources: Path) -> None:
     """
-    Given a valid OMOP ES extract directory that has had the logfile parsed
+    Given a valid OMOP ES extract with 4 procedures, two of which are x-rays.
     When the messages are generated from the directory and the output of logfile parsing
-    Then the messages should match expected values
+    Then two messages should be generated
     """
     # Arrange
     omop_parquet_dir = resources / "omop"
@@ -39,35 +39,19 @@ def test_messages_from_parquet(resources: Path) -> None:
     assert all(isinstance(msg, Message) for msg in messages)
 
     expected_messages = [
-        Message(
-            mrn="12345678",
-            accession_number="12345678",
-            study_date=datetime.date.fromisoformat("2021-07-01"),
-            procedure_occurrence_id=1,
-            project_name="test-extract-uclh-omop-cdm",
-            omop_es_timestamp=datetime.datetime.fromisoformat("2023-12-07T14:08:58"),
-        ),
-        Message(
-            mrn="12345678",
-            accession_number="ABC1234567",
-            study_date=datetime.date.fromisoformat("2021-07-01"),
-            procedure_occurrence_id=2,
-            project_name="test-extract-uclh-omop-cdm",
-            omop_es_timestamp=datetime.datetime.fromisoformat("2023-12-07T14:08:58"),
-        ),
         Message(
             mrn="987654321",
-            accession_number="ABC1234560",
-            study_date=datetime.date.fromisoformat("2020-05-01"),
-            procedure_occurrence_id=3,
+            accession_number="AA12345601",
+            study_date=datetime.date.fromisoformat("2020-05-23"),
+            procedure_occurrence_id=4,
             project_name="test-extract-uclh-omop-cdm",
             omop_es_timestamp=datetime.datetime.fromisoformat("2023-12-07T14:08:58"),
         ),
         Message(
-            mrn="5020765",
-            accession_number="MIG0234560",
-            study_date=datetime.date.fromisoformat("2015-05-01"),
-            procedure_occurrence_id=4,
+            mrn="987654321",
+            accession_number="AA12345605",
+            study_date=datetime.date.fromisoformat("2020-05-23"),
+            procedure_occurrence_id=5,
             project_name="test-extract-uclh-omop-cdm",
             omop_es_timestamp=datetime.datetime.fromisoformat("2023-12-07T14:08:58"),
         ),

test/scripts/check_entry_in_pixl_anon.sh

@@ -17,4 +17,4 @@ set -euxo pipefail
 _sql_command="select * from emap_data.ehr_anon"
 _result=$(docker exec system-test-postgres-1 /bin/bash -c \
   "PGPASSWORD=pixl_db_password psql -U pixl_db_username -d pixl -c \"$_sql_command\"")
-echo "$_result" | grep -q "6 row"
+echo "$_result" | grep -q "2 row"