Mapping stage picks up on lists in schema

src/adtl/autoparser/mapping.py

@@ -102,6 +102,10 @@ def _value_options(f):
                     return ["True", "False", "None"]
                 elif "string" in self.target_types[f]:
                     return self.schema_properties[f].get("enum", np.nan)
+                elif "array" in self.target_types[f]:
+                    return (
+                        self.schema_properties[f].get("items", {}).get("enum", np.nan)
+                    )
                 else:
                     return np.nan
 

tests/test_autoparser/schemas/IB-sample.schema.json

@@ -0,0 +1,110 @@
+{
+  "$schema": "http://json-schema.org/draft-07/schema#",
+  "type": "object",
+  "properties": {
+    "Case ID": {
+      "type": "integer",
+      "description": "Unique identifier for each case",
+      "PrimaryKey": true
+    },
+    "Age": {
+      "type": "integer",
+      "description": "Age of the patient",
+      "minimum": 0
+    },
+    "Gender": {
+      "type": "string",
+      "description": "Gender of the patient",
+      "enum": [
+        "Male",
+        "Female",
+        "Other"
+      ]
+    },
+    "Location": {
+      "type": "string",
+      "description": "City or region where the case was reported"
+    },
+    "Date of Onset": {
+      "type": "string",
+      "format": "date",
+      "description": "Date when symptoms first appeared"
+    },
+    "Symptoms": {
+      "type": [
+        "array",
+        "null"
+      ],
+      "description": "List of symptoms exhibited by the patient",
+      "items": {
+        "type": "string",
+        "enum": [
+          "fever",
+          "cough",
+          "dyspnea",
+          "fatigue",
+          "myalgia",
+          "headache",
+          "anosmia"
+        ]
+      }
+    },
+    "Outcome": {
+      "type": [
+        "string",
+        "null"
+      ],
+      "description": "Final outcome for the patient (recovered or deceased)",
+      "enum": [
+        "Recovered",
+        "Deceased",
+        null
+      ]
+    },
+    "Vaccination Status": {
+      "type": [
+        "string",
+        "null"
+      ],
+      "description": "Vaccination status of the patient",
+      "enum": [
+        "Yes",
+        "No",
+        "Partial",
+        "Unknown"
+      ]
+    },
+    "Days to Recovery": {
+      "type": [
+        "integer",
+        "null"
+      ],
+      "description": "Number of days to recover, null for deceased cases"
+    },
+    "Underlying Conditions": {
+      "type": [
+        "string",
+        "null"
+      ],
+      "description": "Pre-existing health conditions of the patient",
+      "enum": [
+        "None",
+        "Asthma",
+        "Diabetes",
+        "Hypertension",
+        "Heart Disease",
+        "Chronic Lung Disease",
+        "Chronic Kidney Disease",
+        null
+      ]
+    }
+  },
+  "required": [
+    "Case ID",
+    "Age",
+    "Gender",
+    "Location",
+    "Date of Onset"
+  ],
+  "additionalProperties": false
+}

tests/test_autoparser/sources/IB-mapping.csv

@@ -0,0 +1,11 @@
+target_field,source_description,source_field,common_values,target_values,value_mapping
+Case ID,Case Number,CaseNumber,,,
+Age,Person Age,PersonAge,,,
+Gender,Sex,Sex,"Male, F, Female, M","Male, Female, Other","female=Female, f=Female, male=Male, m=Male"
+Location,City,City,"Chicago, Houston, Philly",,
+Date of Onset,Onset Date,OnsetDate,,,
+Symptoms,Reported Symptoms,ReportedSymptoms,"fever, fatigue, cough, short breath, headache, muscle pain, coughing","fever, cough, dyspnea, fatigue, myalgia, headache, anosmia","fever=fever, coughing=cough, cough=cough, fatigue=fatigue, short breath=dyspnea, muscle pain=myalgia, headache=headache"
+Outcome,Health Outcome,HealthOutcome,"recovered, Recov, Dead, Recovered, Died","Recovered, Deceased, None","recov=Recovered, died=Deceased, recovered=Recovered, dead=Deceased"
+Vaccination Status,Vaccination Status,VaxStatus,"yes, no, Partial, No, Yes, none","Yes, No, Partial, Unknown","partial=Partial, none=Unknown, no=No, yes=Yes"
+Days to Recovery,Recovery Days,RecoveryDays,,,
+Underlying Conditions,Pre-existing Conditions,PreexistingConditions,,"None, Asthma, Diabetes, Hypertension, Heart Disease, Chronic Lung Disease, Chronic Kidney Disease, None",

tests/test_autoparser/test_mapper.py

@@ -122,6 +122,50 @@ def test_target_values():
     pd.testing.assert_series_equal(mapper.target_values, target_vals)
 
 
+def test_target_fields_with_enum_lists():
+    mapper = MapperTest(
+        "tests/test_autoparser/sources/IB_sample_dd.csv",
+        Path("tests/test_autoparser/schemas/IB-sample.schema.json"),
+        "fr",
+    )
+    target_vals = pd.Series(
+        data=[
+            np.nan,
+            np.nan,
+            ["Male", "Female", "Other"],
+            np.nan,
+            np.nan,
+            ["fever", "cough", "dyspnea", "fatigue", "myalgia", "headache", "anosmia"],
+            ["Recovered", "Deceased", None],
+            ["Yes", "No", "Partial", "Unknown"],
+            np.nan,
+            [
+                "None",
+                "Asthma",
+                "Diabetes",
+                "Hypertension",
+                "Heart Disease",
+                "Chronic Lung Disease",
+                "Chronic Kidney Disease",
+                None,
+            ],
+        ],
+        index=[
+            "Case ID",
+            "Age",
+            "Gender",
+            "Location",
+            "Date of Onset",
+            "Symptoms",
+            "Outcome",
+            "Vaccination Status",
+            "Days to Recovery",
+            "Underlying Conditions",
+        ],
+    )
+    pd.testing.assert_series_equal(mapper.target_values, target_vals)
+
+
 def test_common_values():
     mapper = ANIMAL_MAPPER