adds expanded lookup to api and tests

domain-cc/cc-app/src/python_src/api.py

@@ -20,7 +20,7 @@
 from .util.lookup_table import ContentionTextLookupTable, DiagnosticCodeLookupTable
 from .util.sanitizer import sanitize_log
 
-exanded_lookup_table = ExpandedLookupTable(
+expanded_lookup_table = ExpandedLookupTable(
     key_text=FILE_READ_HELPER["contention_text"],
     classification_code=FILE_READ_HELPER["classification_code"],
     classification_name=FILE_READ_HELPER["classification_name"],
@@ -242,6 +242,55 @@ def va_gov_claim_classifier(claim: VaGovClaim) -> ClassifierResponse:
     return response
 
 
+def get_expanded_classification(contention: Contention) -> Tuple[int, str]:
+    """
+    Performs the dictionary lookup for the expanded lookup table
+    """
+    classification_code = None
+    classification_name = None
+    if contention.contention_type == "INCREASE":
+        classification = dc_lookup_table.get(contention.diagnostic_code)
+        classification_code = classification["classification_code"]
+        classification_name = classification["classification_name"]
+
+    if contention.contention_text and not classification_code:
+        classification = expanded_lookup_table.get(contention.contention_text)
+        classification_code = classification["classification_code"]
+        classification_name = classification["classification_name"]
+
+    return classification_code, classification_name
+
+
+def classify_contention_expanded_table(
+    contention: Contention, claim: VaGovClaim
+) -> ClassifiedContention:
+    classification_code, classification_name = get_expanded_classification(contention)
+
+    response = ClassifiedContention(
+        classification_code=classification_code,
+        classification_name=classification_name,
+        diagnostic_code=contention.diagnostic_code,
+        contention_type=contention.contention_type,
+    )
+
+    return response
+
+
 @app.post("/expanded-contention-classification")
-def expanded_classifications():
-    return {"length_of_lookup_table": len(exanded_lookup_table)}
+def expanded_classifications(claim: VaGovClaim) -> ClassifierResponse:
+    classified_contentions = []
+    for contention in claim.contentions:
+        classification = classify_contention_expanded_table(contention, claim)
+        classified_contentions.append(classification)
+
+    num_classified = len([c for c in classified_contentions if c.classification_code])
+
+    response = ClassifierResponse(
+        contentions=classified_contentions,
+        claim_id=claim.claim_id,
+        form526_submission_id=claim.form526_submission_id,
+        is_fully_classified=num_classified == len(classified_contentions),
+        num_processed_contentions=len(classified_contentions),
+        num_classified_contentions=num_classified,
+    )
+    return response

domain-cc/cc-app/src/python_src/util/expanded_lookup_config.py

@@ -108,7 +108,7 @@
     "than",
     "with",
     # 'upper',
-    #'low',
+    # 'low',
     "a",
     "va",
     "for",

domain-cc/cc-app/src/python_src/util/expanded_lookup_table.py

@@ -29,7 +29,8 @@ def __init__(
 
     def _musculoskeletal_lookup(self):
         """
-        Creates a lookup table for musculoskeletal conditions with the key a frozenset
+        Creates a lookup table for musculoskeletal conditions with the key a frozenset.
+        The musculoskeletal classifications are stored in the config file and can be added/updated there.
         """
         MUSCULOSKELETAL_LUT_LUT_SET = {}
         for k, v in MUSCULOSKELETAL_LUT.items():
@@ -67,7 +68,7 @@ def _remove_common_words(self, text, common_words: List[str] = COMMON_WORDS):
 
     def _remove_numbers_single_characters(self, text):
         """
-        Removes any non-letter character not already removed from the word and any single character (ex: r)
+        Removes numbers or single character letters
         """
         regex = r"\b[a-zA-Z]{1}\b|\d"
         text = re.sub(regex, " ", text)
@@ -78,8 +79,6 @@ def _removal_pipeline(self, text):
         """
         Pipeline to remove all unwanted characters from the lookup table contention text values
         """
-        # Needs remove punctuation twice to make sure that 's are not removed and then again to ensure that double spaces
-        # are removed. This can definitely be cleaned up later.
         text = self._remove_punctuation(text)
         text = self._remove_numbers_single_characters(text)
         text = self._remove_common_words(text)
@@ -134,8 +133,8 @@ def get(self, input_str: str, default_value=LUT_DEFAULT_VALUE):
         This also process the parenthetical terms in the mappings
         """
         # There is only one case of using due to in the mappings (need to figure out a better way than hard coding it)
-
-        if input_str.strip().lower() == "loss of teeth due to bone loss":
+        input_str = input_str.strip().lower()
+        if input_str == "loss of teeth due to bone loss":
             return {
                 "classification_code": 8967,
                 "classification_name": "Dental and Oral",

domain-cc/cc-app/tests/test_data.py

@@ -1,3 +1,5 @@
+from src.python_src.util.expanded_lookup_config import FILE_READ_HELPER
+from src.python_src.util.expanded_lookup_table import ExpandedLookupTable
 from src.python_src.util.lookup_table import (
     ContentionTextLookupTable,
     DiagnosticCodeLookupTable,
@@ -15,3 +17,12 @@ def test_build_dropdown_lookup_table():
 def test_build_dc_lut():
     dc_lookup_table = DiagnosticCodeLookupTable()
     assert len(dc_lookup_table) == DIAGNOSTIC_CODE_LUT_SIZE
+
+
+def test_build_expanded_table():
+    expanded = ExpandedLookupTable(
+        FILE_READ_HELPER["contention_text"],
+        FILE_READ_HELPER["classification_code"],
+        FILE_READ_HELPER["classification_name"],
+    )
+    assert len(expanded.contention_text_lookup_table) == 1017

domain-cc/cc-app/tests/test_expanded_lookup.py

@@ -155,6 +155,90 @@ def test_lookup_cause_included():
 
 
 def test_api_endpoint(client: TestClient):
-    response = client.post("/expanded-contention-classification")
+    json_post_data = {
+        "claim_id": 100,
+        "form526_submission_id": 500,
+        "contentions": [
+            {
+                "contention_text": "PTSD (post-traumatic stress disorder)",
+                "contention_type": "NEW",
+            },
+            {
+                "contention_text": "",
+                "contention_type": "INCREASE",
+                "diagnostic_code": 5012,
+            },
+            {"contention_text": "anxiety condition", "contention_type": "new"},
+            {"contention_text": "acl tear", "contention_type": "new"},
+            {"contention_text": "totally free text", "contention_type": "new"},
+            {"contention_text": "neuropathy in my hand", "contention_type": "new"},
+            {
+                "contention_text": "knee pain due to something else entirely",
+                "contention_type": "new",
+            },
+            {
+                "contention_text": "left ankle condition",
+                "contention_type": "increase",
+                "diagnostic_code": 1000,
+            },
+        ],
+    }
+    response = client.post("/expanded-contention-classification", json=json_post_data)
     assert response.status_code == 200
-    assert response.json() == {"length_of_lookup_table": 1017}
+    assert response.json() == {
+        "contentions": [
+            {
+                "classification_code": 8989,
+                "classification_name": "Mental Disorders",
+                "diagnostic_code": None,
+                "contention_type": "NEW",
+            },
+            {
+                "classification_code": 8940,
+                "classification_name": "Cancer - Musculoskeletal - Other",
+                "diagnostic_code": 5012,
+                "contention_type": "INCREASE",
+            },
+            {
+                "classification_code": 8989,
+                "classification_name": "Mental Disorders",
+                "diagnostic_code": None,
+                "contention_type": "new",
+            },
+            {
+                "classification_code": 8997,
+                "classification_name": "Musculoskeletal - Knee",
+                "diagnostic_code": None,
+                "contention_type": "new",
+            },
+            {
+                "classification_code": None,
+                "classification_name": None,
+                "diagnostic_code": None,
+                "contention_type": "new",
+            },
+            {
+                "classification_code": None,
+                "classification_name": None,
+                "diagnostic_code": None,
+                "contention_type": "new",
+            },
+            {
+                "classification_code": 8997,
+                "classification_name": "Musculoskeletal - Knee",
+                "diagnostic_code": None,
+                "contention_type": "new",
+            },
+            {
+                "classification_code": 8991,
+                "classification_name": "Musculoskeletal - Ankle",
+                "diagnostic_code": 1000,
+                "contention_type": "increase",
+            },
+        ],
+        "claim_id": 100,
+        "form526_submission_id": 500,
+        "is_fully_classified": False,
+        "num_processed_contentions": 8,
+        "num_classified_contentions": 6,
+    }