From 81342fe8a1f660097b7bc9d7f4569940f748abea Mon Sep 17 00:00:00 2001
From: Dan LaManna <dan.lamanna@kitware.com>
Date: Wed, 15 May 2024 16:28:15 -0400
Subject: [PATCH] Enforce that RCM cases belong to at most one lesion

---
 isic_metadata/metadata.py | 20 ++++++++++++++++++++
 tests/test_batch.py       | 25 +++++++++++++++++++++++++
 2 files changed, 45 insertions(+)

diff --git a/isic_metadata/metadata.py b/isic_metadata/metadata.py
index 0e1e6e5..0dd03b3 100644
--- a/isic_metadata/metadata.py
+++ b/isic_metadata/metadata.py
@@ -164,6 +164,26 @@ def check_rcm_at_most_one_macroscopic_image(self) -> MetadataBatch:
 
         return self
 
+    @model_validator(mode="after")
+    def check_rcm_case_at_most_one_lesion(self) -> MetadataBatch:
+        rcm_case_to_lesions: dict[str, set[str]] = defaultdict(set)
+
+        for item in self.items:
+            if item.rcm_case_id and item.lesion_id:
+                rcm_case_to_lesions[item.rcm_case_id].add(item.lesion_id)
+
+        bad_rcm_cases = [
+            rcm_case for rcm_case in rcm_case_to_lesions if len(rcm_case_to_lesions[rcm_case]) > 1
+        ]
+        if bad_rcm_cases:
+            raise PydanticCustomError(
+                "one_rcm_case_multiple_lesions",
+                "One or more RCM cases belong to multiple lesions.",
+                {"examples": bad_rcm_cases[:5]},
+            )
+
+        return self
+
 
 class MetadataRow(BaseModel):
     model_config = ConfigDict(
diff --git a/tests/test_batch.py b/tests/test_batch.py
index f6337d1..466c9eb 100644
--- a/tests/test_batch.py
+++ b/tests/test_batch.py
@@ -51,3 +51,28 @@ def test_rcm_case_has_at_most_one_macroscopic_image():
             MetadataRow(image_type="RCM: macroscopic", rcm_case_id="bar"),
         ]
     )
+
+
+def test_rcm_cases_belong_to_same_lesion():
+    with pytest.raises(ValidationError) as excinfo:
+        MetadataBatch(
+            items=[
+                MetadataRow(
+                    rcm_case_id="foo", lesion_id="foolesion", _ignore_rcm_model_checks=True
+                ),
+                MetadataRow(
+                    rcm_case_id="foo", lesion_id="barlesion", _ignore_rcm_model_checks=True
+                ),
+            ]
+        )
+    assert len(excinfo.value.errors()) == 1
+    assert "belong to multiple lesions" in excinfo.value.errors()[0]["msg"]
+
+
+def test_blank_rcm_cases_dont_belong_to_same_lesion():
+    MetadataBatch(
+        items=[
+            MetadataRow(rcm_case_id="", lesion_id="foolesion"),
+            MetadataRow(rcm_case_id="", lesion_id="barlesion"),
+        ]
+    )