Merge pull request #57 from ImageMarkup/prepare-for-parquet

Prepare metadata schema for parquet-compatible version

Author: danlamanna

isic_metadata/metadata.py

@@ -215,16 +215,26 @@ class MetadataRow(BaseModel):
     personal_hx_mm: bool | None = None
     family_hx_mm: bool | None = None
     clin_size_long_diam_mm: (
-        Annotated[Decimal, BeforeValidator(ClinSizeLongDiamMm.parse_measurement_str)] | None
+        Annotated[
+            Decimal,
+            BeforeValidator(ClinSizeLongDiamMm.parse_measurement_str),
+            Field(max_digits=5, decimal_places=2),
+        ]
+        | None
     ) = None
     fitzpatrick_skin_type: FitzpatrickSkinType | None = None
     melanocytic: bool | None = None
     concomitant_biopsy: bool | None = None
 
     mel_mitotic_index: MelMitoticIndexEnum | None = None
-    mel_thick_mm: Annotated[Decimal, BeforeValidator(MelThickMm.parse_measurement_str)] | None = (
-        None
-    )
+    mel_thick_mm: (
+        Annotated[
+            Decimal,
+            BeforeValidator(MelThickMm.parse_measurement_str),
+            Field(max_digits=5, decimal_places=2),
+        ]
+        | None
+    ) = None
     mel_ulcer: bool | None = None
 
     patient_id: str | None = None
@@ -239,7 +249,7 @@ class MetadataRow(BaseModel):
 
     rcm_case_id: str | None = None
 
-    unstructured: dict[str, Any] = {}
+    unstructured: dict[str, Any] = Field(default_factory=dict, exclude=True)
 
     # Unused and undocumented
     marker_pen: bool | None = None

tests/test_fields.py

@@ -101,8 +101,8 @@ def test_mel_thick_mm_invalid():
     clin_size=st.one_of(
         # keep max values bounded so they don't generate larger than representable decimals.
         # disallow infinity to avoid the string 'inf'.
-        st.floats(min_value=0, max_value=1_000_000, exclude_min=True, allow_infinity=False),
-        st.integers(min_value=1, max_value=1_000),
+        st.floats(min_value=0, max_value=9999, exclude_min=False, allow_infinity=False),
+        st.integers(min_value=1, max_value=9999),
     ).map(lambda x: f"{x} mm")
 )
 def test_clin_size_long_diam_mm_always_rounded(clin_size: str):