fix tests and serialization

lilac/data/dataset.py

@@ -20,6 +20,7 @@
   StrictFloat,
   StrictInt,
   StrictStr,
+  field_serializer,
   field_validator,
 )
 from pydantic import Field as PydanticField
@@ -115,11 +116,16 @@ class StatsResult(BaseModel):
   # Defined for numeric features.
   min_val: Optional[Union[float, date, datetime]] = None
   max_val: Optional[Union[float, date, datetime]] = None
-  value_samples: Optional[list[float]] = None  # Used for approximating histogram bins
+  value_samples: list[float] = PydanticField(default=[], exclude=True, repr=False) # Samples for calculating histogram bins.
 
   # Defined for text features.
   avg_text_length: Optional[float] = None
 
+  def __eq__(self, other: object) -> bool:
+    if not isinstance(other, StatsResult):
+      return NotImplemented
+    return self.model_dump(exclude={'value_samples'}) == other.model_dump(exclude={'value_samples'})
+
 
 class MediaResult(BaseModel):
   """The result of a media() query."""

lilac/data/dataset_duckdb.py

@@ -3943,12 +3943,9 @@ def _round(value: float) -> float:
     return round(value, round_ndigits)
 
   num_bins = min(int(np.log2(stats.total_count)), MAX_AUTO_BINS)
-  print(stats.value_samples)
   transformer = PowerTransformer().fit(np.array(stats.value_samples).reshape(-1, 1))
-  print(transformer.transform(np.array(stats.value_samples).reshape(-1, 1)).ravel())
   # [-2, 2], assuming post-transform normal distribution, should cover central 95% of the data.
   buckets = transformer.inverse_transform(np.linspace(-2.5, 2.5, num_bins).reshape(-1, 1)).ravel()
-  print(buckets)
   # Sometimes the autogenerated buckets round to the same value.
   # Sometimes PowerTransformer returns NaN for some unusually shaped distributions.
   buckets = sorted(set(_round(val) for val in buckets if not np.isnan(val)))

web/blueprint/src/lib/components/schemaView/Histogram.svelte

@@ -28,7 +28,7 @@
     } else if (end == null) {
       return `≥ ${formatValue(start)}`;
     } else {
-      return `${formatValue(start)} .. ${formatValue(end)}`;
+      return `[${formatValue(start)} .. ${formatValue(end)})`;
     }
   }
   const dispatch = createEventDispatcher();

web/lib/fastapi_client/models/StatsResult.ts

@@ -12,7 +12,7 @@ export type StatsResult = {
     approx_count_distinct: number;
     min_val?: (number | string | null);
     max_val?: (number | string | null);
-    value_samples?: (Array<number> | null);
+    value_samples?: Array<number>;
     avg_text_length?: (number | null);
 };