fix: merge histogram with range

Author: hanchchch

lavender_data/server/dataset/statistics.py

@@ -43,15 +43,23 @@ class CategoricalColumnStatistics(TypedDict):
 
 
 def _merge_histograms(hist: list[float], bin_edges: list[float]) -> Histogram:
-    _map = {}
-    for v, bin_edge in zip(hist, bin_edges):
-        _map[bin_edge] = _map.get(bin_edge, 0) + v
-
     _restored_values = []
-    _bin_edges = sorted(_map.keys())
-    for _value in _bin_edges:
-        h = _map[_value]
-        _restored_values.extend([_value] * int(h))
+    for i in range(len(hist)):
+        _min = bin_edges[i]
+        _max = bin_edges[i + 1]
+        _count = hist[i]
+        if _count == 0:
+            continue
+        elif _count == 1:
+            if i == len(hist) - 1:
+                _restored_values.append(_max)
+            else:
+                _restored_values.append(_min)
+        else:
+            _restored_values.append(_min)
+            _gap = (_max - _min) / (_count - 1)
+            _restored_values.extend([_min + j * _gap for j in range(1, _count - 1)])
+            _restored_values.append(_max)
 
     return get_outlier_aware_hist(_restored_values)
 

lavender_data/server/routes/datasets.py

@@ -138,9 +138,11 @@ def get_dataset_preview(
         raise HTTPException(status_code=404, detail="Dataset not found")
 
     if cache.exists(f"preview:{preview_id}:error"):
+        error = cache.get(f"preview:{preview_id}:error").decode()
+        cache.delete(f"preview:{preview_id}:error")
         raise HTTPException(
             status_code=500,
-            detail=cache.get(f"preview:{preview_id}:error").decode(),
+            detail=error,
         )
 
     if not cache.exists(f"preview:{preview_id}"):