refactor: streamline histogram processing and enhance readability in backend and test files

Author: ElouenGinat

src/khisto/core/backend.py

@@ -176,32 +176,18 @@ def _process_histogram_files(temp_dir: str, base_name: str) -> list[HistogramRes
         khisto_output: _KhistoOutput = _KhistoOutput.from_dict(json.load(f))
 
     histogram_series = khisto_output.histogramSeries
-    n = len(histogram_series.histograms)
     best_idx = histogram_series.interpretableHistogramNumber - 1
-    granularities = (
-        histogram_series.granularities
-        if len(histogram_series.granularities) == n
-        else list(range(n))
-    )
-
-    def _get(lst: list, i: int):
-        if i < len(lst):
-            return lst[i]
-        else:
-            raise ValueError(
-                f"Expected at least {i + 1} values in list, got {len(lst)}"
-            )
 
     return [
         _to_result(
             h,
             is_best=(i == best_idx),
-            granularity=granularities[i],
-            level=_get(histogram_series.levels, i),
-            information_rate=_get(histogram_series.informationRates, i),
-            peak_interval_number=_get(histogram_series.peakIntervalNumbers, i),
-            spike_interval_number=_get(histogram_series.spikeIntervalNumbers, i),
-            empty_interval_number=_get(histogram_series.emptyIntervalNumbers, i),
+            granularity=histogram_series.granularities[i],
+            level=histogram_series.levels[i],
+            information_rate=histogram_series.informationRates[i],
+            peak_interval_number=histogram_series.peakIntervalNumbers[i],
+            spike_interval_number=histogram_series.spikeIntervalNumbers[i],
+            empty_interval_number=histogram_series.emptyIntervalNumbers[i],
         )
         for i, h in enumerate(histogram_series.histograms)
     ]

src/khisto/matplotlib/hist.py

@@ -14,10 +14,9 @@
 from matplotlib.axes import Axes
 
 from khisto.array import histogram as khisto_histogram
-from khisto.array.histogram.api import _apply_cumulative
 
 if TYPE_CHECKING:
-    from numpy.typing import ArrayLike
+    from numpy.typing import ArrayLike, NDArray
 
 
 def _normalize_cumulative(cumulative: bool | float) -> int:
@@ -33,6 +32,24 @@ def _normalize_cumulative(cumulative: bool | float) -> int:
     raise TypeError("cumulative must be a boolean or a number")
 
 
+def _apply_cumulative(
+    hist_values: NDArray[np.float64],
+    bin_edges: NDArray[np.float64],
+    *,
+    density: bool,
+    reverse: bool = False,
+) -> NDArray[np.float64]:
+    """Accumulate histogram values using matplotlib-compatible semantics."""
+    if density:
+        source_values = hist_values * np.diff(bin_edges)
+    else:
+        source_values = hist_values
+
+    if reverse:
+        return np.cumsum(source_values[::-1])[::-1]
+    return np.cumsum(source_values)
+
+
 def hist(
     x: ArrayLike,
     range: Optional[tuple[float, float]] = None,

tests/core/test_backend.py

@@ -274,42 +274,3 @@ def test_series_keeps_finest_histogram_when_all_interpretable(self, tmp_path):
 
         assert len(results) == 2
         assert [result.is_best for result in results] == [False, True]
-
-    def test_best_histogram_is_fallback_without_series(self, tmp_path):
-        """Test that bestHistogram remains the fallback when the series is absent."""
-        payload = {
-            "tool": "Khiops Histogram",
-            "version": "1.1",
-            "bestHistogram": _histogram_payload([0.0, 1.0], [1.0, 3.0], [1, 2]),
-        }
-        self._write_json(tmp_path / "histogram.series.json", payload)
-
-        results = _process_histogram_files(str(tmp_path), "histogram.series")
-
-        assert len(results) == 1
-        assert results[0].is_best is True
-        assert results[0].granularity == 0
-
-    def test_best_histogram_is_fallback_when_series_best_index_is_missing(
-        self, tmp_path
-    ):
-        """Test that bestHistogram matching is used when series metadata is incomplete."""
-        payload = {
-            "tool": "Khiops Histogram",
-            "version": "1.1",
-            "bestHistogram": _histogram_payload([0.0, 1.0], [1.0, 3.0], [1, 2]),
-            "histogramSeries": {
-                "histogramNumber": 2,
-                "granularities": [0, 2],
-                "histograms": [
-                    _histogram_payload([0.0], [3.0], [3]),
-                    _histogram_payload([0.0, 1.0], [1.0, 3.0], [1, 2]),
-                ],
-            },
-        }
-        self._write_json(tmp_path / "histogram.series.json", payload)
-
-        results = _process_histogram_files(str(tmp_path), "histogram.series")
-
-        assert len(results) == 2
-        assert [result.is_best for result in results] == [False, True]