Updated to version 1.9.0

pyproject.toml

@@ -19,7 +19,7 @@ style = "pep440"
 [tool.poetry.group.dev.dependencies]
 wheel = "^0.46.3"
 types-toml = "^0.10.8.20240310"
-pytest = "^9.0.2"
+pytest = "^9.0.3"
 mypy = "^1.19.1"
 flake8 = "^7.3.0"
 toml = "^0.10.2"
@@ -41,7 +41,7 @@ matplotlib = [
 	{ version = ">=3.10.0", python = ">=3.10,<3.14" },
 	{ version = ">=3.10.5", python = ">=3.10,<3.15" },
 ]
-pillow = ">=12.1.1"
+pillow = ">=12.2.0"
 pyyaml = ">=6.0.0"
 
 [tool.poetry.scripts]

src/signaloid/circuitpython/extended_ulab_numpy.py

@@ -261,5 +261,59 @@ def insert(self, arr1, index, arr2):
 
         return res
 
+    def histogram(self, a, bins, weights=None):
+        """
+        Compute the histogram of a dataset.
+
+        Mirrors the subset of ``numpy.histogram`` semantics required by
+        this project: each value is placed in the bin where
+        ``bins[i] <= v < bins[i+1]``, with the last bin closed on the
+        right (``bins[-2] <= v <= bins[-1]``). Values outside
+        ``[bins[0], bins[-1]]`` are excluded. ``bins`` must be a
+        monotonically increasing sequence of bin edges.
+
+        :param a: The input values.
+        :param bins: A monotonically increasing sequence of bin edges
+            (length ``n_bins + 1``).
+        :param weights: Optional per-element weights. If ``None``, each
+            value contributes 1.0 to its bin.
+
+        :return: A tuple ``(hist, bin_edges)`` where ``hist`` has length
+            ``n_bins`` and ``bin_edges`` is the input ``bins`` as an array.
+        """
+
+        edges = self.array(bins)
+        n_bins = len(edges) - 1
+        if n_bins < 1:
+            raise ValueError("histogram requires at least 2 bin edges")
+
+        n_values = len(a)
+        if weights is None:
+            weights = [1.0] * n_values
+
+        hist = [0.0] * n_bins
+        first_edge = edges[0]
+        last_edge = edges[-1]
+
+        for i in range(n_values):
+            v = a[i]
+            if v < first_edge or v > last_edge:
+                continue
+            # bisect_right(edges, v) - 1 gives the numpy-style bin index.
+            lo, hi = 0, n_bins + 1
+            while lo < hi:
+                mid = (lo + hi) // 2
+                if edges[mid] <= v:
+                    lo = mid + 1
+                else:
+                    hi = mid
+            idx = lo - 1
+            if idx == n_bins:
+                # v == edges[-1]: the last bin is closed on the right.
+                idx = n_bins - 1
+            hist[idx] += weights[i]
+
+        return self.array(hist), edges
+
 
 np = NumpyWrapper()

src/signaloid/distributional/dirac_delta.py

@@ -18,7 +18,7 @@
 #   FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER
 #   DEALINGS IN THE SOFTWARE.
 
-from __future__ import annotations
+
 import sys
 import math
 

src/signaloid/distributional/distributional.py

@@ -18,7 +18,7 @@
 #   FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER
 #   DEALINGS IN THE SOFTWARE.
 
-from __future__ import annotations
+
 import math
 import sys
 import re

src/signaloid/distributional/distributional_test.py

@@ -19,8 +19,6 @@
 #   DEALINGS IN THE SOFTWARE.
 
 
-from __future__ import annotations
-
 import csv
 import os
 import unittest

src/signaloid/distributional_information_plotting/plot_histogram_dirac_deltas.py

@@ -18,7 +18,7 @@
 #   FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER
 #   DEALINGS IN THE SOFTWARE.
 
-from __future__ import annotations
+
 import sys
 import math
 
@@ -61,7 +61,7 @@ def from_samples(
         cls,
         samples: np.ndarray | list[float],
         plotting_resolution: int | None = None,
-    ) -> PlotData:
+    ) -> "PlotData":
         """
         Construct a PlotData from an array of float samples.
 
@@ -600,33 +600,48 @@ def _construct_plot_data(self) -> None:
                 "plot_histogram_dirac_deltas: plotting_resolution must be a power of 2!"
             )
 
-        try:
-            # Create the binning such that the average of two bins surrounding a Dirac delta
-            # is the Dirac delta itself.
-            boundary_positions, bin_widths, bin_heights = PlotData.create_binning(
-                finite_dirac_deltas, 0, False
-            )
-
-            # Find the TTR of the created binning. This is always a valid TTR.
-            ttr = PlotData._bin_pdf_to_ttr(
-                boundary_positions, bin_widths, bin_heights, self.plotting_ttr_order
-            )
-
-            # Create the binning from the obtained (valid) TTR using the TTR binning method.
-            boundary_positions, bin_widths, bin_heights = PlotData.create_binning(
-                ttr, self.plotting_ttr_order, True
-            )
+        if self.dist.check_is_full_valid_TTR():
+            try:
+                # Create the binning such that the average of two bins surrounding a Dirac delta
+                # is the Dirac delta itself.
+                boundary_positions, bin_widths, bin_heights = PlotData.create_binning(
+                    finite_dirac_deltas, 0, False
+                )
 
-            self.positions = boundary_positions
-            self.masses = bin_heights
-        except (ValueError, TypeError):
-            positions = np.array([dd.position for dd in finite_dirac_deltas])
-            masses = np.array([dd.mass for dd in finite_dirac_deltas])
+                # Find the TTR of the created binning. This is always a valid TTR.
+                ttr = PlotData._bin_pdf_to_ttr(
+                    boundary_positions,
+                    bin_widths,
+                    bin_heights,
+                    self.plotting_ttr_order,
+                )
 
-            midpoints = (positions[:-1] + positions[1:]) / 2
-            left = 2 * positions[0] - midpoints[0]
-            right = 2 * positions[-1] - midpoints[-1]
-            self.positions = np.concatenate(([left], midpoints, [right]))
+                # Create the binning from the obtained (valid) TTR using the TTR binning method.
+                boundary_positions, bin_widths, bin_heights = PlotData.create_binning(
+                    ttr, self.plotting_ttr_order, True
+                )
 
-            widths = np.diff(self.positions)
-            self.masses = masses / widths
+                self.positions = boundary_positions
+                self.masses = bin_heights
+                return
+            except (ValueError, TypeError):
+                pass
+
+        # Non-TTR input (or TTR pipeline failed): use a uniform-width histogram
+        # at `plotting_resolution`. The TTR binning method assumes the input
+        # forms a valid TTR, so applying it to arbitrary Dirac deltas (e.g. raw
+        # samples) produces meaningless boundaries.
+        positions = np.array([dd.position for dd in finite_dirac_deltas])
+        masses = np.array([dd.mass for dd in finite_dirac_deltas])
+        pos_min = float(positions[0])
+        pos_max = float(positions[-1])
+        edges = np.linspace(pos_min, pos_max, self.plotting_resolution + 1)
+        hist, _ = np.histogram(positions, bins=edges, weights=masses)
+        bin_widths = edges[1:] - edges[:-1]
+        self.positions = edges
+        self.masses = np.divide(
+            hist,
+            bin_widths,
+            out=np.zeros_like(hist, dtype=np.float64),
+            where=bin_widths != 0,
+        )

src/signaloid/distributional_information_plotting/plot_histogram_dirac_deltas_test.py

@@ -19,11 +19,13 @@
 #   DEALINGS IN THE SOFTWARE.
 
 
+import os
 import random
 import unittest
 
 import numpy as np
 from signaloid.distributional.dirac_delta import DiracDelta
+from signaloid.distributional.distributional import DistributionalValue
 from signaloid.distributional_information_plotting.plot_histogram_dirac_deltas import (
     PlotData,
 )
@@ -283,6 +285,65 @@ def test_plot_from_samples_with_special_values_succeeds(self) -> None:
         self.assertTrue(result)
 
 
+class TestNonTTRHistogramFallback(unittest.TestCase):
+    """Covers the uniform-width histogram fallback used when
+    `is_full_valid_TTR` is False or the TTR pipeline raises."""
+
+    @staticmethod
+    def _fixture_path(name: str) -> str:
+        here = os.path.realpath(os.path.dirname(__file__))
+        return os.path.join(here, name)
+
+    def test_invalid_ttr_ux_string_falls_back_to_uniform_histogram(self) -> None:
+        """The fixture is a Ux string that is not a valid TTR; the fallback
+        should produce a uniform-width histogram with shape matching
+        plotting_resolution."""
+        with open(self._fixture_path("invalid_ttr_ux_string.dat")) as f:
+            ux_data = f.read().strip()
+        dist = DistributionalValue.parse(ux_data)
+        self.assertIsNotNone(dist)
+        assert dist is not None
+        self.assertFalse(dist.is_full_valid_TTR)
+
+        pd = PlotData(dist)
+
+        self.assertIsNotNone(pd.plotting_resolution)
+        assert pd.plotting_resolution is not None
+        self.assertEqual(len(pd.positions), pd.plotting_resolution + 1)
+        self.assertEqual(len(pd.masses), pd.plotting_resolution)
+
+        widths = pd.positions[1:] - pd.positions[:-1]
+        self.assertTrue(np.allclose(widths, widths[0]))
+
+        # Density should integrate to ~1 over the finite-mass region.
+        total_area = float(np.sum(widths * pd.masses))
+        self.assertAlmostEqual(total_area, 1.0, places=6)
+
+    def test_clustered_repeated_positions_fallback_succeeds(self) -> None:
+        """Heavy duplication / tight clusters must not crash the fallback;
+        shape and density invariants still hold."""
+        samples = np.concatenate(
+            [
+                np.full(200, 1.0),
+                np.full(200, 2.0),
+                np.full(200, 3.0),
+                np.full(50, 2.0 + 1e-12),
+                np.full(50, 2.0 - 1e-12),
+            ]
+        )
+        pd = PlotData.from_samples(samples)
+
+        self.assertIsNotNone(pd.plotting_resolution)
+        assert pd.plotting_resolution is not None
+        self.assertEqual(len(pd.positions), pd.plotting_resolution + 1)
+        self.assertEqual(len(pd.masses), pd.plotting_resolution)
+
+        widths = pd.positions[1:] - pd.positions[:-1]
+        self.assertTrue(np.allclose(widths, widths[0]))
+        total_area = float(np.sum(widths * pd.masses))
+        self.assertAlmostEqual(total_area, 1.0, places=6)
+
+
 def dirac_deltas_to_ttr(
     dirac_deltas: list[DiracDelta], order: int, count: int = 0
 ) -> list[DiracDelta]:

src/signaloid/distributional_information_plotting/plot_wrapper.py

@@ -18,7 +18,7 @@
 #   FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER
 #   DEALINGS IN THE SOFTWARE.
 
-from __future__ import annotations
+
 import math
 from typing import Any
 import matplotlib

src/signaloid/distributional_information_plotting/plot_wrapper_test.py

@@ -19,8 +19,6 @@
 #   DEALINGS IN THE SOFTWARE.
 
 
-from __future__ import annotations
-
 import csv
 import os
 import shutil

src/signaloid/distributional_information_plotting/sample_generator.py

@@ -19,7 +19,6 @@
 #   DEALINGS IN THE SOFTWARE.
 
 
-from __future__ import annotations
 from signaloid.distributional.distributional import DistributionalValue
 from signaloid.distributional_information_plotting.plot_histogram_dirac_deltas import (
     PlotData,

src/signaloid/uxdata_toolkit.py

@@ -48,7 +48,6 @@
 - pip install git+https://github.com/signaloid/signaloid-python
 """
 
-from __future__ import annotations
 import argparse
 import sys
 import traceback