Refactor: Modularized project into src/ layout and integrated pytest-…

pyproject.toml

@@ -0,0 +1,27 @@
+[build-system]
+requires = ["setuptools>=61.0"]
+build-backend = "setuptools.build_meta"
+
+[project]
+name = "vectfit-python"
+version = "0.1.0"
+description = "Vector Fitting for rational approximation of EMT frequency responses."
+readme = "README.md"
+requires-python = ">=3.8"
+authors = [
+    {name = "Sebastian Loaiza Elejalde", email = "author@example.com"},
+]
+dependencies = [
+    "numpy",
+    "scipy",
+    "pandas",
+]
+
+[project.optional-dependencies]
+test = [
+    "pytest",
+    "pytest-benchmark",
+]
+
+[tool.setuptools.packages.find]
+where = ["src"]

test/conftest.py

@@ -0,0 +1,71 @@
+import sys
+import pytest
+import numpy as np
+import pandas as pd
+from pathlib import Path
+
+
+# Go up one level to the root, then dig into the 'src' folder
+root_dir = Path(__file__).parent.parent
+src_dir = root_dir / "src"
+
+# Tell Python to look inside the 'src' folder for your library
+sys.path.insert(0, str(src_dir))
+
+@pytest.fixture
+def default_opts():
+    """
+    gets the default options for vectfit. 
+    """
+    # Adjust this import based on where your vectfit3.py actually lives
+    # If it's in the same root directory, this will work.
+    from vectfit3 import opts 
+    return opts.copy()
+
+@pytest.fixture
+def generate_synthetic_fs():
+    """
+    Factory fixture: Generates an F(s) frequency response.
+    Supports Scalar (1D) and Vector (2D) responses.
+    """
+    def _generate(s, expected_poles, expected_residues, D=None, E=None):
+        # Determine if we are doing scalar or vector fitting
+        # expected_residues shape: (Nc, n_poles) or (n_poles,)
+        if expected_residues.ndim == 1:
+            F = np.zeros_like(s, dtype=np.complex128)
+            for p, r in zip(expected_poles, expected_residues):
+                F += r / (s - p)
+            if D is not None: F += D
+            if E is not None: F += s * E
+        else:
+            # Vector case: Nc x N
+            Nc = expected_residues.shape[0]
+            N = len(s)
+            F = np.zeros((Nc, N), dtype=np.complex128)
+            for i in range(Nc):
+                for p, r in zip(expected_poles, expected_residues[i]):
+                    F[i, :] += r / (s - p)
+                if D is not None: F[i, :] += D[i]
+                if E is not None: F[i, :] += s * E[i]
+        return F
+    return _generate
+
+
+
+@pytest.fixture
+def load_vectfit_csv():
+    """
+    Clean CSV loader for test datasets. Expects files to be in a 'data' subfolder next to this conftest.py.
+    """
+    data_path = Path(__file__).parent / "data"
+
+    def _loader(filename):
+        file_path = data_path / filename
+        if not file_path.exists():
+            pytest.fail(f"Dataset {filename} not found in {data_path}")
+
+        # We use pandas to read, then convert to numpy for speed
+        df = pd.read_csv(file_path)
+        return df.to_numpy()
+
+    return _loader