Checks for negative entries in close and normalise functions of codata.py

.gitignore

@@ -2,6 +2,7 @@
 __pycache__/
 *.py[cod]
 *$py.class
+.DS_Store
 
 # C extensions
 *.so

pyrolite/comp/codata.py

@@ -1,3 +1,4 @@
+import warnings
 import numpy as np
 import pandas as pd
 import scipy.special
@@ -34,15 +35,21 @@ def close(X: np.ndarray, sumf=np.sum):
 
     Notes
     ------
-        * Does not check for non-positive entries.
+    Checks for non-positive entries and replaces zeros with NaN values.
     """
 
+    if np.any(X <= 0):
+        warnings.warn("Non-positive entries found. Closure operation assumes all positive entries.", UserWarning)
+
     if X.ndim == 2:
-        C = np.array(sumf(X, axis=1))[:, np.newaxis]
+        C = np.array(sumf(X, axis=1), dtype=float)[:, np.newaxis]
     else:
-        C = np.array(sumf(X, axis=0))
+        C = np.array(sumf(X), dtype=float)
 
+    # Replace zero sums with NaN to prevent division by zero
     C[np.isclose(C, 0)] = np.nan
+
+    # Return the array closed to sum to 1
     return np.divide(X, C)
 
 
@@ -66,16 +73,27 @@ def renormalise(df: pd.DataFrame, components: list = [], scale=100.0):
     :class:`pandas.DataFrame`
         Renormalized dataframe.
     """
+
     dfc = df.copy(deep=True)
     if components:
-        cmpnts = [c for c in components if c in dfc.columns]
-        dfc.loc[:, cmpnts] = scale * dfc.loc[:, cmpnts].divide(
-            dfc.loc[:, cmpnts].sum(axis=1).replace(0, np.nan), axis=0
-        )
-        return dfc
-    else:
-        dfc = dfc.divide(dfc.sum(axis=1).replace(0, 100.0), axis=0) * scale
-        return dfc
+        if not all(col in dfc.columns for col in components):
+            raise ValueError("Not all specified components exist in the DataFrame.")
+        dfc = dfc[components]
+
+    # Replace negative values with NaN
+    dfc[dfc < 0] = np.nan
+
+    if (dfc <= 0).any().any():
+        warnings.warn("Non-positive entries found in specified components. "
+                      "Negative values have been replaced with NaN. "
+                      "Renormalisation assumes all positive entries.", UserWarning)
+
+    # Renormalise all columns if no components are specified
+    sum_rows = dfc.sum(axis=1)
+    sum_rows.replace(0, np.nan, inplace=True)  # Handle division by zero by replacing zeros with NaN
+    dfc = dfc.divide(sum_rows, axis=0) * scale
+
+    return dfc
 
 
 def ALR(X: np.ndarray, ind: int = -1, null_col=False):