community/py3-niaarmts: rebuild against python 3.14 and fix tests

community/py3-niaarmts/APKBUILD

@@ -2,7 +2,7 @@
 # Maintainer: Iztok Fister, Jr. <iztok@iztok-jr-fister.eu>
 pkgname=py3-niaarmts
 pkgver=0.2.6
-pkgrel=0
+pkgrel=1
 pkgdesc="Time series numerical association rule mining variants"
 url="https://github.com/firefly-cpp/NiaARMTS"
 arch="noarch"
@@ -11,7 +11,9 @@ depends="python3 py3-matplotlib py3-niapy py3-numpy py3-pandas"
 checkdepends="py3-pytest-xdist"
 makedepends="py3-gpep517 py3-poetry-core"
 subpackages="$pkgname-doc $pkgname-pyc"
-source="https://github.com/firefly-cpp/NiaARMTS/archive/$pkgver/niaarmts-$pkgver.tar.gz"
+source="https://github.com/firefly-cpp/NiaARMTS/archive/$pkgver/niaarmts-$pkgver.tar.gz
+	pandas.patch
+	"
 builddir="$srcdir/NiaARMTS-$pkgver"
 
 build() {
@@ -37,4 +39,5 @@ package() {
 
 sha512sums="
 59f94b4791fa8fae3cc9cb2d16a0c5bea326906d714e126165497112de9b99f0fb87a7403aa09a79eacb09282916bba3b3870a9a65dd29d504bc5a92959e95cf  niaarmts-0.2.6.tar.gz
+246abac646f0a42e7e5298ee891c08d16721ffa760ef3bbc413790deda12e30d28d2976dffd17a03611877625029fd1940115e6ec9b659f9e38cc08741f7cd44  pandas.patch
 "

community/py3-niaarmts/pandas.patch

@@ -0,0 +1,105 @@
+diff --git a/niaarmts/feature.py b/niaarmts/feature.py
+index 106bdfb..e6af172 100644
+--- a/niaarmts/feature.py
++++ b/niaarmts/feature.py
+@@ -1,5 +1,10 @@
+ import pandas as pd
+-import numpy as np
++from pandas.api.types import (
++    is_datetime64_any_dtype,
++    is_numeric_dtype,
++    is_object_dtype,
++    is_string_dtype,
++)
+ 
+ class Feature:
+     def __init__(self, data: pd.DataFrame):
+@@ -22,7 +27,7 @@ class Feature:
+         for column in self.data.columns:
+             col_data = self.data[column]
+ 
+-            if np.issubdtype(col_data.dtype, np.number):
++            if is_numeric_dtype(col_data.dtype):
+                 summary[column] = {
+                     'type': 'Numerical',
+                     'min': col_data.min(),
+@@ -30,13 +35,13 @@ class Feature:
+                     'mean': col_data.mean(),
+                     'std_dev': col_data.std(),
+                 }
+-            elif col_data.dtype == 'object':
++            elif self._is_categorical(col_data):
+                 summary[column] = {
+                     'type': 'Categorical',
+                     'unique_classes': col_data.nunique(),
+                     'classes': col_data.unique()
+                 }
+-            elif np.issubdtype(col_data.dtype, np.datetime64):
++            elif is_datetime64_any_dtype(col_data.dtype):
+                 summary[column] = {
+                     'type': 'Datetime',
+                     'min': col_data.min(),
+@@ -55,7 +60,7 @@ class Feature:
+ 
+         :return: A list of numerical feature names.
+         """
+-        numerical_features = [col for col in self.data.columns if np.issubdtype(self.data[col].dtype, np.number)]
++        numerical_features = [col for col in self.data.columns if is_numeric_dtype(self.data[col].dtype)]
+         return numerical_features
+ 
+     def get_categorical_features(self):
+@@ -64,7 +69,7 @@ class Feature:
+ 
+         :return: A list of categorical feature names.
+         """
+-        categorical_features = [col for col in self.data.columns if self.data[col].dtype == 'object']
++        categorical_features = [col for col in self.data.columns if self._is_categorical(self.data[col])]
+         return categorical_features
+ 
+     def get_datetime_features(self):
+@@ -73,7 +78,7 @@ class Feature:
+ 
+         :return: A list of datetime feature names.
+         """
+-        datetime_features = [col for col in self.data.columns if np.issubdtype(self.data[col].dtype, np.datetime64)]
++        datetime_features = [col for col in self.data.columns if is_datetime64_any_dtype(self.data[col].dtype)]
+         return datetime_features
+ 
+     def get_feature_stats(self, feature_name: str):
+@@ -87,7 +92,7 @@ class Feature:
+             raise ValueError(f"Feature '{feature_name}' not found in the dataset.")
+ 
+         col_data = self.data[feature_name]
+-        if np.issubdtype(col_data.dtype, np.number):
++        if is_numeric_dtype(col_data.dtype):
+             return {
+                 'type': 'Numerical',
+                 'min': col_data.min(),
+@@ -95,13 +100,13 @@ class Feature:
+                 'mean': col_data.mean(),
+                 'std_dev': col_data.std(),
+             }
+-        elif col_data.dtype == 'object':
++        elif self._is_categorical(col_data):
+             return {
+                 'type': 'Categorical',
+                 'unique_classes': col_data.nunique(),
+                 'classes': col_data.unique(),
+             }
+-        elif np.issubdtype(col_data.dtype, np.datetime64):
++        elif is_datetime64_any_dtype(col_data.dtype):
+             return {
+                 'type': 'Datetime',
+                 'min': col_data.min(),
+@@ -111,3 +116,11 @@ class Feature:
+             return {
+                 'type': 'Unknown',
+             }
++
++    @staticmethod
++    def _is_categorical(series: pd.Series):
++        return (
++            is_string_dtype(series.dtype)
++            or is_object_dtype(series.dtype)
++            or isinstance(series.dtype, pd.CategoricalDtype)
++        ) and not is_datetime64_any_dtype(series.dtype)