Replace print statements with logging messages

pyhctsa/calculator.py

@@ -4,6 +4,7 @@
 from itertools import product
 from pathlib import Path
 from typing import Union, Any, Callable
+import logging
 
 import numpy as np
 import pandas as pd
@@ -38,8 +39,8 @@ def classify_output(res) -> int:
         out = 4
     return out
 
-def _apply_selection_wrapper(func:Callable, filter_keys:Union[str, list[str]],
-                             keep:bool=True) -> Callable:
+def _apply_selection_wrapper(func: Callable, filter_keys: Union[str, list[str]],
+                             keep: bool = True) -> Callable:
     """
     Wraps a function to selectively filter keys from its dict output.
 
@@ -69,7 +70,7 @@ def wrapper(*args, **kwargs):
         if isinstance(result, dict):
             missing = [k for k in keys if k not in result] # log all of the missing keys
             if missing:
-                print(f'Warning: time-series features for func {func} not found {missing}')
+                logging.info(f'Warning: time-series features for func {func} not found {missing}')
             if keep:
                 return {k: result[k] for k in keys if k in result}
             else:
@@ -86,7 +87,7 @@ def _standardise_inputs(data) -> list[np.ndarray]:
         elif data.ndim == 2:
             if data.shape[0] > data.shape[1]:
                 # notify the user to check that the shapes make sense
-                print(f"Check that the shape of the 2D input is such "
+                logging.warning(f"Check that the shape of the 2D input is such "
                       f"that (n_series, n_samples). Got shape: {data.shape}")
             return [np.asarray(row, dtype=float) for row in data]
         else:
@@ -216,7 +217,7 @@ def _check_deps(self, module_key, feature_name, config):
         missing = [dep for dep in deps_to_check if not _check_optional_deps(dep)]
         if missing:
             full_name = f"{module_key}.{feature_name}"
-            print(f"Skipping function '{full_name}' - missing dependencies: {', '.join(missing)}")
+            logging.info(f"Skipping function '{full_name}' - missing dependencies: {', '.join(missing)}")
             self._skipped_functions.append((full_name, missing))
             return False
         return True
@@ -229,7 +230,7 @@ def _build_feature_funcs(self):
             try:
                 module = importlib.import_module(f"{self._operations_package}.{module_key}")
             except ImportError as e:
-                print(f"Failed to import module '{module_key}': {e}")
+                logging.warning(f"Failed to import module '{module_key}': {e}")
                 # Skip all functions in this module since we can't import it
                 for feature_name in self.config[module_key].keys():
                     skipped_functions.append((f"{module_key}.{feature_name}", ["import_error"]))
@@ -273,7 +274,7 @@ def _build_feature_funcs(self):
         # store information about skipped functions for later reference
         self._skipped_functions = skipped_functions
         if skipped_functions:
-            print(f"Total functions skipped due to missing dependencies: {len(skipped_functions)}")
+            logging.info(f"Total functions skipped due to missing dependencies: {len(skipped_functions)}")
 
         return feature_funcs
 

pyhctsa/distribute.py

@@ -60,7 +60,7 @@ class LocalDistributor(BaseDistributor):
         Number of worker processes to use. If ``None``, defaults to
         ``multiprocessing.cpu_count()``.
     """
-    def __init__(self, n_workers : Union[int, None] = None):
+    def __init__(self, n_workers: Union[int, None] = None):
         self.n_workers = n_workers or mp.cpu_count()
         try:
             pathos_mp.set_start_method('spawn', force=True)

pyhctsa/operations/correlation.py

@@ -157,7 +157,7 @@ def add_noise(y: ArrayLike, tau: Union[int, str] = 1, ami_method: str = 'even',
 
     return out
 
-def first_under_fn(x : ArrayLike, m : ArrayLike, p : ArrayLike) -> float:
+def first_under_fn(x: ArrayLike, m: ArrayLike, p: ArrayLike) -> float:
     """
     Find the value of m for the first time p goes under the threshold, x. 
     p and m are vectors of the same length
@@ -916,7 +916,7 @@ def stick_angles(y: ArrayLike) -> dict:
 
     return out
 
-def _sub_statav(x: ArrayLike, n : int) -> tuple:
+def _sub_statav(x: ArrayLike, n: int) -> tuple:
     # helper function
     nn = len(x)
     if nn < 2 * n: # not long enough
@@ -1280,7 +1280,7 @@ def embed2_shapes(y: ArrayLike, tau: Union[str, int, None] = 'tau',
     counts -= 1 # ignore self counts
 
     if np.all(counts == 0):
-        print("No counts detected!")
+        logging.warning("embed2_shapes: no counts detected!")
         return np.nan
 
     # Return basic statistics on the counts
@@ -1535,7 +1535,7 @@ def acf_y(t):
         for i, t in enumerate(tau):
             if np.any(np.isnan(y)):
                 good_r = (~np.isnan(y[:N-t])) & (~np.isnan(y[t:]))
-                print(f'NaNs in time series, computing for {np.sum(good_r)}/{len(good_r)} pairs of points')
+                logging.info(f'NaNs in time series, computing for {np.sum(good_r)}/{len(good_r)} pairs of points.')
                 y1 = y[:N-t]
                 y1n = y1[good_r] - np.mean(y1[good_r])
                 y2 = y[t:]
@@ -1719,7 +1719,7 @@ def _stat_av(y: ArrayLike, window_stat: str = 'mean', num_seg: int = 5, inc_move
         logging.warning(f"Time-series of length {len(y)} is too short for {num_seg} windows")
         return np.nan
     inc = np.floor(win_length/inc_move) # increment to move at each step
-    # if incrment rounded down to zero, prop it up
+    # if increment rounded down to zero, prop it up
     if inc == 0:
         inc = 1
 
@@ -1829,7 +1829,7 @@ def autocorr_shape(y: ArrayLike, stop_when: Union[int, str] = 'pos_drown') -> di
 
     # Check for good behavior
     if np.any(np.isnan(acf)):
-        # This is an anomalous time series (e.g., all constant, or conatining NaNs)
+        # This is an anomalous time series (e.g., all constant, or containing NaNs)
         out = np.nan
 
     out = {}

pyhctsa/operations/distribution.py

@@ -73,11 +73,11 @@ def compare_ks_fit(x: ArrayLike, what_distn: str) -> dict:
     elif what_distn == 'exp':
         # Check positivity
         if np.any(x < 0):
-            print("The data contains negative values, but Exponential is a positive-only distribution.")
+            logging.warning("The data contains negative values, but Exponential is a positive-only distribution.")
             return np.nan
         # Check constant
         if np.all(x == x[0]):
-            print("Data are a constant")
+            logging.warning("Data are a constant.")
             return np.nan
         # Fit Exponential distribution (equivalent to expfit in MATLAB)
         _, lam = expon.fit(x, floc=0)  # force support at 0
@@ -91,7 +91,7 @@ def compare_ks_fit(x: ArrayLike, what_distn: str) -> dict:
     elif what_distn == 'logn':
         # Check positivity
         if np.any(x <= 0):
-            print("The data are not positive, but Log-Normal is a positive-only distribution.")
+            logging.warning("The data are not positive, but Log-Normal is a positive-only distribution.")
             return np.nan
         # Fit log-normal distribution
         shape, loc, scale = lognorm.fit(x, floc=0)  # sigma, 0, exp(mu)
@@ -107,7 +107,7 @@ def compare_ks_fit(x: ArrayLike, what_distn: str) -> dict:
         ffit_func = lambda xi: lognorm.pdf(xi, s=sigma, loc=0, scale=np.exp(mu))
 
     else:
-        raise ValueError(f"Unknown distribution:  {what_distn}.")
+        raise ValueError(f"Unknown distribution: {what_distn}.")
 
     # ----------------------------
     # Estimate smoothed empirical distribution

pyhctsa/operations/entropy.py

@@ -1,5 +1,6 @@
 from math import factorial
 from typing import Optional, Union
+import logging
 
 import numpy as np
 from numpy.typing import ArrayLike
@@ -265,7 +266,7 @@ def multi_scale_entropy(
             pp_text = f"after {pre_process_how} pre-processing"
         else:
             pp_text = ""
-        print(f"Warning: Not enough samples ({len(y)} {pp_text}) to compute SampEn at multiple scales")
+        logging.warning(f"Not enough samples ({len(y)} {pp_text}) to compute sample entropy at multiple scales")
         return {'out': np.nan}
 
     # Output raw values

pyhctsa/operations/graph.py

@@ -3,6 +3,7 @@
 import scipy
 from scipy.stats import expon, norm
 from ts2vg import NaturalVG
+import logging
 
 from pyhctsa.operations.correlation import autocorr, first_crossing
 from pyhctsa.operations.entropy import distribution_entropy
@@ -87,7 +88,7 @@ def visibility_graph(y: ArrayLike, meth: str = 'horiz', max_l: int = 5000) -> di
     N = len(y)
     if N > max_l:
         # too long to store in memory
-        print(f"Time series ({N} > {max_l}) is too long for visibility graph."
+        logging.info(f"Time series ({N} > {max_l}) is too long for visibility graph."
               f"Analyzing the first {max_l} samples.")
         y = y[:max_l]
         N = len(y)

pyhctsa/operations/information.py

@@ -189,7 +189,7 @@ def _mi_bin(v1: ArrayLike, v2: ArrayLike, r1: Union[str, list] = 'range',
     if np.any(mask):
         mi = np.sum(p_ij[mask] * np.log(p_ij[mask] / p_ixp_j[mask]))
     else:
-        print("The histograms aren't catching any points. Perhaps due to an inappropriate custom range for binning the data.")
+        logging.warning("The histograms aren't catching any points. Perhaps due to an inappropriate custom range for binning the data.")
         mi = np.nan
 
     return mi
@@ -403,7 +403,7 @@ def automutual_info(
     for k, delay in enumerate(time_delay):
         # check enough samples to compute automutual info
         if delay > n - min_samples:
-            # time sereis too short - keep the remaining values as NaNs
+            # time series too short - keep the remaining values as NaNs
             break
 
         # form the time-delay vectors y1 and y2
@@ -424,8 +424,8 @@ def automutual_info(
             amis[k] = mi_calc.computeAverageLocalOfObservations()
 
     if np.isnan(amis).any():
-        print(
-            f"Warning: Time series (n={n}) is too short for automutual information calculations "
+        logging.warning(
+            f"Time series (n={n}) is too short for automutual information calculations "
             f"up to lags of {max(time_delay)}"
         )
 
@@ -457,9 +457,9 @@ def mutual_info(
 
     Parameters
     ----------
-    y1 : ArrayLike
+    y1 : array-like
         First input time series.
-    y2 : ArrayLike
+    y2 : array-like
         Second input time series.
     est_method : str, optional
         Estimation method to use:
@@ -679,7 +679,7 @@ def _rm_info(*args):
         return
 
     # some initial tests on the input arguments
-    x = np.array(args[0])  # make sure the imputs are in numpy array form
+    x = np.array(args[0])  # make sure the inputs are in numpy array form
     y = np.array(args[1])
 
     x_shape = x.shape
@@ -689,23 +689,23 @@ def _rm_info(*args):
     len_y = y_shape[0]
 
     if len(x_shape) != 1:  # makes sure x is a row vector
-        print("Error: invalid dimension of x")
+        logging.warning("Invalid dimension of x")
         return
 
     if len(y_shape) != 1:
-        print("Error: invalid dimension of y")
+        logging.warning("Invalid dimension of y")
         return
 
     if len_x != len_y:  # makes sure x and y have the same amount of elements
-        print("Error: unequal length of x and y")
+        logging.warning("Unequal length of x and y")
         return
 
     if n_args > 5:
-        print("Error: too many arguments")
+        logging.warning("Too many arguments")
         return
 
     if n_args < 2:
-        print("Error: not enough arguments")
+        logging.warning("Not enough arguments")
         return
 
     # setting up variables depending on amount of inputs
@@ -869,19 +869,19 @@ def _rm_histogram_2(*args):
     leny = yshape[0]
 
     if len(xshape) != 1:  # makes sure x is a row vector
-        print("Error: invalid dimension of x")
+        logging.warning("Invalid dimension of x")
         return
 
     if len(yshape) != 1:
-        print("Error: invalid dimension of y")
+        logging.warning("Invalid dimension of y")
         return
 
     if lenx != leny:  # makes sure x and y have the same amount of elements
-        print("Error: unequal length of x and y")
+        logging.warning("Unequal length of x and y")
         return
 
     if nargin > 3:
-        print("Error: too many arguments")
+        logging.warning("Too many arguments")
         return
 
     if nargin == 2:
@@ -909,16 +909,16 @@ def _rm_histogram_2(*args):
     # checking descriptor to make sure it is valid, otherwise print an error
 
     if ncellx < 1:
-        print("Error: invalid number of cells in X dimension")
+        logging.warning("Invalid number of cells in X dimension")
 
     if ncelly < 1:
-        print("Error: invalid number of cells in Y dimension")
+        logging.warning("Invalid number of cells in Y dimension")
 
     if upperx <= lowerx:
-        print("Error: invalid bounds in X dimension")
+        logging.warning("Invalid bounds in X dimension")
 
     if uppery <= lowery:
-        print("Error: invalid bounds in Y dimension")
+        logging.warning("Invalid bounds in Y dimension")
 
     result = np.zeros([int(ncellx), int(ncelly)],
                       dtype=int)  # should do the same thing as matlab: result(1:ncellx,1:ncelly) = 0;

pyhctsa/operations/medical.py

@@ -271,7 +271,6 @@ def pol_var(x: ArrayLike, d: float = 1, D: int = 6) -> float:
     i = 0
     pc = 0
 
-    # seqcnt = 0
     while i <= (N-D):
         x_seq = x_sym[i:(i+D)]
         if np.array_equal(x_seq, z_seq) or np.array_equal(x_seq, o_seq):