dragon-ml-toolbox 19.13.0__py3-none-any.whl → 20.0.0__py3-none-any.whl

ml_tools/optimization_tools/_optimization_plots.py

@@ -0,0 +1,218 @@
+from typing import Union, Optional
+from pathlib import Path
+import pandas as pd
+import matplotlib.pyplot as plt
+import seaborn as sns
+
+from ..utilities import yield_dataframes_from_dir
+
+from ..path_manager import sanitize_filename, make_fullpath, list_csv_paths
+from .._core import get_logger
+
+
+_LOGGER = get_logger("Optimization Plots")
+
+
+__all__ = [
+    "plot_optimal_feature_distributions",
+    "plot_optimal_feature_distributions_from_dataframe",
+]
+
+
+def plot_optimal_feature_distributions(results_dir: Union[str, Path], 
+                                       verbose: bool=False,
+                                       target_columns: Optional[list[str]] = None):
+    """
+    Analyzes optimization results and plots the distribution of optimal values.
+
+    This function is compatible with mixed-type CSVs (strings for
+    categorical features, numbers for continuous). It automatically
+    detects the data type for each feature and generates:
+    
+    - A Bar Plot for categorical (string) features.
+    - A KDE Plot for continuous (numeric) features.
+    
+    Plots are saved in a subdirectory inside the source directory.
+
+    Parameters
+    ----------
+    results_dir : str | Path
+        The path to the directory containing the optimization result CSV files.
+    target_columns (list[str] | None): 
+        A list of target column names to explicitly exclude from plotting. If None, it defaults to excluding only the last column (assumed as the target).
+    """
+    # Check results_dir and create output path
+    results_path = make_fullpath(results_dir, enforce="directory")
+    output_path = make_fullpath(results_path / "DistributionPlots", make=True)
+    
+    # Check that the directory contains csv files
+    list_csv_paths(results_path, verbose=False, raise_on_empty=True)
+
+    # --- Data Loading and Preparation ---
+    _LOGGER.debug(f"📁 Starting analysis from results in: '{results_dir}'")
+    
+    data_to_plot = []
+    for df, df_name in yield_dataframes_from_dir(results_path, verbose=True):
+        if df.shape[1] < 2:
+            _LOGGER.warning(f"Skipping '{df_name}': must have at least 2 columns (feature + target).")
+            continue
+        
+        # --- Column selection logic ---
+        if target_columns:
+            # 1. Explicitly drop known targets to isolate features
+            existing_targets = [c for c in target_columns if c in df.columns]
+            features_df = df.drop(columns=existing_targets)
+            
+            if features_df.empty:
+                _LOGGER.warning(f"Skipping '{df_name}': All columns were dropped based on target_columns list.")
+                continue
+        else:
+            # 2. Fallback: Assume the last column is the only target
+            features_df = df.iloc[:, :-1]
+        
+        # 3. Melt the filtered dataframe
+        melted_df = features_df.melt(var_name='feature', value_name='value')
+        
+        # Set target as the filename (or joined target names) to differentiate sources
+        melted_df['target'] = '\n'.join(target_columns) if target_columns else df_name
+        data_to_plot.append(melted_df)
+    
+    if not data_to_plot:
+        _LOGGER.error("No valid data to plot after processing all CSVs.")
+        return
+        
+    long_df = pd.concat(data_to_plot, ignore_index=True)
+
+    # --- Delegate to Helper ---
+    _generate_and_save_feature_plots(long_df, output_path, verbose)
+
+
+def plot_optimal_feature_distributions_from_dataframe(dataframe: pd.DataFrame,
+                                                      save_dir: Union[str, Path],
+                                                      verbose: bool=False,
+                                                      target_columns: Optional[list[str]] = None):
+    """
+    Analyzes a single dataframe of optimization results and plots the distribution of optimal values.
+
+    This function is compatible with mixed-type data (strings for categorical features, 
+    numbers for continuous). It automatically detects the data type for each feature 
+    and generates:
+    
+    - A Bar Plot for categorical (string) features.
+    - A KDE Plot for continuous (numeric) features.
+    
+    Plots are saved in a 'DistributionPlots' subdirectory inside the save_dir.
+
+    Parameters
+    ----------
+    dataframe : pd.DataFrame
+        The dataframe containing the optimization results (features + target/s).
+    save_dir : str | Path
+        The directory where the 'DistributionPlots' folder will be created.
+    verbose : bool, optional
+        If True, logs details about which plot type is chosen for each feature.
+    target_columns : list[str] | None
+        A list of target column names to explicitly exclude from plotting. 
+        If None, it defaults to excluding only the last column (assumed as the target).
+    """
+    # Check results_dir and create output path
+    root_path = make_fullpath(save_dir, make=True, enforce="directory")
+    output_path = make_fullpath(root_path / "DistributionPlots", make=True, enforce="directory")
+    
+    _LOGGER.debug(f"📁 Starting analysis from provided DataFrame. Output: '{output_path}'")
+
+    if dataframe.empty:
+        _LOGGER.error("Provided dataframe is empty.")
+        return
+
+    if dataframe.shape[1] < 2:
+        _LOGGER.warning("DataFrame has fewer than 2 columns. Expecting at least one feature and one target.")
+
+    # --- Data Preparation ---
+    if target_columns:
+        # Explicitly drop known targets to isolate features
+        existing_targets = [c for c in target_columns if c in dataframe.columns]
+        features_df = dataframe.drop(columns=existing_targets)
+        target_label = '\n'.join(target_columns)
+    else:
+        # Fallback: Assume the last column is the only target
+        features_df = dataframe.iloc[:, :-1]
+        target_label = "Optimization Result"
+
+    if features_df.empty:
+        _LOGGER.warning("Skipping plotting: All columns were dropped based on target_columns list.")
+        return
+
+    # Melt and assign static target label
+    long_df = features_df.melt(var_name='feature', value_name='value')
+    long_df['target'] = target_label
+
+    # --- Delegate to Helper ---
+    _generate_and_save_feature_plots(long_df, output_path, verbose)
+
+
+def _generate_and_save_feature_plots(long_df: pd.DataFrame, output_path: Path, verbose: bool) -> None:
+    """
+    Private helper: iterates over a melted DataFrame (columns: feature, value, target)
+    and generates/saves the appropriate plot (Bar or KDE) for each feature.
+    """
+    features = long_df['feature'].unique()
+    unique_targets = long_df['target'].unique()
+    
+    _LOGGER.info(f"📊 Found data for {len(features)} features. Generating plots...")
+
+    for feature_name in features:
+        plt.figure(figsize=(12, 7))
+        
+        # .copy() to ensure we are working with a distinct object
+        feature_df = long_df[long_df['feature'] == feature_name].copy()
+
+        # --- Type-checking logic ---
+        feature_df['numeric_value'] = pd.to_numeric(feature_df['value'], errors='coerce')
+        
+        # If *any* value failed conversion (is NaN), treat it as categorical.
+        if feature_df['numeric_value'].isna().any():
+            
+            # --- PLOT 1: CATEGORICAL (String-based) ---
+            if verbose:
+                print(f"    Plotting '{feature_name}' as categorical (bar plot).")
+            
+            # Calculate percentages for a clean bar plot
+            norm_df = (feature_df.groupby('target')['value']
+                       .value_counts(normalize=True)
+                       .mul(100)
+                       .rename('percent')
+                       .reset_index())
+            
+            ax = sns.barplot(data=norm_df, x='value', y='percent', hue='target')
+            plt.ylabel("Frequency (%)", fontsize=12)
+            ax.set_ylim(0, 100) 
+            
+            # always rotate x-ticks for categorical clarity
+            plt.xticks(rotation=45, ha='right')
+
+        else:
+            # --- PLOT 2: CONTINUOUS (Numeric-based) ---
+            if verbose:
+                print(f"    Plotting '{feature_name}' as continuous (KDE plot).")
+            
+            ax = sns.kdeplot(data=feature_df, x='numeric_value', hue='target',
+                             fill=True, alpha=0.1, warn_singular=False)
+            
+            plt.xlabel("Feature Value", fontsize=12)
+            plt.ylabel("Density", fontsize=12)
+
+        # --- Common settings for both plot types ---
+        plt.title(f"Optimal Value Distribution for '{feature_name}'", fontsize=16)
+        plt.grid(axis='y', alpha=0.5, linestyle='--')
+        
+        legend = ax.get_legend()
+        if legend:
+            legend.set_title('Target')
+
+        sanitized_feature_name = sanitize_filename(feature_name)
+        plot_filename = output_path / f"Distribution_{sanitized_feature_name}.svg"
+        plt.savefig(plot_filename, bbox_inches='tight')
+        plt.close()
+
+    _LOGGER.info(f"All plots saved successfully to: '{output_path}'")

ml_tools/{path_manager.py → path_manager/__init__.py}

@@ -1,5 +1,6 @@
-from ._core._path_manager import (
-    DragonPathManager,
+from ._dragonmanager import DragonPathManager
+
+from ._path_tools import (
     make_fullpath,
     sanitize_filename,
     list_csv_paths,
@@ -7,9 +8,11 @@ from ._core._path_manager import (
     list_subdirectories,
     clean_directory,
     safe_move,
-    info
 )
 
+from ._imprimir import info
+
+
 __all__ = [
     "DragonPathManager",
     "make_fullpath",

ml_tools/{_core/_path_manager.py → path_manager/_dragonmanager.py}

@@ -1,25 +1,17 @@
-from typing import Optional, List, Dict, Union, Literal
+from typing import Optional, Union
 from pathlib import Path
-import re
 import sys
-import shutil
 
-from ._script_info import _script_info
-from ._logger import get_logger
+from .._core import get_logger
 
+from ._path_tools import sanitize_filename
 
-_LOGGER = get_logger("Path Manager")
+
+_LOGGER = get_logger("DragonPathManager")
 
 
 __all__ = [
-    "DragonPathManager",
-    "make_fullpath",
-    "sanitize_filename",
-    "list_csv_paths",
-    "list_files_by_extension",
-    "list_subdirectories",
-    "clean_directory",
-    "safe_move",
+    "DragonPathManager"
 ]
 
 
@@ -37,7 +29,7 @@ class DragonPathManager:
     def __init__(
         self,
         anchor_file: str,
-        base_directories: Optional[List[str]] = None,
+        base_directories: Optional[list[str]] = None,
         strict_to_root: bool = True
     ):
         """
@@ -61,7 +53,7 @@ class DragonPathManager:
         resolved_anchor_path = Path(anchor_file).resolve()
         self._package_name = resolved_anchor_path.parent.name
         self._is_bundled, bundle_root = self._get_bundle_root()
-        self._paths: Dict[str, Path] = {}
+        self._paths: dict[str, Path] = {}
         self._strict_to_root = strict_to_root
 
         if self._is_bundled:
@@ -111,12 +103,12 @@ class DragonPathManager:
             _LOGGER.error(f"Path key '{key}' cannot start with underscores.")
             raise ValueError()
 
-    def update(self, new_paths: Dict[str, Union[str, Path]]) -> None:
+    def update(self, new_paths: dict[str, Union[str, Path]]) -> None:
         """
         Adds new paths in the manager.
 
         Args:
-            new_paths (Dict[str, Union[str, Path]]): A dictionary where keys are
+            new_paths (dict[str, Union[str, Path]]): A dictionary where keys are
                                     the identifiers and values are the
                                     Path objects to store.
         """
@@ -135,7 +127,7 @@ class DragonPathManager:
     def _sanitize_key(self, key: str):
         return sanitize_filename(key)
         
-    def make_dirs(self, keys: Optional[List[str]] = None, verbose: bool = False) -> None:
+    def make_dirs(self, keys: Optional[list[str]] = None, verbose: bool = False) -> None:
         """
         Creates directory structures for registered paths in writable locations.
 
@@ -325,331 +317,3 @@ class DragonPathManager:
         # Store absolute Path.
         self._paths[sanitized_name] = new_path
 
-
-def make_fullpath(
-        input_path: Union[str, Path],
-        make: bool = False,
-        verbose: bool = False,
-        enforce: Optional[Literal["directory", "file"]] = None
-    ) -> Path:
-    """
-    Resolves a string or Path into an absolute Path, optionally creating it.
-
-    - If the path exists, it is returned.
-    - If the path does not exist and `make=True`, it will:
-        - Create the file if the path has a suffix
-        - Create the directory if it has no suffix
-    - If `make=False` and the path does not exist, an error is raised.
-    - If `enforce`, raises an error if the resolved path is not what was enforced.
-    - Optionally prints whether the resolved path is a file or directory.
-
-    Parameters:
-        input_path (str | Path): 
-            Path to resolve.
-        make (bool): 
-            If True, attempt to create file or directory.
-        verbose (bool): 
-            Print classification after resolution.
-        enforce ("directory" | "file" | None):
-            Raises an error if the resolved path is not what was enforced.
-
-    Returns:
-        Path: Resolved absolute path.
-
-    Raises:
-        ValueError: If the path doesn't exist and can't be created.
-        TypeError: If the final path does not match the `enforce` parameter.
-        
-    ## 🗒️ Note:
-    
-    Directories with dots will be treated as files.
-    
-    Files without extension will be treated as directories.
-    """
-    path = Path(input_path).expanduser()
-
-    is_file = path.suffix != ""
-
-    try:
-        resolved = path.resolve(strict=True)
-    except FileNotFoundError:
-        if not make:
-            _LOGGER.error(f"Path does not exist: '{path}'.")
-            raise FileNotFoundError()
-
-        try:
-            if is_file:
-                # Create parent directories first
-                path.parent.mkdir(parents=True, exist_ok=True)
-                path.touch(exist_ok=False)
-            else:
-                path.mkdir(parents=True, exist_ok=True)
-            resolved = path.resolve(strict=True)
-        except Exception:
-            _LOGGER.exception(f"Failed to create {'file' if is_file else 'directory'} '{path}'.")
-            raise IOError()
-    
-    if enforce == "file" and not resolved.is_file():
-        _LOGGER.error(f"Path was enforced as a file, but it is not: '{resolved}'")
-        raise TypeError()
-    
-    if enforce == "directory" and not resolved.is_dir():
-        _LOGGER.error(f"Path was enforced as a directory, but it is not: '{resolved}'")
-        raise TypeError()
-
-    if verbose:
-        if resolved.is_file():
-            print("📄 Path is a File")
-        elif resolved.is_dir():
-            print("📁 Path is a Directory")
-        else:
-            print("❓ Path exists but is neither file nor directory")
-
-    return resolved
-
-
-def sanitize_filename(filename: str) -> str:
-    """
-    Sanitizes the name by:
-    - Stripping leading/trailing whitespace.
-    - Replacing all internal whitespace characters with underscores.
-    - Removing or replacing characters invalid in filenames.
-
-    Args:
-        filename (str): Base filename.
-
-    Returns:
-        str: A sanitized string suitable to use as a filename.
-    """
-    # Strip leading/trailing whitespace
-    sanitized = filename.strip()
-    
-    # Replace all whitespace sequences (space, tab, etc.) with underscores
-    sanitized = re.sub(r'\s+', '_', sanitized)
-
-    # Conservative filter to keep filenames safe across platforms
-    sanitized = re.sub(r'[^\w\-.]', '', sanitized)
-    
-    # Check for empty string after sanitization
-    if not sanitized:
-        _LOGGER.error("The sanitized filename is empty. The original input may have contained only invalid characters.")
-        raise ValueError()
-
-    return sanitized
-
-
-def list_csv_paths(directory: Union[str, Path], verbose: bool = True, raise_on_empty: bool = True) -> dict[str, Path]:
-    """
-    Lists all `.csv` files in the specified directory and returns a mapping: filenames (without extensions) to their absolute paths.
-
-    Parameters:
-        directory (str | Path): Path to the directory containing `.csv` files.
-        verbose (bool): If True, prints found files.
-        raise_on_empty (bool): If True, raises IOError if no files are found.
-
-    Returns:
-        (dict[str, Path]): Dictionary mapping {filename: filepath}.
-    """
-    # wraps the more general function
-    return list_files_by_extension(directory=directory, extension="csv", verbose=verbose, raise_on_empty=raise_on_empty)
-
-
-def list_files_by_extension(
-    directory: Union[str, Path], 
-    extension: str, 
-    verbose: bool = True,
-    raise_on_empty: bool = True
-) -> dict[str, Path]:
-    """
-    Lists all files with the specified extension in the given directory and returns a mapping: 
-    filenames (without extensions) to their absolute paths.
-
-    Parameters:
-        directory (str | Path): Path to the directory to search in.
-        extension (str): File extension to search for (e.g., 'json', 'txt').
-        verbose (bool): If True, logs the files found.
-        raise_on_empty (bool): If True, raises IOError if no matching files are found.
-
-    Returns:
-        (dict[str, Path]): Dictionary mapping {filename: filepath}. Returns empty dict if none found and raise_on_empty is False.
-    """
-    dir_path = make_fullpath(directory, enforce="directory")
-    
-    # Normalize the extension (remove leading dot if present)
-    normalized_ext = extension.lstrip(".").lower()
-    pattern = f"*.{normalized_ext}"
-    
-    matched_paths = list(dir_path.glob(pattern))
-    
-    if not matched_paths:
-        msg = f"No '.{normalized_ext}' files found in directory: {dir_path}."
-        if raise_on_empty:
-            _LOGGER.error(msg)
-            raise IOError()
-        else:
-            if verbose:
-                _LOGGER.warning(msg)
-            return {}
-
-    name_path_dict = {p.stem: p for p in matched_paths}
-    
-    if verbose:
-        _LOGGER.info(f"📂 '{normalized_ext.upper()}' files found:")
-        for name in name_path_dict:
-            print(f"\t{name}")
-    
-    return name_path_dict
-
-
-def list_subdirectories(
-    root_dir: Union[str, Path], 
-    verbose: bool = True, 
-    raise_on_empty: bool = True
-) -> dict[str, Path]:
-    """
-    Scans a directory and returns a dictionary of its immediate subdirectories.
-
-    Args:
-        root_dir (str | Path): The path to the directory to scan.
-        verbose (bool): If True, prints the number of directories found. 
-        raise_on_empty (bool): If True, raises IOError if no subdirectories are found.
-
-    Returns:
-        dict[str, Path]: A dictionary mapping subdirectory names (str) to their full Path objects.
-    """
-    root_path = make_fullpath(root_dir, enforce="directory")
-    
-    directories = [p.resolve() for p in root_path.iterdir() if p.is_dir()]
-    
-    if len(directories) < 1:
-        msg = f"No subdirectories found inside '{root_path}'"
-        if raise_on_empty:
-            _LOGGER.error(msg)
-            raise IOError()
-        else:
-            if verbose:
-                _LOGGER.warning(msg)
-            return {}
-    
-    if verbose:
-        count = len(directories)
-        # Use pluralization for better readability
-        plural = 'ies' if count != 1 else 'y'
-        print(f"Found {count} subdirector{plural} in '{root_path.name}'.")
-    
-    # Create a dictionary where the key is the directory's name (a string)
-    # and the value is the full Path object.
-    dir_map = {p.name: p for p in directories}
-    
-    return dir_map
-
-
-def clean_directory(directory: Union[str, Path], verbose: bool = False) -> None:
-    """
-    ⚠️  DANGER: DESTRUCTIVE OPERATION ⚠️
-
-    Deletes all files and subdirectories inside the specified directory. It is designed to empty a folder, not delete the folder itself.
-
-    Safety: It skips hidden files and directories (those starting with a period '.'). This works for macOS/Linux hidden files and dot-config folders on Windows.
-
-    Args:
-        directory (str | Path): The directory path to clean.
-        verbose (bool): If True, prints the name of each top-level item deleted.
-    """
-    target_dir = make_fullpath(directory, enforce="directory")
-
-    if verbose:
-        _LOGGER.warning(f"Starting cleanup of directory: {target_dir}")
-
-    for item in target_dir.iterdir():
-        # Safety Check: Skip hidden files/dirs
-        if item.name.startswith("."):
-            continue
-
-        try:
-            if item.is_file() or item.is_symlink():
-                item.unlink()
-                if verbose:
-                    print(f"    🗑️  Deleted file: {item.name}")
-            elif item.is_dir():
-                shutil.rmtree(item)
-                if verbose:
-                    print(f"    🗑️  Deleted directory: {item.name}")
-        except Exception as e:
-            _LOGGER.warning(f"Failed to delete item '{item.name}': {e}")
-            continue
-
-
-def safe_move(
-    source: Union[str, Path], 
-    final_destination: Union[str, Path], 
-    rename: Optional[str] = None, 
-    overwrite: bool = False
-) -> Path:
-    """
-    Moves a file or directory to a destination directory with safety checks.
-
-    Features:
-    - Supports optional renaming (sanitized automatically).
-    - PRESERVES file extensions during renaming (cannot be modified).
-    - Prevents accidental overwrites unless explicit.
-
-    Args:
-        source (str | Path): The file or directory to move.
-        final_destination (str | Path): The destination DIRECTORY where the item will be moved. It will be created if it does not exist.
-        rename (Optional[str]): If provided, the moved item will be renamed to this. Note: For files, the extension is strictly preserved.
-        overwrite (bool): If True, overwrites the destination path if it exists.
-    
-    Returns:
-        Path: The new absolute path of the moved item.
-    """
-    # 1. Validation and Setup
-    src_path = make_fullpath(source, make=False)
-
-    # Ensure destination directory exists
-    dest_dir_path = make_fullpath(final_destination, make=True, enforce="directory")
-
-    # 2. Determine Target Name
-    if rename:
-        sanitized_name = sanitize_filename(rename)
-        if src_path.is_file():
-            # Strict Extension Preservation
-            final_name = f"{sanitized_name}{src_path.suffix}"
-        else:
-            final_name = sanitized_name
-    else:
-        final_name = src_path.name
-
-    final_path = dest_dir_path / final_name
-
-    # 3. Safety Checks (Collision Detection)
-    if final_path.exists():
-        if not overwrite:
-            _LOGGER.error(f"Destination already exists: '{final_path}'. Use overwrite=True to force.")
-            raise FileExistsError()
-        
-        # Smart Overwrite Handling
-        if final_path.is_dir():
-            if src_path.is_file():
-                _LOGGER.error(f"Cannot overwrite directory '{final_path}' with file '{src_path}'")
-                raise IsADirectoryError()
-            # If overwriting a directory, we must remove the old one first to avoid nesting/errors
-            shutil.rmtree(final_path)
-        else:
-            # Destination is a file
-            if src_path.is_dir():
-                _LOGGER.error(f"Cannot overwrite file '{final_path}' with directory '{src_path}'")
-                raise FileExistsError()
-            final_path.unlink()
-
-    # 4. Perform Move
-    try:
-        shutil.move(str(src_path), str(final_path))
-        return final_path
-    except Exception as e:
-        _LOGGER.exception(f"Failed to move '{src_path}' to '{final_path}'")
-        raise e
-
-
-def info():
-    _script_info(__all__)

ml_tools/path_manager/_imprimir.py

@@ -0,0 +1,15 @@
+from .._core import _imprimir_disponibles
+
+_GRUPOS = [
+    "DragonPathManager",
+    "make_fullpath",
+    "sanitize_filename",
+    "list_csv_paths",
+    "list_files_by_extension",
+    "list_subdirectories",
+    "clean_directory",
+    "safe_move",
+]
+
+def info():
+    _imprimir_disponibles(_GRUPOS)