Semapp 1.0.5__py3-none-any.whl

semapp/Processing/rename_tif.py

@@ -0,0 +1,498 @@
+"""
+Module for renaming TIFF files based on coordinates from KLARF files.
+Supports KRONOS, COMPLUS4T, and normal modes.
+"""
+
+import os
+
+# Try to import QApplication for processEvents (optional, won't break if not available)
+try:
+    from PyQt5.QtWidgets import QApplication
+    HAS_QT = True
+except ImportError:
+    HAS_QT = False
+
+
+def rename_files(output_dir, coordinates, settings, is_kronos=False, is_complus4t=False):
+    """
+    Rename TIFF files based on coordinates.
+    Main function that detects the mode and calls the appropriate renamer.
+    
+    Args:
+        output_dir: Directory containing the TIFF files to rename
+        coordinates: DataFrame with columns ["defect_id", "X", "Y", "defect_size"]
+        settings: List of settings dictionaries (for normal mode)
+        is_kronos: Boolean indicating if this is KRONOS mode
+        is_complus4t: Boolean indicating if this is COMPLUS4T mode
+    
+    Returns:
+        int: Number of files renamed
+    """
+    if is_kronos:
+        return _rename_kronos(output_dir, coordinates)
+    elif is_complus4t:
+        return _rename_complus4t(output_dir, coordinates)
+    else:
+        return _rename_normal(output_dir, coordinates, settings)
+
+
+def rename_files_all(dirname, coordinates_dict, settings, is_kronos=False, is_complus4t=False):
+    """
+    Rename all TIFF files in all subdirectories based on coordinates.
+    Main function that detects the mode and calls the appropriate renamer.
+    
+    Args:
+        dirname: Base directory containing subdirectories
+        coordinates_dict: Dictionary mapping wafer_num to coordinates DataFrame
+        settings: List of settings dictionaries (for normal mode)
+        is_kronos: Boolean indicating if this is KRONOS mode
+        is_complus4t: Boolean indicating if this is COMPLUS4T mode
+    
+    Returns:
+        int: Total number of files renamed across all directories
+    """
+    if is_kronos:
+        return _rename_all_kronos(dirname, coordinates_dict)
+    elif is_complus4t:
+        return _rename_all_complus4t(dirname, coordinates_dict)
+    else:
+        return _rename_all_normal(dirname, coordinates_dict, settings)
+
+
+def _rename_kronos(output_dir, coordinates):
+    """
+    Rename TIFF files for KRONOS mode.
+    
+    Args:
+        output_dir: Directory containing the TIFF files to rename
+        coordinates: DataFrame with columns ["defect_id", "X", "Y", "defect_size"]
+    
+    Returns:
+        int: Number of files renamed
+    """
+    if not os.path.exists(output_dir):
+        print(f"[DEBUG] output_dir does not exist: {output_dir}")
+        return 0
+    
+    tiff_files = [f for f in os.listdir(output_dir)
+                  if "page" in f and f.lower().endswith(('.tiff', '.tif'))]
+    tiff_files.sort()
+    
+    print(f"[DEBUG] KRONOS: Found {len(tiff_files)} TIFF files in output_dir")
+    
+    has_qt = HAS_QT
+    
+    renamed_count = 0
+    for idx, file in enumerate(tiff_files):
+        if has_qt and idx % 10 == 0:
+            QApplication.processEvents()
+        
+        try:
+            file_number = int(file.split('_')[2].split('.')[0])
+        except (ValueError, IndexError) as e:
+            print(f"[DEBUG] Error extracting file_number from {file}: {e}")
+            continue
+        
+        # KRONOS: file_number directly corresponds to the row index (0-based)
+        csv_row_index = file_number - 1
+        
+        if csv_row_index >= len(coordinates):
+            print(f"[DEBUG] Warning: csv_row_index {csv_row_index} is out of bounds. Total defects: {len(coordinates)}")
+            continue
+        
+        x = coordinates.iloc[csv_row_index, 1]  # Column 1 is X
+        y = coordinates.iloc[csv_row_index, 2]  # Column 2 is Y
+        new_name = f"{x}_{y}.tif"
+        
+        old_path = os.path.join(output_dir, file)
+        new_path = os.path.join(output_dir, new_name)
+        
+        # Handle file name collisions
+        if os.path.exists(new_path):
+            base_name = new_name.rsplit('.', 1)[0]
+            extension = new_name.rsplit('.', 1)[1] if '.' in new_name else 'tif'
+            counter = 1
+            while os.path.exists(new_path):
+                new_name = f"{base_name}_{counter}.{extension}"
+                new_path = os.path.join(output_dir, new_name)
+                counter += 1
+        
+        try:
+            os.rename(old_path, new_path)
+            renamed_count += 1
+        except Exception as e:
+            print(f"[DEBUG] Error renaming {file}: {e}")
+    
+    return renamed_count
+
+
+def _rename_complus4t(output_dir, coordinates):
+    """
+    Rename TIFF files for COMPLUS4T mode.
+    
+    Args:
+        output_dir: Directory containing the TIFF files to rename
+        coordinates: DataFrame with columns ["defect_id", "X", "Y", "defect_size"]
+    
+    Returns:
+        int: Number of files renamed
+    """
+    if not os.path.exists(output_dir):
+        print(f"[DEBUG] output_dir does not exist: {output_dir}")
+        return 0
+    
+    tiff_files = [f for f in os.listdir(output_dir)
+                  if "page" in f and f.lower().endswith(('.tiff', '.tif'))]
+    tiff_files.sort()
+    
+    print(f"[DEBUG] COMPLUS4T: Found {len(tiff_files)} TIFF files in output_dir")
+    
+    has_qt = HAS_QT
+    
+    renamed_count = 0
+    for idx, file in enumerate(tiff_files):
+        if has_qt and idx % 10 == 0:
+            QApplication.processEvents()
+        
+        try:
+            file_number = int(file.split('_')[2].split('.')[0])
+        except (ValueError, IndexError) as e:
+            print(f"[DEBUG] Error extracting file_number from {file}: {e}")
+            continue
+        
+        # COMPLUS4T: file_number is the defect_id
+        matching_rows = coordinates[coordinates['defect_id'] == file_number]
+        
+        if matching_rows.empty:
+            print(f"[DEBUG] Warning: defect_id {file_number} not found in coordinates")
+            continue
+        
+        row = matching_rows.iloc[0]
+        x = row['X']
+        y = row['Y']
+        new_name = f"{x}_{y}.tif"
+        
+        old_path = os.path.join(output_dir, file)
+        new_path = os.path.join(output_dir, new_name)
+        
+        # Handle file name collisions
+        if os.path.exists(new_path):
+            base_name = new_name.rsplit('.', 1)[0]
+            extension = new_name.rsplit('.', 1)[1] if '.' in new_name else 'tif'
+            counter = 1
+            while os.path.exists(new_path):
+                new_name = f"{base_name}_{counter}.{extension}"
+                new_path = os.path.join(output_dir, new_name)
+                counter += 1
+        
+        try:
+            os.rename(old_path, new_path)
+            renamed_count += 1
+        except Exception as e:
+            print(f"[DEBUG] Error renaming {file}: {e}")
+    
+    return renamed_count
+
+
+def _rename_normal(output_dir, coordinates, settings):
+    """
+    Rename TIFF files for normal mode.
+    
+    Args:
+        output_dir: Directory containing the TIFF files to rename
+        coordinates: DataFrame with columns ["defect_id", "X", "Y", "defect_size"]
+        settings: List of settings dictionaries
+    
+    Returns:
+        int: Number of files renamed
+    """
+    if not os.path.exists(output_dir):
+        print(f"[DEBUG] output_dir does not exist: {output_dir}")
+        return 0
+    
+    tiff_files = [f for f in os.listdir(output_dir)
+                  if "page" in f and f.lower().endswith(('.tiff', '.tif'))]
+    tiff_files.sort()
+    
+    print(f"[DEBUG] Normal: Found {len(tiff_files)} TIFF files in output_dir")
+    
+    has_qt = HAS_QT
+    
+    renamed_count = 0
+    for idx, file in enumerate(tiff_files):
+        if has_qt and idx % 10 == 0:
+            QApplication.processEvents()
+        
+        try:
+            file_number = int(file.split('_')[2].split('.')[0])
+        except (ValueError, IndexError) as e:
+            print(f"[DEBUG] Error extracting file_number from {file}: {e}")
+            continue
+        
+        # Normal mode: calculate row using settings
+        csv_row_index = (file_number - 1) // len(settings)
+        remainder = (file_number - 1) % len(settings)
+        
+        if csv_row_index >= len(coordinates):
+            print(f"[DEBUG] Warning: csv_row_index {csv_row_index} is out of bounds. Total defects: {len(coordinates)}")
+            continue
+        
+        x = coordinates.iloc[csv_row_index, 1]  # Column 1 is X
+        y = coordinates.iloc[csv_row_index, 2]  # Column 2 is Y
+        
+        scale = settings[remainder]["Scale"]
+        image_type = settings[remainder]["Image Type"]
+        new_name = f"{scale}_{x}_{y}_{image_type}.tif"
+        
+        old_path = os.path.join(output_dir, file)
+        new_path = os.path.join(output_dir, new_name)
+        
+        # Handle file name collisions
+        if os.path.exists(new_path):
+            base_name = new_name.rsplit('.', 1)[0]
+            extension = new_name.rsplit('.', 1)[1] if '.' in new_name else 'tif'
+            counter = 1
+            while os.path.exists(new_path):
+                new_name = f"{base_name}_{counter}.{extension}"
+                new_path = os.path.join(output_dir, new_name)
+                counter += 1
+        
+        try:
+            os.rename(old_path, new_path)
+            renamed_count += 1
+        except Exception as e:
+            print(f"[DEBUG] Error renaming {file}: {e}")
+    
+    return renamed_count
+
+
+def _rename_all_kronos(dirname, coordinates_dict):
+    """
+    Rename all TIFF files for KRONOS mode across all subdirectories.
+    
+    Args:
+        dirname: Base directory containing subdirectories
+        coordinates_dict: Dictionary mapping wafer_num to coordinates DataFrame
+    
+    Returns:
+        int: Total number of files renamed
+    """
+    total_renamed = 0
+    has_qt = HAS_QT
+    
+    for subdir, _, _ in os.walk(dirname):
+        if subdir == dirname:
+            continue
+        
+        output_dir = os.path.join(dirname, os.path.basename(subdir))
+        
+        try:
+            wafer_num = int(os.path.basename(subdir))
+        except ValueError:
+            continue
+        
+        if wafer_num not in coordinates_dict:
+            continue
+        
+        coordinates = coordinates_dict[wafer_num]
+        
+        tiff_files = [f for f in os.listdir(output_dir)
+                      if "page" in f and f.lower().endswith(('.tiff', '.tif'))]
+        tiff_files.sort()
+        
+        renamed_in_dir = 0
+        for idx, file in enumerate(tiff_files):
+            if has_qt and idx % 10 == 0:
+                QApplication.processEvents()
+            
+            try:
+                file_number = int(file.split('_')[2].split('.')[0])
+            except (ValueError, IndexError):
+                continue
+            
+            csv_row_index = file_number - 1
+            
+            if csv_row_index >= len(coordinates):
+                continue
+            
+            x = coordinates.iloc[csv_row_index, 1]
+            y = coordinates.iloc[csv_row_index, 2]
+            new_name = f"{x}_{y}.tif"
+            
+            old_path = os.path.join(output_dir, file)
+            new_path = os.path.join(output_dir, new_name)
+            
+            if os.path.exists(new_path):
+                base_name = new_name.rsplit('.', 1)[0]
+                extension = new_name.rsplit('.', 1)[1] if '.' in new_name else 'tif'
+                counter = 1
+                while os.path.exists(new_path):
+                    new_name = f"{base_name}_{counter}.{extension}"
+                    new_path = os.path.join(output_dir, new_name)
+                    counter += 1
+            
+            try:
+                os.rename(old_path, new_path)
+                renamed_in_dir += 1
+                total_renamed += 1
+            except Exception:
+                pass
+    
+    return total_renamed
+
+
+def _rename_all_complus4t(dirname, coordinates_dict):
+    """
+    Rename all TIFF files for COMPLUS4T mode across all subdirectories.
+    
+    Args:
+        dirname: Base directory containing subdirectories
+        coordinates_dict: Dictionary mapping wafer_num to coordinates DataFrame
+    
+    Returns:
+        int: Total number of files renamed
+    """
+    total_renamed = 0
+    has_qt = HAS_QT
+    
+    for subdir, _, _ in os.walk(dirname):
+        if subdir == dirname:
+            continue
+        
+        output_dir = os.path.join(dirname, os.path.basename(subdir))
+        
+        try:
+            wafer_num = int(os.path.basename(subdir))
+        except ValueError:
+            continue
+        
+        if wafer_num not in coordinates_dict:
+            continue
+        
+        coordinates = coordinates_dict[wafer_num]
+        
+        tiff_files = [f for f in os.listdir(output_dir)
+                      if "page" in f and f.lower().endswith(('.tiff', '.tif'))]
+        tiff_files.sort()
+        
+        renamed_in_dir = 0
+        for idx, file in enumerate(tiff_files):
+            if has_qt and idx % 10 == 0:
+                QApplication.processEvents()
+            
+            try:
+                file_number = int(file.split('_')[2].split('.')[0])
+            except (ValueError, IndexError):
+                continue
+            
+            matching_rows = coordinates[coordinates['defect_id'] == file_number]
+            
+            if matching_rows.empty:
+                continue
+            
+            row = matching_rows.iloc[0]
+            x = row['X']
+            y = row['Y']
+            new_name = f"{x}_{y}.tif"
+            
+            old_path = os.path.join(output_dir, file)
+            new_path = os.path.join(output_dir, new_name)
+            
+            if os.path.exists(new_path):
+                base_name = new_name.rsplit('.', 1)[0]
+                extension = new_name.rsplit('.', 1)[1] if '.' in new_name else 'tif'
+                counter = 1
+                while os.path.exists(new_path):
+                    new_name = f"{base_name}_{counter}.{extension}"
+                    new_path = os.path.join(output_dir, new_name)
+                    counter += 1
+            
+            try:
+                os.rename(old_path, new_path)
+                renamed_in_dir += 1
+                total_renamed += 1
+            except Exception:
+                pass
+    
+    return total_renamed
+
+
+def _rename_all_normal(dirname, coordinates_dict, settings):
+    """
+    Rename all TIFF files for normal mode across all subdirectories.
+    
+    Args:
+        dirname: Base directory containing subdirectories
+        coordinates_dict: Dictionary mapping wafer_num to coordinates DataFrame
+        settings: List of settings dictionaries
+    
+    Returns:
+        int: Total number of files renamed
+    """
+    total_renamed = 0
+    has_qt = HAS_QT
+    
+    for subdir, _, _ in os.walk(dirname):
+        if subdir == dirname:
+            continue
+        
+        output_dir = os.path.join(dirname, os.path.basename(subdir))
+        
+        try:
+            wafer_num = int(os.path.basename(subdir))
+        except ValueError:
+            continue
+        
+        if wafer_num not in coordinates_dict:
+            continue
+        
+        coordinates = coordinates_dict[wafer_num]
+        
+        tiff_files = [f for f in os.listdir(output_dir)
+                      if "page" in f and f.lower().endswith(('.tiff', '.tif'))]
+        tiff_files.sort()
+        
+        renamed_in_dir = 0
+        for idx, file in enumerate(tiff_files):
+            if has_qt and idx % 10 == 0:
+                QApplication.processEvents()
+            
+            try:
+                file_number = int(file.split('_')[2].split('.')[0])
+            except (ValueError, IndexError):
+                continue
+            
+            csv_row_index = (file_number - 1) // len(settings)
+            remainder = (file_number - 1) % len(settings)
+            
+            if csv_row_index >= len(coordinates):
+                continue
+            
+            x = coordinates.iloc[csv_row_index, 1]
+            y = coordinates.iloc[csv_row_index, 2]
+            
+            scale = settings[remainder]["Scale"]
+            image_type = settings[remainder]["Image Type"]
+            new_name = f"{scale}_{x}_{y}_{image_type}.tif"
+            
+            old_path = os.path.join(output_dir, file)
+            new_path = os.path.join(output_dir, new_name)
+            
+            if os.path.exists(new_path):
+                base_name = new_name.rsplit('.', 1)[0]
+                extension = new_name.rsplit('.', 1)[1] if '.' in new_name else 'tif'
+                counter = 1
+                while os.path.exists(new_path):
+                    new_name = f"{base_name}_{counter}.{extension}"
+                    new_path = os.path.join(output_dir, new_name)
+                    counter += 1
+            
+            try:
+                os.rename(old_path, new_path)
+                renamed_in_dir += 1
+                total_renamed += 1
+            except Exception:
+                pass
+    
+    return total_renamed
+