Semapp 1.0.5__py3-none-any.whl

semapp/Processing/split_tif.py

@@ -0,0 +1,323 @@
+"""
+Module for splitting multi-page TIFF files into individual TIFF files.
+Supports KRONOS, COMPLUS4T, and normal modes.
+"""
+
+import os
+import glob
+from PIL import Image
+
+
+def split_tiff(tiff_path, output_dir, coordinates=None, is_kronos=False, is_complus4t=False):
+    """
+    Split a merged TIFF file into individual TIFF files.
+    Main function that detects the mode and calls the appropriate splitter.
+    
+    Args:
+        tiff_path: Path to the multi-page TIFF file
+        output_dir: Directory where split files will be saved
+        coordinates: DataFrame with columns ["defect_id", "X", "Y", "defect_size"] (for COMPLUS4T)
+        is_kronos: Boolean indicating if this is KRONOS mode
+        is_complus4t: Boolean indicating if this is COMPLUS4T mode
+    
+    Returns:
+        list: List of file paths of the generated TIFF files
+    """
+    if is_complus4t:
+        return _split_tiff_complus4t(tiff_path, output_dir, coordinates)
+    elif is_kronos:
+        return _split_tiff_kronos(tiff_path, output_dir)
+    else:
+        return _split_tiff_normal(tiff_path, output_dir)
+
+
+def split_tiff_all(dirname, coordinates_dict=None, is_kronos=False, is_complus4t=False):
+    """
+    Split all merged TIFF files in all subdirectories into individual TIFF files.
+    Main function that detects the mode and calls the appropriate splitter.
+    
+    Args:
+        dirname: Base directory containing subdirectories
+        coordinates_dict: Dictionary mapping wafer_num to coordinates DataFrame (for COMPLUS4T)
+        is_kronos: Boolean indicating if this is KRONOS mode
+        is_complus4t: Boolean indicating if this is COMPLUS4T mode
+    
+    Returns:
+        list: List of all generated TIFF file paths
+    """
+    if is_complus4t:
+        return _split_tiff_all_complus4t(dirname, coordinates_dict)
+    elif is_kronos:
+        return _split_tiff_all_kronos(dirname)
+    else:
+        return _split_tiff_all_normal(dirname)
+
+
+def _split_tiff_kronos(tiff_path, output_dir):
+    """
+    Split TIFF file for KRONOS mode.
+    Splits all pages sequentially.
+    
+    Args:
+        tiff_path: Path to the multi-page TIFF file
+        output_dir: Directory where split files will be saved
+    
+    Returns:
+        list: List of file paths of the generated TIFF files
+    """
+    if not os.path.exists(tiff_path):
+        return []
+    
+    os.makedirs(output_dir, exist_ok=True)
+    output_files = []
+    
+    try:
+        img = Image.open(tiff_path)
+        page_index = 0
+        
+        while True:
+            output_file = os.path.join(output_dir, f"data_page_{page_index + 1}.tiff")
+            img.save(output_file, format="TIFF")
+            output_files.append(output_file)
+            
+            try:
+                img.seek(page_index + 1)
+                page_index += 1
+            except EOFError:
+                break
+    except Exception as error:
+        raise RuntimeError(f"Error splitting TIFF file: {error}") from error
+    
+    return output_files
+
+
+def _split_tiff_complus4t(tiff_path, output_dir, coordinates):
+    """
+    Split TIFF file for COMPLUS4T mode.
+    Only splits pages corresponding to defect_id from coordinates.
+    
+    Args:
+        tiff_path: Path to the multi-page TIFF file
+        output_dir: Directory where split files will be saved
+        coordinates: DataFrame with columns ["defect_id", "X", "Y", "defect_size"]
+    
+    Returns:
+        list: List of file paths of the generated TIFF files
+    """
+    if not os.path.exists(tiff_path):
+        return []
+    
+    if coordinates is None or coordinates.empty:
+        print(f"Warning: No coordinates provided for COMPLUS4T split")
+        return []
+    
+    os.makedirs(output_dir, exist_ok=True)
+    output_files = []
+    
+    try:
+        img = Image.open(tiff_path)
+        
+        # COMPLUS4T: only split pages corresponding to defect_id
+        # Pages in TIFF correspond to defect_id (331, 332, 333...)
+        # defect_id are 1-based in KLARF, but TIFF pages are 0-based
+        # So defect_id 331 = page index 330 (331 - 1)
+        for idx, row in coordinates.iterrows():
+            defect_id = int(row['defect_id'])
+            page_index = defect_id - 1  # Convert to 0-based index
+            
+            try:
+                img.seek(page_index)
+                output_file = os.path.join(output_dir, f"data_page_{defect_id}.tiff")
+                img.save(output_file, format="TIFF")
+                output_files.append(output_file)
+            except EOFError:
+                print(f"Warning: Page {page_index} (defect_id {defect_id}) not found in TIFF")
+                continue
+    except Exception as error:
+        raise RuntimeError(f"Error splitting TIFF file: {error}") from error
+    
+    return output_files
+
+
+def _split_tiff_normal(tiff_path, output_dir):
+    """
+    Split TIFF file for normal mode.
+    Splits all pages sequentially.
+    
+    Args:
+        tiff_path: Path to the multi-page TIFF file
+        output_dir: Directory where split files will be saved
+    
+    Returns:
+        list: List of file paths of the generated TIFF files
+    """
+    if not os.path.exists(tiff_path):
+        return []
+    
+    os.makedirs(output_dir, exist_ok=True)
+    output_files = []
+    
+    try:
+        img = Image.open(tiff_path)
+        page_index = 0
+        
+        while True:
+            output_file = os.path.join(output_dir, f"data_page_{page_index + 1}.tiff")
+            img.save(output_file, format="TIFF")
+            output_files.append(output_file)
+            
+            try:
+                img.seek(page_index + 1)
+                page_index += 1
+            except EOFError:
+                break
+    except Exception as error:
+        raise RuntimeError(f"Error splitting TIFF file: {error}") from error
+    
+    return output_files
+
+
+def _split_tiff_all_kronos(dirname):
+    """
+    Split all TIFF files for KRONOS mode across all subdirectories.
+    
+    Args:
+        dirname: Base directory containing subdirectories
+    
+    Returns:
+        list: List of all generated TIFF file paths
+    """
+    all_output_files = []
+    
+    for subdir, _, _ in os.walk(dirname):
+        if subdir == dirname:
+            continue
+        
+        tiff_path = os.path.join(subdir, "data.tif")
+        
+        if not os.path.exists(tiff_path):
+            continue
+        
+        try:
+            img = Image.open(tiff_path)
+            page_index = 0
+            
+            while True:
+                output_file = tiff_path.replace(".tif", "") + f"_page_{page_index + 1}.tiff"
+                img.save(output_file, format="TIFF")
+                all_output_files.append(output_file)
+                
+                try:
+                    img.seek(page_index + 1)
+                    page_index += 1
+                except EOFError:
+                    break
+        except Exception as error:
+            print(f"Error splitting TIFF file {tiff_path}: {error}")
+            continue
+    
+    return all_output_files
+
+
+def _split_tiff_all_complus4t(dirname, coordinates_dict):
+    """
+    Split 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:
+        list: List of all generated TIFF file paths
+    """
+    all_output_files = []
+    
+    # For COMPLUS4T, the TIFF file is in the parent directory
+    tiff_files = glob.glob(os.path.join(dirname, '*.tiff'))
+    if not tiff_files:
+        tiff_files = glob.glob(os.path.join(dirname, '*.tif'))
+    
+    if not tiff_files:
+        print("Warning: No TIFF file found in parent directory for COMPLUS4T mode")
+        return []
+    
+    parent_tiff_path = tiff_files[0]
+    
+    for subdir, _, _ in os.walk(dirname):
+        if subdir == dirname:
+            continue
+        
+        try:
+            wafer_num = int(os.path.basename(subdir))
+        except ValueError:
+            continue
+        
+        if wafer_num not in coordinates_dict:
+            continue
+        
+        coordinates = coordinates_dict[wafer_num]
+        output_dir = os.path.join(dirname, str(wafer_num))
+        os.makedirs(output_dir, exist_ok=True)
+        
+        try:
+            img = Image.open(parent_tiff_path)
+            
+            for idx, row in coordinates.iterrows():
+                defect_id = int(row['defect_id'])
+                page_index = defect_id - 1
+                
+                try:
+                    img.seek(page_index)
+                    output_file = os.path.join(output_dir, f"data_page_{defect_id}.tiff")
+                    img.save(output_file, format="TIFF")
+                    all_output_files.append(output_file)
+                except EOFError:
+                    print(f"Warning: Page {page_index} (defect_id {defect_id}) not found in TIFF for wafer {wafer_num}")
+                    continue
+        except Exception as error:
+            print(f"Error splitting TIFF file for wafer {wafer_num}: {error}")
+            continue
+    
+    return all_output_files
+
+
+def _split_tiff_all_normal(dirname):
+    """
+    Split all TIFF files for normal mode across all subdirectories.
+    
+    Args:
+        dirname: Base directory containing subdirectories
+    
+    Returns:
+        list: List of all generated TIFF file paths
+    """
+    all_output_files = []
+    
+    for subdir, _, _ in os.walk(dirname):
+        if subdir == dirname:
+            continue
+        
+        tiff_path = os.path.join(subdir, "data.tif")
+        
+        if not os.path.exists(tiff_path):
+            continue
+        
+        try:
+            img = Image.open(tiff_path)
+            page_index = 0
+            
+            while True:
+                output_file = tiff_path.replace(".tif", "") + f"_page_{page_index + 1}.tiff"
+                img.save(output_file, format="TIFF")
+                all_output_files.append(output_file)
+                
+                try:
+                    img.seek(page_index + 1)
+                    page_index += 1
+                except EOFError:
+                    break
+        except Exception as error:
+            print(f"Error splitting TIFF file {tiff_path}: {error}")
+            continue
+    
+    return all_output_files