Init CBZ Volume Combiner v0.2.0

cbz_volume_combiner/file_utils.py

@@ -0,0 +1,125 @@
+# File: /home/code/projects/manga-organizer-1/cbz-volume-combiner/cbz_volume_combiner/file_utils.py
+import os
+import re
+from .parsing import parse_manga_filename
+
+def find_cbz_files(folder_path, recursive=False, extra_verbose=False):
+    """Find all CBZ files in the given folder."""
+    cbz_files = []
+    
+    if recursive:
+        if extra_verbose:
+            print(f"Recursively searching for CBZ files in {folder_path}")
+            
+        for root, _, files in os.walk(folder_path):
+            for file in files:
+                if file.lower().endswith('.cbz'):
+                    cbz_files.append(os.path.join(root, file))
+    else:
+        if extra_verbose:
+            print(f"Searching for CBZ files in {folder_path} (non-recursive)")
+            
+        for file in os.listdir(folder_path):
+            if file.lower().endswith('.cbz'):
+                cbz_files.append(os.path.join(folder_path, file))
+    
+    if extra_verbose:
+        print(f"Found {len(cbz_files)} CBZ files")
+        
+    return cbz_files
+
+def fix_missing_files(chapter_infos, folder_path, extra_verbose=False):
+    """Attempt to find missing files by searching for similar filenames."""
+    fixed_chapters = []
+    unresolved_chapters = []
+    
+    for chapter_info in chapter_infos:
+        # Check if file exists first
+        if os.path.exists(chapter_info['filename']):
+            fixed_chapters.append(chapter_info)
+            continue
+            
+        # If we're here, the file doesn't exist - get the problematic filename
+        base_filename = os.path.basename(chapter_info['filename'])
+        directory = os.path.dirname(chapter_info['filename'])
+        
+        if extra_verbose:
+            print(f"Trying to find replacement for: {base_filename}")
+        
+        # First approach: Direct check with normalized path
+        # This is for cases where Python's path handling might be different from the filesystem
+        found_replacement = False
+        
+        # Look for similar files in the directory
+        try:
+            # Get all CBZ files in the directory
+            cbz_files_in_dir = [f for f in os.listdir(directory) if f.lower().endswith('.cbz')]
+            
+            # First try: Look for exact matches with volume and chapter numbers
+            vol_num = chapter_info['volume']
+            chap_num = chapter_info['chapter_str']
+            
+            # Improved pattern matching for volume and chapter numbers
+            vol_pattern = f"v{vol_num:02d}"  # e.g., "v07"
+            chap_pattern = f"c{chap_num}"    # e.g., "c037"
+            
+            # Try to find a direct match first
+            for file in cbz_files_in_dir:
+                # Check volume and chapter patterns
+                if vol_pattern in file and chap_pattern in file:
+                    potential_path = os.path.join(directory, file)
+                    if os.path.exists(potential_path):
+                        if extra_verbose:
+                            print(f"  Found direct match: {file}")
+                        new_chapter_info = dict(chapter_info)
+                        new_chapter_info['filename'] = potential_path
+                        fixed_chapters.append(new_chapter_info)
+                        found_replacement = True
+                        break
+            
+            # If no direct match, try more flexible matching
+            if not found_replacement:
+                # Fuzzy match approach
+                chapter_pattern = rf"v0*{vol_num}\s+c0*{chap_num.lstrip('0')}"
+                
+                for file in cbz_files_in_dir:
+                    # Remove special characters for comparison
+                    clean_file = file
+                    for char in "'?,":
+                        clean_file = clean_file.replace(char, '')
+                    
+                    # Strip special characters from the pattern too
+                    clean_pattern = chapter_pattern
+                    for char in "'?,":
+                        clean_pattern = clean_pattern.replace(char, '')
+                    
+                    # Try matching with cleaned strings
+                    if re.search(chapter_pattern, file, re.IGNORECASE) or re.search(clean_pattern, clean_file, re.IGNORECASE):
+                        potential_path = os.path.join(directory, file)
+                        if os.path.exists(potential_path):
+                            if extra_verbose:
+                                print(f"  Found fuzzy match: {file}")
+                            new_chapter_info = dict(chapter_info)
+                            new_chapter_info['filename'] = potential_path
+                            fixed_chapters.append(new_chapter_info)
+                            found_replacement = True
+                            break
+            
+            # Last resort: List all files and let the user see what's available
+            if not found_replacement and extra_verbose:
+                print("  No match found. Available files in directory:")
+                for idx, file in enumerate(sorted(cbz_files_in_dir)):
+                    if idx < 20:  # Limit to first 20 files to avoid spam
+                        print(f"    - {file}")
+                    else:
+                        print(f"    ... and {len(cbz_files_in_dir) - 20} more files")
+                        break
+                        
+        except Exception as e:
+            if extra_verbose:
+                print(f"  Error while searching for replacement: {str(e)}")
+                
+        if not found_replacement:
+            unresolved_chapters.append(chapter_info)
+            
+    return fixed_chapters, unresolved_chapters