sh1 map compare

compare.py

# Save script as asm/maps/compare.py
# Then run
#   python compare.py map0_s01 map0_s02 0
# This will compare line count between files in each maps nonmatchings folder
# For each pair that matches it removes comments & data/func refs and uses Levenshtein to compare them
# Last param is max distance to show, use 0 to get exact matches, but can tweak it to find others if wanted.
# (since this only works with asm files in nonmatchings folder, it can't compare map0_s00 since we've matched some of them)
import os
import re
import Levenshtein

def count_lines_in_file(file_path):
    try:
        with open(file_path, 'r') as file:
            return sum(1 for _ in file)
    except Exception as e:
        print(f"Error reading {file_path}: {e}")
        return -1

def read_and_clean_file(file_path):
    try:
        with open(file_path, 'r') as file:
            content = file.read()
            # Remove /* blockquote comments */
            content = re.sub(r'/\*.*?\*/', '', content, flags=re.DOTALL)
            # Remove func_[8 characters] and D_[8 characters]
            content = re.sub(r'func_[a-fA-F0-9]{8}', '', content)
            content = re.sub(r'D_[a-fA-F0-9]{8}', '', content)
            return content
    except Exception as e:
        print(f"Error reading {file_path}: {e}")
        return None

def find_matching_line_counts(map1, map2, maxdistance):
    
    path1 = map1 + "/nonmatchings/" + map1
    path2 = map2 + "/nonmatchings/" + map2
    
    if not os.path.isdir(path1) or not os.path.isdir(path2):
        print("One or both paths are invalid directories.")
        return
    
    # Count lines in .s files from path1
    line_counts1 = {
        os.path.basename(f): count_lines_in_file(os.path.join(path1, f))
        for f in os.listdir(path1) if f.endswith('.s')
    }
    
    # Count lines in .s files from path2
    line_counts2 = {
        os.path.basename(f): count_lines_in_file(os.path.join(path2, f))
        for f in os.listdir(path2) if f.endswith('.s')
    }
    
    # Find matching line counts
    matched_files = []
    for file1, count1 in line_counts1.items():
        for file2, count2 in line_counts2.items():
            if count1 == count2 and count1 != -1:
                matched_files.append((file1, file2, count1))
    
    if matched_files:
        print("Distances:")
        for file1, file2, count in matched_files:
            #print(f"{file1} ↔ {file2} → {count} lines")
            
            # Read and clean the contents
            content1 = read_and_clean_file(os.path.join(path1, file1))
            content2 = read_and_clean_file(os.path.join(path2, file2))
            
            if content1 is not None and content2 is not None:
                # Compute Levenshtein distance
                distance = Levenshtein.distance(content1, content2)
                if distance <= maxdistance:
                    print(f"  {map1}/{file1} and {map2}/{file2}: {distance}")
    else:
        print("\nNo matching files found.")

if __name__ == "__main__":
    import argparse
    
    parser = argparse.ArgumentParser(description="Find matching line counts in .s files across two folders and compare contents.")
    parser.add_argument("map1", type=str, help="First map folder")
    parser.add_argument("map2", type=str, help="Second map folder")
    parser.add_argument("distance", type=int, help="Max distance")
    
    args = parser.parse_args()

    
    find_matching_line_counts(args.map1, args.map2, args.distance)