#!/usr/bin/env python3
"""Compare two .set files to find differences when adding a mapping"""
import sys
from pathlib import Path

def compare_binary_files(file1_path, file2_path):
    """Compare two binary files and show differences"""

    # Read both files
    with open(file1_path, 'rb') as f1:
        data1 = f1.read()

    with open(file2_path, 'rb') as f2:
        data2 = f2.read()

    print(f"File 1 (original): {len(data1):,} bytes")
    print(f"File 2 (added):    {len(data2):,} bytes")
    print(f"Size difference:   {len(data2) - len(data1):,} bytes")
    print("=" * 80)

    # Find first difference
    min_len = min(len(data1), len(data2))
    first_diff = None

    for i in range(min_len):
        if data1[i] != data2[i]:
            first_diff = i
            break

    if first_diff is not None:
        print(f"\nFirst difference at offset: {first_diff} (0x{first_diff:X})")

        # Show context around first difference
        start = max(0, first_diff - 50)
        end = min(min_len, first_diff + 200)

        print(f"\nOriginal file bytes [{start}:{end}]:")
        print_hex_dump(data1[start:end], start)

        print(f"\nAdded mapping file bytes [{start}:{end}]:")
        print_hex_dump(data2[start:end], start)

    else:
        print("\nFiles are identical up to the shorter length")
        if len(data1) != len(data2):
            print(f"Extra data at end of {'file1' if len(data1) > len(data2) else 'file2'}")

    # Find all difference regions
    print("\n" + "=" * 80)
    print("FINDING ALL DIFFERENCE REGIONS:")
    print("=" * 80)

    diff_regions = []
    in_diff = False
    diff_start = None

    for i in range(min_len):
        if data1[i] != data2[i]:
            if not in_diff:
                diff_start = i
                in_diff = True
        else:
            if in_diff:
                diff_regions.append((diff_start, i))
                in_diff = False

    if in_diff:
        diff_regions.append((diff_start, min_len))

    print(f"\nFound {len(diff_regions)} different regions:")

    for idx, (start, end) in enumerate(diff_regions[:10], 1):  # Show first 10
        length = end - start
        print(f"\nRegion {idx}: offset {start} (0x{start:X}), length {length} bytes")

        # Show bytes
        region_start = max(0, start - 20)
        region_end = min(min_len, end + 20)

        print(f"  Original [{region_start}:{region_end}]:")
        print_hex_dump(data1[region_start:region_end], region_start, highlight_start=start, highlight_end=end)

        print(f"  Added    [{region_start}:{region_end}]:")
        print_hex_dump(data2[region_start:region_end], region_start, highlight_start=start, highlight_end=end)

    # Count "ules" markers in each file
    print("\n" + "=" * 80)
    print("ACTION MAPPING MARKERS:")
    print("=" * 80)

    ules_count1 = count_markers(data1, b'ules')
    ules_count2 = count_markers(data2, b'ules')

    print(f"'ules' markers in original: {ules_count1}")
    print(f"'ules' markers in added:    {ules_count2}")
    print(f"Difference:                 {ules_count2 - ules_count1}")

    # Find positions of ules markers
    print("\nPositions of 'ules' markers:")
    print("Original file:")
    find_and_print_markers(data1, b'ules')

    print("\nAdded mapping file:")
    find_and_print_markers(data2, b'ules')

def print_hex_dump(data, offset=0, highlight_start=None, highlight_end=None):
    """Print hex dump of data with optional highlighting"""
    for i in range(0, len(data), 16):
        chunk = data[i:i+16]
        addr = offset + i

        # Hex part
        hex_parts = []
        for j, byte in enumerate(chunk):
            byte_offset = addr + j
            if highlight_start and highlight_end and highlight_start <= byte_offset < highlight_end:
                hex_parts.append(f"\033[91m{byte:02X}\033[0m")  # Red highlight
            else:
                hex_parts.append(f"{byte:02X}")

        hex_str = " ".join(hex_parts)

        # ASCII part
        ascii_parts = []
        for j, byte in enumerate(chunk):
            byte_offset = addr + j
            if 32 <= byte <= 126:
                char = chr(byte)
            else:
                char = "."

            if highlight_start and highlight_end and highlight_start <= byte_offset < highlight_end:
                ascii_parts.append(f"\033[91m{char}\033[0m")
            else:
                ascii_parts.append(char)

        ascii_str = "".join(ascii_parts)

        print(f"  {addr:06X}: {hex_str:<48}  {ascii_str}")

def count_markers(data, marker):
    """Count occurrences of marker in data"""
    count = 0
    pos = 0
    while True:
        pos = data.find(marker, pos)
        if pos == -1:
            break
        count += 1
        pos += len(marker)
    return count

def find_and_print_markers(data, marker):
    """Find and print all marker positions"""
    positions = []
    pos = 0
    while True:
        pos = data.find(marker, pos)
        if pos == -1:
            break
        positions.append(pos)
        pos += len(marker)

    for i, pos in enumerate(positions[:20], 1):  # Show first 20
        # Check if it's preceded by 0x05 (marker byte)
        if pos > 0 and data[pos-1] == 0x05:
            print(f"  {i:2d}. Offset {pos:6d} (0x{pos:05X}) - marker type 0x05")
        else:
            print(f"  {i:2d}. Offset {pos:6d} (0x{pos:05X})")

    if len(positions) > 20:
        print(f"  ... and {len(positions) - 20} more")

if __name__ == '__main__':
    original = 'TestMKS_original.set'
    added = 'TestMKS_added_mapping.set'

    if not Path(original).exists():
        print(f"Error: {original} not found")
        sys.exit(1)

    if not Path(added).exists():
        print(f"Error: {added} not found")
        sys.exit(1)

    compare_binary_files(original, added)