geutebruck/GeViSetEditor/GeViSetEditor.Core/Parsers/AdvancedBinaryAnalyzer.cs

using System;
using System.Collections.Generic;
using System.Linq;
using System.Text;

namespace GeViSetEditor.Core.Parsers
{
    /// <summary>
    /// Advanced binary analyzer to understand .set file structure completely
    /// </summary>
    public class AdvancedBinaryAnalyzer
    {
        private byte[] _data;
        private int _position;

        /// <summary>
        /// Analyze binary structure and report findings
        /// </summary>
        public BinaryStructureReport Analyze(byte[] data)
        {
            _data = data;
            _position = 0;

            var report = new BinaryStructureReport
            {
                FileSize = data.Length
            };

            Console.WriteLine("\n=== Advanced Binary Structure Analysis ===\n");

            // Read header
            AnalyzeHeader(report);

            // Find all markers
            FindAllMarkers(report);

            // Analyze data types distribution
            AnalyzeDataTypes(report);

            // Find repeated patterns
            FindPatterns(report);

            // Identify section structure
            AnalyzeSections(report);

            return report;
        }

        private void AnalyzeHeader(BinaryStructureReport report)
        {
            Console.WriteLine("=== Header Analysis ===");

            // Skip null byte if present
            if (_position < _data.Length && _data[_position] == 0x00)
            {
                report.HeaderHasNullPrefix = true;
                _position++;
                Console.WriteLine($"Null byte at position 0");
            }

            // Read header length
            if (_position < _data.Length)
            {
                byte headerLen = _data[_position++];
                if (_position + headerLen <= _data.Length)
                {
                    report.Header = Encoding.UTF8.GetString(_data, _position, headerLen);
                    _position += headerLen;
                    Console.WriteLine($"Header: '{report.Header}' (length: {headerLen})");
                    Console.WriteLine($"Header ends at offset: {_position} (0x{_position:X})\n");
                    report.HeaderEndOffset = _position;
                }
            }
        }

        private void FindAllMarkers(BinaryStructureReport report)
        {
            Console.WriteLine("=== Finding Known Markers ===");

            // Known markers
            var markers = new Dictionary<string, byte[]>
            {
                { "Rules", new byte[] { 0x05, 0x52, 0x75, 0x6C, 0x65, 0x73 } },
                { "Description", new byte[] { 0x07, 0x0B, 0x44, 0x65, 0x73, 0x63, 0x72, 0x69, 0x70, 0x74, 0x69, 0x6F, 0x6E } },
                { "Name", new byte[] { 0x07, 0x04, 0x4E, 0x61, 0x6D, 0x65 } },
                { "Enabled", new byte[] { 0x07, 0x07, 0x45, 0x6E, 0x61, 0x62, 0x6C, 0x65, 0x64 } }
            };

            foreach (var marker in markers)
            {
                var offsets = FindPattern(_data, marker.Value);
                report.Markers[marker.Key] = offsets;
                Console.WriteLine($"{marker.Key}: {offsets.Count} occurrences");
            }

            Console.WriteLine();
        }

        private void AnalyzeDataTypes(BinaryStructureReport report)
        {
            Console.WriteLine("=== Data Type Distribution ===");

            int boolCount = 0, intCount = 0, stringCount = 0, actionCount = 0;

            for (int i = 0; i < _data.Length; i++)
            {
                if (_data[i] == 0x01) boolCount++;
                else if (_data[i] == 0x04) intCount++;
                else if (_data[i] == 0x07) stringCount++;

                // Action string pattern
                if (i + 2 < _data.Length &&
                    _data[i] == 0x07 && _data[i+1] == 0x01 && _data[i+2] == 0x40)
                {
                    actionCount++;
                }
            }

            Console.WriteLine($"0x01 (Boolean marker): {boolCount}");
            Console.WriteLine($"0x04 (Integer marker): {intCount}");
            Console.WriteLine($"0x07 (String marker): {stringCount}");
            Console.WriteLine($"Action pattern (07 01 40): {actionCount}\n");

            report.DataTypeCounts["boolean"] = boolCount;
            report.DataTypeCounts["integer"] = intCount;
            report.DataTypeCounts["string"] = stringCount;
            report.DataTypeCounts["action"] = actionCount;
        }

        private void FindPatterns(BinaryStructureReport report)
        {
            Console.WriteLine("=== Repeated Byte Patterns ===");

            // Find sequences that repeat
            var sequences = new Dictionary<string, int>();

            // Check 4-byte patterns
            for (int i = 0; i < _data.Length - 4; i++)
            {
                var seq = $"{_data[i]:X2} {_data[i+1]:X2} {_data[i+2]:X2} {_data[i+3]:X2}";
                if (sequences.ContainsKey(seq))
                    sequences[seq]++;
                else
                    sequences[seq] = 1;
            }

            var topPatterns = sequences
                .Where(kv => kv.Value > 50) // More than 50 occurrences
                .OrderByDescending(kv => kv.Value)
                .Take(10);

            foreach (var pattern in topPatterns)
            {
                Console.WriteLine($"{pattern.Key}: {pattern.Value} times");
            }

            Console.WriteLine();
        }

        private void AnalyzeSections(BinaryStructureReport report)
        {
            Console.WriteLine("=== Section Structure Analysis ===");

            // Start after header
            _position = report.HeaderEndOffset;

            int sectionCount = 0;
            while (_position < _data.Length - 100)
            {
                var section = TryParseSection(_position);
                if (section != null)
                {
                    report.Sections.Add(section);
                    sectionCount++;
                    _position = section.EndOffset;
                }
                else
                {
                    _position++;
                }
            }

            Console.WriteLine($"Identified {sectionCount} potential sections\n");

            // Show first few sections
            Console.WriteLine("=== First 10 Sections ===");
            foreach (var section in report.Sections.Take(10))
            {
                Console.WriteLine($"Offset {section.StartOffset}-{section.EndOffset}: {section.Name}");
                Console.WriteLine($"  Items: {section.Items.Count}, Size: {section.Size} bytes");
            }

            Console.WriteLine();
        }

        private SectionInfo TryParseSection(int offset)
        {
            // Try to identify section start
            // Common pattern: Pascal string followed by items

            if (offset + 10 > _data.Length)
                return null;

            // Look for string marker
            if (_data[offset] != 0x07)
                return null;

            byte nameLen = _data[offset + 1];
            if (nameLen < 3 || nameLen > 50 || offset + 2 + nameLen > _data.Length)
                return null;

            string name = Encoding.UTF8.GetString(_data, offset + 2, nameLen);

            // Validate name
            if (!name.All(c => char.IsLetterOrDigit(c) || c == '_' || c == ' '))
                return null;

            var section = new SectionInfo
            {
                StartOffset = offset,
                Name = name,
                NameLength = nameLen
            };

            int pos = offset + 2 + nameLen;

            // Parse items in section until we hit another section or Rules
            int itemCount = 0;
            while (pos < _data.Length - 10 && itemCount < 100)
            {
                var item = TryParseItem(pos);
                if (item != null)
                {
                    section.Items.Add(item);
                    pos = item.EndOffset;
                    itemCount++;
                }
                else
                {
                    // Check if we hit next section or Rules
                    if (IsLikelyNextSection(pos))
                        break;
                    pos++;
                }
            }

            section.EndOffset = pos;
            section.Size = pos - offset;

            return section.Items.Count > 0 ? section : null;
        }

        private ItemInfo TryParseItem(int offset)
        {
            if (offset + 5 > _data.Length)
                return null;

            // Item pattern: 07 <len> <key> <type> <value>
            if (_data[offset] != 0x07)
                return null;

            byte keyLen = _data[offset + 1];
            if (keyLen < 2 || keyLen > 40 || offset + 2 + keyLen + 2 > _data.Length)
                return null;

            string key = Encoding.UTF8.GetString(_data, offset + 2, keyLen);

            // Validate key
            if (!key.All(c => char.IsLetterOrDigit(c) || c == '_' || c == '-'))
                return null;

            int valueOffset = offset + 2 + keyLen;
            byte typeMarker = _data[valueOffset];

            var item = new ItemInfo
            {
                StartOffset = offset,
                Key = key,
                TypeMarker = typeMarker
            };

            // Parse value based on type
            switch (typeMarker)
            {
                case 0x01: // Boolean
                    if (valueOffset + 2 <= _data.Length)
                    {
                        item.Value = _data[valueOffset + 1] != 0;
                        item.ValueType = "boolean";
                        item.EndOffset = valueOffset + 2;
                        return item;
                    }
                    break;

                case 0x04: // Integer
                    if (valueOffset + 5 <= _data.Length)
                    {
                        item.Value = BitConverter.ToInt32(_data, valueOffset + 1);
                        item.ValueType = "integer";
                        item.EndOffset = valueOffset + 5;
                        return item;
                    }
                    break;

                case 0x07: // String
                    if (valueOffset + 2 <= _data.Length)
                    {
                        byte strLen = _data[valueOffset + 1];
                        if (strLen > 0 && valueOffset + 2 + strLen <= _data.Length)
                        {
                            item.Value = Encoding.UTF8.GetString(_data, valueOffset + 2, strLen);
                            item.ValueType = "string";
                            item.EndOffset = valueOffset + 2 + strLen;
                            return item;
                        }
                    }
                    break;
            }

            return null;
        }

        private bool IsLikelyNextSection(int pos)
        {
            if (pos + 10 > _data.Length)
                return false;

            // Check for Rules marker
            if (_data[pos] == 0x05 && pos + 6 <= _data.Length)
            {
                string marker = Encoding.UTF8.GetString(_data, pos + 1, 5);
                if (marker == "Rules")
                    return true;
            }

            // Check for section name pattern
            if (_data[pos] == 0x07)
            {
                byte len = _data[pos + 1];
                if (len >= 5 && len <= 30 && pos + 2 + len <= _data.Length)
                {
                    string name = Encoding.UTF8.GetString(_data, pos + 2, len);
                    return name.All(c => char.IsLetterOrDigit(c) || c == '_');
                }
            }

            return false;
        }

        private List<int> FindPattern(byte[] data, byte[] pattern)
        {
            var offsets = new List<int>();

            for (int i = 0; i <= data.Length - pattern.Length; i++)
            {
                bool match = true;
                for (int j = 0; j < pattern.Length; j++)
                {
                    if (data[i + j] != pattern[j])
                    {
                        match = false;
                        break;
                    }
                }

                if (match)
                    offsets.Add(i);
            }

            return offsets;
        }
    }

    public class BinaryStructureReport
    {
        public int FileSize { get; set; }
        public bool HeaderHasNullPrefix { get; set; }
        public string Header { get; set; } = "";
        public int HeaderEndOffset { get; set; }
        public Dictionary<string, List<int>> Markers { get; set; } = new();
        public Dictionary<string, int> DataTypeCounts { get; set; } = new();
        public List<SectionInfo> Sections { get; set; } = new();
    }

    public class SectionInfo
    {
        public int StartOffset { get; set; }
        public int EndOffset { get; set; }
        public int Size { get; set; }
        public string Name { get; set; } = "";
        public int NameLength { get; set; }
        public List<ItemInfo> Items { get; set; } = new();
    }

    public class ItemInfo
    {
        public int StartOffset { get; set; }
        public int EndOffset { get; set; }
        public string Key { get; set; } = "";
        public byte TypeMarker { get; set; }
        public string ValueType { get; set; } = "";
        public object Value { get; set; }
    }
}