DCIT/DCIT.Core/Services/DiffSerializationService.cs

using System;
using System.IO;
using System.Reflection;
using System.Text;
using DCIT.Core.Models;
using DCIT.Core.Utilities;
using Newtonsoft.Json;

namespace DCIT.Core.Services
{
    public class DiffSerializationService
    {
        private static readonly byte[] PayloadCiphertextPropertyMarker = Encoding.UTF8.GetBytes("\"PayloadCiphertext\":");
        private static readonly byte[] PayloadCiphertextValueMarker = Encoding.UTF8.GetBytes("\"PayloadCiphertext\":\"");
        private static readonly byte[] MetadataSyntheticSuffix = Encoding.UTF8.GetBytes("null,\"Payload\":null}");
        private const int MetadataScanCapBytes = 1 << 20;

        private readonly BmpDiffCodec _bmpDiffCodec;
        private readonly string _appVersion;

        public DiffSerializationService(BmpDiffCodec bmpDiffCodec)
        {
            _bmpDiffCodec = bmpDiffCodec;
            _appVersion = Assembly.GetExecutingAssembly().GetName().Version != null
                ? Assembly.GetExecutingAssembly().GetName().Version.ToString()
                : "1.0.0";
        }

        public string AppVersion
        {
            get { return _appVersion; }
        }

        public string GetBmpPath(string diffPath)
        {
            return Path.ChangeExtension(diffPath, ".bmp");
        }

        public string Write(DiffDocument plainDiff, string diffPath, bool encryptJson)
        {
            if (plainDiff == null)
            {
                throw new ArgumentNullException("plainDiff");
            }

            if (plainDiff.Payload == null)
            {
                throw new InvalidOperationException("差异载荷不能为空。");
            }

            var originalPayload = plainDiff.Payload;
            var originalEncrypted = plainDiff.Encrypted;
            var originalEncryption = plainDiff.Encryption;
            var originalPayloadCiphertext = plainDiff.PayloadCiphertext;

            string payloadTmpPath = null;
            string cipherTmpPath = null;

            try
            {
                payloadTmpPath = Path.Combine(Path.GetTempPath(), "dcit-payload-" + Guid.NewGuid().ToString("N") + ".bin");
                SerializeInstanceToUtf8File(plainDiff.Payload, payloadTmpPath);
                plainDiff.Integrity.PayloadHash = HashUtility.ComputeFileSha256Base64(payloadTmpPath);
                plainDiff.Integrity.TamperProtectionValue = HashUtility.ComputeFileHmacBase64(payloadTmpPath);

                plainDiff.Encrypted = false;
                plainDiff.Encryption = null;
                plainDiff.PayloadCiphertext = null;

                Directory.CreateDirectory(Path.GetDirectoryName(diffPath) ?? string.Empty);
                var bmpPath = GetBmpPath(diffPath);

                // Serialize the full plainDiff (with Payload attached) to BMP. Use a temp file so we
                // never hold the whole BMP-bound JSON in memory.
                var bmpTmpPath = Path.Combine(Path.GetTempPath(), "dcit-bmp-" + Guid.NewGuid().ToString("N") + ".json");
                try
                {
                    SerializeInstanceToUtf8File(plainDiff, bmpTmpPath);
                    using (var bmpFile = File.OpenRead(bmpTmpPath))
                    {
                        _bmpDiffCodec.EncodeFromStream(bmpFile, bmpPath);
                    }
                }
                finally
                {
                    TryDeleteFile(bmpTmpPath);
                }

                if (encryptJson)
                {
                    cipherTmpPath = Path.Combine(Path.GetTempPath(), "dcit-cipher-" + Guid.NewGuid().ToString("N") + ".bin");
                    EncryptedPayloadHeader header;
                    using (var payloadStream = File.OpenRead(payloadTmpPath))
                    using (var cipherStream = File.Create(cipherTmpPath))
                    {
                        header = HashUtility.EncryptStream(payloadStream, cipherStream);
                    }

                    plainDiff.Encrypted = true;
                    plainDiff.Encryption = new DiffEncryptionInfo
                    {
                        Algorithm = "AES-256-CBC+HMAC-SHA256",
                        KeyId = "default",
                        Nonce = header.NonceBase64,
                        Tag = header.TagBase64,
                    };
                    plainDiff.PayloadCiphertext = null;
                    plainDiff.Payload = null;

                    WriteEncryptedDiffToFile(plainDiff, diffPath, cipherTmpPath);
                }
                else
                {
                    SerializeToFile(plainDiff, diffPath);
                }

                return bmpPath;
            }
            finally
            {
                plainDiff.Payload = originalPayload;
                plainDiff.Encrypted = originalEncrypted;
                plainDiff.Encryption = originalEncryption;
                plainDiff.PayloadCiphertext = originalPayloadCiphertext;

                TryDeleteFile(payloadTmpPath);
                TryDeleteFile(cipherTmpPath);
            }
        }

        private static void SerializeInstanceToUtf8File(object instance, string path)
        {
            using (var stream = File.Create(path))
            using (var streamWriter = new StreamWriter(stream, new UTF8Encoding(false), bufferSize: 65536))
            using (var jsonWriter = new JsonTextWriter(streamWriter))
            {
                var serializer = JsonSerializer.CreateDefault();
                serializer.Formatting = Formatting.None;
                serializer.Serialize(jsonWriter, instance);
                jsonWriter.Flush();
            }
        }

        private static void WriteEncryptedDiffToFile(DiffDocument document, string diffPath, string cipherTempPath)
        {
            var metadataJson = JsonConvert.SerializeObject(document, Formatting.None);
            const string placeholder = "\"PayloadCiphertext\":null";
            var placeholderIndex = metadataJson.IndexOf(placeholder, StringComparison.Ordinal);
            if (placeholderIndex < 0)
            {
                throw new InvalidOperationException("加密差异元数据缺少 PayloadCiphertext 占位符。");
            }

            using (var fileStream = File.Create(diffPath))
            using (var writer = new StreamWriter(fileStream, new UTF8Encoding(false), bufferSize: 65536))
            {
                writer.Write(metadataJson.Substring(0, placeholderIndex));
                writer.Write("\"PayloadCiphertext\":\"");
                using (var cipherStream = File.OpenRead(cipherTempPath))
                {
                    var buffer = new byte[30720];
                    int read;
                    while ((read = cipherStream.Read(buffer, 0, buffer.Length)) > 0)
                    {
                        writer.Write(Convert.ToBase64String(buffer, 0, read));
                    }
                }

                writer.Write('"');
                writer.Write(metadataJson.Substring(placeholderIndex + placeholder.Length));
            }
        }

        private static void TryDeleteFile(string path)
        {
            if (string.IsNullOrWhiteSpace(path))
            {
                return;
            }

            try
            {
                if (File.Exists(path))
                {
                    File.Delete(path);
                }
            }
            catch
            {
                // Best-effort cleanup; temp files live in the system temp dir.
            }
        }

        public LoadedDiffDocument Load(string diffOrBmpPath)
        {
            if (diffOrBmpPath == null)
            {
                throw new ArgumentNullException("diffOrBmpPath");
            }

            if (Path.GetExtension(diffOrBmpPath).Equals(".bmp", StringComparison.OrdinalIgnoreCase))
            {
                return LoadFromBmp(diffOrBmpPath);
            }

            return LoadFromDiff(diffOrBmpPath);
        }

        public DiffDocument ReadMetadata(string diffOrBmpPath)
        {
            if (diffOrBmpPath == null)
            {
                throw new ArgumentNullException("diffOrBmpPath");
            }

            if (Path.GetExtension(diffOrBmpPath).Equals(".bmp", StringComparison.OrdinalIgnoreCase))
            {
                var json = _bmpDiffCodec.DecodeJson(diffOrBmpPath);
                return JsonConvert.DeserializeObject<DiffDocument>(json) ?? new DiffDocument();
            }

            return ReadDiffMetadata(diffOrBmpPath);
        }

        public DiffDocument CreateBaseDocument()
        {
            return new DiffDocument
            {
                App = new DiffAppInfo
                {
                    Name = "DCIT",
                    Version = _appVersion,
                },
                Algorithm = new DiffAlgorithmInfo
                {
                    DiffVersion = "1.0",
                    BmpCodecVersion = "1.0",
                },
            };
        }

        // BMP diffs use a legacy in-memory load path. Large BMP-encoded diffs remain a known
        // OOM risk; the default and reported path is .diff, which is fully streaming below.
        private LoadedDiffDocument LoadFromBmp(string bmpPath)
        {
            var json = _bmpDiffCodec.DecodeJson(bmpPath);
            var document = JsonConvert.DeserializeObject<DiffDocument>(json);
            if (document == null)
            {
                throw new InvalidDataException("差异文件解析失败。");
            }

            if (document.Encrypted)
            {
                if (document.Encryption == null || string.IsNullOrWhiteSpace(document.PayloadCiphertext))
                {
                    throw new InvalidDataException("加密差异文件缺少必要字段。");
                }

                var payloadBytes = HashUtility.Decrypt(
                    document.PayloadCiphertext,
                    document.Encryption.Nonce,
                    document.Encryption.Tag);
                var payloadJson = Encoding.UTF8.GetString(payloadBytes);
                document.Payload = JsonConvert.DeserializeObject<DiffPayload>(payloadJson);
                ValidateIntegrity(document, payloadBytes);
            }
            else
            {
                if (document.Payload == null)
                {
                    throw new InvalidDataException("明文差异文件缺少 payload。");
                }

                ValidateIntegrity(document, Encoding.UTF8.GetBytes(SerializeCanonical(document.Payload)));
            }

            return new LoadedDiffDocument
            {
                Document = document,
            };
        }

        private LoadedDiffDocument LoadFromDiff(string diffPath)
        {
            var document = ReadDiffMetadata(diffPath);
            string payloadTmpPath = null;
            string cipherTmpPath = null;

            try
            {
                if (document.Encrypted)
                {
                    if (document.Encryption == null
                        || string.IsNullOrWhiteSpace(document.Encryption.Nonce)
                        || string.IsNullOrWhiteSpace(document.Encryption.Tag))
                    {
                        throw new InvalidDataException("加密差异文件缺少加密元数据。");
                    }

                    cipherTmpPath = Path.Combine(Path.GetTempPath(), "dcit-rcipher-" + Guid.NewGuid().ToString("N") + ".bin");
                    payloadTmpPath = Path.Combine(Path.GetTempPath(), "dcit-rpayload-" + Guid.NewGuid().ToString("N") + ".bin");

                    ExtractBase64RegionToFile(diffPath, PayloadCiphertextValueMarker, cipherTmpPath);

                    var iv = Convert.FromBase64String(document.Encryption.Nonce);
                    var tag = Convert.FromBase64String(document.Encryption.Tag);
                    using (var cipherStream = File.OpenRead(cipherTmpPath))
                    using (var payloadStream = File.Create(payloadTmpPath))
                    {
                        HashUtility.DecryptStream(cipherStream, payloadStream, iv, tag);
                    }

                    ValidateIntegrityFromPayloadFile(document, payloadTmpPath);
                    document.Payload = DeserializePayloadFromFile(payloadTmpPath);
                }
                else
                {
                    // Stream-deserialize the whole document. PayloadCiphertext is null (small) and
                    // Payload is a large nested object that Newtonsoft builds incrementally without
                    // buffering the whole JSON text or any single huge string token.
                    var full = StreamDeserializeDiffDocument(diffPath);
                    if (full == null)
                    {
                        throw new InvalidDataException("差异文件解析失败。");
                    }

                    if (full.Payload == null)
                    {
                        throw new InvalidDataException("明文差异文件缺少 payload。");
                    }

                    document = full;
                    payloadTmpPath = Path.Combine(Path.GetTempPath(), "dcit-rpayload-" + Guid.NewGuid().ToString("N") + ".bin");
                    SerializeInstanceToUtf8File(document.Payload, payloadTmpPath);
                    ValidateIntegrityFromPayloadFile(document, payloadTmpPath);
                }

                return new LoadedDiffDocument
                {
                    Document = document,
                };
            }
            finally
            {
                TryDeleteFile(payloadTmpPath);
                TryDeleteFile(cipherTmpPath);
            }
        }

        // Reads only the metadata portion of a .diff file. The huge fields (PayloadCiphertext value
        // and Payload value) sit after the metadata in declaration order; we byte-scan up to the
        // "PayloadCiphertext" property marker (a few KB), synthesize null values for the two huge
        // fields, and parse the resulting small metadata JSON. The huge content is never read.
        private DiffDocument ReadDiffMetadata(string diffPath)
        {
            var prefix = ReadPrefixThroughMarker(diffPath, PayloadCiphertextPropertyMarker, MetadataScanCapBytes);
            string metadataJson;
            if (prefix == null)
            {
                metadataJson = File.ReadAllText(diffPath, Encoding.UTF8);
            }
            else
            {
                var combined = new byte[prefix.Length + MetadataSyntheticSuffix.Length];
                Buffer.BlockCopy(prefix, 0, combined, 0, prefix.Length);
                Buffer.BlockCopy(MetadataSyntheticSuffix, 0, combined, prefix.Length, MetadataSyntheticSuffix.Length);
                metadataJson = Encoding.UTF8.GetString(combined);
            }

            return JsonConvert.DeserializeObject<DiffDocument>(metadataJson) ?? new DiffDocument();
        }

        private static DiffDocument StreamDeserializeDiffDocument(string diffPath)
        {
            using (var stream = File.OpenRead(diffPath))
            using (var streamReader = new StreamReader(stream, new UTF8Encoding(false)))
            using (var jsonReader = new JsonTextReader(streamReader))
            {
                return JsonSerializer.CreateDefault().Deserialize<DiffDocument>(jsonReader);
            }
        }

        private static DiffPayload DeserializePayloadFromFile(string payloadPath)
        {
            using (var stream = File.OpenRead(payloadPath))
            using (var streamReader = new StreamReader(stream, new UTF8Encoding(false)))
            using (var jsonReader = new JsonTextReader(streamReader))
            {
                return JsonSerializer.CreateDefault().Deserialize<DiffPayload>(jsonReader);
            }
        }

        private static void ValidateIntegrity(DiffDocument document, byte[] payloadBytes)
        {
            var actualHash = HashUtility.ComputeSha256Base64(payloadBytes);
            var actualTamper = HashUtility.ComputeHmacBase64(payloadBytes);
            if (!string.Equals(actualHash, document.Integrity == null ? null : document.Integrity.PayloadHash, StringComparison.Ordinal))
            {
                throw new InvalidDataException("差异文件载荷哈希校验失败。");
            }

            if (!string.Equals(actualTamper, document.Integrity == null ? null : document.Integrity.TamperProtectionValue, StringComparison.Ordinal))
            {
                throw new InvalidDataException("差异文件防篡改校验失败。");
            }
        }

        private static void ValidateIntegrityFromPayloadFile(DiffDocument document, string payloadPath)
        {
            var actualHash = HashUtility.ComputeFileSha256Base64(payloadPath);
            var actualTamper = HashUtility.ComputeFileHmacBase64(payloadPath);
            if (!string.Equals(actualHash, document.Integrity == null ? null : document.Integrity.PayloadHash, StringComparison.Ordinal))
            {
                throw new InvalidDataException("差异文件载荷哈希校验失败。");
            }

            if (!string.Equals(actualTamper, document.Integrity == null ? null : document.Integrity.TamperProtectionValue, StringComparison.Ordinal))
            {
                throw new InvalidDataException("差异文件防篡改校验失败。");
            }
        }

        private static string SerializeCanonical(object instance)
        {
            return JsonConvert.SerializeObject(instance, Formatting.None);
        }

        private static void SerializeToFile(object instance, string path)
        {
            using (var stream = File.Create(path))
            using (var streamWriter = new StreamWriter(stream, new UTF8Encoding(false), bufferSize: 65536))
            using (var jsonWriter = new JsonTextWriter(streamWriter))
            {
                var serializer = JsonSerializer.CreateDefault();
                serializer.Formatting = Formatting.None;
                serializer.Serialize(jsonWriter, instance);
                jsonWriter.Flush();
            }
        }

        // Reads bytes from the start of the file up to and including the first occurrence of marker.
        // Returns null if the marker is not found within cap bytes. Caller chooses cap so the marker
        // (which sits after the small metadata block) is found without reading huge payload content.
        private static byte[] ReadPrefixThroughMarker(string path, byte[] marker, int cap)
        {
            using (var stream = File.OpenRead(path))
            {
                var ms = new MemoryStream();
                var buffer = new byte[65536];
                int read;
                while ((read = stream.Read(buffer, 0, buffer.Length)) > 0)
                {
                    ms.Write(buffer, 0, read);
                    if (ms.Length >= cap)
                    {
                        break;
                    }
                }

                var arr = ms.ToArray();
                var idx = IndexOfSubarray(arr, marker);
                if (idx < 0)
                {
                    return null;
                }

                var result = new byte[idx + marker.Length];
                Buffer.BlockCopy(arr, 0, result, 0, result.Length);
                return result;
            }
        }

        // Locates valueMarker in the file, then stream-decodes the base64 region between the marker
        // and the next closing double-quote into outputPath. Decodes in 4096-char chunks (a multiple
        // of 4) so there is no mid-chunk base64 alignment error.
        private static void ExtractBase64RegionToFile(string diffPath, byte[] valueMarker, string outputPath)
        {
            long markerEndPosition;
            using (var probe = File.OpenRead(diffPath))
            {
                markerEndPosition = FindMarkerEnd(probe, valueMarker);
            }

            if (markerEndPosition < 0)
            {
                throw new InvalidDataException("差异文件缺少密文字段标记。");
            }

            using (var stream = File.OpenRead(diffPath))
            {
                stream.Position = markerEndPosition;
                using (var outStream = File.Create(outputPath))
                {
                    var charBuf = new char[4096];
                    var charLen = 0;
                    var byteBuf = new byte[65536];
                    int read;
                    var done = false;
                    while (!done && (read = stream.Read(byteBuf, 0, byteBuf.Length)) > 0)
                    {
                        for (var i = 0; i < read; i++)
                        {
                            var b = byteBuf[i];
                            if (b == '"')
                            {
                                done = true;
                                break;
                            }

                            charBuf[charLen++] = (char)b;
                            if (charLen == charBuf.Length)
                            {
                                var decoded = Convert.FromBase64CharArray(charBuf, 0, charLen);
                                outStream.Write(decoded, 0, decoded.Length);
                                charLen = 0;
                            }
                        }
                    }

                    if (charLen > 0)
                    {
                        var decoded = Convert.FromBase64CharArray(charBuf, 0, charLen);
                        outStream.Write(decoded, 0, decoded.Length);
                    }
                }
            }
        }

        private static long FindMarkerEnd(Stream stream, byte[] marker)
        {
            var ms = new MemoryStream();
            var buffer = new byte[65536];
            int read;
            while ((read = stream.Read(buffer, 0, buffer.Length)) > 0)
            {
                ms.Write(buffer, 0, read);
                var arr = ms.ToArray();
                var idx = IndexOfSubarray(arr, marker);
                if (idx >= 0)
                {
                    return idx + marker.Length;
                }

                if (ms.Length > MetadataScanCapBytes)
                {
                    return -1;
                }
            }

            return -1;
        }

        private static int IndexOfSubarray(byte[] haystack, byte[] needle)
        {
            if (needle == null || needle.Length == 0)
            {
                return 0;
            }

            if (haystack == null || haystack.Length < needle.Length)
            {
                return -1;
            }

            for (var i = 0; i <= haystack.Length - needle.Length; i++)
            {
                var match = true;
                for (var j = 0; j < needle.Length; j++)
                {
                    if (haystack[i + j] != needle[j])
                    {
                        match = false;
                        break;
                    }
                }

                if (match)
                {
                    return i;
                }
            }

            return -1;
        }
    }

    public class LoadedDiffDocument
    {
        public DiffDocument Document { get; set; }
    }
}