/// <summary>
/// Delete an object stored in the deduplication index.
/// This method will use the callbacks supplied in the method signature.
/// </summary>
/// <param name="objectName">The name of the object.</param>
/// <param name="callbacks">CallbackMethods object containing callback methods.</param>
/// <returns>True if successful.</returns>
public bool DeleteObject(string objectName, CallbackMethods callbacks)
{
if (String.IsNullOrEmpty(objectName))
{
throw new ArgumentNullException(nameof(objectName));
}
if (callbacks == null)
{
throw new ArgumentNullException(nameof(callbacks));
}
if (callbacks.DeleteChunk == null)
{
throw new ArgumentException("DeleteChunk callback must be specified.");
}
objectName = DedupeCommon.SanitizeString(objectName);
List <string> garbageCollectChunks = null;
lock (_ChunkLock)
{
_Database.DeleteObjectChunks(objectName, out garbageCollectChunks);
if (garbageCollectChunks != null && garbageCollectChunks.Count > 0)
{
foreach (string key in garbageCollectChunks)
{
if (!callbacks.DeleteChunk(key))
{
Log("Unable to delete chunk: " + key);
}
}
}
}
return(true);
}
/// <summary>
/// Initialize an existing index using an external database. Tables must be created ahead of time.
/// </summary>
/// <param name="database">Database provider implemented using the Database.DbProvider class.</param>
/// <param name="writeChunkMethod">Method to call to write a chunk to storage.</param>
/// <param name="readChunkMethod">Method to call to read a chunk from storage.</param>
/// <param name="deleteChunkMethod">Method to call to delete a chunk from storage.</param>
/// <param name="debugDedupe">Enable console logging for deduplication operations.</param>
/// <param name="debugSql">Enable console logging for SQL operations.</param>
public DedupeLibrary(DbProvider database, Func <Chunk, bool> writeChunkMethod, Func <string, byte[]> readChunkMethod, Func <string, bool> deleteChunkMethod, bool debugDedupe, bool debugSql)
{
if (writeChunkMethod == null)
{
throw new ArgumentNullException(nameof(writeChunkMethod));
}
if (readChunkMethod == null)
{
throw new ArgumentNullException(nameof(readChunkMethod));
}
if (deleteChunkMethod == null)
{
throw new ArgumentNullException(nameof(deleteChunkMethod));
}
_Database = database ?? throw new ArgumentNullException(nameof(database));
Callbacks = new CallbackMethods
{
WriteChunk = writeChunkMethod,
ReadChunk = readChunkMethod,
DeleteChunk = deleteChunkMethod
};
DebugDedupe = debugDedupe;
DebugSql = debugSql;
_ChunkLock = new object();
InitFromExistingIndex();
}
/// <summary>
/// Retrieve a read-only stream over an object that has been stored.
/// </summary>
/// <param name="objectName">The name of the object.</param>
/// <param name="callbacks">CallbackMethods object containing callback methods.</param>
/// <param name="stream">Read-only stream.</param>
/// <returns>True if successful.</returns>
public bool RetrieveObjectStream(string objectName, CallbackMethods callbacks, out DedupeStream stream)
{
stream = null;
if (String.IsNullOrEmpty(objectName))
{
throw new ArgumentNullException(nameof(objectName));
}
if (callbacks == null)
{
throw new ArgumentNullException(nameof(callbacks));
}
if (callbacks.ReadChunk == null)
{
throw new ArgumentException("ReadChunk callback must be specified.");
}
objectName = DedupeCommon.SanitizeString(objectName);
if (!RetrieveObjectMetadata(objectName, out ObjectMetadata md))
{
return(false);
}
stream = new DedupeStream(md, _Database, callbacks);
return(true);
}
/// <summary>
/// Retrieve an object from the deduplication index.
/// This method will use the callbacks supplied in the method signature.
/// </summary>
/// <param name="objectName">The name of the object.</param>
/// <param name="callbacks">CallbackMethods object containing callback methods.</param>
/// <param name="data">The byte data from the object.</param>
/// <returns>True if successful.</returns>
public bool RetrieveObject(string objectName, CallbackMethods callbacks, out byte[] data)
{
bool success = RetrieveObject(objectName, callbacks, out long contentLength, out Stream stream);
data = DedupeCommon.StreamToBytes(stream);
return(success);
}
/// <summary>
/// Store an object within a container in the deduplication index if it doesn't already exist, or, replace the object if it does.
/// This method will use the callbacks supplied in the method signature.
/// </summary>
/// <param name="objectName">The name of the object. Must be unique in the index.</param>
/// <param name="callbacks">CallbackMethods object containing callback methods.</param>
/// <param name="data">The byte data for the object.</param>
/// <param name="chunks">The list of chunks identified during the deduplication operation.</param>
/// <returns>True if successful.</returns>
public bool StoreOrReplaceObject(string objectName, CallbackMethods callbacks, byte[] data, out List <Chunk> chunks)
{
if (data == null || data.Length < 1)
{
throw new ArgumentNullException(nameof(data));
}
return(StoreOrReplaceObject(objectName, callbacks, data.Length, DedupeCommon.BytesToStream(data), out chunks));
}
/// <summary>
/// Retrieve an object from the deduplication index.
/// This method will use the callbacks supplied in the method signature.
/// </summary>
/// <param name="objectName">The name of the object.</param>
/// <param name="callbacks">CallbackMethods object containing callback methods.</param>
/// <param name="contentLength">The length of the data.</param>
/// <param name="stream">The stream containing the data.</param>
/// <returns>True if successful.</returns>
public bool RetrieveObject(string objectName, CallbackMethods callbacks, out long contentLength, out Stream stream)
{
stream = null;
contentLength = 0;
if (String.IsNullOrEmpty(objectName))
{
throw new ArgumentNullException(nameof(objectName));
}
if (callbacks == null)
{
throw new ArgumentNullException(nameof(callbacks));
}
if (callbacks.ReadChunk == null)
{
throw new ArgumentException("ReadChunk callback must be specified.");
}
objectName = DedupeCommon.SanitizeString(objectName);
ObjectMetadata md = null;
lock (_ChunkLock)
{
if (!_Database.GetObjectMetadata(objectName, out md))
{
Log("Unable to retrieve object metadata for object " + objectName);
return(false);
}
if (md.Chunks == null || md.Chunks.Count < 1)
{
Log("No chunks returned");
return(false);
}
stream = new MemoryStream();
foreach (Chunk curr in md.Chunks)
{
byte[] chunkData = callbacks.ReadChunk(curr.Key);
if (chunkData == null || chunkData.Length < 1)
{
Log("Unable to read chunk " + curr.Key);
return(false);
}
stream.Write(chunkData, 0, chunkData.Length);
contentLength += chunkData.Length;
}
if (contentLength > 0)
{
stream.Seek(0, SeekOrigin.Begin);
}
}
return(true);
}
/// <summary>
/// Store an object within a container in the deduplication index if it doesn't already exist, or, replace the object if it does.
/// This method will use the callbacks supplied in the method signature.
/// </summary>
/// <param name="objectName">The name of the object. Must be unique in the index.</param>
/// <param name="callbacks">CallbackMethods object containing callback methods.</param>
/// <param name="contentLength">The length of the data.</param>
/// <param name="stream">The stream containing the data.</param>
/// <param name="chunks">The list of chunks identified during the deduplication operation.</param>
/// <returns>True if successful.</returns>
public bool StoreOrReplaceObject(string objectName, CallbackMethods callbacks, long contentLength, Stream stream, out List <Chunk> chunks)
{
#region Initialize
chunks = new List <Chunk>();
if (String.IsNullOrEmpty(objectName))
{
throw new ArgumentNullException(nameof(objectName));
}
if (callbacks == null)
{
throw new ArgumentNullException(nameof(callbacks));
}
if (callbacks.WriteChunk == null)
{
throw new ArgumentException("WriteChunk callback must be specified.");
}
if (callbacks.DeleteChunk == null)
{
throw new ArgumentException("DeleteChunk callback must be specified.");
}
if (stream == null)
{
throw new ArgumentNullException(nameof(stream));
}
if (!stream.CanRead)
{
throw new ArgumentException("Cannot read from supplied stream.");
}
objectName = DedupeCommon.SanitizeString(objectName);
#endregion
#region Delete-if-Exists
if (_Database.ObjectExists(objectName))
{
Log("Object " + objectName + " already exists, deleting");
if (!DeleteObject(objectName))
{
Log("Unable to delete existing object");
return(false);
}
else
{
Log("Successfully deleted object for replacement");
}
}
#endregion
return(StoreObject(objectName, callbacks, contentLength, stream, out chunks));
}
internal DedupeStream(ObjectMetadata md, DbProvider db, CallbackMethods callbacks)
{
if (md == null)
{
throw new ArgumentNullException(nameof(md));
}
if (db == null)
{
throw new ArgumentNullException(nameof(db));
}
if (callbacks == null)
{
throw new ArgumentNullException(nameof(callbacks));
}
_Metadata = md;
_Database = db;
_Callbacks = callbacks;
}
/// <summary>
/// Initialize an existing index using an internal Sqlite database.
/// </summary>
/// <param name="indexFile">Path and filename.</param>
/// <param name="writeChunkMethod">Method to call to write a chunk to storage.</param>
/// <param name="readChunkMethod">Method to call to read a chunk from storage.</param>
/// <param name="deleteChunkMethod">Method to call to delete a chunk from storage.</param>
/// <param name="debugDedupe">Enable console logging for deduplication operations.</param>
/// <param name="debugSql">Enable console logging for SQL operations.</param>
public DedupeLibrary(string indexFile, Func <Chunk, bool> writeChunkMethod, Func <string, byte[]> readChunkMethod, Func <string, bool> deleteChunkMethod, bool debugDedupe, bool debugSql)
{
if (String.IsNullOrEmpty(indexFile))
{
throw new ArgumentNullException(nameof(indexFile));
}
if (!File.Exists(indexFile))
{
throw new FileNotFoundException("Index file not found.");
}
if (writeChunkMethod == null)
{
throw new ArgumentNullException(nameof(writeChunkMethod));
}
if (readChunkMethod == null)
{
throw new ArgumentNullException(nameof(readChunkMethod));
}
if (deleteChunkMethod == null)
{
throw new ArgumentNullException(nameof(deleteChunkMethod));
}
_IndexFile = DedupeCommon.SanitizeString(indexFile);
Callbacks = new CallbackMethods
{
WriteChunk = writeChunkMethod,
ReadChunk = readChunkMethod,
DeleteChunk = deleteChunkMethod
};
DebugDedupe = debugDedupe;
DebugSql = debugSql;
_ChunkLock = new object();
_Database = new SqliteProvider(_IndexFile, DebugSql);
InitFromExistingIndex();
}
/// <summary>
/// Store an object in the deduplication index.
/// This method will use the callbacks supplied in the method signature.
/// </summary>
/// <param name="objectName">The name of the object. Must be unique in the index.</param>
/// <param name="callbacks">CallbackMethods object containing callback methods.</param>
/// <param name="contentLength">The length of the data.</param>
/// <param name="stream">The stream containing the data.</param>
/// <param name="chunks">The list of chunks identified during the deduplication operation.</param>
/// <returns>True if successful.</returns>
public bool StoreObject(string objectName, CallbackMethods callbacks, long contentLength, Stream stream, out List <Chunk> chunks)
{
#region Initialize
chunks = new List <Chunk>();
if (String.IsNullOrEmpty(objectName))
{
throw new ArgumentNullException(nameof(objectName));
}
if (callbacks == null)
{
throw new ArgumentNullException(nameof(callbacks));
}
if (callbacks.WriteChunk == null)
{
throw new ArgumentException("WriteChunk callback must be specified.");
}
if (callbacks.DeleteChunk == null)
{
throw new ArgumentException("DeleteChunk callback must be specified.");
}
if (contentLength < 1)
{
throw new ArgumentException("Content length must be at least one byte.");
}
if (stream == null)
{
throw new ArgumentNullException(nameof(stream));
}
if (!stream.CanRead)
{
throw new ArgumentException("Cannot read from supplied stream.");
}
objectName = DedupeCommon.SanitizeString(objectName);
if (_Database.ObjectExists(objectName))
{
Log("Object " + objectName + " already exists");
return(false);
}
bool garbageCollectionRequired = false;
#endregion
#region Chunk-Data
try
{
Func <Chunk, bool> processChunk = delegate(Chunk chunk)
{
if (chunk == null)
{
return(false);
}
lock (_ChunkLock)
{
if (!_Database.AddObjectChunk(objectName, contentLength, chunk))
{
Log("Unable to add chunk key " + chunk.Key);
garbageCollectionRequired = true;
return(false);
}
if (!callbacks.WriteChunk(chunk))
{
Log("Unable to write chunk key " + chunk.Key);
garbageCollectionRequired = true;
return(false);
}
}
return(true);
};
if (!ChunkStream(contentLength, stream, processChunk, out chunks))
{
Log("Unable to chunk object " + objectName);
garbageCollectionRequired = true;
return(false);
}
}
finally
{
if (garbageCollectionRequired)
{
List <string> garbageCollectKeys = new List <string>();
_Database.DeleteObjectChunks(objectName, out garbageCollectKeys);
if (garbageCollectKeys != null && garbageCollectKeys.Count > 0)
{
foreach (string key in garbageCollectKeys)
{
if (!callbacks.DeleteChunk(key))
{
Log("Unable to garbage collect chunk " + key);
}
}
}
}
}
#endregion
return(true);
}
/// <summary>
/// Create a new index using an external database. Tables must be created ahead of time.
/// </summary>
/// <param name="database">Database provider implemented using the Database.DbProvider class.</param>
/// <param name="minChunkSize">Minimum chunk size, must be divisible by 8, divisible by 64, and 128 or greater.</param>
/// <param name="maxChunkSize">Maximum chunk size, must be divisible by 8, divisible by 64, and at least 8 times larger than minimum chunk size.</param>
/// <param name="shiftCount">Number of bytes to shift while identifying chunk boundaries, must be less than or equal to minimum chunk size.</param>
/// <param name="boundaryCheckBytes">Number of bytes to examine while checking for a chunk boundary, must be 8 or fewer.</param>
/// <param name="writeChunkMethod">Method to call to write a chunk to storage.</param>
/// <param name="readChunkMethod">Method to call to read a chunk from storage.</param>
/// <param name="deleteChunkMethod">Method to call to delete a chunk from storage.</param>
/// <param name="debugDedupe">Enable console logging for deduplication operations.</param>
/// <param name="debugSql">Enable console logging for SQL operations.</param>
public DedupeLibrary(
DbProvider database,
int minChunkSize,
int maxChunkSize,
int shiftCount,
int boundaryCheckBytes,
Func <Chunk, bool> writeChunkMethod,
Func <string, byte[]> readChunkMethod,
Func <string, bool> deleteChunkMethod,
bool debugDedupe,
bool debugSql)
{
if (minChunkSize % 8 != 0)
{
throw new ArgumentException("Value for minChunkSize must be evenly divisible by 8.");
}
if (maxChunkSize % 8 != 0)
{
throw new ArgumentException("Value for maxChunkSize must be evenly divisible by 8.");
}
if (minChunkSize % 64 != 0)
{
throw new ArgumentException("Value for minChunkSize must be evenly divisible by 64.");
}
if (maxChunkSize % 64 != 0)
{
throw new ArgumentException("Value for maxChunkSize must be evenly divisible by 64.");
}
if (minChunkSize < 1024)
{
throw new ArgumentOutOfRangeException("Value for minChunkSize must be 256 or greater.");
}
if (maxChunkSize <= minChunkSize)
{
throw new ArgumentOutOfRangeException("Value for maxChunkSize must be greater than minChunkSize and " + (8 * minChunkSize) + " or less.");
}
if (maxChunkSize < (8 * minChunkSize))
{
throw new ArgumentOutOfRangeException("Value for maxChunkSize must be " + (8 * minChunkSize) + " or greater.");
}
if (shiftCount > minChunkSize)
{
throw new ArgumentOutOfRangeException("Value for shiftCount must be less than or equal to minChunkSize.");
}
if (writeChunkMethod == null)
{
throw new ArgumentNullException(nameof(writeChunkMethod));
}
if (readChunkMethod == null)
{
throw new ArgumentNullException(nameof(readChunkMethod));
}
if (deleteChunkMethod == null)
{
throw new ArgumentNullException(nameof(deleteChunkMethod));
}
if (boundaryCheckBytes < 1 || boundaryCheckBytes > 8)
{
throw new ArgumentNullException(nameof(boundaryCheckBytes));
}
_Database = database ?? throw new ArgumentNullException(nameof(database));
_MinChunkSize = minChunkSize;
_MaxChunkSize = maxChunkSize;
_ShiftCount = shiftCount;
_BoundaryCheckBytes = boundaryCheckBytes;
Callbacks = new CallbackMethods
{
WriteChunk = writeChunkMethod,
ReadChunk = readChunkMethod,
DeleteChunk = deleteChunkMethod
};
DebugDedupe = debugDedupe;
DebugSql = debugSql;
_ChunkLock = new object();
InitNewIndex();
}
/// <summary>
/// Create a new index using an internal Sqlite database.
/// </summary>
/// <param name="indexFile">Path and filename.</param>
/// <param name="minChunkSize">Minimum chunk size, must be divisible by 8, divisible by 64, and 128 or greater.</param>
/// <param name="maxChunkSize">Maximum chunk size, must be divisible by 8, divisible by 64, and at least 8 times larger than minimum chunk size.</param>
/// <param name="shiftCount">Number of bytes to shift while identifying chunk boundaries, must be less than or equal to minimum chunk size.</param>
/// <param name="boundaryCheckBytes">Number of bytes to examine while checking for a chunk boundary, must be 8 or fewer.</param>
/// <param name="writeChunkMethod">Method to call to write a chunk to storage.</param>
/// <param name="readChunkMethod">Method to call to read a chunk from storage.</param>
/// <param name="deleteChunkMethod">Method to call to delete a chunk from storage.</param>
/// <param name="debugDedupe">Enable console logging for deduplication operations.</param>
/// <param name="debugSql">Enable console logging for SQL operations.</param>
public DedupeLibrary(
string indexFile,
int minChunkSize,
int maxChunkSize,
int shiftCount,
int boundaryCheckBytes,
Func <Chunk, bool> writeChunkMethod,
Func <string, byte[]> readChunkMethod,
Func <string, bool> deleteChunkMethod,
bool debugDedupe,
bool debugSql)
{
if (String.IsNullOrEmpty(indexFile))
{
throw new ArgumentNullException(nameof(indexFile));
}
if (minChunkSize % 8 != 0)
{
throw new ArgumentException("Value for minChunkSize must be evenly divisible by 8.");
}
if (maxChunkSize % 8 != 0)
{
throw new ArgumentException("Value for maxChunkSize must be evenly divisible by 8.");
}
if (minChunkSize % 64 != 0)
{
throw new ArgumentException("Value for minChunkSize must be evenly divisible by 64.");
}
if (maxChunkSize % 64 != 0)
{
throw new ArgumentException("Value for maxChunkSize must be evenly divisible by 64.");
}
if (minChunkSize < 1024)
{
throw new ArgumentOutOfRangeException("Value for minChunkSize must be 256 or greater.");
}
if (maxChunkSize <= minChunkSize)
{
throw new ArgumentOutOfRangeException("Value for maxChunkSize must be greater than minChunkSize and " + (8 * minChunkSize) + " or less.");
}
if (maxChunkSize < (8 * minChunkSize))
{
throw new ArgumentOutOfRangeException("Value for maxChunkSize must be " + (8 * minChunkSize) + " or greater.");
}
if (shiftCount > minChunkSize)
{
throw new ArgumentOutOfRangeException("Value for shiftCount must be less than or equal to minChunkSize.");
}
if (writeChunkMethod == null)
{
throw new ArgumentNullException(nameof(writeChunkMethod));
}
if (readChunkMethod == null)
{
throw new ArgumentNullException(nameof(readChunkMethod));
}
if (deleteChunkMethod == null)
{
throw new ArgumentNullException(nameof(deleteChunkMethod));
}
if (boundaryCheckBytes < 1 || boundaryCheckBytes > 8)
{
throw new ArgumentNullException(nameof(boundaryCheckBytes));
}
if (File.Exists(indexFile))
{
throw new IOException("Index file already exists.");
}
_IndexFile = DedupeCommon.SanitizeString(indexFile);
_MinChunkSize = minChunkSize;
_MaxChunkSize = maxChunkSize;
_ShiftCount = shiftCount;
_BoundaryCheckBytes = boundaryCheckBytes;
Callbacks = new CallbackMethods();
Callbacks.WriteChunk = writeChunkMethod;
Callbacks.ReadChunk = readChunkMethod;
Callbacks.DeleteChunk = deleteChunkMethod;
DebugDedupe = debugDedupe;
DebugSql = debugSql;
_ChunkLock = new object();
_Database = new SqliteProvider(_IndexFile, DebugSql);
InitNewIndex();
}
