/Utilities/Compression/Deflaters/Deflater.cs
C# | 655 lines | 283 code | 57 blank | 315 comment | 46 complexity | 2b5ce61d6bec7fd66112f0c1b20b87aa MD5 | raw file
Possible License(s): Apache-2.0
- // Based on Mike Krueger's SharpZipLib, Copyright (C) 2001 (GNU license).
- // Authors of the original java version: Jochen Hoenicke, John Leuner
- // See http://www.ISeeSharpCode.com for more information.
-
- using System;
- using Delta.Utilities.Helpers;
-
- namespace Delta.Utilities.Compression.Deflaters
- {
- /// <summary>
- /// This is the Deflater class. The deflater class compresses input
- /// with the deflate algorithm described in RFC 1951. It has several
- /// compression levels and three different strategies described below.
- /// <para />
- /// This class is <i>not</i> thread safe. This is inherent in the API, due
- /// to the split of deflate and setInput.
- /// <para />
- /// author of the original java version : Jochen Hoenicke
- /// </summary>
- public class Deflater
- {
- #region Constants
- /// <summary>
- /// The best and slowest compression level.
- /// This tries to find very long and distant string repetitions.
- /// </summary>
- public const int BestCompression = 9;
-
- /// <summary>
- /// The worst but fastest compression level.
- /// </summary>
- public const int BestSpeed = 1;
-
- /// <summary>
- /// The default compression level.
- /// </summary>
- public const int DefaultCompression = MathHelper.InvalidIndex;
-
- /// <summary>
- /// This level won't compress at all but output uncompressed blocks.
- /// </summary>
- public const int NoCompression = 0;
-
- /// <summary>
- /// The compression method. This is the only method supported so far.
- /// There is no need to use this constant at all.
- /// </summary>
- public const int Deflated = 8;
-
- /*
- * The Deflater can do the following state transitions:
- * (1) -> InitializeState ----> InitializeFinishingState ---.
- * / | (2) (5) |
- * / v (5) |
- * (3)| SetDictionaryState ---> SetDictionaryFinishingState |(3)
- * \ | (3) | ,-------'
- * | | | (3) /
- * v v (5) v v
- * (1) -> BusyState ----> FinishingState
- * | (6)
- * v
- * FinishedState
- * \_____________________________________/
- * | (7)
- * v
- * ClosedState
-
- * (1) If we should produce a header we start in InitializeState, otherwise
- * we start in BusyState.
- * (2) A dictionary may be set only when we are in InitializeState, then
- * we change the state as indicated.
- * (3) Whether a dictionary is set or not, on the first call of deflate
- * we change to BusyState.
- * (4) -- intentionally left blank -- :)
- * (5) FinishingState is entered, when flush() is called to indicate that
- * there is no more INPUT. There are also states indicating, that
- * the header wasn't written yet.
- * (6) FinishedState is entered, when everything has been flushed to the
- * internal pending output buffer.
- * (7) At any time (7)
- */
-
- /// <summary>
- /// Is set dictionary
- /// </summary>
- /// <returns>0x01</returns>
- private const int IsSetDictionary = 0x01;
-
- /// <summary>
- /// Is flushing
- /// </summary>
- /// <returns>0x04</returns>
- private const int IsFlushing = 0x04;
-
- /// <summary>
- /// Is finishing
- /// </summary>
- /// <returns>0x08</returns>
- private const int IsFinishing = 0x08;
-
- /// <summary>
- /// Initialize state
- /// </summary>
- /// <returns>0x00</returns>
- private const int InitializeState = 0x00;
-
- /// <summary>
- /// Set dictionary state
- /// </summary>
- /// <returns>0x01</returns>
- private const int SetDictionaryState = 0x01;
-
- /*unused
- /// <summary>
- /// Initialize finishing state
- /// </summary>
- /// <returns>0x08</returns>
- //private static int InitializeFinishingState = 0x08;
- /// <summary>
- /// Set dictionary finishing state
- /// </summary>
- /// <returns>0x09</returns>
- //private static int SetDictionaryFinishingState = 0x09;
- */
-
- /// <summary>
- /// Busy state
- /// </summary>
- /// <returns>0x10</returns>
- private const int BusyState = 0x10;
-
- /// <summary>
- /// Flushing state
- /// </summary>
- /// <returns>0x14</returns>
- private const int FlushingState = 0x14;
-
- /// <summary>
- /// Finishing state
- /// </summary>
- /// <returns>0x1c</returns>
- private const int FinishingState = 0x1c;
-
- /// <summary>
- /// Finished state
- /// </summary>
- /// <returns>0x 1e</returns>
- private const int FinishedState = 0x1e;
-
- /// <summary>
- /// Closed state
- /// </summary>
- /// <returns>0x7f</returns>
- private const int ClosedState = 0x7f;
- #endregion
-
- #region Adler (Public)
- /// <summary>
- /// Gets the current adler checksum of the data that was processed so far.
- /// </summary>
- public int Adler
- {
- get
- {
- return engine.Adler;
- }
- }
- #endregion
-
- #region TotalIn (Public)
- /// <summary>
- /// Gets the number of input bytes processed so far.
- /// </summary>
- public int TotalIn
- {
- get
- {
- return engine.TotalIn;
- }
- }
- #endregion
-
- #region TotalOut (Public)
- /// <summary>
- /// Gets the number of output bytes so far.
- /// </summary>
- public long TotalOut
- {
- get
- {
- return totalOut;
- }
- }
- #endregion
-
- #region IsFinished (Public)
- /// <summary>
- /// Returns true if the stream was finished and no more output bytes
- /// are available.
- /// </summary>
- public bool IsFinished
- {
- get
- {
- return state == FinishedState && pending.IsFlushed;
- }
- }
- #endregion
-
- #region IsNeedingInput (Public)
- /// <summary>
- /// Returns true, if the input buffer is empty.
- /// You should then call setInput().
- /// NOTE: This method can also return true when the stream
- /// was finished.
- /// </summary>
- public bool IsNeedingInput
- {
- get
- {
- return engine.NeedsInput();
- }
- }
- #endregion
-
- #region Private
-
- #region level (Private)
- /// <summary>
- /// Compression level.
- /// </summary>
- private int level;
- #endregion
-
- #region noZlibHeaderOrFooter (Private)
- /// <summary>
- /// If true no Zlib/RFC1950 headers or footers are generated
- /// </summary>
- private readonly bool noZlibHeaderOrFooter;
- #endregion
-
- #region state (Private)
- /// <summary>
- /// The current state.
- /// </summary>
- private int state;
- #endregion
-
- #region totalOut (Private)
- /// <summary>
- /// The total bytes of output written.
- /// </summary>
- private long totalOut;
- #endregion
-
- #region pending (Private)
- /// <summary>
- /// The pending output.
- /// </summary>
- private readonly DeflaterPending pending;
- #endregion
-
- #region engine (Private)
- /// <summary>
- /// The deflater engine.
- /// </summary>
- private readonly DeflaterEngine engine;
- #endregion
-
- #endregion
-
- #region Constructors
- /// <summary>
- /// Creates a new deflater with default compression level.
- /// </summary>
- public Deflater()
- : this(DefaultCompression, false)
- {
- }
-
- /// <summary>
- /// Creates a new deflater with given compression level.
- /// </summary>
- /// <param name="level">The compression level, a value between
- /// NoCompression and BestCompression, or DefaultCompression.
- /// </param>
- /// <exception cref="System.ArgumentOutOfRangeException">
- /// If level is out of range.</exception>
- public Deflater(int level)
- : this(level, false)
- {
- }
-
- /// <summary>
- /// Creates a new deflater with given compression level.
- /// </summary>
- /// <param name="level">
- /// the compression level, a value between NoCompression
- /// and BestCompression.
- /// </param>
- /// <param name="noZlibHeaderOrFooter">
- /// true, if we should suppress the Zlib/RFC1950 header at the
- /// beginning and the adler checksum at the end of the output.
- /// This is useful for the GZIP/PKZIP formats.
- /// </param>
- /// <exception cref="System.ArgumentOutOfRangeException">
- /// If level is out of range.</exception>
- public Deflater(int level, bool noZlibHeaderOrFooter)
- {
- if (level == DefaultCompression)
- {
- level = 6;
- }
- else if (level < NoCompression ||
- level > BestCompression)
- {
- throw new ArgumentOutOfRangeException("level");
- }
-
- pending = new DeflaterPending();
- engine = new DeflaterEngine(pending);
- this.noZlibHeaderOrFooter = noZlibHeaderOrFooter;
- SetStrategy(DeflateStrategy.Default);
- SetLevel(level);
- Reset();
- }
- #endregion
-
- #region Reset (Public)
- /// <summary>
- /// Resets the deflater. The deflater acts afterwards as if it was
- /// just created with the same compression level and strategy as it
- /// had before.
- /// </summary>
- public void Reset()
- {
- state = (noZlibHeaderOrFooter
- ? BusyState
- : InitializeState);
- totalOut = 0;
- pending.Reset();
- engine.Reset();
- }
- #endregion
-
- #region Flush (Public)
- /// <summary>
- /// Flushes the current input block. Further calls to deflate() will
- /// produce enough output to inflate everything in the current input
- /// block. This is not part of Sun's JDK so I have made it package
- /// private. It is used by DeflaterOutputStream to implement
- /// flush().
- /// </summary>
- public void Flush()
- {
- state |= IsFlushing;
- }
- #endregion
-
- #region Finish (Public)
- /// <summary>
- /// Finishes the deflater with the current input block. It is an error
- /// to give more input after this method was called. This method must
- /// be called to force all bytes to be flushed.
- /// </summary>
- public void Finish()
- {
- state |= IsFlushing | IsFinishing;
- }
- #endregion
-
- #region SetInput (Public)
- /// <summary>
- /// Sets the data which should be compressed next. This should be only
- /// called when needsInput indicates that more input is needed.
- /// If you call setInput when needsInput() returns false, the
- /// previous input that is still pending will be thrown away.
- /// The given byte array should not be changed, before needsInput()
- /// returns true again.
- /// This call is equivalent to <code>setInput(input, 0, input.length)
- /// </code>.
- /// </summary>
- /// <param name="input">
- /// the buffer containing the input data.
- /// </param>
- /// <exception cref="System.InvalidOperationException">
- /// if the buffer was finished() or ended().
- /// </exception>
- public void SetInput(byte[] input)
- {
- SetInput(input, 0, input.Length);
- }
-
- /// <summary>
- /// Sets the data which should be compressed next. This should be
- /// only called when needsInput indicates that more input is needed.
- /// The given byte array should not be changed, before needsInput() returns
- /// true again.
- /// </summary>
- /// <param name="input">The buffer containing the input data.</param>
- /// <param name="off">The start of the data.</param>
- /// <param name="len">The length of the data.</param>
- /// <exception cref="System.InvalidOperationException">
- /// if the buffer was finished() or ended() or if previous input is still
- /// pending.
- /// </exception>
- public void SetInput(byte[] input, int off, int len)
- {
- if ((state & IsFinishing) != 0)
- {
- throw new InvalidOperationException("finish()/end() already called");
- }
- engine.SetInput(input, off, len);
- }
- #endregion
-
- #region SetLevel (Public)
- /// <summary>
- /// Sets the compression level. There is no guarantee of the exact
- /// position of the change, but if you call this when needsInput is
- /// true the change of compression level will occur somewhere near
- /// before the end of the so far given input.
- /// </summary>
- /// <param name="lvl">
- /// the new compression level.
- /// </param>
- public void SetLevel(int lvl)
- {
- if (lvl == DefaultCompression)
- {
- lvl = 6;
- }
- else if (lvl < NoCompression || lvl > BestCompression)
- {
- throw new ArgumentOutOfRangeException("lvl");
- }
-
- if (level != lvl)
- {
- level = lvl;
- engine.SetLevel(lvl);
- }
- }
- #endregion
-
- #region GetLevel (Public)
- /// <summary>
- /// Get current compression level
- /// </summary>
- /// <returns>Returns the current compression level</returns>
- public int GetLevel()
- {
- return level;
- }
- #endregion
-
- #region SetStrategy (Public)
- /// <summary>
- /// Sets the compression strategy. Strategy is one of
- /// DEFAULT_STRATEGY, HUFFMAN_ONLY and FILTERED. For the exact
- /// position where the strategy is changed, the same as for
- /// setLevel() applies.
- /// </summary>
- /// <param name="strategy">
- /// The new compression strategy.
- /// </param>
- public void SetStrategy(DeflateStrategy strategy)
- {
- engine.Strategy = strategy;
- }
- #endregion
-
- #region Deflate (Public)
- /// <summary>
- /// Deflates the current input block with to the given array.
- /// </summary>
- /// <param name="output">
- /// The buffer where compressed data is stored
- /// </param>
- /// <returns>
- /// The number of compressed bytes added to the output, or 0 if either
- /// needsInput() or finished() returns true or length is zero.
- /// </returns>
- public int Deflate(byte[] output)
- {
- return Deflate(output, 0, output.Length);
- }
-
- /// <summary>
- /// Deflates the current input block to the given array.
- /// </summary>
- /// <param name="output">Buffer to store the compressed data.</param>
- /// <param name="offset">Offset into the output array.</param>
- /// <param name="length">The maximum number of bytes that may be stored.
- /// </param>
- /// <returns>
- /// The number of compressed bytes added to the output, or 0 if either
- /// needsInput() or finished() returns true or length is zero.
- /// </returns>
- /// <exception cref="System.InvalidOperationException">
- /// If end() was previously called.
- /// </exception>
- /// <exception cref="System.ArgumentOutOfRangeException">
- /// If offset and/or length don't match the array length.
- /// </exception>
- public int Deflate(byte[] output, int offset, int length)
- {
- int origLength = length;
-
- if (state == ClosedState)
- {
- throw new InvalidOperationException("Deflater closed");
- }
-
- if (state < BusyState)
- {
- /* output header */
- int header = (Deflated +
- ((DeflaterConstants.MaxWBits - 8) << 4)) << 8;
- int level_flags = (level - 1) >> 1;
- if (level_flags < 0 || level_flags > 3)
- {
- level_flags = 3;
- }
- header |= level_flags << 6;
- if ((state & IsSetDictionary) != 0)
- {
- // Dictionary was set
- header |= DeflaterConstants.PresetDictionary;
- }
- header += 31 - (header % 31);
-
- pending.WriteShortMsb(header);
- if ((state & IsSetDictionary) != 0)
- {
- int chksum = engine.Adler;
- engine.ResetAdler();
- pending.WriteShortMsb(chksum >> 16);
- pending.WriteShortMsb(chksum & 0xffff);
- }
-
- state = BusyState | (state & (IsFlushing | IsFinishing));
- }
-
- for (;;)
- {
- int count = pending.Flush(output, offset, length);
- offset += count;
- totalOut += count;
- length -= count;
-
- if (length == 0 || state == FinishedState)
- {
- break;
- }
-
- if (!engine.Deflate((state & IsFlushing) != 0,
- (state & IsFinishing) != 0))
- {
- if (state == BusyState)
- {
- /* We need more input now */
- return origLength - length;
- }
- else if (state == FlushingState)
- {
- if (level != NoCompression)
- {
- // We have to supply some lookahead. 8 bit lookahead
- // is needed by the zlib inflater, and we must fill
- // the next byte, so that all bits are flushed.
- int neededbits = 8 + ((-pending.BitCount) & 7);
- while (neededbits > 0)
- {
- /* write a static tree block consisting solely of
- * an EOF:
- */
- pending.WriteBits(2, 10);
- neededbits -= 10;
- }
- }
- state = BusyState;
- }
- else if (state == FinishingState)
- {
- pending.AlignToByte();
-
- // Compressed data is complete.
- // Write footer information if required.
- if (!noZlibHeaderOrFooter)
- {
- int adler = engine.Adler;
- pending.WriteShortMsb(adler >> 16);
- pending.WriteShortMsb(adler & 0xffff);
- }
- state = FinishedState;
- }
- }
- }
- return origLength - length;
- }
- #endregion
-
- #region SetDictionary (Public)
- /// <summary>
- /// Sets the dictionary which should be used in the deflate process.
- /// This call is equivalent to <code>setDictionary(dict, 0, dict.Length)
- /// </code>.
- /// </summary>
- /// <param name="dict">
- /// the dictionary.
- /// </param>
- /// <exception cref="System.InvalidOperationException">
- /// if setInput () or deflate () were already called or another dictionary
- /// was already set.
- /// </exception>
- public void SetDictionary(byte[] dict)
- {
- SetDictionary(dict, 0, dict.Length);
- }
-
- /// <summary>
- /// Sets the dictionary which should be used in the deflate process.
- /// The dictionary is a byte array containing strings that are
- /// likely to occur in the data which should be compressed. The
- /// dictionary is not stored in the compressed output, only a
- /// checksum. To decompress the output you need to supply the same
- /// dictionary again.
- /// </summary>
- /// <param name="dict">
- /// The dictionary data
- /// </param>
- /// <param name="offset">
- /// An offset into the dictionary.
- /// </param>
- /// <param name="length">
- /// The length of the dictionary data to use
- /// </param>
- /// <exception cref="System.InvalidOperationException">
- /// If setInput () or deflate () were already called or another
- /// dictionary was already set.
- /// </exception>
- public void SetDictionary(byte[] dict, int offset, int length)
- {
- if (state != InitializeState)
- {
- throw new InvalidOperationException();
- }
-
- state = SetDictionaryState;
- engine.SetDictionary(dict, offset, length);
- }
- #endregion
- }
- }