using System;
using System.IO;

namespace Gif.Components
{
 public class LZWEncoder
 {

 private static readonly int EOF = -1;

 private int imgW, imgH;
 private byte[] pixAry;
 private int initCodeSize;
 private int remaining;
 private int curPixel;

 // GIFCOMPR.C    - GIF Image compression routines
 //
 // Lempel-Ziv compression based on 'compress'. GIF modifications by
 // David Rowley (mgardi@watdcsu.waterloo.edu)

 // General DEFINEs

 static readonly int BITS = 12;

 static readonly int HSIZE = 5003; // 80% occupancy

 // GIF Image compression - modified 'compress'
 //
 // Based on: compress.c - File compression ala IEEE Computer, June 1984.
 //
 // By Authors: Spencer W. Thomas   (decvax!harpo!utah-cs!utah-gr!thomas)
 //       Jim McKie       (decvax!mcvax!jim)
 //       Steve Davies      (decvax!vax135!petsd!peora!srd)
 //       Ken Turkowski     (decvax!decwrl!turtlevax!ken)
 //       James A. Woods     (decvax!ihnp4!ames!jaw)
 //       Joe Orost       (decvax!vax135!petsd!joe)

 int n_bits; // number of bits/code
 int maxbits = BITS; // user settable max # bits/code
 int maxcode; // maximum code, given n_bits
 int maxmaxcode = 1 << BITS; // should NEVER generate this code

 int[] htab = new int[HSIZE];//這個是放hash的筒子,在這里面可以很快的找到1個key
 int[] codetab = new int[HSIZE];

 int hsize = HSIZE; // for dynamic table sizing

 int free_ent = 0; // first unused entry

 // block compression parameters -- after all codes are used up,
 // and compression rate changes, start over.
 bool clear_flg = false;

 // Algorithm: use open addressing double hashing (no chaining) on the
 // prefix code / next character combination. We do a variant of Knuth's
 // algorithm D (vol. 3, sec. 6.4) along with G. Knott's relatively-prime
 // secondary probe. Here, the modular division first probe is gives way
 // to a faster exclusive-or manipulation. Also do block compression with
 // an adaptive reset, whereby the code table is cleared when the compression
 // ratio decreases, but after the table fills. The variable-length output
 // codes are re-sized at this point, and a special CLEAR code is generated
 // for the decompressor. Late addition: construct the table according to
 // file size for noticeable speed improvement on small files. Please direct
 // questions about this implementation to ames!jaw.

 int g_init_bits;

 int ClearCode;
 int EOFCode;

 // output
 //
 // Output the given code.
 // Inputs:
 //   code:  A n_bits-bit integer. If == -1, then EOF. This assumes
 //       that n_bits =< wordsize - 1.
 // Outputs:
 //   Outputs code to the file.
 // Assumptions:
 //   Chars are 8 bits long.
 // Algorithm:
 //   Maintain a BITS character long buffer (so that 8 codes will
 // fit in it exactly). Use the VAX insv instruction to insert each
 // code in turn. When the buffer fills up empty it and start over.

 int cur_accum = 0;
 int cur_bits = 0;

 int [] masks =
 {
  0x0000,
  0x0001,
  0x0003,
  0x0007,
  0x000F,
  0x001F,
  0x003F,
  0x007F,
  0x00FF,
  0x01FF,
  0x03FF,
  0x07FF,
  0x0FFF,
  0x1FFF,
  0x3FFF,
  0x7FFF,
  0xFFFF };

 // Number of characters so far in this 'packet'
 int a_count;

 // Define the storage for the packet accumulator
 byte[] accum = new byte[256];

 //----------------------------------------------------------------------------
 public LZWEncoder(int width, int height, byte[] pixels, int color_depth)
 {
  imgW = width;
  imgH = height;
  pixAry = pixels;
  initCodeSize = Math.Max(2, color_depth);
 }
 
 // Add a character to the end of the current packet, and if it is 254
 // characters, flush the packet to disk.
 void Add(byte c, Stream outs)
 {
  accum[a_count++] = c;
  if (a_count >= 254)
  Flush(outs);
 }
 
 // Clear out the hash table

 // table clear for block compress
 void ClearTable(Stream outs)
 {
  ResetCodeTable(hsize);
  free_ent = ClearCode + 2;
  clear_flg = true;

  Output(ClearCode, outs);
 }
 
 // reset code table
    // 全部初始化為-1
 void ResetCodeTable(int hsize)
 {
  for (int i = 0; i < hsize; ++i)
  htab[i] = -1;
 }
 
 void Compress(int init_bits, Stream outs)
 {
  int fcode;
  int i /* = 0 */;
  int c;
  int ent;
  int disp;
  int hsize_reg;
  int hshift;

  // Set up the globals: g_init_bits - initial number of bits
      //原始數(shù)據(jù)的字長,在gif文件中，原始數(shù)據(jù)的字長可以為1(單色圖),4(16色)，和8(256色)
      //開始的時候先加上1
      //但是當(dāng)原始數(shù)據(jù)長度為1的時候，開始為3
      //因此原始長度1->3,4->5,8->9

      //?為何原始數(shù)據(jù)字長為1的時候，開始長度為3呢？?
      //如果+1=2，只能表示四種狀態(tài)，加上clearcode和endcode就用完了。所以必須擴展到3
  g_init_bits = init_bits;

  // Set up the necessary values
      //是否需要加清除標(biāo)志
      //GIF為了提高壓縮率，采用的是變長的字長(VCL)。比如說原始數(shù)據(jù)是8位，那么開始先加上1位(8+1=9)
      //當(dāng)標(biāo)號到2^9=512的時候，超過了當(dāng)前長度9所能表現(xiàn)的最大值，此時后面的標(biāo)號就必須用10位來表示
      //以此類推，當(dāng)標(biāo)號到2^12的時候，因為最大為12,不能繼續(xù)擴展了，需要在2^12=4096的位置上插入一個ClearCode,表示從這往后，從9位重新再來了    
  clear_flg = false;
  n_bits = g_init_bits;
      //獲得n位數(shù)能表述的最大值(gif圖像中開始一般為3,5,9，故maxcode一般為7,31,511)
  maxcode = MaxCode(n_bits);
      //表示從這里我重新開始構(gòu)造字典字典了，以前的所有標(biāo)記作廢，
      //開始使用新的標(biāo)記。這個標(biāo)號集的大小多少比較合適呢？據(jù)說理論上是越大壓縮率越高（我個人感覺太大了也不見得就好），
      //不過處理的開銷也呈指數(shù)增長
      //gif規(guī)定，clearcode的值為原始數(shù)據(jù)最大字長所能表達的數(shù)值+1;比如原始數(shù)據(jù)長度為8,則clearcode=1<<(9-1)=256
  ClearCode = 1 << (init_bits - 1);
      //結(jié)束標(biāo)志為clearcode+1
  EOFCode = ClearCode + 1;
      //這個是解除結(jié)束的
  free_ent = ClearCode + 2;
      //清楚數(shù)量
  a_count = 0; // clear packet
      //從圖像中獲得下一個像素
  ent = NextPixel();

  hshift = 0;
  for (fcode = hsize; fcode < 65536; fcode *= 2)
  ++hshift;
      //設(shè)置hash碼范圍
  hshift = 8 - hshift; // set hash code range bound

  hsize_reg = hsize;
      //清除固定大小的hash表，用于存儲標(biāo)記，這個相當(dāng)于字典
  ResetCodeTable(hsize_reg); // clear hash table

  Output(ClearCode, outs);

  outer_loop : while ((c = NextPixel()) != EOF)
    {
    fcode = (c << maxbits) + ent;              
    i = (c << hshift) ^ ent; // xor hashing
               //嘿嘿,小樣,又來了,我認識你
    if (htab[i] == fcode)
    {
     ent = codetab[i];
     continue;
    }
               //這小子,新來的
    else if (htab[i] >= 0) // non-empty slot
    {
     disp = hsize_reg - i; // secondary hash (after G. Knott)
     if (i == 0)
     disp = 1;
     do
     {
     if ((i -= disp) < 0)
      i += hsize_reg;

     if (htab[i] == fcode)
     {
      ent = codetab[i];
      goto outer_loop;
     }
     } while (htab[i] >= 0);
    }
     Output(ent, outs);
               //從這里可以看出,ent就是前綴（prefix）,而當(dāng)前正在處理的字符標(biāo)志就是后綴（suffix）
    ent = c;
               //判斷終止結(jié)束符是否超過當(dāng)前位數(shù)所能表述的范圍
    if (free_ent < maxmaxcode)
    {
                 //如果沒有超
     codetab[i] = free_ent++; // code -> hashtable
                 //hash表里面建立相應(yīng)索引
     htab[i] = fcode;
    }
    else
                 //說明超過了當(dāng)前所能表述的范圍,清空字典,重新再來
     ClearTable(outs);
    }
  // Put out the final code.
  Output(ent, outs);
  Output(EOFCode, outs);
 }
 
 //----------------------------------------------------------------------------
 public void Encode( Stream os)
 {
  os.WriteByte( Convert.ToByte( initCodeSize) ); // write "initial code size" byte
      //這個圖像包含多少個像素
  remaining = imgW * imgH; // reset navigation variables
      //當(dāng)前處理的像素索引
  curPixel = 0;

  Compress(initCodeSize + 1, os); // compress and write the pixel data

  os.WriteByte(0); // write block terminator
 }
 
 // Flush the packet to disk, and reset the accumulator
 void Flush(Stream outs)
 {
  if (a_count > 0)
  {
  outs.WriteByte( Convert.ToByte( a_count ));
  outs.Write(accum, 0, a_count);
  a_count = 0;
  }
 } 
   
    /// <summary>
    /// 獲得n位數(shù)所能表達的最大數(shù)值
    /// </summary>
    /// <param name="n_bits">位數(shù)，一般情況下n_bits = 9</param>
    /// <returns>最大值,例如n_bits=8,則返回值就為2^8-1=255</returns>
 int MaxCode(int n_bits)
 {
  return (1 << n_bits) - 1;
 }
 
 //----------------------------------------------------------------------------
 // Return the next pixel from the image
 //----------------------------------------------------------------------------
    /// <summary>
    /// 從圖像中獲得下一個像素
    /// </summary>
    /// <returns></returns>
 private int NextPixel()
 {
      //還剩多少個像素沒有處理
      //如果沒有了,返回結(jié)束標(biāo)志
  if (remaining == 0)
  return EOF;
      //否則處理下一個,并將未處理像素數(shù)目-1
  --remaining;
      //當(dāng)前處理的像素
  int temp = curPixel + 1;
      //如果當(dāng)前處理像素在像素范圍之內(nèi)
  if ( temp < pixAry.GetUpperBound( 0 ))
  {
        //下一個像素
  byte pix = pixAry[curPixel++];
  return pix & 0xff;
  }
  return 0xff;
 }
   /// <summary>
   /// 輸出字到輸出流
   /// </summary>
   /// <param name="code">要輸出的字</param>
   /// <param name="outs">輸出流</param>
 void Output(int code, Stream outs)
 {
      //得到當(dāng)前標(biāo)志位所能表示的最大標(biāo)志值
  cur_accum &= masks[cur_bits];

  if (cur_bits > 0)
  cur_accum |= (code << cur_bits);
  else
        //如果標(biāo)志位為0,就將當(dāng)前標(biāo)號為輸入流
  cur_accum = code;
      //當(dāng)前能標(biāo)志的最大字長度(9-10-11-12-9-10。。。。。。。)
  cur_bits += n_bits;
      //如果當(dāng)前最大長度大于8
  while (cur_bits >= 8)
  {
        //向流中輸出一個字節(jié)
  Add((byte) (cur_accum & 0xff), outs);
        //將當(dāng)前標(biāo)號右移8位
  cur_accum >>= 8;
  cur_bits -= 8;
  }

  // If the next entry is going to be too big for the code size,
  // then increase it, if possible.
  if (free_ent > maxcode || clear_flg)
  {
  if (clear_flg)
  {
   maxcode = MaxCode(n_bits = g_init_bits);
   clear_flg = false;
  }
  else
  {
   ++n_bits;
   if (n_bits == maxbits)
   maxcode = maxmaxcode;
   else
   maxcode = MaxCode(n_bits);
  }
  }

  if (code == EOFCode)
  {
  // At EOF, write the rest of the buffer.
  while (cur_bits > 0)
  {
   Add((byte) (cur_accum & 0xff), outs);
   cur_accum >>= 8;
   cur_bits -= 8;
  }

  Flush(outs);
  }
 }
 }
}