| 网站首页 | 业界新闻 | 小组 | 威客 | 人才 | 下载频道 | 博客 | 代码贴 | 在线编程 | 编程论坛
欢迎加入我们,一同切磋技术
用户名:   
 
密 码:  
共有 635 人关注过本帖
标题:哪位大侠能帮我解释这个LZ77压缩算法的代码 (本人感激不尽)
只看楼主 加入收藏
maoniu
Rank: 1
等 级:新手上路
帖 子:25
专家分:0
注 册:2010-4-7
结帖率:57.14%
收藏
 问题点数:0 回复次数:0 
哪位大侠能帮我解释这个LZ77压缩算法的代码 (本人感激不尽)
/**************************************************************
    LZSS.C -- A Data Compression Program
    (tab = 4 spaces)
***************************************************************
    4/6/1989 Haruhiko Okumura
    Use, distribute, and modify this program freely.
    Please send me your improved versions.
        PC-VAN        SCIENCE
        NIFTY-Serve    PAF01022
        CompuServe    74050,1022
**************************************************************/
#include <stdio.h>
#include <stdlib.h>
#include <string.h>
#include <ctype.h>

#define N         4096    /* size of ring buffer */
#define F           18    /* upper limit for match_length */
#define THRESHOLD    2   /* encode string into position and length
                           if match_length is greater than this */
#define NIL            N    /* index for root of binary search trees */

unsigned long int
        textsize = 0,    /* text size counter */
        codesize = 0,    /* code size counter */
        printcount = 0;    /* counter for reporting progress every 1K bytes */
unsigned char
        text_buf[N + F - 1];    /* ring buffer of size N,
            with extra F-1 bytes to facilitate string comparison */
int        match_position, match_length,  /* of longest match.  These are
            set by the InsertNode() procedure. */
        lson[N + 1], rson[N + 257], dad[N + 1];  /* left & right children &
            parents -- These constitute binary search trees. */
FILE    *infile, *outfile;  /* input & output files */

void InitTree(void)  /* initialize trees */
{
    int  i;

    /* For i = 0 to N - 1, rson[i] and lson[i] will be the right and
       left children of node i.  These nodes need not be initialized.
       Also, dad[i] is the parent of node i.  These are initialized to
       NIL (= N), which stands for 'not used.'
       For i = 0 to 255, rson[N + i + 1] is the root of the tree
       for strings that begin with character i.  These are initialized
       to NIL.  Note there are 256 trees. */

    for (i = N + 1; i <= N + 256; i++) rson[i] = NIL;
    for (i = 0; i < N; i++) dad[i] = NIL;
}

void InsertNode(int r)
    /* Inserts string of length F, text_buf[r..r+F-1], into one of the
       trees (text_buf[r]'th tree) and returns the longest-match position
       and length via the global variables match_position and match_length.
       If match_length = F, then removes the old node in favor of the new
       one, because the old one will be deleted sooner.
       Note r plays double role, as tree node and position in buffer. */
{
    int  i, p, cmp;
    unsigned char  *key;

    cmp = 1;  key = &text_buf[r];  p = N + 1 + key[0];
    rson[r] = lson[r] = NIL;  match_length = 0;
    for ( ; ; ) {
        if (cmp >= 0) {
            if (rson[p] != NIL) p = rson[p];
            else {  rson[p] = r;  dad[r] = p;  return;  }
        } else {
            if (lson[p] != NIL) p = lson[p];
            else {  lson[p] = r;  dad[r] = p;  return;  }
        }
        for (i = 1; i < F; i++)
            if ((cmp = key[i] - text_buf[p + i]) != 0)  break;
        if (i > match_length) {
            match_position = p;
            if ((match_length = i) >= F)  break;
        }
    }
    dad[r] = dad[p];  lson[r] = lson[p];  rson[r] = rson[p];
    dad[lson[p]] = r;  dad[rson[p]] = r;
    if (rson[dad[p]] == p) rson[dad[p]] = r;
    else                   lson[dad[p]] = r;
    dad[p] = NIL;  /* remove p */
}

void DeleteNode(int p)  /* deletes node p from tree */
{
    int  q;
   
    if (dad[p] == NIL) return;  /* not in tree */
    if (rson[p] == NIL) q = lson[p];
    else if (lson[p] == NIL) q = rson[p];
    else {
        q = lson[p];
        if (rson[q] != NIL) {
            do {  q = rson[q];  } while (rson[q] != NIL);
            rson[dad[q]] = lson[q];  dad[lson[q]] = dad[q];
            lson[q] = lson[p];  dad[lson[p]] = q;
        }
        rson[q] = rson[p];  dad[rson[p]] = q;
    }
    dad[q] = dad[p];
    if (rson[dad[p]] == p) rson[dad[p]] = q;  else lson[dad[p]] = q;
    dad[p] = NIL;
}

void Encode(void)
{
    int  i, c, len, r, s, last_match_length, code_buf_ptr;
    unsigned char  code_buf[17], mask;
   
    InitTree();  /* initialize trees */
    code_buf[0] = 0;  /* code_buf[1..16] saves eight units of code, and
        code_buf[0] works as eight flags, "1" representing that the unit
        is an unencoded letter (1 byte), "0" a position-and-length pair
        (2 bytes).  Thus, eight units require at most 16 bytes of code. */
    code_buf_ptr = mask = 1;
    s = 0;  r = N - F;
    for (i = s; i < r; i++) text_buf[i] = ' ';  /* Clear the buffer with
        any character that will appear often. */
    for (len = 0; len < F && (c = getc(infile)) != EOF; len++)
        text_buf[r + len] = c;  /* Read F bytes into the last F bytes of
            the buffer */
    if ((textsize = len) == 0) return;  /* text of size zero */
    for (i = 1; i <= F; i++) InsertNode(r - i);  /* Insert the F strings,
        each of which begins with one or more 'space' characters.  Note
        the order in which these strings are inserted.  This way,
        degenerate trees will be less likely to occur. */
    InsertNode(r);  /* Finally, insert the whole string just read.  The
        global variables match_length and match_position are set. */
    do {
        if (match_length > len) match_length = len;  /* match_length
            may be spuriously long near the end of text. */
        if (match_length <= THRESHOLD) {
            match_length = 1;  /* Not long enough match.  Send one byte. */
            code_buf[0] |= mask;  /* 'send one byte' flag */
            code_buf[code_buf_ptr++] = text_buf[r];  /* Send uncoded. */
        } else {
            code_buf[code_buf_ptr++] = (unsigned char) match_position;
            code_buf[code_buf_ptr++] = (unsigned char)
                (((match_position >> 4) & 0xf0)
              | (match_length - (THRESHOLD + 1)));  /* Send position and
                    length pair. Note match_length > THRESHOLD. */
        }
        if ((mask <<= 1) == 0) {  /* Shift mask left one bit. */
            for (i = 0; i < code_buf_ptr; i++)  /* Send at most 8 units of */
                putc(code_buf[i], outfile);     /* code together */
            codesize += code_buf_ptr;
            code_buf[0] = 0;  code_buf_ptr = mask = 1;
        }
        last_match_length = match_length;
        for (i = 0; i < last_match_length &&
                (c = getc(infile)) != EOF; i++) {
            DeleteNode(s);        /* Delete old strings and */
            text_buf[s] = c;    /* read new bytes */
            if (s < F - 1) text_buf[s + N] = c;  /* If the position is
                near the end of buffer, extend the buffer to make
                string comparison easier. */
            s = (s + 1) & (N - 1);  r = (r + 1) & (N - 1);
                /* Since this is a ring buffer, increment the position
                   modulo N. */
            InsertNode(r);    /* Register the string in text_buf[r..r+F-1] */
        }
        if ((textsize += i) > printcount) {
            printf("%12ld\r", textsize);  printcount += 1024;
                /* Reports progress each time the textsize exceeds
                   multiples of 1024. */
        }
        while (i++ < last_match_length) {    /* After the end of text, */
            DeleteNode(s);                    /* no need to read, but */
            s = (s + 1) & (N - 1);  r = (r + 1) & (N - 1);
            if (--len) InsertNode(r);        /* buffer may not be empty. */
        }
    } while (len > 0);    /* until length of string to be processed is zero */
    if (code_buf_ptr > 1) {        /* Send remaining code. */
        for (i = 0; i < code_buf_ptr; i++) putc(code_buf[i], outfile);
        codesize += code_buf_ptr;
    }
    printf("In : %ld bytes\n", textsize);    /* Encoding is done. */
    printf("Out: %ld bytes\n", codesize);
    printf("Out/In: %.3f\n", (double)codesize / textsize);
}

void Decode(void)    /* Just the reverse of Encode(). */
{
    int  i, j, k, r, c;
    unsigned int  flags;
   
    for (i = 0; i < N - F; i++) text_buf[i] = ' ';
    r = N - F;  flags = 0;
    for ( ; ; ) {
        if (((flags >>= 1) & 256) == 0) {
            if ((c = getc(infile)) == EOF) break;
            flags = c | 0xff00;        /* uses higher byte cleverly */
        }                            /* to count eight */
        if (flags & 1) {
            if ((c = getc(infile)) == EOF) break;
            putc(c, outfile);  text_buf[r++] = c;  r &= (N - 1);
        } else {
            if ((i = getc(infile)) == EOF) break;
            if ((j = getc(infile)) == EOF) break;
            i |= ((j & 0xf0) << 4);  j = (j & 0x0f) + THRESHOLD;
            for (k = 0; k <= j; k++) {
                c = text_buf[(i + k) & (N - 1)];
                putc(c, outfile);  text_buf[r++] = c;  r &= (N - 1);
            }
        }
    }
}

int main(int argc, char *argv[])
{
    char  *s;
   
    if (argc != 4) {
        printf("'lzss e file1 file2' encodes file1 into file2.\n"
               "'lzss d file2 file1' decodes file2 into file1.\n"
               "%d\n",argc);
        return EXIT_FAILURE;
    }
    printf("ok\n");
    if ((s = argv[1], s[1] || strpbrk(s, "DEde") == NULL)
     || (s = argv[2], (infile  = fopen(s, "rb")) == NULL)
     || (s = argv[3], (outfile = fopen(s, "wb")) == NULL)) {
        printf("??? %s\n", s);  return EXIT_FAILURE;
    }
    if (toupper(*argv[1]) == 'E') Encode();  else Decode();
    fclose(infile);  fclose(outfile);
    return EXIT_SUCCESS;
}


[ 本帖最后由 maoniu 于 2010-4-7 19:03 编辑 ]
搜索更多相关主题的帖子: 感激不尽 算法 代码 压缩 
2010-04-07 19:01
快速回复:哪位大侠能帮我解释这个LZ77压缩算法的代码 (本人感激不尽)
数据加载中...
 
   



关于我们 | 广告合作 | 编程中国 | 清除Cookies | TOP | 手机版

编程中国 版权所有,并保留所有权利。
Powered by Discuz, Processed in 0.019531 second(s), 7 queries.
Copyright©2004-2024, BCCN.NET, All Rights Reserved