Programmer 7500/MAX_PROGRAMMERS.iso/INFO/COMPRESS/LZ13.ARC/LZDCMP1.ASM

Assembly Source File | 1988-09-23 | 6.1 KB | 272 lines

    title    lzdcmp - file decompressor using limpel-ziev algorithm

;Tom Pfau
;Digital Equipment Corporation
;Parsippany, NJ

;Constants
CLEAR        equ    256
EOF        equ    257
FIRST_FREE    equ    258
;MAXMAX        equ    4096

    include    macros1.mlb

;Hash table entry
hash_rec    struc
next    dw    ?            ; prefix code
char    db    ?            ; suffix char
hash_rec    ends

;Declare segments
code    segment byte public 'code'
code    ends
stack    segment word stack 'data'
    dw    128 dup (?)
stack    ends
data    segment word public 'data'
data    ends
memory    segment para public 'memory'
hash    label    hash_rec
memory    ends

;Start coding
code    segment
    assume    CS:code,DS:data,ss:stack,ES:data

start    proc    far
    mov    bx,seg hash        ;End of program
    mov    ax,DS            ;Beginning of program
    sub    bx,ax            ;Size of program
    inc    bx            ;Make sure
    setmem    bx            ;Set program size
    mov    bx,data            ;Address data segment
    mov    ES,bx
    mov    DS,bx
    print    input_prompt        ;Get file names
    input    input_file
    print    crlf
    print    output_prompt
    input    output_file
    print    crlf
    mov    al,input_file+1        ;Terminate with nulls
    xor    ah,ah
    mov    si,ax
    mov    input_file+2[si],ah    ;TH 0
    mov    al,output_file+1
    mov    si,ax
    mov    output_file+2[si],ah    ;TH 0
    hopen    input_file+2,0        ;Open input and output
    mov    input_handle,ax
    hcreat    output_file+2,0
    mov    output_handle,ax
    call    decompress        ;Decompress files
    hclose    input_handle
    hclose    output_handle
    exit                ;Done
start    endp

data    segment
input_prompt    db    'Input file: $'
output_prompt    db    'Output file: $'
input_file    db    80,0,80 dup (?)
output_file    db    80,0,80 dup (?)
crlf        db    13,10,'$'
input_handle    dw    ?
output_handle    dw    ?
data    ends

decompress    proc    near
    malloc    768            ;Allocate space for hash table
    mov    hash_seg,ax        ;Save segment address
    hread    input_handle,input_buffer,1024    ;Read from input
l1:    call    read_code        ;Get a code
    cmp    ax,EOF            ;End of file?
    jne    l2            ;no

    cmp    output_offset,0        ;Data in output buffer?
    je    l1a            ;no
     hwrite    output_handle,output_buffer,output_offset    ;Flush buffer
l1a:    ret                ;done

l2:    cmp    ax,CLEAR        ;Clear code?
    jne    l7            ;no

    call    init_tab        ;Initialize table
    call    read_code        ;Read next code
    mov    cur_code,ax        ;Initialize variables
    mov    old_code,ax
    mov    k,al
    mov    fin_char,al
;    mov    al,k
    call    write_char        ;Write character
    jmp    l1            ;Get next code

l7:    mov    cur_code,ax        ;Save new code
    mov    in_code,ax
    mov    ES,hash_seg        ;Point to hash table
    cmp    ax,free_code        ;Code in table? (k<w>k<w>k)
    jl    l11            ;yes
     mov    ax,old_code        ;get previous code
     mov    cur_code,ax        ;make current
     mov    al,fin_char        ;get old last char
     push    ax            ;push it
     inc    stack_count
l11:    cmp    cur_code,255        ;Code or character?
    jle    l15            ;Char

    mov    bx,cur_code        ;Convert code to address
    call    index
    mov    al,ES:2[bx]        ;Get suffix char
    push    ax            ;push it
    inc    stack_count
    mov    ax,ES:[bx]        ;Get prefix code
    mov    cur_code,ax        ;Save it
    jmp    l11            ;Translate again

l15:
;    push    DS            ;Restore seg reg
;    pop    ES
    mov    ax,DS
    mov    ES,ax
    mov    ax,cur_code        ;Get code
    mov    fin_char,al        ;Save as final, k
    mov    k,al
    push    ax            ;Push it
    inc    stack_count
    mov    cx,stack_count        ;Pop stack
    jcxz    l18            ;If anything there
l17:     pop    ax
     call    write_char
     loop    l17
l18:    mov    stack_count,cx        ;Clear count on stack
    call    add_code        ;Add new code to table
    mov    ax,in_code        ;Save input code
    mov    old_code,ax
    mov    bx,free_code        ;Hit table limit?
    cmp    bx,max_code
    jl    l23            ;Less means no
     cmp    nbits,12        ;Still within twelve bits?
     je    l23            ;no (next code should be clear)
      inc    nbits            ;Increase code size
      shl    max_code,1        ;Double max code
l23:    jmp    l1            ;Get next code
decompress    endp

data    segment
hash_seg    dw    ?
cur_code    dw    ?
old_code    dw    ?
in_code        dw    ?
free_code    dw    FIRST_FREE
stack_count    dw    0
nbits        dw    9
max_code    dw    512
fin_char    db    ?
k        db    ?
data    ends

read_code    proc    near
    mov    ax,bit_offset        ;Get bit offset
    add    ax,nbits        ;Adjust by code size
    xchg    bit_offset,ax        ;Swap
    mov    cx,8            ;Calculate byte offset
    xor    dx,dx
    div    cx
    cmp    ax,1021            ;Approaching end of buffer?
    jl    rd0            ;no

    push    dx            ;Save offset in byte
    add    dx,nbits        ;Calculate new bit offset
    mov    bit_offset,dx
    mov    cx,1024            ;1k buffer
    mov    bp,ax            ;save byte offset
    sub    cx,ax            ;Calculate bytes left
;    add    ax,offset input_buffer    ;Point to char
    mov    di,offset input_buffer    ;TH Point to beginning of buffer
    add    ax,di            ;TH Point to char

    mov    si,ax
;    lea    di,input_buffer        ;Point to beginning of buffer
rep    movsb                ;Move last chars down
    hread    input_handle,[di],bp    ;Fill rest of buffer
    xor    ax,ax            ;Clear ax
    pop    dx            ;Restore offset in byte
rd0:    add    ax,offset input_buffer    ;Point to char
    mov    si,ax
    lodsw                ;Get word
    mov    bx,ax            ;Save in AX
    lodsb                ;Next byte
    mov    cx,dx            ;Offset in byte
    jcxz    rd2            ;If zero, skip shifts
rd1:     shr    al,1            ;Put code in low (code size) bits of BX
     rcr    bx,1
     loop    rd1
rd2:    mov    ax,bx            ;put code in ax
    mov    bx,nbits        ;mask off unwanted bits
    sub    bx,9
    shl    bx,1
    and    ax,masks[bx]
    ret
read_code    endp

data    segment
masks        dw    1ffh,3ffh,7ffh,0fffh
input_buffer    db    1024 dup (?)
bit_offset    dw    0
data    ends

init_tab    proc    near
    mov    nbits,9            ;Initialize variables
    mov    max_code,512
    mov    free_code,FIRST_FREE
    ret
init_tab    endp

write_char    proc    near
    mov    di,output_offset    ;Get offset in buffer
    cmp    di,1024            ;Full?
    jl    wc1            ;no
     push    ax            ;Save registers
     push    cx
     hwrite    output_handle,output_buffer,di    ;Write buffer to file
     pop    cx
     pop    ax
     xor    di,di    ;TH        ;Point to beginning of buffer
     mov    output_offset,di
wc1:    lea    di,output_buffer[di]    ;Point into buffer
    stosb                ;Store char
    inc    output_offset        ;Increment number of chars in buffer
    ret
write_char    endp

data    segment
output_offset    dw    0
output_buffer    db    1024 dup (?)
data    ends

index        proc    near
    mov    bp,bx            ;bx = bx * 3 (3 byte entries)
    shl    bx,1            ;bp = bx
    add    bx,bp            ;bx = bx * 2 + bp
    ret
index        endp

add_code    proc    near
    mov    bx,free_code        ;Get new code
    call    index            ;convert to address
    push    ES            ;point to hash table
    mov    ES,hash_seg
    mov    al,k            ;get suffix char
    mov    ES:[bx].char,al        ;save it
    mov    ax,old_code        ;get prefix code
    mov    ES:[bx].next,ax        ;save it
    pop    ES
    inc    free_code        ;set next code
    ret
add_code    endp

code    ends

    end    start