diff --git a/asm/z80/unlzsa1_fast.asm b/asm/z80/unlzsa1_fast.asm new file mode 100644 index 0000000..111f2eb --- /dev/null +++ b/asm/z80/unlzsa1_fast.asm @@ -0,0 +1,163 @@ +; +; Speed-optimized LZSA1 decompressor by spke & uniabis (111 bytes) +; +; ver.00 by spke for LZSA 0.5.4 (03-24/04/2019, 134 bytes); +; ver.01 by spke for LZSA 0.5.6 (25/04/2019, -24 bytes, +0.2% speed); +; ver.02 by spke for LZSA 1.0.5 (24/07/2019, added support for backward decompression); +; ver.03 by uniabis (30/07/2019, -1 byte, +3.5% speed); +; ver.04 by spke (31/07/2019, small re-organization of macros); +; ver.05 by uniabis (22/08/2019, -2 bytes, same speed); +; ver.06 by spke for LZSA 1.0.7 (27/08/2019, +4 bytes, +2.1% speed); +; ver.07 by spke for LZSA 1.1.0 (25/09/2019, added full revision history) +; +; The data must be compressed using the command line compressor by Emmanuel Marty +; The compression is done as follows: +; +; lzsa.exe -f1 -r +; +; where option -r asks for the generation of raw (frame-less) data. +; +; The decompression is done in the standard way: +; +; ld hl,FirstByteOfCompressedData +; ld de,FirstByteOfMemoryForDecompressedData +; call DecompressLZSA1 +; +; Backward compression is also supported; you can compress files backward using: +; +; lzsa.exe -f1 -r -b +; +; and decompress the resulting files using: +; +; ld hl,LastByteOfCompressedData +; ld de,LastByteOfMemoryForDecompressedData +; call DecompressLZSA1 +; +; (do not forget to uncomment the BACKWARD_DECOMPRESS option in the decompressor). +; +; Of course, LZSA compression algorithms are (c) 2019 Emmanuel Marty, +; see https://github.com/emmanuel-marty/lzsa for more information +; +; Drop me an email if you have any comments/ideas/suggestions: zxintrospec@gmail.com +; +; This software is provided 'as-is', without any express or implied +; warranty. In no event will the authors be held liable for any damages +; arising from the use of this software. +; +; Permission is granted to anyone to use this software for any purpose, +; including commercial applications, and to alter it and redistribute it +; freely, subject to the following restrictions: +; +; 1. The origin of this software must not be misrepresented; you must not +; claim that you wrote the original software. If you use this software +; in a product, an acknowledgment in the product documentation would be +; appreciated but is not required. +; 2. Altered source versions must be plainly marked as such, and must not be +; misrepresented as being the original software. +; 3. This notice may not be removed or altered from any source distribution. + +; DEFINE BACKWARD_DECOMPRESS + + IFNDEF BACKWARD_DECOMPRESS + + MACRO NEXT_HL + inc hl + ENDM + + MACRO ADD_OFFSET + ex de,hl : add hl,de + ENDM + + MACRO COPY_MATCH + ldi : ldi : ldir + ENDM + + MACRO BLOCKCOPY + ldir + ENDM + + ELSE + + MACRO NEXT_HL + dec hl + ENDM + + MACRO ADD_OFFSET + ex de,hl : ld a,e : sub l : ld l,a + ld a,d : sbc h : ld h,a ; 4*4+3*4 = 28t / 7 bytes + ENDM + + MACRO COPY_MATCH + ldd : ldd : lddr + ENDM + + MACRO BLOCKCOPY + lddr + ENDM + + ENDIF + +@DecompressLZSA1: + ld b,0 : jr ReadToken + +NoLiterals: xor (hl) + push de : NEXT_HL : ld e,(hl) : jp m,LongOffset + + ; short matches have length 0+3..14+3 +ShortOffset: ld d,#FF : add 3 : cp 15+3 : jr nc,LongerMatch + + ; placed here this saves a JP per iteration +CopyMatch: ld c,a +.UseC NEXT_HL : ex (sp),hl ; BC = len, DE = offset, HL = dest, SP ->[dest,src] + ADD_OFFSET ; BC = len, DE = dest, HL = dest-offset, SP->[src] + COPY_MATCH : pop hl ; BC = 0, DE = dest, HL = src + +ReadToken: ; first a byte token "O|LLL|MMMM" is read from the stream, + ; where LLL is the number of literals and MMMM is + ; a length of the match that follows after the literals + ld a,(hl) : and #70 : jr z,NoLiterals + + cp #70 : jr z,MoreLiterals ; LLL=7 means 7+ literals... + rrca : rrca : rrca : rrca ; LLL<7 means 0..6 literals... + + ld c,a : ld a,(hl) + NEXT_HL : BLOCKCOPY + + ; next we read the first byte of the offset + push de : ld e,(hl) + ; the top bit of token is set if the offset contains two bytes + and #8F : jp p,ShortOffset + +LongOffset: ; read second byte of the offset + NEXT_HL : ld d,(hl) + add -128+3 : cp 15+3 : jp c,CopyMatch + + ; MMMM=15 indicates a multi-byte number of literals +LongerMatch: NEXT_HL : add (hl) : jr nc,CopyMatch + + ; the codes are designed to overflow; + ; the overflow value 1 means read 1 extra byte + ; and overflow value 0 means read 2 extra bytes +.code1 ld b,a : NEXT_HL : ld c,(hl) : jr nz,CopyMatch.UseC +.code0 NEXT_HL : ld b,(hl) + + ; the two-byte match length equal to zero + ; designates the end-of-data marker + ld a,b : or c : jr nz,CopyMatch.UseC + pop de : ret + +MoreLiterals: ; there are three possible situations here + xor (hl) : exa + ld a,7 : NEXT_HL : add (hl) : jr c,ManyLiterals + +CopyLiterals: ld c,a +.UseC NEXT_HL : BLOCKCOPY + + push de : ld e,(hl) + exa : jp p,ShortOffset : jr LongOffset + +ManyLiterals: +.code1 ld b,a : NEXT_HL : ld c,(hl) : jr nz,CopyLiterals.UseC +.code0 NEXT_HL : ld b,(hl) : jr CopyLiterals.UseC + + diff --git a/asm/z80/unlzsa1_small.asm b/asm/z80/unlzsa1_small.asm new file mode 100644 index 0000000..28603c1 --- /dev/null +++ b/asm/z80/unlzsa1_small.asm @@ -0,0 +1,135 @@ +; +; Size-optimized LZSA1 decompressor by spke & uniabis (67 bytes) +; +; ver.00 by spke for LZSA 0.5.4 (23/04/2019, 69 bytes); +; ver.01 by spke for LZSA 1.0.5 (24/07/2019, added support for backward decompression); +; ver.02 by uniabis (30/07/2019, -1 byte, +3.2% speed); +; ver.03 by spke for LZSA 1.0.7 (31/07/2019, small re-organization of macros); +; ver.04 by spke (06/08/2019, -1 byte, -1.2% speed); +; ver.05 by spke for LZSA 1.1.0 (25/09/2019, added full revision history) +; +; The data must be compressed using the command line compressor by Emmanuel Marty +; The compression is done as follows: +; +; lzsa.exe -f1 -r +; +; where option -r asks for the generation of raw (frame-less) data. +; +; The decompression is done in the standard way: +; +; ld hl,FirstByteOfCompressedData +; ld de,FirstByteOfMemoryForDecompressedData +; call DecompressLZSA1 +; +; Backward compression is also supported; you can compress files backward using: +; +; lzsa.exe -f1 -r -b +; +; and decompress the resulting files using: +; +; ld hl,LastByteOfCompressedData +; ld de,LastByteOfMemoryForDecompressedData +; call DecompressLZSA1 +; +; (do not forget to uncomment the BACKWARD_DECOMPRESS option in the decompressor). +; +; Of course, LZSA compression algorithms are (c) 2019 Emmanuel Marty, +; see https://github.com/emmanuel-marty/lzsa for more information +; +; Drop me an email if you have any comments/ideas/suggestions: zxintrospec@gmail.com +; +; This software is provided 'as-is', without any express or implied +; warranty. In no event will the authors be held liable for any damages +; arising from the use of this software. +; +; Permission is granted to anyone to use this software for any purpose, +; including commercial applications, and to alter it and redistribute it +; freely, subject to the following restrictions: +; +; 1. The origin of this software must not be misrepresented; you must not +; claim that you wrote the original software. If you use this software +; in a product, an acknowledgment in the product documentation would be +; appreciated but is not required. +; 2. Altered source versions must be plainly marked as such, and must not be +; misrepresented as being the original software. +; 3. This notice may not be removed or altered from any source distribution. + +; DEFINE BACKWARD_DECOMPRESS + + IFNDEF BACKWARD_DECOMPRESS + + MACRO NEXT_HL + inc hl + ENDM + + MACRO ADD_OFFSET + ex de,hl : add hl,de + ENDM + + MACRO BLOCKCOPY + ldir + ENDM + + ELSE + + MACRO NEXT_HL + dec hl + ENDM + + MACRO ADD_OFFSET + push hl : or a : sbc hl,de : pop de ; 11+4+15+10 = 40t / 5 bytes + ENDM + + MACRO BLOCKCOPY + lddr + ENDM + + ENDIF + +@DecompressLZSA1: + ld b,0 + + ; first a byte token "O|LLL|MMMM" is read from the stream, + ; where LLL is the number of literals and MMMM is + ; a length of the match that follows after the literals +ReadToken: ld a,(hl) : NEXT_HL : push af + and #70 : jr z,NoLiterals + + rrca : rrca : rrca : rrca ; LLL<7 means 0..6 literals... + cp #07 : call z,ReadLongBA ; LLL=7 means 7+ literals... + + ld c,a : BLOCKCOPY + + ; next we read the low byte of the -offset +NoLiterals: pop af : push de : ld e,(hl) : NEXT_HL : ld d,#FF + ; the top bit of token is set if + ; the offset contains the high byte as well + or a : jp p,ShortOffset + +LongOffset: ld d,(hl) : NEXT_HL + + ; last but not least, the match length is read +ShortOffset: and #0F : add 3 ; MMMM<15 means match lengths 0+3..14+3 + cp 15+3 : call z,ReadLongBA ; MMMM=15 means lengths 14+3+ + ld c,a + + ex (sp),hl ; BC = len, DE = -offset, HL = dest, SP -> [src] + ADD_OFFSET ; BC = len, DE = dest, HL = dest+(-offset), SP -> [src] + BLOCKCOPY ; BC = 0, DE = dest + pop hl : jr ReadToken ; HL = src + + ; a standard routine to read extended codes + ; into registers B (higher byte) and A (lower byte). +ReadLongBA: add (hl) : NEXT_HL : ret nc + + ; the codes are designed to overflow; + ; the overflow value 1 means read 1 extra byte + ; and overflow value 0 means read 2 extra bytes +.code1: ld b,a : ld a,(hl) : NEXT_HL : ret nz +.code0: ld c,a : ld b,(hl) : NEXT_HL + + ; the two-byte match length equal to zero + ; designates the end-of-data marker + or b : ld a,c : ret nz + pop de : pop de : ret + diff --git a/asm/z80/unlzsa2_fast.asm b/asm/z80/unlzsa2_fast.asm new file mode 100644 index 0000000..f9535bf --- /dev/null +++ b/asm/z80/unlzsa2_fast.asm @@ -0,0 +1,249 @@ +; +; Speed-optimized LZSA2 decompressor by spke (v.2 02-27/08/2019, 216 bytes); +; with improvements by uniabis (30/07/2019, -5 bytes, +3% speed and support for Hitachi HD64180). +; +; The data must be compressed using the command line compressor by Emmanuel Marty +; The compression is done as follows: +; +; lzsa.exe -f2 -r +; +; where option -r asks for the generation of raw (frame-less) data. +; +; The decompression is done in the standard way: +; +; ld hl,FirstByteOfCompressedData +; ld de,FirstByteOfMemoryForDecompressedData +; call DecompressLZSA2 +; +; Backward compression is also supported; you can compress files backward using: +; +; lzsa.exe -f2 -r -b +; +; and decompress the resulting files using: +; +; ld hl,LastByteOfCompressedData +; ld de,LastByteOfMemoryForDecompressedData +; call DecompressLZSA2 +; +; (do not forget to uncomment the BACKWARD_DECOMPRESS option in the decompressor). +; +; Of course, LZSA2 compression algorithms are (c) 2019 Emmanuel Marty, +; see https://github.com/emmanuel-marty/lzsa for more information +; +; Drop me an email if you have any comments/ideas/suggestions: zxintrospec@gmail.com +; +; This software is provided 'as-is', without any express or implied +; warranty. In no event will the authors be held liable for any damages +; arising from the use of this software. +; +; Permission is granted to anyone to use this software for any purpose, +; including commercial applications, and to alter it and redistribute it +; freely, subject to the following restrictions: +; +; 1. The origin of this software must not be misrepresented; you must not +; claim that you wrote the original software. If you use this software +; in a product, an acknowledgment in the product documentation would be +; appreciated but is not required. +; 2. Altered source versions must be plainly marked as such, and must not be +; misrepresented as being the original software. +; 3. This notice may not be removed or altered from any source distribution. + +; DEFINE BACKWARD_DECOMPRESS ; uncomment for data compressed with option -b +; DEFINE HD64180 ; uncomment for systems using Hitachi HD64180 + + IFNDEF BACKWARD_DECOMPRESS + + MACRO NEXT_HL + inc hl + ENDM + + MACRO ADD_OFFSET + ex de,hl : add hl,de + ENDM + + MACRO BLOCKCOPY + ldir + ENDM + + MACRO COPY_MATCH + ldi : ldir + ENDM + + ELSE + + MACRO NEXT_HL + dec hl + ENDM + + MACRO ADD_OFFSET + ex de,hl : ld a,e : sub l : ld l,a + ld a,d : sbc h : ld h,a ; 4*4+3*4 = 28t / 7 bytes + ENDM + + MACRO BLOCKCOPY + lddr + ENDM + + MACRO COPY_MATCH + ldd : lddr + ENDM + + ENDIF + + IFNDEF HD64180 + + MACRO LD_IX_DE + ld ixl,e : ld ixh,d + ENDM + + MACRO LD_DE_IX + ld e,ixl : ld d,ixh + ENDM + + ELSE + + MACRO LD_IX_DE + push de : pop ix + ENDM + + MACRO LD_DE_IX + push ix : pop de + ENDM + + ENDIF + +@DecompressLZSA2: + ; A' stores next nibble as %1111.... or assumed to contain trash + ; B is assumed to be 0 + ld b,0 : scf : exa : jr ReadToken + + + + +ManyLiterals: ld a,18 : add (hl) : NEXT_HL : jr nc,CopyLiterals + ld c,(hl) : NEXT_HL + ld a,b : ld b,(hl) + jr ReadToken.NEXTHLuseBC + + + + +MoreLiterals: ld b,(hl) : NEXT_HL + scf : exa : jr nc,.noUpdate + + ld a,(hl) : or #F0 : exa + ld a,(hl) : NEXT_HL : or #0F + rrca : rrca : rrca : rrca + +.noUpdate ;sub #F0-3 : cp 15+3 : jr z,ManyLiterals + inc a : jr z,ManyLiterals : sub #F0-3+1 + +CopyLiterals: ld c,a : ld a,b : ld b,0 + BLOCKCOPY + push de : or a : jp p,CASE0xx ;: jr CASE1xx + + cp %11000000 : jr c,CASE10x + +CASE11x cp %11100000 : jr c,CASE110 + + ; "111": repeated offset +CASE111: LD_DE_IX : jr MatchLen + + + + +Literals0011: jr nz,MoreLiterals + + ; if "LL" of the byte token is equal to 0, + ; there are no literals to copy +NoLiterals: or (hl) : NEXT_HL + push de : jp m,CASE1xx + + ; short (5 or 9 bit long) offsets +CASE0xx ld d,#FF : cp %01000000 : jr c,CASE00x + + ; "01x": the case of the 9-bit offset +CASE01x: cp %01100000 : rl d + +ReadOffsetE ld e,(hl) : NEXT_HL + +SaveOffset: LD_IX_DE + +MatchLen: inc a : and %00000111 : jr z,LongerMatch : inc a + +CopyMatch: ld c,a +.useC ex (sp),hl ; BC = len, DE = offset, HL = dest, SP ->[dest,src] + ADD_OFFSET ; BC = len, DE = dest, HL = dest-offset, SP->[src] + COPY_MATCH : pop hl + + ; compressed data stream contains records + ; each record begins with the byte token "XYZ|LL|MMM" +ReadToken: ld a,(hl) : and %00011000 : jp pe,Literals0011 ; process the cases 00 and 11 separately + + rrca : rrca : rrca + + ld c,a : ld a,(hl) ; token is re-read for further processing +.NEXTHLuseBC NEXT_HL + BLOCKCOPY + + ; the token and literals are followed by the offset + push de : or a : jp p,CASE0xx + +CASE1xx cp %11000000 : jr nc,CASE11x + + ; "10x": the case of the 5-bit offset +CASE10x: ld c,a : exa : jr nc,.noUpdate + + ld a,(hl) : or #F0 : exa + ld a,(hl) : NEXT_HL : or #0F + rrca : rrca : rrca : rrca + +.noUpdate ld d,a : ld a,c + cp %10100000 : rl d + dec d : dec d : jr ReadOffsetE + + + + ; "110": 16-bit offset +CASE110: ld d,(hl) : NEXT_HL : jr ReadOffsetE + + + + + ; "00x": the case of the 5-bit offset +CASE00x: ld c,a : exa : jr nc,.noUpdate + + ld a,(hl) : or #F0 : exa + ld a,(hl) : NEXT_HL : or #0F + rrca : rrca : rrca : rrca + +.noUpdate ld e,a : ld a,c + cp %00100000 : rl e : jr SaveOffset + + + + + +LongerMatch: scf : exa : jr nc,.noUpdate + + ld a,(hl) : or #F0 : exa + ld a,(hl) : NEXT_HL : or #0F + rrca : rrca : rrca : rrca + +.noUpdate sub #F0-9 : cp 15+9 : jr c,CopyMatch + +LongMatch: add (hl) : NEXT_HL : jr nc,CopyMatch + ld c,(hl) : NEXT_HL + ld b,(hl) : NEXT_HL : jr nz,CopyMatch.useC + pop de : ret + + + + + + + + + + + diff --git a/asm/z80/unlzsa2_small.asm b/asm/z80/unlzsa2_small.asm new file mode 100644 index 0000000..b65bd52 --- /dev/null +++ b/asm/z80/unlzsa2_small.asm @@ -0,0 +1,181 @@ +; +; Size-optimized LZSA2 decompressor by spke (v.1 02-09/06/2019, 140 bytes); +; with improvements by uniabis (30/07/2019, -1 byte, +3% speed and support for Hitachi HD64180). +; +; The data must be compressed using the command line compressor by Emmanuel Marty +; The compression is done as follows: +; +; lzsa.exe -f2 -r +; +; where option -r asks for the generation of raw (frame-less) data. +; +; The decompression is done in the standard way: +; +; ld hl,FirstByteOfCompressedData +; ld de,FirstByteOfMemoryForDecompressedData +; call DecompressLZSA2 +; +; Backward compression is also supported; you can compress files backward using: +; +; lzsa.exe -f2 -r -b +; +; and decompress the resulting files using: +; +; ld hl,LastByteOfCompressedData +; ld de,LastByteOfMemoryForDecompressedData +; call DecompressLZSA2 +; +; (do not forget to uncomment the BACKWARD_DECOMPRESS option in the decompressor). +; +; Of course, LZSA2 compression algorithms are (c) 2019 Emmanuel Marty, +; see https://github.com/emmanuel-marty/lzsa for more information +; +; Drop me an email if you have any comments/ideas/suggestions: zxintrospec@gmail.com +; +; This software is provided 'as-is', without any express or implied +; warranty. In no event will the authors be held liable for any damages +; arising from the use of this software. +; +; Permission is granted to anyone to use this software for any purpose, +; including commercial applications, and to alter it and redistribute it +; freely, subject to the following restrictions: +; +; 1. The origin of this software must not be misrepresented; you must not +; claim that you wrote the original software. If you use this software +; in a product, an acknowledgment in the product documentation would be +; appreciated but is not required. +; 2. Altered source versions must be plainly marked as such, and must not be +; misrepresented as being the original software. +; 3. This notice may not be removed or altered from any source distribution. +; + +; DEFINE BACKWARD_DECOMPRESS ; uncomment for data compressed with option -b +; DEFINE HD64180 ; uncomment for systems using Hitachi HD64180 + + IFNDEF BACKWARD_DECOMPRESS + + MACRO NEXT_HL + inc hl + ENDM + + MACRO ADD_OFFSET + ex de,hl : add hl,de + ENDM + + MACRO BLOCKCOPY + ldir + ENDM + + ELSE + + MACRO NEXT_HL + dec hl + ENDM + + MACRO ADD_OFFSET + push hl : or a : sbc hl,de : pop de ; 11+4+15+10 = 40t / 5 bytes + ENDM + + MACRO BLOCKCOPY + lddr + ENDM + + ENDIF + + IFNDEF HD64180 + + MACRO LD_IY_DE + ld iyl,e : ld iyh,d + ENDM + + MACRO LD_DE_IY + ld e,iyl : ld d,iyh + ENDM + + ELSE + + MACRO LD_IY_DE + push de : pop iy + ENDM + + MACRO LD_DE_IY + push iy : pop de + ENDM + + ENDIF + +@DecompressLZSA2: + xor a : ld b,a : exa : jr ReadToken + +CASE00x: call ReadNibble + ld e,a : ld a,c + cp %00100000 : rl e : jr SaveOffset + +CASE0xx ld d,#FF : cp %01000000 : jr c,CASE00x + +CASE01x: cp %01100000 : rl d + +OffsetReadE: ld e,(hl) : NEXT_HL + +SaveOffset: LD_IY_DE + +MatchLen: and %00000111 : add 2 : cp 9 : call z,ExtendedCode + +CopyMatch: ld c,a + ex (sp),hl ; BC = len, DE = -offset, HL = dest, SP -> [src] + ADD_OFFSET ; BC = len, DE = dest, HL = dest+(-offset), SP -> [src] + BLOCKCOPY ; BC = 0, DE = dest + pop hl ; HL = src + +ReadToken: ld a,(hl) : NEXT_HL : push af + and %00011000 : jr z,NoLiterals + + rrca : rrca : rrca + call pe,ExtendedCode + + ld c,a + BLOCKCOPY + +NoLiterals: pop af : push de + or a : jp p,CASE0xx + +CASE1xx cp %11000000 : jr nc,CASE11x + +CASE10x: call ReadNibble + ld d,a : ld a,c + cp %10100000 : rl d + dec d : dec d : DB #CA ; jr OffsetReadE ; #CA is JP Z,.. to skip all commands in CASE110 before jr OffsetReadE + +CASE110: ld d,(hl) : NEXT_HL : jr OffsetReadE + +CASE11x cp %11100000 : jr c,CASE110 + +CASE111: LD_DE_IY : jr MatchLen + +ExtendedCode: call ReadNibble : inc a : jr z,ExtraByte + sub #F0+1 : add c : ret +ExtraByte ld a,15 : add c : add (hl) : NEXT_HL : ret nc + ld a,(hl) : NEXT_HL + ld b,(hl) : NEXT_HL : ret nz + pop de : pop de ; RET is not needed, because RET from ReadNibble is sufficient + +ReadNibble: ld c,a : xor a : exa : ret m +UpdateNibble ld a,(hl) : or #F0 : exa + ld a,(hl) : NEXT_HL : or #0F + rrca : rrca : rrca : rrca : ret + + + + + + + + + + + + + + + +