uxn-utils/cli/lz/ulzenc.tal

( ulz encoder )

|10 @Console &vector $2 &read $1 &pad $5 &write $1 &err $1
|a0 @File &vector $2 &success $2 &stat $2 &delete $1 &append $1 &name $2 &length $2 &read $2 &write $2

|0000

	@src $30
	@dst $30
	@ptr $1
	@match-len $2
	@output-ptr $2
	@match-ctl $2
	@dict-best $2
	@combine $2
	@dict $2
	@dict-len $2

|0100

@ready-src ( -> )
	;&await .Console/vector DEO2
	BRK
	&await ( -> )
		.Console/read DEI .src skey ?ready-dst
		BRK

@ready-dst ( -> )
	;&await .Console/vector DEO2
	BRK
	&await ( -> )
		.Console/read DEI .dst skey ?on-ready
		BRK

@on-ready ( -> )
	( | load raw )
	;src .File/name DEO2
	#4000 .File/length DEO2
	;raw .File/read DEO2
	;raw .File/success DEI2 uxn_lz_compress
	( write )
	;dst .File/name DEO2
	.output-ptr LDZ2 ;compressed SUB2 .File/length DEO2
	;compressed .File/write DEO2
	( halt ) #800f DEO
	BRK

@<append-byte> ( byte -- )
	.output-ptr LDZ2 INC2k .output-ptr STZ2 STA
	JMP2r

@uxn_lz_compress ( input* length* -- )
	( fill variables )
	;compressed .output-ptr STZ2


	ADD2k NIP2 SWP2
	&w ( end* start* -- )
		EQU2k ?&end
		( | get available dictionary size )
		DUP2 ;raw SUB2 #0100 LTH2k ?{ SWP2 }
		POP2 .dict-len STZ2

		( | size of the string to search for )
		SUB2k #3fff #0004 ADD2 LTH2k ?{ SWP2 }
		POP2 ,&string-len STR2

		( | debug )
		.dict-len LDZ2 phex #2018 DEO ,&string-len LDR2 phex #2018 DEO

		( | itterate through the dictionary )
		#0000 .match-len STZ2
		DUP2 .dict-len LDZ2 SUB2 .dict STZ2

		LIT "[ #18 DEO
		LDAk #18 DEO
		LIT "] #18 DEO
		#0a18 DEO

		&for1 ( for ; dict_len; dict++, dict_len-- )
			.dict-len LDZ2 #0000 EQU2 ?&end-for1

			( Find common prefix length with the string )
			#0000
			&for2 ( for i = 0;; i++ )
				( If we reach the end of the string, it's the best possible match. )
				DUP2 [ LIT2 &string-len $2 ] NEQ2 ?{
					DUP2 .match-len STZ2
					.dict LDZ2 .dict-best STZ2
					POP2 !&done-search }

				( in[i] != dict[i % dict_len] break; )
				( a ) ADD2k LDA STH
				( b ) DUP2 .dict-len LDZ2 DIV2k MUL2 SUB2 .dict LDZ2 ADD2 LDA STHr
					NEQ ?&end-for2
				INC2 ORAk ?&for2
			&end-for2

			( i > match_len )
			DUP2 .match-len LDZ2 LTH2 ?{
				DUP2 .match-len STZ2
				.dict LDZ2 .dict-best STZ2 }
			POP2

			.dict LDZ2 INC2 .dict STZ2
			.dict-len LDZ2 #0001 SUB2 .dict-len STZ2 !&for1
		&end-for1
	&done-search ( -- )
		( CPY ) .match-len LDZ2 #0003 GTH2 ?op-cpy
		( LIT ) !op-lit
	&end
	POP2 POP2 JMP2r


(
@|opcodes )

@op-cpy ( in* -- )
	LIT "* #18 DEO
	( More numeric range: treat 0 as 4, 1 as 5, etc. )
	.match-len LDZ2 #0004 SUB2 .match-ctl STZ2
	( CPY2 )
	.match-ctl LDZ2 #003f GTH2 ?&cpy2
		( CPY1 )
		( *output_ptr++ = match_ctl | 0x80; )
		.match-ctl LDZ2 NIP #80 ORA <append-byte>
		!&cpy-resume
		&cpy2
		.match-ctl LDZ2
		( match_ctl >> 8 | 0xc0 ) SWP #c0 ORA <append-byte>
		( match_ctl ) <append-byte>
	&cpy-resume
	( *output_ptr++ = in - dict_best - 1; )
	DUP2 .dict-best LDZ2 SUB2 #0001 SUB2 NIP <append-byte>
	( in += match_len; Advance input by size of the match )
	.match-len LDZ2 ADD2
	( Disable combining previous literal, if any )
	#0000 .combine STZ2
	!uxn_lz_compress/w

@op-lit ( in* -- )
	.combine LDZ2 ORA ?&combine
	( start a new literal )
	( Store this address, and later use it to increment the literal size. )
	( combine = output_ptr++; )
	.output-ptr LDZ2
		INC2k .output-ptr STZ2
		.combine STZ2
	( *combine = 0; )
	#00 .combine LDZ2 LDA2 STA
	( *output_ptr++ = *in++; )
	LDAk <append-byte>
	INC2 !uxn_lz_compress/w
&combine ( -- )
	( if ++*combine == 127 )
	.combine LDZ2 LDA INC DUP #7f NEQ ?{ POP #00 }
		.combine LDZ2 STA
	LDAk <append-byte>
	INC2 !uxn_lz_compress/w

(
@|stdlib )

@skey ( key buf -- proc )
	OVR #21 LTH ?&eval
	#00 SWP sput #00 JMP2r
	&eval POP2 #01 JMP2r

@scap ( str* -- end* )
	&w INC2 & LDAk ?&w
	JMP2r

@sput ( chr str* -- )
	scap/ INC2k #00 ROT ROT STA
	STA
	JMP2r

@getc ( -- b )
	#0001 .File/length DEO2
	;&b .File/read DEO2
	[ LIT &b $1 ] JMP2r

@phex ( short* -- )
	SWP phex/b
	&b ( -- )
		DUP #04 SFT phex/c
	&c ( -- )
		#0f AND DUP #09 GTH #27 MUL ADD [ LIT "0 ] ADD #18 DEO
		JMP2r

@BREAK
	#0a18 DEO
	#010e DEO #010f DEO BRK

@DEBUG1
	LIT "> #18 DEO
	DUP phex/b #0a18 DEO
	JMP2r

(
@|memory )

@raw $4000

@compressed $4000