From 9101bda743a2c23758cc51a7f76038ea290271ce Mon Sep 17 00:00:00 2001
From: neauoire <aliceffekt@gmail.com>
Date: Sat, 18 Nov 2023 12:19:30 -0800
Subject: [PATCH] (lz) *

---
 cli/lz/ulzenc.c   |  6 +++--
 cli/lz/ulzenc.tal | 66 ++++++++++++++++++++++++++++++++++++++++++++---
 2 files changed, 67 insertions(+), 5 deletions(-)

diff --git a/cli/lz/ulzenc.c b/cli/lz/ulzenc.c
index 5621ee9..38ba59b 100644
--- a/cli/lz/ulzenc.c
+++ b/cli/lz/ulzenc.c
@@ -36,15 +36,17 @@ uxn_lz_compress(const void *input, int input_size)
 		/* DEBUG: printf("%04x %04x \n", dict_len, string_len); */
 
 		/* Iterate through the dictionary */
-		for(dict = in - dict_len, match_len = 0; dict_len; dict++, dict_len -= 1) {
+		match_len = 0;
+		dict = in - dict_len;
+		for(; dict_len; dict++, dict_len -= 1) {
 			/* Find common prefix length with the string */
+			/* If we reach the end of the string, this is the best possible match. End. */
 			for(i = 0;; i++) {
 				if(i == string_len) {
 					match_len = i;
 					dict_best = dict;
 					goto done_search;
 				}
-				/* ^ If we reach the end of the string, this is the best possible match. End. */
 				/* Dictionary repeats if we hit the end */
 				if(in[i] != dict[i % dict_len])
 					break;
diff --git a/cli/lz/ulzenc.tal b/cli/lz/ulzenc.tal
index bd69ce6..89c1500 100644
--- a/cli/lz/ulzenc.tal
+++ b/cli/lz/ulzenc.tal
@@ -8,6 +8,7 @@
 	@src $30
 	@dst $30
 	@ptr $1
+	@match-len $2
 
 |0100
 
@@ -34,6 +35,10 @@
 	( halt ) #800f DEO
 	BRK
 
+@<append-byte> ( byte -- )
+	.output-ptr LDZ2 INC2k .output-ptr STZ2 STA
+	JMP2r
+
 @uxn_lz_compress ( input* length* -- )
 	ADD2k NIP2 SWP2
 	&w ( -- )
@@ -44,8 +49,63 @@
 		SUB2k #3fff #0004 ADD2 LTH2k ?{ SWP2 }
 		POP2 ,&string-len STR2
 		( | itterate through the dictionary )
-		[ LIT2 &dict-len $2 ] phex #2018 DEO
-		[ LIT2 &string-len $2 ] phex #0a18 DEO
+		#0000 .match-len STZ2
+		[ LIT2 &dict-len $2 ] SUB2k .dict STZ2
+		( | create a few inlines )
+		DUP2 ,&in STR2
+		&for1 ( dict-len != 0 )
+			( Find common prefix length with the string )
+			#0000
+			&for2 ( i != 0 )
+				( 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 POP2 !&done-search }
+				( Dictionary repeats if we hit the end )
+				( in[i] != dict[i % dict_len] break; )
+				DUP2 [ LIT2 &in $2 ] ADD2 LDA2
+				OVR2 ,&dict-len LDR2 DIV2k MUL2 SUB2
+					.dict LDZ2 ADD2 LDA2 NEQ2 ?&end
+				INC2 ORAk ?&for2
+			&end
+			( i > match_len )
+			DUP2 .match-len LDZ2 GTH2 #00 EQU ?{
+				DUP2 .match-len STZ2
+				.dict-len LDZ2 .dict STZ2 }
+			POP2
+			.dict LDZ2 INC2 .dict STZ2
+			#0001 SUB2 ORAk ?&for1
+		POP2
+
+	&done-search ( -- )
+
+		( CPY )
+		.match-len LDZ2 #0004 LTH2 ?&no-cpy
+			( 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 #0080 ORA2 <append-byte>
+				!&cpy-resume
+				&cpy2
+				( TODO )
+			&cpy-resume
+			( *output_ptr++ = in - dict_best - 1; )
+			DUP2 .dict-best LDZ2 #0001 SUB2 SUB2 <append-byte>
+
+
+			&no-cpy
+
+
+
+		( LIT )
+
+
+		phex #2018 DEO
+		phex #0a18 DEO
 		( TODO: remove me!! ) INC2 NEQ2k ?&w
 	JMP2r
 
@@ -78,7 +138,7 @@
 	&c ( -- )
 		#0f AND DUP #09 GTH #27 MUL ADD [ LIT "0 ] ADD #18 DEO
 		JMP2r
-	
+
 (
 @|memory )