From 6cafb40142fca520cc21eb213dab2bc08b545e8d Mon Sep 17 00:00:00 2001
From: JalonSolov <JalonSolov@gmail.com>
Date: Tue, 12 May 2026 14:21:27 -0400
Subject: [PATCH] lz: add lzma2 variation (#27145)

* lz: add lzma2 variation

* add lzma2 to interop
---
 vlib/compress/lz/README.md            |  1 +
 vlib/compress/lz/interop/lz_interop.v |  2 +-
 vlib/compress/lz/lz.v                 |  4 ++++
 vlib/compress/lz/lz_test.v            |  7 +++++--
 vlib/compress/lz/lzma2.v              | 18 ++++++++++++++++++
 5 files changed, 29 insertions(+), 3 deletions(-)
 create mode 100644 vlib/compress/lz/lzma2.v

diff --git a/vlib/compress/lz/README.md b/vlib/compress/lz/README.md
index c4b1d0667..a13bde0b0 100644
--- a/vlib/compress/lz/README.md
+++ b/vlib/compress/lz/README.md
@@ -10,6 +10,7 @@ Supported formats:
 - `lz4`
 - `lzss`
 - `lzma`
+- `lzma2`
 - `lzjb`
 
 Use the generic API when selecting a format dynamically:
diff --git a/vlib/compress/lz/interop/lz_interop.v b/vlib/compress/lz/interop/lz_interop.v
index 5c010d08f..3d02a1384 100644
--- a/vlib/compress/lz/interop/lz_interop.v
+++ b/vlib/compress/lz/interop/lz_interop.v
@@ -17,7 +17,7 @@ fn main() {
 
 	println('LZ interop input: ${data.len} bytes')
 
-	for format in [lz.Format.lz77, .lz78, .lzw, .lz4, .lzss, .lzma, .lzjb] {
+	for format in [lz.Format.lz77, .lz78, .lzw, .lz4, .lzss, .lzma, .lzma2, .lzjb] {
 		validate_v_roundtrip(data, format) or {
 			eprintln('V validation failed for ${format}: ${err.msg()}')
 			exit(1)
diff --git a/vlib/compress/lz/lz.v b/vlib/compress/lz/lz.v
index cd9c44135..8bddee6fb 100644
--- a/vlib/compress/lz/lz.v
+++ b/vlib/compress/lz/lz.v
@@ -8,6 +8,7 @@ pub enum Format {
 	lz4
 	lzss
 	lzma
+	lzma2
 	lzjb
 }
 
@@ -21,6 +22,7 @@ pub fn format_from_string(name string) !Format {
 		'lz4' { .lz4 }
 		'lzss' { .lzss }
 		'lzma' { .lzma }
+		'lzma2' { .lzma2 }
 		'lzjb' { .lzjb }
 		else { return error('unknown lz format: ${name}') }
 	}
@@ -35,6 +37,7 @@ pub fn compress(data []u8, format Format) ![]u8 {
 		.lz4 { compress_lz4(data) }
 		.lzss { compress_lzss(data) }
 		.lzma { compress_lzma(data) }
+		.lzma2 { compress_lzma2(data) }
 		.lzjb { compress_lzjb(data) }
 	}
 }
@@ -48,6 +51,7 @@ pub fn decompress(data []u8, format Format) ![]u8 {
 		.lz4 { decompress_lz4(data) }
 		.lzss { decompress_lzss(data) }
 		.lzma { decompress_lzma(data) }
+		.lzma2 { decompress_lzma2(data) }
 		.lzjb { decompress_lzjb(data) }
 	}
 }
diff --git a/vlib/compress/lz/lz_test.v b/vlib/compress/lz/lz_test.v
index 577d9abc3..f30df8c13 100644
--- a/vlib/compress/lz/lz_test.v
+++ b/vlib/compress/lz/lz_test.v
@@ -4,7 +4,7 @@ const sample_data = ('The quick brown fox jumps over the lazy dog. '.repeat(12)
 	'aaaaaaaaabbbbbbbbbcccccccccdddddddddeeeeeeeee').bytes()
 
 fn test_roundtrip_all_formats() {
-	formats := [Format.lz77, .lz78, .lzw, .lz4, .lzss, .lzma, .lzjb]
+	formats := [Format.lz77, .lz78, .lzw, .lz4, .lzss, .lzma, .lzma2, .lzjb]
 	for format in formats {
 		compressed := compress(sample_data, format)!
 		decompressed := decompress(compressed, format)!
@@ -31,6 +31,9 @@ fn test_format_specific_api_roundtrip() {
 	lzma_data := compress_lzma(sample_data)!
 	assert decompress_lzma(lzma_data)! == sample_data
 
+	lzma2_data := compress_lzma2(sample_data)!
+	assert decompress_lzma2(lzma2_data)! == sample_data
+
 	lzjb_data := compress_lzjb(sample_data)!
 	assert decompress_lzjb(lzjb_data)! == sample_data
 }
@@ -82,7 +85,7 @@ fn test_high_entropy_roundtrip_large_window_formats() {
 		data[i] = u8(state >> 24)
 	}
 
-	for format in [Format.lz4, .lzma] {
+	for format in [Format.lz4, .lzma, .lzma2] {
 		compressed := compress(data, format)!
 		decompressed := decompress(compressed, format)!
 		assert decompressed == data
diff --git a/vlib/compress/lz/lzma2.v b/vlib/compress/lz/lzma2.v
new file mode 100644
index 000000000..cd3c794db
--- /dev/null
+++ b/vlib/compress/lz/lzma2.v
@@ -0,0 +1,18 @@
+module lz
+
+const lzma2_profile = MatchProfile{
+	window:      65535
+	min_match:   3
+	max_match:   130
+	max_literal: 128
+}
+
+// compress_lzma2 compresses data using a pure-V LZMA2-like stream.
+pub fn compress_lzma2(data []u8) ![]u8 {
+	return compress_with_profile(data, lzma2_profile, .lzma2)
+}
+
+// decompress_lzma2 decompresses data produced by compress_lzma2.
+pub fn decompress_lzma2(data []u8) ![]u8 {
+	return decompress_with_profile(data, lzma2_profile, .lzma2)
+}
-- 
2.39.5