16 Commits

Author SHA1 Message Date
f647449497 add --raw-cost and --hexdump flags to help text 2023-08-25 19:42:59 +02:00
7186188118 add --raw-cost option for heatmap reporting 2023-08-25 19:34:17 +02:00
588c1452b9 strip release executables 2023-08-25 00:03:51 +02:00
1e0c8bb5a7 update readme, change --unpack to --decompress, put progress bar on stderr 2023-08-24 22:59:27 +02:00
e11622202b add support to read/write from/to stdin/stdout 2023-08-24 00:00:20 +02:00
795e6c3090 add basic example for compiling upkr to a c library 2023-03-12 13:45:23 +01:00
080db40d00 deny missing docs 2023-01-27 20:26:36 +01:00
5684185+vsariola@users.noreply.github.com
f502bf4e28 optimize dos stub size to save 2 bytes (discovered by qkumba) 2023-01-26 22:04:59 +01:00
2eb8f230ba add documentation, make pbr optional as well 2022-10-26 23:40:41 +02:00
4eab36b9d9 add some api documentation 2022-10-25 23:33:32 +02:00
7cec54f62b make crossterm dependency optional 2022-10-25 22:40:56 +02:00
7fa6be6ff4 implement printing heatmap as hexdump 2022-10-24 23:34:07 +02:00
cab51e06ff implement heatmap calculation 2022-10-23 23:06:09 +02:00
c4fce626da some clean up - move dos unpacker, fix arm32 unpacker formatting 2022-10-19 22:32:57 +02:00
0d7cda06bb Merge pull request #6 from vsariola/dev/x86
implement three versions of a decompression stub for 16-bit x86 DOS
2022-10-19 22:11:04 +02:00
5684185+vsariola@users.noreply.github.com
140678ae20 implement three versions of a decompression stub for 16-bit x86 DOS 2022-10-17 15:50:09 +03:00
22 changed files with 1630 additions and 92 deletions

255
Cargo.lock generated
View File

@@ -10,9 +10,15 @@ checksum = "38d9ff5d688f1c13395289f67db01d4826b46dd694e7580accdc3e8430f2d98e"
[[package]]
name = "autocfg"
version = "1.0.1"
version = "1.1.0"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "cdb031dd78e28731d87d56cc8ffef4a8f36ca26c38fe2de700543e627f8a464a"
checksum = "d468802bab17cbc0cc575e9b053f41e72aa36bfa6b7f55e3529ffa43161b97fa"
[[package]]
name = "bitflags"
version = "1.3.2"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "bef38d45163c2f1dde094a7dfd33ccf595c92905c8f8f4fdc18d06fb1037718a"
[[package]]
name = "cc"
@@ -56,6 +62,31 @@ dependencies = [
"lazy_static",
]
[[package]]
name = "crossterm"
version = "0.25.0"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "e64e6c0fbe2c17357405f7c758c1ef960fce08bdfb2c03d88d2a18d7e09c4b67"
dependencies = [
"bitflags",
"crossterm_winapi",
"libc",
"mio",
"parking_lot",
"signal-hook",
"signal-hook-mio",
"winapi",
]
[[package]]
name = "crossterm_winapi"
version = "0.9.0"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "2ae1b35a484aa10e07fe0638d02301c5ad24de82d310ccbd2f3693da5f09bf1c"
dependencies = [
"winapi",
]
[[package]]
name = "lazy_static"
version = "1.4.0"
@@ -70,9 +101,40 @@ checksum = "478ee9e62aaeaf5b140bd4138753d1f109765488581444218d3ddda43234f3e8"
[[package]]
name = "libc"
version = "0.2.108"
version = "0.2.135"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "8521a1b57e76b1ec69af7599e75e38e7b7fad6610f037db8c79b127201b5d119"
checksum = "68783febc7782c6c5cb401fbda4de5a9898be1762314da0bb2c10ced61f18b0c"
[[package]]
name = "lock_api"
version = "0.4.9"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "435011366fe56583b16cf956f9df0095b405b82d76425bc8981c0e22e60ec4df"
dependencies = [
"autocfg",
"scopeguard",
]
[[package]]
name = "log"
version = "0.4.17"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "abb12e687cfb44aa40f41fc3978ef76448f9b6038cad6aef4259d3c095a2382e"
dependencies = [
"cfg-if",
]
[[package]]
name = "mio"
version = "0.8.4"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "57ee1c23c7c63b0c9250c339ffdc69255f110b298b901b9f6c82547b7b87caaf"
dependencies = [
"libc",
"log",
"wasi 0.11.0+wasi-snapshot-preview1",
"windows-sys 0.36.1",
]
[[package]]
name = "num-traits"
@@ -83,6 +145,29 @@ dependencies = [
"autocfg",
]
[[package]]
name = "parking_lot"
version = "0.12.1"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "3742b2c103b9f06bc9fff0a37ff4912935851bee6d36f3c02bcc755bcfec228f"
dependencies = [
"lock_api",
"parking_lot_core",
]
[[package]]
name = "parking_lot_core"
version = "0.9.4"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "4dc9e0dc2adc1c69d09143aff38d3d30c5c3f0df0dad82e6d25547af174ebec0"
dependencies = [
"cfg-if",
"libc",
"redox_syscall",
"smallvec",
"windows-sys 0.42.0",
]
[[package]]
name = "pbr"
version = "1.0.4"
@@ -113,6 +198,15 @@ dependencies = [
"proc-macro2",
]
[[package]]
name = "redox_syscall"
version = "0.2.16"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "fb5a58c1855b4b6819d59012155603f0b22ad30cad752600aadfcb695265519a"
dependencies = [
"bitflags",
]
[[package]]
name = "sacabase"
version = "2.0.0"
@@ -122,6 +216,48 @@ dependencies = [
"num-traits",
]
[[package]]
name = "scopeguard"
version = "1.1.0"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "d29ab0c6d3fc0ee92fe66e2d99f700eab17a8d57d1c1d3b748380fb20baa78cd"
[[package]]
name = "signal-hook"
version = "0.3.14"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "a253b5e89e2698464fc26b545c9edceb338e18a89effeeecfea192c3025be29d"
dependencies = [
"libc",
"signal-hook-registry",
]
[[package]]
name = "signal-hook-mio"
version = "0.2.3"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "29ad2e15f37ec9a6cc544097b78a1ec90001e9f71b81338ca39f430adaca99af"
dependencies = [
"libc",
"mio",
"signal-hook",
]
[[package]]
name = "signal-hook-registry"
version = "1.4.0"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "e51e73328dc4ac0c7ccbda3a494dfa03df1de2f46018127f60c693f2648455b0"
dependencies = [
"libc",
]
[[package]]
name = "smallvec"
version = "1.10.0"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "a507befe795404456341dfab10cef66ead4c041f62b8b11bbb92bffe5d0953e0"
[[package]]
name = "syn"
version = "1.0.101"
@@ -160,7 +296,7 @@ source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "6db9e6914ab8b1ae1c260a4ae7a49b6c5611b40328a735b21862567685e73255"
dependencies = [
"libc",
"wasi",
"wasi 0.10.0+wasi-snapshot-preview1",
"winapi",
]
@@ -172,10 +308,11 @@ checksum = "dcc811dc4066ac62f84f11307873c4850cb653bfa9b1719cee2bd2204a4bc5dd"
[[package]]
name = "upkr"
version = "0.2.0"
version = "0.2.2"
dependencies = [
"anyhow",
"cdivsufsort",
"crossterm",
"lexopt",
"pbr",
"thiserror",
@@ -187,6 +324,12 @@ version = "0.10.0+wasi-snapshot-preview1"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "1a143597ca7c7793eff794def352d41792a93c481eb1042423ff7ff72ba2c31f"
[[package]]
name = "wasi"
version = "0.11.0+wasi-snapshot-preview1"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "9c8d87e72b64a3b4db28d11ce29237c246188f4f51057d65a7eab63b7987e423"
[[package]]
name = "winapi"
version = "0.3.9"
@@ -208,3 +351,103 @@ name = "winapi-x86_64-pc-windows-gnu"
version = "0.4.0"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "712e227841d057c1ee1cd2fb22fa7e5a5461ae8e48fa2ca79ec42cfc1931183f"
[[package]]
name = "windows-sys"
version = "0.36.1"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "ea04155a16a59f9eab786fe12a4a450e75cdb175f9e0d80da1e17db09f55b8d2"
dependencies = [
"windows_aarch64_msvc 0.36.1",
"windows_i686_gnu 0.36.1",
"windows_i686_msvc 0.36.1",
"windows_x86_64_gnu 0.36.1",
"windows_x86_64_msvc 0.36.1",
]
[[package]]
name = "windows-sys"
version = "0.42.0"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "5a3e1820f08b8513f676f7ab6c1f99ff312fb97b553d30ff4dd86f9f15728aa7"
dependencies = [
"windows_aarch64_gnullvm",
"windows_aarch64_msvc 0.42.0",
"windows_i686_gnu 0.42.0",
"windows_i686_msvc 0.42.0",
"windows_x86_64_gnu 0.42.0",
"windows_x86_64_gnullvm",
"windows_x86_64_msvc 0.42.0",
]
[[package]]
name = "windows_aarch64_gnullvm"
version = "0.42.0"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "41d2aa71f6f0cbe00ae5167d90ef3cfe66527d6f613ca78ac8024c3ccab9a19e"
[[package]]
name = "windows_aarch64_msvc"
version = "0.36.1"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "9bb8c3fd39ade2d67e9874ac4f3db21f0d710bee00fe7cab16949ec184eeaa47"
[[package]]
name = "windows_aarch64_msvc"
version = "0.42.0"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "dd0f252f5a35cac83d6311b2e795981f5ee6e67eb1f9a7f64eb4500fbc4dcdb4"
[[package]]
name = "windows_i686_gnu"
version = "0.36.1"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "180e6ccf01daf4c426b846dfc66db1fc518f074baa793aa7d9b9aaeffad6a3b6"
[[package]]
name = "windows_i686_gnu"
version = "0.42.0"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "fbeae19f6716841636c28d695375df17562ca208b2b7d0dc47635a50ae6c5de7"
[[package]]
name = "windows_i686_msvc"
version = "0.36.1"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "e2e7917148b2812d1eeafaeb22a97e4813dfa60a3f8f78ebe204bcc88f12f024"
[[package]]
name = "windows_i686_msvc"
version = "0.42.0"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "84c12f65daa39dd2babe6e442988fc329d6243fdce47d7d2d155b8d874862246"
[[package]]
name = "windows_x86_64_gnu"
version = "0.36.1"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "4dcd171b8776c41b97521e5da127a2d86ad280114807d0b2ab1e462bc764d9e1"
[[package]]
name = "windows_x86_64_gnu"
version = "0.42.0"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "bf7b1b21b5362cbc318f686150e5bcea75ecedc74dd157d874d754a2ca44b0ed"
[[package]]
name = "windows_x86_64_gnullvm"
version = "0.42.0"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "09d525d2ba30eeb3297665bd434a54297e4170c7f1a44cad4ef58095b4cd2028"
[[package]]
name = "windows_x86_64_msvc"
version = "0.36.1"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "c811ca4a8c853ef420abd8592ba53ddbbac90410fab6903b3e79972a631f7680"
[[package]]
name = "windows_x86_64_msvc"
version = "0.42.0"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "f40009d85759725a34da6d89a94e63d7bdc50a862acf0dbc7c8e488f1edcb6f5"

View File

@@ -1,14 +1,18 @@
[package]
name = "upkr"
version = "0.2.0"
version = "0.2.2"
edition = "2021"
[profile.release]
strip = "debuginfo"
[features]
terminal = ["crossterm", "pbr"]
[dependencies]
cdivsufsort = "2"
lexopt = "0.2.1"
anyhow = "1"
thiserror = "1.0.36"
pbr = "1"
pbr = { version = "1", optional = true }
crossterm = { version = "0.25.0", default-features = false, optional = true }

View File

@@ -20,20 +20,35 @@ is both about twice as fast and smaller than the Shrinkler unpacker.
The release includes a reference c unpacker, as well as some optimized asm unpackers (arm and riscv). The unpckers in
c_unpacker and asm_unpackers unpack the default upkr compressed format. The z80_unpacker
is based on some variations to the compressed format. (Use `upkr --z80` to select those variations.)
An optimized x86 (DOS) unpacker is currently being worked on out of tree.
The 16 bit dos unpacker also uses some variations. (`upkr --x86`)
### More unpackers outside this repository
* [Atari Lynx](https://github.com/42Bastian/new_bll/blob/master/demos/depacker/unupkr.asm)
* [Atari Jaguar](https://github.com/42Bastian/new_bjl/blob/main/exp/depacker/unupkr.js)
* [8080, R800](https://github.com/ivagorRetrocomp/DeUpkr)
## Usage
```
upkr [-l level(0-9)] [config options] <infile> [<outfile>]
upkr -u [config options] <infile> [<outfile>]
upkr --heatmap [config options] <infile> [<outfile>]
upkr --margin [config options] <infile>
-l, --level N compression level 0-9
-0, ..., -9 short form for setting compression level
-u, --unpack unpack infile
-d, --decompress decompress infile
--heatmap calculate heatmap from compressed file
--raw-cost report raw cost of literals in heatmap
(the cost of literals is spread across all matches
that reference the literal by default.)
--hexdump print heatmap as colored hexdump
--margin calculate margin for overlapped unpacking of a packed file
When no infile is given, or the infile is '-', read from stdin.
When no outfile is given and reading from stdin, or when outfile is '-', write to stdout.
Config presets for specific unpackers:
--z80 --big-endian-bitstream --invert-bit-encoding --simplified-prob-update -9
--x86 --bitstream --invert-is-match-bit --invert-continue-value-bit --invert-new-offset-bit
@@ -56,3 +71,13 @@ Config options to tailor output to specific optimized unpackers:
--max-offset N
--max-length N
```
## Heatmap
By default, the `--heatmap` flag writes out the heatmap data as a binary file. The heatmap file is
the same size as the unpacked data. Each byte can be interpreted like this:
```
is_literal = byte & 1; // whether the byte was encoded as a literal (as opposed to a match)
size_in_bits = 2.0 ** (((byte >> 1) - 64) / 8.0); // the size this byte takes up in the compressed data
```

View File

@@ -17,7 +17,7 @@
upkr_unpack:
push { r3-r11, lr }
mov r2, #384
mov r2, #384
mov r3, #128
.Lclear:
subs r2, r2, #1
@@ -29,7 +29,7 @@ upkr_unpack:
bl upkr_decode_bit
bcc .Ldata
.Lmatch:
mov r5, #256
mov r5, #256
rsbs r6, r4, #0
blcc upkr_decode_bit
bcc .Lskip_offset
@@ -39,7 +39,7 @@ upkr_unpack:
popeq { r3-r11, pc }
.Lskip_offset:
mov r5, #256+64
mov r5, #256+64
bl upkr_decode_length
.Lcopy_loop:
ldrb r5, [r0, r3]
@@ -55,46 +55,46 @@ upkr_unpack:
.Ldata_loop:
bl upkr_decode_bit
adc r5, r5, r5
movs r4, r5, lsr #8
movs r4, r5, lsr #8
beq .Ldata_loop
b .Lstore
b .Lstore
.type upkr_decode_length, %function
upkr_decode_length:
mov r12, lr
mov r12, lr
mov r4, #0
mvn r6, #0
mvn r6, #0
.Lbit_loop:
bl upkr_decode_bit_inc
addcc r4, r4, r6
movcc pc, r12
addcc r4, r4, r6
movcc pc, r12
bl upkr_decode_bit_inc
addcs r4, r4, r6
mov r6, r6, lsl #1
mov r6, r6, lsl #1
b .Lbit_loop
.type upkr_decode_bit, %function
upkr_decode_bit_inc:
add r5, r5, #1
add r5, r5, #1
upkr_decode_bit:
cmp r2, #4096
ldrltb r8, [r1], #1
orrlt r2, r8, r2, lsl#8
blt upkr_decode_bit
cmp r2, #4096
ldrltb r8, [r1], #1
orrlt r2, r8, r2, lsl#8
blt upkr_decode_bit
ldrb r8, [sp, -r5]
and r9, r2, #255
add r9, r9, #1
cmp r8, r9
rsbcs r8, r8, #256
mvn r9, r2, lsr#8
addcs r9, r9, #1
mla r2, r8, r9, r2
add r9, r8, #8
sub r8, r8, r9, lsr#4
rsbcs r8, r8, #256
strb r8, [sp, -r5]
mov pc, r14
and r9, r2, #255
add r9, r9, #1
cmp r8, r9
rsbcs r8, r8, #256
mvn r9, r2, lsr#8
addcs r9, r9, #1
mla r2, r8, r9, r2
add r9, r8, #8
sub r8, r8, r9, lsr#4
rsbcs r8, r8, #256
strb r8, [sp, -r5]
mov pc, r14

2
c_library/.gitignore vendored Normal file
View File

@@ -0,0 +1,2 @@
/target/
/upkr

127
c_library/Cargo.lock generated Normal file
View File

@@ -0,0 +1,127 @@
# This file is automatically @generated by Cargo.
# It is not intended for manual editing.
version = 3
[[package]]
name = "anyhow"
version = "1.0.69"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "224afbd727c3d6e4b90103ece64b8d1b67fbb1973b1046c2281eed3f3803f800"
[[package]]
name = "autocfg"
version = "1.1.0"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "d468802bab17cbc0cc575e9b053f41e72aa36bfa6b7f55e3529ffa43161b97fa"
[[package]]
name = "cc"
version = "1.0.79"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "50d30906286121d95be3d479533b458f87493b30a4b5f79a607db8f5d11aa91f"
[[package]]
name = "cdivsufsort"
version = "2.0.0"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "edefce019197609da416762da75bb000bbd2224b2d89a7e722c2296cbff79b8c"
dependencies = [
"cc",
"sacabase",
]
[[package]]
name = "lexopt"
version = "0.2.1"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "478ee9e62aaeaf5b140bd4138753d1f109765488581444218d3ddda43234f3e8"
[[package]]
name = "num-traits"
version = "0.2.15"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "578ede34cf02f8924ab9447f50c28075b4d3e5b269972345e7e0372b38c6cdcd"
dependencies = [
"autocfg",
]
[[package]]
name = "proc-macro2"
version = "1.0.51"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "5d727cae5b39d21da60fa540906919ad737832fe0b1c165da3a34d6548c849d6"
dependencies = [
"unicode-ident",
]
[[package]]
name = "quote"
version = "1.0.23"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "8856d8364d252a14d474036ea1358d63c9e6965c8e5c1885c18f73d70bff9c7b"
dependencies = [
"proc-macro2",
]
[[package]]
name = "sacabase"
version = "2.0.0"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "9883fc3d6ce3d78bb54d908602f8bc1f7b5f983afe601dabe083009d86267a84"
dependencies = [
"num-traits",
]
[[package]]
name = "syn"
version = "1.0.109"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "72b64191b275b66ffe2469e8af2c1cfe3bafa67b529ead792a6d0160888b4237"
dependencies = [
"proc-macro2",
"quote",
"unicode-ident",
]
[[package]]
name = "thiserror"
version = "1.0.39"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "a5ab016db510546d856297882807df8da66a16fb8c4101cb8b30054b0d5b2d9c"
dependencies = [
"thiserror-impl",
]
[[package]]
name = "thiserror-impl"
version = "1.0.39"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "5420d42e90af0c38c3290abcca25b9b3bdf379fc9f55c528f53a269d9c9a267e"
dependencies = [
"proc-macro2",
"quote",
"syn",
]
[[package]]
name = "unicode-ident"
version = "1.0.8"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "e5464a87b239f13a63a501f2701565754bae92d243d4bb7eb12f6d57d2269bf4"
[[package]]
name = "upkr"
version = "0.2.1"
dependencies = [
"anyhow",
"cdivsufsort",
"lexopt",
"thiserror",
]
[[package]]
name = "upkr_c"
version = "0.0.1"
dependencies = [
"upkr",
]

17
c_library/Cargo.toml Normal file
View File

@@ -0,0 +1,17 @@
[package]
name = "upkr_c"
version = "0.0.1"
edition = "2021"
[lib]
name = "upkr"
crate-type = ["staticlib"]
[profile.release]
opt-level = "s"
strip = "debuginfo"
lto = true
panic = "abort"
[dependencies]
upkr = { path="..", default-features=false }

8
c_library/Makefile Normal file
View File

@@ -0,0 +1,8 @@
upkr: upkr.c upkr.h target/release/libupkr.a
gcc -O2 -Ltarget/release -o upkr upkr.c -lupkr -lm
strip upkr
target/release/libupkr.a: cargo
cargo build --release
.PHONY: cargo

11
c_library/Readme.md Normal file
View File

@@ -0,0 +1,11 @@
This is a simple example of compiling upkr to a library that can be linked in a
c program. It consists of a small rust crate which implements the c api and
compiles to a static library and a matching c header file. As is, the rust
crate offers two simple functions to compress/uncompress data with the default
upkr config.
The provided makefile will only work on linux. Building the example upkr.c on
other platforms is left as an exercise for the reader ;)
On Windows you might have to make sure to install and use the correct rust
toolchain version (mingw vs. msvc) to match your c compiler.

42
c_library/src/lib.rs Normal file
View File

@@ -0,0 +1,42 @@
use std::ffi::c_int;
// the upkr config to use, this can be modified to use other configs
fn config() -> upkr::Config {
upkr::Config::default()
}
#[no_mangle]
pub extern "C" fn upkr_compress(
output_buffer: *mut u8,
output_buffer_size: usize,
input_buffer: *const u8,
input_size: usize,
compression_level: c_int,
) -> usize {
let output_buffer = unsafe { std::slice::from_raw_parts_mut(output_buffer, output_buffer_size) };
let input_buffer = unsafe { std::slice::from_raw_parts(input_buffer, input_size) };
let packed_data = upkr::pack(input_buffer, compression_level.max(0).min(9) as u8, &config(), None);
let copy_size = packed_data.len().min(output_buffer.len());
output_buffer[..copy_size].copy_from_slice(&packed_data[..copy_size]);
packed_data.len()
}
#[no_mangle]
pub extern "C" fn upkr_uncompress(output_buffer: *mut u8, output_buffer_size: usize, input_buffer: *const u8, input_size: usize) -> isize {
let output_buffer = unsafe { std::slice::from_raw_parts_mut(output_buffer, output_buffer_size)};
let input_buffer = unsafe { std::slice::from_raw_parts(input_buffer, input_size)};
match upkr::unpack(input_buffer, &config(), output_buffer.len()) {
Ok(unpacked_data) => {
output_buffer[..unpacked_data.len()].copy_from_slice(&unpacked_data);
unpacked_data.len() as isize
}
Err(upkr::UnpackError::OverSize { size, .. }) => size as isize,
Err(other) => {
eprintln!("[upkr] compressed data corrupt: {}", other);
-1
}
}
}

99
c_library/upkr.c Normal file
View File

@@ -0,0 +1,99 @@
#include "upkr.h"
#include <stdio.h>
#include <string.h>
#include <stdlib.h>
int main(int argc, char** argv) {
if(argc < 2) {
fprintf(stdout, "Usage:\n upkr [compress] [-0 .. -9] <file> [<out-file>]\n upkr [uncompress] <file> [<out-file>]\n");
return 1;
}
int argi = 1;
int uncompress = 0;
int compression_level = 4;
if(strcmp(argv[argi], "compress") == 0) {
++argi;
} else if(strcmp(argv[argi], "uncompress") == 0) {
uncompress = 1;
++argi;
}
if(argi < argc && argv[argi][0] == '-') {
compression_level = atoi(argv[argi] + 1);
++argi;
}
if(argi == argc) {
fprintf(stdout, "intput filename missing\n");
return 1;
}
const char* input_name = argv[argi++];
char* output_name;
if(argi < argc) {
output_name = argv[argi];
} else {
output_name = malloc(strlen(input_name) + 5);
strcpy(output_name, input_name);
strcat(output_name, uncompress ? ".unp" : ".upk");
}
FILE* file = fopen(input_name, "rb");
if(file == 0) {
fprintf(stdout, "failed to open input file '%s'\n", file);
return 1;
}
fseek(file, 0, SEEK_END);
long input_size = ftell(file);
rewind(file);
char* input_buffer = (char*)malloc(input_size);
long offset = 0;
while(offset < input_size) {
long read_size = fread(input_buffer + offset, 1, input_size - offset, file);
if(read_size <= 0) {
fprintf(stdout, "error reading input file\n");
return 1;
}
offset += read_size;
}
fclose(file);
long output_buffer_size = input_size * 8;
long output_size;
char* output_buffer = (char*)malloc(output_buffer_size);
for(;;) {
if(uncompress) {
output_size = upkr_uncompress(output_buffer, output_buffer_size, input_buffer, input_size);
} else {
output_size = upkr_compress(output_buffer, output_buffer_size, input_buffer, input_size, compression_level);
}
if(output_size < 0) {
return 1;
}
if(output_size <= output_buffer_size) {
break;
}
output_buffer = (char*)realloc(output_buffer, output_size);
output_buffer_size = output_size;
}
file = fopen(output_name, "wb");
if(file == 0) {
fprintf(stdout, "failed to open output file '%s'\n", output_name);
return 1;
}
offset = 0;
while(offset < output_size) {
long written_size = fwrite(output_buffer + offset, 1, output_size - offset, file);
if(written_size <= 0) {
fprintf(stdout, "error writing output file\n");
return 1;
}
offset += written_size;
}
fclose(file);
return 0;
}

25
c_library/upkr.h Normal file
View File

@@ -0,0 +1,25 @@
#ifndef UPKR_H_INCLUDED
#include <stddef.h>
#ifdef __cplusplus
extern "C" {
#endif
// input_buffer/input_size: input data to compress
// output_buffer/output_buffer_size: buffer to compress into
// compression_level: 0-9
// returns the size of the compressed data, even if it didn't fit into the output buffer
size_t upkr_compress(void* output_buffer, size_t output_buffer_size, void* input_buffer, size_t input_size, int compression_level);
// input_buffer/input_size: compressed data
// output_buffer/output_buffer_size: buffer to uncompress into
// return value:
// >= 0 : size of uncompressed data, even if it didn't fit into the output buffer
// < 0 : input data corrupt, unable to decompress
ptrdiff_t upkr_uncompress(void* output_buffer, size_t output_buffer_size, void* input_buffer, size_t input_size);
#ifdef __cplusplus
}
#endif
#endif

13
dos_unpacker/readme.txt Normal file
View File

@@ -0,0 +1,13 @@
16 bit DOS executable stubs
---------------------------
by pestis and TomCat
unpack_x86_16_DOS.asm:
maximum compatibility, relocates unpacked code to normal start address
unpack_x86_16_DOS_no_relocation.asm:
saves some bytes by not relocating, unpacked code needs to be assembled to
start at 0x3FFE
unpack_x86_16_DOS_no_repeated_offset.asm:
removes support for repeated offsets, potentially at the cost of some compression ratio.
most likely only a win in very narrow circumstances around the 1kb mark

View File

@@ -0,0 +1,160 @@
; Contributions from pestis, TomCat and exoticorn
;
; This is the 16-bit DOS x86 decompression stub for upkr, which is designed for
; maximum compatibility: it relocates the compressed data so it can be
; decompressed starting at the normal .COM starting address. In other words,
; many of the already existing .COM files should be compressable using this
; stub.
;
; How to use:
; 1) Pack your intro using upkr into data.bin with the --x86 command line
; argument:
;
; $ upkr --x86 intro.com data.bin
;
; 2) Compile this .asm file using nasm (or any compatible assembler):
;
; $ nasm unpack_x86_16_DOS.asm -fbin -o intropck.com
;
; The packed size of the intro+stub is limited by max_len (see below) bytes.
;
; In specific cases, the unpacker stub can be further optimized to save a byte
; or two:
; 1) You can remove CLC before RET, if you don't mind carry being set upon
; program entry
; 2) You can also move PUSHA before PUSH SI and put POPA as the first
; operation of the compressed code.
max_len equ 16384
prog_start equ (0x100+max_len+510+relocation-upkr_unpack)
probs equ (((prog_start+max_len+510)+255)/256)*256
org 0x100
; This is will be loaded at 0x100, but relocates the code and data to prog_start
relocation:
push si ; si = 0x100 at DOS start, so save it for later ret
pusha ; pusha to recall all registers before starting intro
push si ; for pop di to start writing the output
mov di, prog_start ; the depacker & data are relocated from 0x100 to prog_start
mov ch, max_len/512
rep movsw
jmp si ; jump to relocated upkr_unpack
; upkr_unpack unpacks the code to 0x100 and runs it when done.
upkr_unpack:
xchg ax, bp ; position in input bitstream (bp) = 0
cwd ; upkr_state (dx) = 0;
xchg ax, cx ; cx = 0x9XX
mov al, 128 ; for(int i = 0; i < sizeof(upkr_probs); ++i) upkr_probs[i] = 128;
rep stosb
pop di ; u8* write_ptr = (u8*)destination;
.mainloop:
mov bx, probs
call upkr_decode_bit
jc .else ; if(upkr_decode_bit(0)) {
mov bh, (probs+256)/256
jcxz .skip_call
call upkr_decode_bit
jc .skipoffset
.skip_call:
stc
call upkr_decode_number ; offset = upkr_decode_length(258) - 1;
loop .notdone ; if(offset == 0)
popa
clc
ret
.notdone:
mov si, di
.sub:
dec si
loop .sub
.skipoffset:
mov bl, 128 ; int length = upkr_decode_length(384);
call upkr_decode_number
rep movsb ; *write_ptr = write_ptr[-offset];
jmp .mainloop
.byteloop:
call upkr_decode_bit ; int bit = upkr_decode_bit(byte);
.else:
adc bl, bl ; byte = (byte << 1) + bit;
jnc .byteloop
xchg ax, bx
stosb
inc si
mov cl, 1
jmp .mainloop ; prev_was_match = 0;
; upkr_decode_bit decodes one bit from the rANS entropy encoded bit stream.
; parameters:
; bx = memory address of the context probability
; dx = decoder state
; bp = bit position in input stream
; returns:
; dx = new decoder state
; bp = new bit position in input stream
; carry = bit
; trashes ax
upkr_load_bit:
bt [compressed_data-relocation+prog_start], bp
inc bp
adc dx, dx
upkr_decode_bit:
inc dx ; inc dx, dec dx is used to test the top (sign) bit of dx
dec dx
jns upkr_load_bit
movzx ax, byte [bx] ; u16 prob = upkr_probs[context_index]
neg byte [bx]
push ax ; save prob, tmp = prob
cmp dl, al ; int bit = (upkr_state & 255) < prob ? 1 : 0; (carry = bit)
pushf ; save bit flags
jc .bit ; (skip if bit)
xchg [bx], al ; tmp = 256 - tmp;
.bit:
shr byte [bx], 4 ; upkr_probs[context_index] = tmp + (256 - tmp + 8) >> 4;
adc [bx], al
mul dh ; upkr_state = tmp * (upkr_state >> 8) + (upkr_state & 255);
mov dh, 0
add dx, ax
popf
pop ax
jc .bit2 ; (skip if bit)
neg byte [bx] ; tmp = 256 - tmp;
sub dx, ax ; upkr_state -= prob; note that this will also leave carry always unset, which is what we want
.bit2:
ret ; return the bit in carry
; upkr_decode_number loads a variable length encoded number (up to 16 bits) from
; the compressed stream. Only numbers 1..65535 can be encoded. If the encoded
; number has 4 bits and is 1ABC, it is encoded using a kind of an "interleaved
; elias code": 0A0B0C1. The 1 in the end implies that no more bits are coming.
; parameters:
; cx = must be 0
; bx = memory address of the context probability
; dx = decoder state
; bp = bit position in input stream
; carry = must be 1
; returns:
; cx = length
; dx = new decoder state
; bp = new bit position in input stream
; carry = 1
; trashes bl, ax
upkr_decode_number_loop:
inc bx
call upkr_decode_bit
upkr_decode_number:
rcr cx, 1
inc bx
call upkr_decode_bit
jnc upkr_decode_number_loop ; 0 = there's more bits coming, 1 = no more bits
.loop2:
rcr cx, 1
jnc .loop2
ret
compressed_data:
incbin "data.bin"

View File

@@ -0,0 +1,151 @@
; Contributions from pestis, TomCat and exoticorn
;
; This is the 16-bit DOS x86 decompression stub for upkr, which decompresses the
; code starting at address 0x3FFE (or whatever is defined by the entrypoint
; below). Thus, the packed code needs to be assembled with org 0x3FFE to work.
;
; How to use:
; 1) Put POPA as the first instruction of your compiled code and use org
; 0x3FFE
; 2) Pack your intro using upkr into data.bin with the --x86 command line
; argument:
;
; $ upkr --x86 intro.com data.bin
;
; 2) Compile this .asm file using nasm (or any compatible assembler) e.g.
;
; $ nasm unpack_x86_16_DOS_no_relocation.asm -fbin -o intropck.com
;
; In specific cases, the unpacker stub can be further optimized to save a byte
; or two:
; 1) If your stub+compressed code is 2k or smaller, you can save 1 byte by
; putting probs at 0x900 and initializing DI with SALC; XCHG AX, DI instead
; of MOV DI, probs
; 2) If you remove the PUSHA (and POPA in the compressed code), then you can
; assume the registers as follows: AX = 0x00XX, BX = probs + 0x1XX, CX = 0
; DX = (trash), SI = DI = right after your program, SP = as it was when the
; program started, flags = carry set
;
; Note that even with the PUSHA / POPA, carry will be set (!) unlike normal dos
; program.
entry equ 0x3FFE
probs equ entry - 0x1FE ; must be aligned to 256
org 0x100
; This is will be loaded at 0x100, but relocates the code and data to prog_start
upkr_unpack:
pusha
xchg ax, bp ; position in bitstream = 0
cwd ; upkr_state = 0;
mov di, probs
mov ax, 0x8080 ; for(int i = 0; i < sizeof(upkr_probs); ++i) upkr_probs[i] = 128;
rep stosw
push di
.mainloop:
mov bx, probs
call upkr_decode_bit
jc .else ; if(upkr_decode_bit(0)) {
mov bh, (probs+256)/256
jcxz .skip_call ; if(prev_was_match || upkr_decode_bit(257)) {
call upkr_decode_bit
jc .skipoffset
.skip_call:
stc
call upkr_decode_number ; offset = upkr_decode_number(258) - 1;
mov si, di
loop .sub ; if(offset == 0)
ret
.sub:
dec si
loop .sub
.skipoffset:
mov bl, 128 ; int length = upkr_decode_number(384);
call upkr_decode_number
rep movsb ; *write_ptr = write_ptr[-offset];
jmp .mainloop
.byteloop:
call upkr_decode_bit ; int bit = upkr_decode_bit(byte);
.else:
adc bl, bl ; byte = (byte << 1) + bit;
jnc .byteloop
xchg ax, bx
stosb
inc si
mov cl, 1
jmp .mainloop ; prev_was_match = 0;
; upkr_decode_bit decodes one bit from the rANS entropy encoded bit stream.
; parameters:
; bx = memory address of the context probability
; dx = decoder state
; bp = bit position in input stream
; returns:
; dx = new decoder state
; bp = new bit position in input stream
; carry = bit
; trashes ax
upkr_load_bit:
bt [compressed_data], bp
inc bp
adc dx, dx
upkr_decode_bit:
inc dx
dec dx ; inc dx, dec dx is used to test the top (sign) bit of dx
jns upkr_load_bit
movzx ax, byte [bx] ; u16 prob = upkr_probs[context_index]
neg byte [bx]
push ax ; save prob, tmp = prob
cmp dl, al ; int bit = (upkr_state & 255) < prob ? 1 : 0; (carry = bit)
pushf ; save bit flags
jc .bit ; (skip if bit)
xchg [bx], al ; tmp = 256 - tmp;
.bit:
shr byte [bx], 4 ; upkr_probs[context_index] = tmp + (256 - tmp + 8) >> 4;
adc [bx], al
mul dh ; upkr_state = tmp * (upkr_state >> 8) + (upkr_state & 255);
mov dh, 0
add dx, ax
popf
pop ax
jc .bit2 ; (skip if bit)
neg byte [bx] ; tmp = 256 - tmp;
sub dx, ax ; upkr_state -= prob; note that this will also leave carry always unset, which is what we want
.bit2:
ret ; flags = bit
; upkr_decode_number loads a variable length encoded number (up to 16 bits) from
; the compressed stream. Only numbers 1..65535 can be encoded. If the encoded
; number has 4 bits and is 1ABC, it is encoded using a kind of an "interleaved
; elias code": 0A0B0C1. The 1 in the end implies that no more bits are coming.
; parameters:
; cx = must be 0
; bx = memory address of the context probability
; dx = decoder state
; bp = bit position in input stream
; carry = must be 1
; returns:
; cx = length
; dx = new decoder state
; bp = new bit position in input stream
; carry = 1
; trashes bl, ax
upkr_decode_number_loop:
inc bx
call upkr_decode_bit
upkr_decode_number:
rcr cx, 1
inc bx
call upkr_decode_bit
jnc upkr_decode_number_loop ; while(upkr_decode_bit(context_index)) {
.loop2:
rcr cx, 1
jnc .loop2
ret
compressed_data:
incbin "data.bin"

View File

@@ -0,0 +1,154 @@
; Contributions from pestis, TomCat and exoticorn
;
; This is the 16-bit DOS x86 decompression stub for upkr, which is designed for
; the --no-repeated-offsets option of upkr. The decompression stub is slightly
; smaller, but the compressed data might be bigger, so you have to test if
; --no-repeated-offsets pays off in the end. This stub relocates the compressed
; data so it can be decompressed starting at the normal .COM starting address.
;
; How to use:
; 1) Pack your intro using upkr into data.bin with the --x86b command line
; argument: (notice the --x86b, not --x86!)
;
; $ upkr --x86b intro.com data.bin
;
; 2) Compile this .asm file using nasm (or any compatible assembler):
;
; $ nasm unpack_x86_16_DOS_no_repeated_offsets.asm -fbin -o intropck.com
;
; The packed size of the intro+stub is limited by max_len (see below) bytes.
;
; In specific cases, the unpacker stub can be further optimized to save a byte
; or two:
; 1) You can remove CLC before RET, if you don't mind carry being set upon
; program entry
; 2) You can also move PUSHA before PUSH SI and put POPA as the first
; operation of the compressed code.
max_len equ 16384
prog_start equ (0x100+max_len+510+relocation-upkr_unpack)
probs equ (((prog_start+max_len+510)+255)/256)*256
org 0x100
; This is will be loaded at 0x100, but relocates the code and data to prog_start
relocation:
push si ; si = 0x100 at DOS start, so save it for later ret
pusha ; pusha to recall all registers before starting intro
push si ; for pop di to start writing the output
mov di, prog_start ; the depacker & data are relocated from 0x100 to prog_start
mov ch, max_len/512
rep movsw
jmp si ; jump to relocated upkr_unpack
; upkr_unpack unpacks the code to 0x100 and runs it when done.
upkr_unpack:
xchg ax, bp ; position in bitstream = 0
cwd ; upkr_state = 0;
xchg cx, ax ; cx > 0x0200
mov al, 128 ; for(int i = 0; i < sizeof(upkr_probs); ++i) upkr_probs[i] = 128;
rep stosb
pop di ; u8* write_ptr = (u8*)destination;
.mainloop:
mov bx, probs
call upkr_decode_bit
jnc .else ; if(upkr_decode_bit(0)) {
inc bh
call upkr_decode_number ; offset = upkr_decode_number(258) - 1;
loop .notdone ; if(offset == 0)
popa
clc
ret
.notdone:
mov si, di
.sub:
dec si
loop .sub
mov bl, 128 ; int length = upkr_decode_number(384);
call upkr_decode_number
rep movsb ; *write_ptr = write_ptr[-offset];
jmp .mainloop
.else:
inc bx
.byteloop:
call upkr_decode_bit ; int bit = upkr_decode_bit(byte);
adc bl, bl ; byte = (byte << 1) + bit;
jnc .byteloop
xchg ax, bx
stosb
jmp .mainloop ; prev_was_match = 0;
; upkr_decode_bit decodes one bit from the rANS entropy encoded bit stream.
; parameters:
; bx = memory address of the context probability
; dx = decoder state
; bp = bit position in input stream
; returns:
; dx = new decoder state
; bp = new bit position in input stream
; carry = bit
; trashes ax
upkr_load_bit:
bt [compressed_data-relocation+prog_start], bp
inc bp
adc dx, dx
upkr_decode_bit:
inc dx
dec dx ; or whatever other test for the top bit there is
jns upkr_load_bit
movzx ax, byte [bx] ; u16 prob = upkr_probs[context_index]
neg byte [bx]
push ax ; save prob, tmp = prob
cmp dl, al ; int bit = (upkr_state & 255) < prob ? 1 : 0; (carry = bit)
pushf ; save bit flags
jc .bit ; (skip if bit)
xchg [bx], al ; tmp = 256 - tmp;
.bit:
shr byte [bx], 4 ; upkr_probs[context_index] = tmp + (256 - tmp + 8) >> 4;
adc [bx], al ; upkr_probs[context_index] = tmp;
mul dh ; upkr_state = tmp * (upkr_state >> 8) + (upkr_state & 255);
mov dh, 0
add dx, ax
popf
pop ax
jc .bit2 ; (skip if bit)
neg byte [bx] ; tmp = 256 - tmp;
sub dx, ax ; upkr_state -= prob; note that this will also leave carry always unset, which is what we want
.bit2:
ret ; flags = bit
; upkr_decode_number loads a variable length encoded number (up to 16 bits) from
; the compressed stream. Only numbers 1..65535 can be encoded. If the encoded
; number has 4 bits and is 1ABC, it is encoded using a kind of an "interleaved
; elias code": 0A0B0C1. The 1 in the end implies that no more bits are coming.
; parameters:
; cx = must be 0
; bx = memory address of the context probability
; dx = decoder state
; bp = bit position in input stream
; carry = must be 1
; returns:
; cx = length
; dx = new decoder state
; bp = new bit position in input stream
; carry = 1
; trashes bl, ax
upkr_decode_number_loop:
inc bx
call upkr_decode_bit
upkr_decode_number:
rcr cx, 1
inc bx
call upkr_decode_bit
jnc upkr_decode_number_loop ; 0 = there's more bits coming, 1 = no more bits
.loop2:
rcr cx, 1
jnc .loop2
ret
compressed_data:
incbin "data.bin"

View File

@@ -23,13 +23,15 @@ upkr-windows-$(VERSION).zip: upkr-windows/upkr.exe PHONY
zip -r -9 $@ upkr-windows
upkr-linux/upkr:
cargo build --target x86_64-unknown-linux-musl --release
cargo build --target x86_64-unknown-linux-musl --release -F terminal
mkdir -p upkr-linux
cp ../target/x86_64-unknown-linux-musl/release/upkr upkr-linux/
strip upkr-linux/upkr
upkr-windows/upkr.exe:
cargo build --target x86_64-pc-windows-gnu --release
cargo build --target x86_64-pc-windows-gnu --release -F terminal
mkdir -p upkr-windows
cp ../target/x86_64-pc-windows-gnu/release/upkr.exe upkr-windows/
x86_64-w64-mingw32-strip upkr-windows/upkr.exe
PHONY:

208
src/heatmap.rs Normal file
View File

@@ -0,0 +1,208 @@
/// Heatmap information about a compressed block of data.
///
/// For each byte in the uncompressed data, the heatmap provides two pieces of intormation:
/// 1. whether this byte was encoded as a literal or as part of a match
/// 2. how many (fractional) bits where spend on encoding this byte
///
/// For the sake of the heatmap, the cost of literals are spread out across all matches
/// that reference the literal.
///
/// If the `terminal` feature is enabled, there is a function to write out the
/// heatmap as a colored hexdump.
pub struct Heatmap {
data: Vec<u8>,
cost: Vec<f32>,
raw_cost: Vec<f32>,
literal_index: Vec<usize>,
}
impl Heatmap {
pub(crate) fn new() -> Heatmap {
Heatmap {
data: Vec::new(),
cost: Vec::new(),
raw_cost: Vec::new(),
literal_index: Vec::new(),
}
}
pub(crate) fn add_literal(&mut self, byte: u8, cost: f32) {
self.data.push(byte);
self.cost.push(cost);
self.literal_index.push(self.literal_index.len());
}
pub(crate) fn add_match(&mut self, offset: usize, length: usize, mut cost: f32) {
cost /= length as f32;
for _ in 0..length {
self.data.push(self.data[self.data.len() - offset]);
self.literal_index
.push(self.literal_index[self.literal_index.len() - offset]);
self.cost.push(cost);
}
}
pub(crate) fn finish(&mut self) {
self.raw_cost = self.cost.clone();
let mut ref_count = vec![0usize; self.literal_index.len()];
for &index in &self.literal_index {
ref_count[index] += 1;
}
let mut shifted = vec![];
for (&index, &cost) in self.literal_index.iter().zip(self.cost.iter()) {
let delta = (self.cost[index] - cost) / ref_count[index] as f32;
shifted.push(delta);
shifted[index] -= delta;
}
for (cost, delta) in self.cost.iter_mut().zip(shifted.into_iter()) {
*cost += delta;
}
}
/// Reverses the heatmap
pub fn reverse(&mut self) {
self.data.reverse();
self.cost.reverse();
self.literal_index.reverse();
for index in self.literal_index.iter_mut() {
*index = self.data.len() - *index;
}
}
/// The number of (uncompressed) bytes of data in this heatmap
pub fn len(&self) -> usize {
self.cost.len()
}
/// Returns whether the byte at `index` was encoded as a literal
pub fn is_literal(&self, index: usize) -> bool {
self.literal_index[index] == index
}
/// Returns the cost of encoding the byte at `index` in (fractional) bits.
/// The cost of literal bytes is spread across the matches that reference it.
/// See `raw_cost` for the raw encoding cost of each byte.
pub fn cost(&self, index: usize) -> f32 {
self.cost[index]
}
/// Returns the raw cost of encoding the byte at `index` in (fractional) bits
pub fn raw_cost(&self, index: usize) -> f32 {
self.raw_cost[index]
}
/// Returns the uncompressed data byte at `index`
pub fn byte(&self, index: usize) -> u8 {
self.data[index]
}
#[cfg(feature = "crossterm")]
/// Print the heatmap as a colored hexdump
pub fn print_as_hex(&self) -> std::io::Result<()> {
self.print_as_hex_internal(false)
}
#[cfg(feature = "crossterm")]
/// Print the heatmap as a colored hexdump, based on `raw_cost`.
pub fn print_as_hex_raw_cost(&self) -> std::io::Result<()> {
self.print_as_hex_internal(true)
}
#[cfg(feature = "crossterm")]
fn print_as_hex_internal(&self, report_raw_cost: bool) -> std::io::Result<()> {
use crossterm::{
style::{Attribute, Color, Print, SetAttribute, SetBackgroundColor},
QueueableCommand,
};
use std::io::{stdout, Write};
fn set_color(
mut out: impl QueueableCommand,
heatmap: &Heatmap,
index: usize,
num_colors: u16,
report_raw_cost: bool,
) -> std::io::Result<()> {
let cost = if report_raw_cost {
heatmap.raw_cost(index)
} else {
heatmap.cost(index)
};
if num_colors < 256 {
let colors = [
Color::Red,
Color::Yellow,
Color::Green,
Color::Cyan,
Color::Blue,
Color::DarkBlue,
Color::Black,
];
let color_index = (3. - cost.log2())
.round()
.max(0.)
.min((colors.len() - 1) as f32) as usize;
out.queue(SetBackgroundColor(colors[color_index]))?;
} else {
let colors = [
196, 166, 136, 106, 76, 46, 41, 36, 31, 26, 21, 20, 19, 18, 17, 16,
];
let color_index = ((3. - cost.log2()) * 2.5)
.round()
.max(0.)
.min((colors.len() - 1) as f32) as usize;
out.queue(SetBackgroundColor(Color::AnsiValue(colors[color_index])))?;
}
out.queue(SetAttribute(if heatmap.is_literal(index) {
Attribute::Underlined
} else {
Attribute::NoUnderline
}))?;
Ok(())
}
let num_colors = crossterm::style::available_color_count();
let term_width = crossterm::terminal::size()?.0.min(120) as usize;
let bytes_per_row = (term_width - 8) / 4;
for row_start in (0..self.data.len()).step_by(bytes_per_row) {
let row_range = row_start..self.data.len().min(row_start + bytes_per_row);
let mut stdout = stdout();
stdout.queue(Print(&format!("{:04x} ", row_start)))?;
for i in row_range.clone() {
set_color(&mut stdout, self, i, num_colors, report_raw_cost)?;
stdout.queue(Print(&format!("{:02x} ", self.data[i])))?;
}
let num_spaces = 1 + (bytes_per_row - (row_range.end - row_range.start)) * 3;
let gap: String = std::iter::repeat(' ').take(num_spaces).collect();
stdout
.queue(SetAttribute(Attribute::Reset))?
.queue(Print(&gap))?;
for i in row_range.clone() {
set_color(&mut stdout, self, i, num_colors, report_raw_cost)?;
let byte = self.data[i];
if byte >= 32 && byte < 127 {
stdout.queue(Print(format!("{}", byte as char)))?;
} else {
stdout.queue(Print("."))?;
}
}
stdout
.queue(SetAttribute(Attribute::Reset))?
.queue(Print("\n"))?;
stdout.flush()?;
}
Ok(())
}
}

View File

@@ -1,31 +1,75 @@
#![deny(missing_docs)]
//! Compression and decompression of the upkr format and variants.
//!
//! Upkr is a compression format initially designed for the MicroW8 fantasy console,
//! with design goals being a competitive compression ratio, reasonable fast
//! decompression, low memory overhead and very small decompression code
//! when handoptimized in assembler. (An optimized DOS execuable decompressor is <140 bytes.)
mod context_state;
mod greedy_packer;
mod heatmap;
mod lz;
mod match_finder;
mod parsing_packer;
mod rans;
pub use lz::{calculate_margin, unpack, UnpackError};
pub use heatmap::Heatmap;
pub use lz::{calculate_margin, create_heatmap, unpack, UnpackError};
/// The type of a callback function to be given to the `pack` function.
///
/// It will be periodically called with the number of bytes of the input already processed.
pub type ProgressCallback<'a> = &'a mut dyn FnMut(usize);
/// A configuration of which compression format variation to use.
///
/// Use `Config::default()` for the standard upkr format.
///
/// Compression format variants exist to help with micro-optimizations in uncompression
/// code on specific platforms.
#[derive(Debug)]
pub struct Config {
/// Shift in bits from a bitstream into the rANS state, rather than whole bytes.
/// This decreases the size of the rNAS state to 16 bits which is very useful on
/// 8 bit platforms.
pub use_bitstream: bool,
/// The number of parity contexts (usually 1, 2 or 4). This can improve compression
/// on data that consists of regular groups of 2 or 4 bytes. One example is 32bit ARM
/// code, where each instruction is 4 bytes, so `parity_contexts = 4` improves compression
/// quite a bit. Defaults to `1`.
pub parity_contexts: usize,
/// Invert the encoding of bits in the rANS coder. `bit = state_lo >= prob` instead of
/// `bit = state_lo < prob`.
pub invert_bit_encoding: bool,
/// The boolean value which encodes a match. Defaults to `true`.
pub is_match_bit: bool,
/// The boolean value which encodes a new offset (rather than re-using the previous offset).
/// Defaults to `true`.
pub new_offset_bit: bool,
/// The boolean value which encodes that there are more bits comming for length/offset values.
/// Defaults to `true`.
pub continue_value_bit: bool,
/// Reverses the bits in the bitstream.
pub bitstream_is_big_endian: bool,
/// A slightly less accurate, but slightly simpler variation of the prob update in the
/// rANS coder, Used for the z80 uncompressor.
pub simplified_prob_update: bool,
/// Disables support for re-using the last offset in the compression format.
/// This might save a few bytes when working with very small data.
pub no_repeated_offsets: bool,
/// Standard upkr encodes the EOF marker in the offset. This encodes it in the match length
/// instead.
pub eof_in_length: bool,
/// The maximum match offset value to encode when compressing.
pub max_offset: usize,
/// The maximum match length value to encode when compressing.
pub max_length: usize,
}
@@ -53,7 +97,7 @@ impl Default for Config {
}
impl Config {
pub fn min_length(&self) -> usize {
fn min_length(&self) -> usize {
if self.eof_in_length {
2
} else {
@@ -62,6 +106,21 @@ impl Config {
}
}
/// Compresses the given data.
///
/// # Arguments
/// - `data`: The data to compress
/// - `level`: The compression level (0-9). Increasing the level by one roughly halves the
/// compression speed.
/// - `config`: The compression format variant to use.
/// - `progress_callback`: An optional callback which will periodically be called with
/// the number of bytes already processed.
///
/// # Example
/// ```rust
/// let compressed_data = upkr::pack(b"Hello, World! Yellow world!", 0, &upkr::Config::default(), None);
/// assert!(compressed_data.len() < 27);
/// ```
pub fn pack(
data: &[u8],
level: u8,
@@ -75,6 +134,9 @@ pub fn pack(
}
}
/// Estimate the exact (fractional) size of upkr compressed data.
///
/// Note that this currently does NOT work for the bitstream variant.
pub fn compressed_size(mut data: &[u8]) -> f32 {
let mut state = 0;
while state < 4096 {

View File

@@ -1,4 +1,5 @@
use crate::context_state::ContextState;
use crate::heatmap::Heatmap;
use crate::rans::{EntropyCoder, RansDecoder};
use crate::Config;
use thiserror::Error;
@@ -132,42 +133,91 @@ impl CoderState {
}
}
/// The error type for the uncompressing related functions
#[derive(Error, Debug)]
pub enum UnpackError {
/// a match offset pointing beyond the start of the unpacked data was encountered
#[error("match offset out of range: {offset} > {position}")]
OffsetOutOfRange { offset: usize, position: usize },
OffsetOutOfRange {
/// the match offset
offset: usize,
/// the current position in the uncompressed stream
position: usize,
},
/// The passed size limit was exceeded
#[error("Unpacked data over size limit: {size} > {limit}")]
OverSize { size: usize, limit: usize },
OverSize {
/// the size of the uncompressed data
size: usize,
/// the size limit passed into the function
limit: usize,
},
/// The end of the packed data was reached without an encoded EOF marker
#[error("Unexpected end of input data")]
UnexpectedEOF {
#[from]
/// the underlying EOF error in the rANS decoder
source: crate::rans::UnexpectedEOF,
},
/// An offset or length value was found that exceeded 32bit
#[error("Overflow while reading value")]
ValueOverflow,
}
/// Uncompress a piece of compressed data
///
/// Returns either the uncompressed data, or an `UnpackError`
///
/// # Parameters
///
/// - `packed_data`: the compressed data
/// - `config`: the exact compression format config used to compress the data
/// - `max_size`: the maximum size of uncompressed data to return. When this is exceeded,
/// `UnpackError::OverSize` is returned
pub fn unpack(
packed_data: &[u8],
config: &Config,
max_size: usize,
) -> Result<Vec<u8>, UnpackError> {
let mut result = vec![];
let _ = unpack_internal(Some(&mut result), packed_data, config, max_size)?;
let _ = unpack_internal(Some(&mut result), None, packed_data, config, max_size)?;
Ok(result)
}
/// Calculates the minimum margin when overlapping buffers.
///
/// Returns the minimum margin needed between the end of the compressed data and the
/// end of the uncompressed data when overlapping the two buffers to save on RAM.
pub fn calculate_margin(packed_data: &[u8], config: &Config) -> Result<isize, UnpackError> {
unpack_internal(None, packed_data, config, usize::MAX)
unpack_internal(None, None, packed_data, config, usize::MAX)
}
pub fn unpack_internal(
/// Calculates a `Heatmap` from compressed data.
///
/// # Parameters
///
/// - `packed_data`: the compressed data
/// - `config`: the exact compression format config used to compress the data
/// - `max_size`: the maximum size of the heatmap to return. When this is exceeded,
/// `UnpackError::OverSize` is returned
pub fn create_heatmap(
packed_data: &[u8],
config: &Config,
max_size: usize,
) -> Result<Heatmap, UnpackError> {
let mut heatmap = Heatmap::new();
let _ = unpack_internal(None, Some(&mut heatmap), packed_data, config, max_size)?;
Ok(heatmap)
}
fn unpack_internal(
mut result: Option<&mut Vec<u8>>,
mut heatmap: Option<&mut Heatmap>,
packed_data: &[u8],
config: &Config,
max_size: usize,
) -> Result<isize, UnpackError> {
let mut decoder = RansDecoder::new(packed_data, &config);
let mut decoder = RansDecoder::new(packed_data, &config)?;
let mut contexts = ContextState::new((1 + 255) * config.parity_contexts + 1 + 64 + 64, &config);
let mut offset = usize::MAX;
let mut position = 0usize;
@@ -198,6 +248,7 @@ pub fn unpack_internal(
}
loop {
let prev_decoder = decoder.clone();
margin = margin.max(position as isize - decoder.pos() as isize);
let literal_base = position % config.parity_contexts * 256;
if decoder.decode_with_context(&mut contexts.context_mut(literal_base))?
@@ -231,6 +282,9 @@ pub fn unpack_internal(
if offset > position {
return Err(UnpackError::OffsetOutOfRange { offset, position });
}
if let Some(ref mut heatmap) = heatmap {
heatmap.add_match(offset, length, decoder.cost(&prev_decoder));
}
if let Some(ref mut result) = result {
for _ in 0..length {
if result.len() < max_size {
@@ -251,6 +305,9 @@ pub fn unpack_internal(
context_index = (context_index << 1) | bit as usize;
byte |= (bit as u8) << i;
}
if let Some(ref mut heatmap) = heatmap {
heatmap.add_literal(byte, decoder.cost(&prev_decoder));
}
if let Some(ref mut result) = result {
if result.len() < max_size {
result.push(byte);
@@ -261,6 +318,10 @@ pub fn unpack_internal(
}
}
if let Some(heatmap) = heatmap {
heatmap.finish();
}
if position > max_size {
return Err(UnpackError::OverSize {
size: position,

View File

@@ -9,6 +9,10 @@ fn main() -> Result<()> {
let mut reverse = false;
let mut unpack = false;
let mut calculate_margin = false;
let mut create_heatmap = false;
let mut report_raw_cost = false;
#[allow(unused_mut)]
let mut do_hexdump = false;
let mut level = 2;
let mut infile: Option<PathBuf> = None;
let mut outfile: Option<PathBuf> = None;
@@ -56,8 +60,12 @@ fn main() -> Result<()> {
level = 9;
}
Short('u') | Long("unpack") => unpack = true,
Short('u') | Long("unpack") | Short('d') | Long("decompress") => unpack = true,
Long("margin") => calculate_margin = true,
Long("heatmap") => create_heatmap = true,
Long("raw-cost") => report_raw_cost = true,
#[cfg(feature = "crossterm")]
Long("hexdump") => do_hexdump = true,
Short('l') | Long("level") => level = parser.value()?.parse()?,
Short(n) if n.is_ascii_digit() => level = n as u8 - b'0',
Short('h') | Long("help") => print_help(0),
@@ -72,64 +80,52 @@ fn main() -> Result<()> {
}
}
let infile = infile.unwrap_or_else(|| print_help(1));
let outfile = outfile.unwrap_or_else(|| {
let mut name = infile.clone();
if unpack {
if name.extension().filter(|&e| e == "upk").is_some() {
name.set_extension("");
} else {
name.set_extension("bin");
}
} else {
let mut filename = name
.file_name()
.unwrap_or_else(|| OsStr::new(""))
.to_os_string();
filename.push(".upk");
name.set_file_name(filename);
}
name
});
let infile = IoTarget::from_filename(infile);
let outfile = |tpe: OutFileType| infile.output(tpe, &outfile);
if config.parity_contexts != 1 && config.parity_contexts != 2 && config.parity_contexts != 4 {
eprintln!("--parity has to be 1, 2, or 4");
process::exit(1);
}
if !unpack && !calculate_margin {
let mut data = vec![];
File::open(infile)?.read_to_end(&mut data)?;
if !unpack && !calculate_margin && !create_heatmap {
let mut data = infile.read()?;
if reverse {
data.reverse();
}
let mut pb = pbr::ProgressBar::new(data.len() as u64);
pb.set_units(pbr::Units::Bytes);
let mut packed_data = upkr::pack(
&data,
level,
&config,
Some(&mut |pos| {
pb.set(pos as u64);
}),
);
pb.finish();
#[cfg(feature = "terminal")]
let mut packed_data = {
let mut pb = pbr::ProgressBar::on(std::io::stderr(), data.len() as u64);
pb.set_units(pbr::Units::Bytes);
let packed_data = upkr::pack(
&data,
level,
&config,
Some(&mut |pos| {
pb.set(pos as u64);
}),
);
pb.finish();
eprintln!();
packed_data
};
#[cfg(not(feature = "terminal"))]
let mut packed_data = upkr::pack(&data, level, &config, None);
if reverse {
packed_data.reverse();
}
println!(
eprintln!(
"Compressed {} bytes to {} bytes ({}%)",
data.len(),
packed_data.len(),
packed_data.len() as f32 * 100. / data.len() as f32
);
File::create(outfile)?.write_all(&packed_data)?;
outfile(OutFileType::Packed).write(&packed_data)?;
} else {
let mut data = vec![];
File::open(infile)?.read_to_end(&mut data)?;
let mut data = infile.read()?;
if reverse {
data.reverse();
}
@@ -138,7 +134,36 @@ fn main() -> Result<()> {
if reverse {
unpacked_data.reverse();
}
File::create(outfile)?.write_all(&unpacked_data)?;
outfile(OutFileType::Unpacked).write(&unpacked_data)?;
}
if create_heatmap {
let mut heatmap = upkr::create_heatmap(&data, &config, max_unpacked_size)?;
if reverse {
heatmap.reverse();
}
match do_hexdump {
#[cfg(feature = "crossterm")]
true => {
if report_raw_cost {
heatmap.print_as_hex_raw_cost()?
} else {
heatmap.print_as_hex()?
}
}
_ => {
let mut heatmap_bin = Vec::with_capacity(heatmap.len());
for i in 0..heatmap.len() {
let cost = if report_raw_cost {
heatmap.raw_cost(i)
} else {
heatmap.cost(i)
};
let cost = (cost.log2() * 8. + 64.).round().max(0.).min(127.) as u8;
heatmap_bin.push((cost << 1) | heatmap.is_literal(i) as u8);
}
outfile(OutFileType::Heatmap).write(&heatmap_bin)?;
}
}
}
if calculate_margin {
println!("{}", upkr::calculate_margin(&data, &config)?);
@@ -148,17 +173,102 @@ fn main() -> Result<()> {
Ok(())
}
enum OutFileType {
Packed,
Unpacked,
Heatmap,
}
enum IoTarget {
StdInOut,
File(PathBuf),
}
impl IoTarget {
fn from_filename(filename: Option<PathBuf>) -> IoTarget {
if let Some(path) = filename {
if path.as_os_str() == "-" {
IoTarget::StdInOut
} else {
IoTarget::File(path)
}
} else {
IoTarget::StdInOut
}
}
fn read(&self) -> Result<Vec<u8>> {
let mut buffer = vec![];
match *self {
IoTarget::StdInOut => std::io::stdin().read_to_end(&mut buffer)?,
IoTarget::File(ref path) => File::open(path)?.read_to_end(&mut buffer)?,
};
Ok(buffer)
}
fn write(&self, data: &[u8]) -> Result<()> {
match *self {
IoTarget::StdInOut => std::io::stdout().write_all(data)?,
IoTarget::File(ref path) => File::create(path)?.write_all(data)?,
};
Ok(())
}
fn output(&self, tpe: OutFileType, outname: &Option<PathBuf>) -> IoTarget {
if outname.is_some() {
return IoTarget::from_filename(outname.clone());
}
match *self {
IoTarget::StdInOut => IoTarget::StdInOut,
IoTarget::File(ref path) => {
let mut name = path.clone();
match tpe {
OutFileType::Packed => {
let mut filename = name
.file_name()
.unwrap_or_else(|| OsStr::new(""))
.to_os_string();
filename.push(".upk");
name.set_file_name(filename);
}
OutFileType::Unpacked => {
if name.extension().filter(|&e| e == "upk").is_some() {
name.set_extension("");
} else {
name.set_extension("bin");
}
}
OutFileType::Heatmap => {
name.set_extension("heatmap");
}
}
IoTarget::File(name)
}
}
}
}
fn print_help(exit_code: i32) -> ! {
eprintln!("Usage:");
eprintln!(" upkr [-l level(0-9)] [config options] <infile> [<outfile>]");
eprintln!(" upkr -u [config options] <infile> [<outfile>]");
eprintln!(" upkr --heatmap [config options] <infile> [<outfile>]");
eprintln!(" upkr --margin [config options] <infile>");
eprintln!();
eprintln!(" -l, --level N compression level 0-9");
eprintln!(" -0, ..., -9 short form for setting compression level");
eprintln!(" -u, --unpack unpack infile");
eprintln!(" -d, --decompress decompress infile");
eprintln!(" --heatmap calculate heatmap from compressed file");
eprintln!(" --raw-cost report raw cost of literals in heatmap");
#[cfg(feature = "crossterm")]
eprintln!(" --hexdump print heatmap as colored hexdump");
eprintln!(" --margin calculate margin for overlapped unpacking of a packed file");
eprintln!();
eprintln!("When no infile is given, or the infile is '-', read from stdin.");
eprintln!(
"When no outfile is given and reading from stdin, or when outfile is '-', write to stdout."
);
eprintln!();
eprintln!("Version: {}", env!("CARGO_PKG_VERSION"));
eprintln!();
eprintln!("Config presets for specific unpackers:");

View File

@@ -148,6 +148,7 @@ impl EntropyCoder for CostCounter {
}
}
#[derive(Clone)]
pub struct RansDecoder<'a> {
data: &'a [u8],
pos: usize,
@@ -166,8 +167,8 @@ const PROB_MASK: u32 = ONE_PROB - 1;
pub struct UnexpectedEOF;
impl<'a> RansDecoder<'a> {
pub fn new(data: &'a [u8], config: &Config) -> RansDecoder<'a> {
RansDecoder {
pub fn new(data: &'a [u8], config: &Config) -> Result<RansDecoder<'a>, UnexpectedEOF> {
let mut decoder = RansDecoder {
data,
pos: 0,
state: 0,
@@ -176,7 +177,9 @@ impl<'a> RansDecoder<'a> {
bits_left: 0,
invert_bit_encoding: config.invert_bit_encoding,
bitstream_is_big_endian: config.bitstream_is_big_endian,
}
};
decoder.refill()?;
Ok(decoder)
}
pub fn pos(&self) -> usize {
@@ -189,8 +192,7 @@ impl<'a> RansDecoder<'a> {
Ok(bit)
}
pub fn decode_bit(&mut self, prob: u16) -> Result<bool, UnexpectedEOF> {
let prob = prob as u32;
fn refill(&mut self) -> Result<(), UnexpectedEOF> {
if self.use_bitstream {
while self.state < 32768 {
if self.bits_left == 0 {
@@ -219,6 +221,13 @@ impl<'a> RansDecoder<'a> {
self.pos += 1;
}
}
Ok(())
}
pub fn decode_bit(&mut self, prob: u16) -> Result<bool, UnexpectedEOF> {
self.refill()?;
let prob = prob as u32;
let bit = (self.state & PROB_MASK) < prob;
@@ -231,4 +240,9 @@ impl<'a> RansDecoder<'a> {
Ok(bit ^ self.invert_bit_encoding)
}
pub fn cost(&self, prev: &RansDecoder) -> f32 {
f32::log2(prev.state as f32) - f32::log2(self.state as f32)
+ (self.pos - prev.pos) as f32 * 8.
}
}