dcx: repack from existing instance

README.md

@@ -4,11 +4,12 @@ Rusted Iron Ring
 Low-level library for exploring From Software games files.
 
 This project is mainly to play with the Rust language, Nom parser, FFI, etc; if
-you need an actually used and tested library, see [SoulsFormats][soulsformats].
-The main target has been Dark Souls 1 PTDE, but checkout the features section
-below.
+you need an actually used and tested library, see [SoulsFormats][soulsformats]
+(C#) or [soulstruct][soulstruct] (Python). The main target has been Dark Souls 1
+PTDE, but checkout the features section below.
 
 [soulsformats]: https://github.com/JKAnderson/SoulsFormats
+[soulstruct]: https://github.com/Grimrukh/soulstruct
 
 
 
@@ -54,7 +55,7 @@ Features
 | Type     | Games | Features                                 |
 |----------|-------|------------------------------------------|
 | BHD5/BDT | DS1   | Load, extract                            |
-| DCX      | DS1   | Load, extract                            |
+| DCX      | DS1   | Load, extract, repack (untested)         |
 | BND3     | DS1   | Load, extract                            |
 | BHF3     | DS1   | Load, extract                            |
 | DAT      | KF4   | Load, extract, repack                    |
@@ -77,6 +78,7 @@ check out [SiegLib][sieglib].
 - There is a demo Python binding for some `name_hashes` features in the
     `bindings/python` dir, that uses [PyO3][pyo3] and thus requires nightly
     rustc to build.
+- There are a few scripts useful for some testing/modding tasks.
 
 [pyo3]: https://pyo3.rs/
 

src/formats/common.rs

@@ -5,6 +5,7 @@ use encoding_rs::SHIFT_JIS;
 use nom::IResult;
 use nom::bytes::complete::take_while;
 
+/// Trait for structs that are easy to pack to bytes.
 pub trait Pack {
     /// Write the entirety of `self` as bytes to the write buffer `f`.
     fn write(&self, f: &mut dyn io::Write) -> io::Result<usize>;

src/formats/dat.rs

@@ -18,6 +18,11 @@ pub struct DatHeader {
     pub num_files: u32,
 }
 
+fn parse_header(i: &[u8]) -> IResult<&[u8], DatHeader> {
+    let (i, (unk00, num_files)) = tuple((le_u32, le_u32))(i)?;
+    Ok((i, DatHeader { unk00, num_files }))
+}
+
 impl Pack for DatHeader {
     fn write(&self, f: &mut dyn io::Write) -> io::Result<usize> {
         f.write_all(&self.unk00.to_le_bytes())?;
@@ -26,11 +31,6 @@ impl Pack for DatHeader {
     }
 }
 
-fn parse_header(i: &[u8]) -> IResult<&[u8], DatHeader> {
-    let (i, (unk00, num_files)) = tuple((le_u32, le_u32))(i)?;
-    Ok((i, DatHeader { unk00, num_files }))
-}
-
 pub const FILE_ENTRY_SIZE: usize = 0x40;
 pub const FILE_ENTRY_NAME_MAXLEN: usize = 0x34;
 
@@ -42,6 +42,13 @@ pub struct DatFileEntry {
     pub ofs_data: u32,
 }
 
+fn parse_file_entry(i: &[u8]) -> IResult<&[u8], DatFileEntry> {
+    let (i, name) = take_cstring_from(i, FILE_ENTRY_NAME_MAXLEN)?;
+    let name = String::from_utf8_lossy(name).to_string();
+    let (i, (size, padded_size, ofs_data)) = tuple((le_u32, le_u32, le_u32))(i)?;
+    Ok((i, DatFileEntry { name, size, padded_size, ofs_data }))
+}
+
 impl Pack for DatFileEntry {
     fn write(&self, f: &mut dyn io::Write) -> io::Result<usize> {
         let name_bytes = self.name.as_bytes();
@@ -54,13 +61,6 @@ impl Pack for DatFileEntry {
     }
 }
 
-fn parse_file_entry(i: &[u8]) -> IResult<&[u8], DatFileEntry> {
-    let (i, name) = take_cstring_from(i, FILE_ENTRY_NAME_MAXLEN)?;
-    let name = String::from_utf8_lossy(name).to_string();
-    let (i, (size, padded_size, ofs_data)) = tuple((le_u32, le_u32, le_u32))(i)?;
-    Ok((i, DatFileEntry { name, size, padded_size, ofs_data }))
-}
-
 pub const INTERNAL_PATH_SEP: char = '/';
 pub const DATA_ALIGN: usize = 0x8000;
 

src/formats/dcx.rs

@@ -1,3 +1,9 @@
+//! DCX format.
+//!
+//! Support DFLT method only.
+
+use std::io;
+
 use nom::IResult;
 use nom::branch::alt;
 use nom::bytes::complete::tag;
@@ -5,6 +11,11 @@ use nom::multi::count;
 use nom::number::complete::*;
 use nom::sequence::tuple;
 
+use crate::formats::common::Pack;
+
+pub const HEADER_MAGIC: &[u8] = b"DCX\0";
+pub const HEADER_SIZE: usize = 0x18;
+
 #[derive(Debug)]
 pub struct DcxHeader {
     pub magic: Vec<u8>,
@@ -17,10 +28,25 @@ pub struct DcxHeader {
 
 fn parse_header(i: &[u8]) -> IResult<&[u8], DcxHeader> {
     let (i, (magic, unk04, ofs_dcs, ofs_dcp, unk10, unk14)) =
-        tuple((tag(b"DCX\0"), be_u32, be_u32, be_u32, be_u32, be_u32))(i)?;
+        tuple((tag(HEADER_MAGIC), be_u32, be_u32, be_u32, be_u32, be_u32))(i)?;
     Ok((i, DcxHeader { magic: magic.to_vec(), unk04, ofs_dcs, ofs_dcp, unk10, unk14 }))
 }
 
+impl Pack for DcxHeader {
+    fn write(&self, f: &mut dyn io::Write) -> io::Result<usize> {
+        f.write_all(&self.magic)?;
+        f.write_all(&self.unk04.to_be_bytes())?;
+        f.write_all(&self.ofs_dcs.to_be_bytes())?;
+        f.write_all(&self.ofs_dcp.to_be_bytes())?;
+        f.write_all(&self.unk10.to_be_bytes())?;
+        f.write_all(&self.unk14.to_be_bytes())?;
+        Ok(HEADER_SIZE)
+    }
+}
+
+pub const SIZES_CHUNK_MAGIC: &[u8] = b"DCS\0";
+pub const SIZES_CHUNK_SIZE: usize = 0xC;
+
 #[derive(Debug)]
 pub struct DcxSizes {
     pub magic: Vec<u8>,
@@ -30,10 +56,22 @@ pub struct DcxSizes {
 
 fn parse_sizes(i: &[u8]) -> IResult<&[u8], DcxSizes> {
     let (i, (magic, uncompressed_size, compressed_size)) =
-        tuple((tag(b"DCS\0"), be_u32, be_u32))(i)?;
+        tuple((tag(SIZES_CHUNK_MAGIC), be_u32, be_u32))(i)?;
     Ok((i, DcxSizes { magic: magic.to_vec(), uncompressed_size, compressed_size }))
 }
 
+impl Pack for DcxSizes {
+    fn write(&self, f: &mut dyn io::Write) -> io::Result<usize> {
+        f.write_all(&self.magic)?;
+        f.write_all(&self.uncompressed_size.to_be_bytes())?;
+        f.write_all(&self.compressed_size.to_be_bytes())?;
+        Ok(SIZES_CHUNK_SIZE)
+    }
+}
+
+pub const PARAMS_CHUNK_MAGIC: &[u8] = b"DCP\0";
+pub const PARAMS_CHUNK_SIZE: usize = 0x32;
+
 #[derive(Debug)]
 pub struct DcxParams {
     pub magic: Vec<u8>,
@@ -52,7 +90,7 @@ pub struct DcxParams {
 fn parse_params(i: &[u8]) -> IResult<&[u8], DcxParams> {
     let (i, (magic, method, ofs_dca, flags, unk10, unk14, unk18, unk1C)) =
         tuple((
-            tag(b"DCP\0"),
+            tag(PARAMS_CHUNK_MAGIC),
             alt((tag(b"DFLT"), tag(b"EDGE"), tag(b"KRAK"))),
             be_u32,
             count(be_u8, 4),
@@ -79,6 +117,26 @@ fn parse_params(i: &[u8]) -> IResult<&[u8], DcxParams> {
     ))
 }
 
+impl Pack for DcxParams {
+    fn write(&self, f: &mut dyn io::Write) -> io::Result<usize> {
+        f.write_all(&self.magic)?;
+        f.write_all(&self.method)?;
+        f.write_all(&self.ofs_dca.to_be_bytes())?;
+        f.write_all(&self.unk0C.to_be_bytes())?;
+        f.write_all(&self.unk0D.to_be_bytes())?;
+        f.write_all(&self.unk0E.to_be_bytes())?;
+        f.write_all(&self.unk0F.to_be_bytes())?;
+        f.write_all(&self.unk10.to_be_bytes())?;
+        f.write_all(&self.unk14.to_be_bytes())?;
+        f.write_all(&self.unk18.to_be_bytes())?;
+        f.write_all(&self.unk1C.to_be_bytes())?;
+        Ok(PARAMS_CHUNK_SIZE)
+    }
+}
+
+pub const ARCHIVE_CHUNK_MAGIC: &[u8] = b"DCA\0";
+pub const ARCHIVE_CHUNK_SIZE: usize = 0x8;
+
 #[derive(Debug)]
 pub struct DcxArchive {
     pub magic: Vec<u8>,
@@ -86,10 +144,18 @@ pub struct DcxArchive {
 }
 
 fn parse_archive(i: &[u8]) -> IResult<&[u8], DcxArchive> {
-    let (i, (magic, ofs_data)) = tuple((tag(b"DCA\0"), be_u32))(i)?;
+    let (i, (magic, ofs_data)) = tuple((tag(ARCHIVE_CHUNK_MAGIC), be_u32))(i)?;
     Ok((i, DcxArchive { magic: magic.to_vec(), ofs_data }))
 }
 
+impl Pack for DcxArchive {
+    fn write(&self, f: &mut dyn io::Write) -> io::Result<usize> {
+        f.write_all(&self.magic)?;
+        f.write_all(&self.ofs_data.to_be_bytes())?;
+        Ok(ARCHIVE_CHUNK_SIZE)
+    }
+}
+
 #[derive(Debug)]
 pub struct Dcx {
     pub header: DcxHeader,

src/lib.rs

@@ -1,9 +1,6 @@
 #![allow(non_snake_case)]
 
 pub mod name_hashes;
-pub mod repackers {
-    pub mod dat;
-}
 pub mod formats {
     pub mod bhd;
     pub mod bhf;
@@ -14,6 +11,11 @@ pub mod formats {
     pub mod param;
     pub mod paramdef;
 }
+pub mod repackers {
+    pub mod dat;
+    pub mod dcx;
+    pub mod errors;
+}
 pub mod unpackers {
     pub mod bhd;
     pub mod bhf;

src/repackers/dcx.rs

@@ -0,0 +1,45 @@
+use std::fs;
+use std::io::Write;
+
+use flate2::Compression;
+use flate2::write::ZlibEncoder;
+
+use crate::formats::common::Pack;
+use crate::formats::dcx;
+use crate::repackers::errors::PackError;
+
+/// Repack a previously unpacked DCX with this new data.
+///
+/// Params that are not well understood are reused, others are replaced
+/// with accurate values.
+pub fn pack_dcx(dcx: &mut dcx::Dcx, data: &[u8], output_path: &str) -> Result<(), PackError> {
+    dcx.sizes.uncompressed_size = data.len() as u32;
+    let compressed = compress(dcx, data)?;
+    dcx.sizes.compressed_size = compressed.len() as u32;
+
+    let mut output_file = fs::File::create(output_path)?;
+    dcx.header.write(&mut output_file)?;
+    dcx.sizes.write(&mut output_file)?;
+    dcx.params.write(&mut output_file)?;
+    dcx.archive.write(&mut output_file)?;
+    output_file.write_all(&compressed)?;
+    Ok(())
+}
+
+/// Compress data using DCX params.
+pub fn compress(dcx: &dcx::Dcx, data: &[u8]) -> Result<Vec<u8>, PackError> {
+    let method: &[u8] = dcx.params.method.as_slice();
+    if method == b"DFLT" {
+        compress_dflt(dcx, data)
+    } else {
+        let method_string = String::from_utf8_lossy(method).to_string();
+        Err(PackError::Compression(format!("Method unknown: {}", method_string)))
+    }
+}
+
+fn compress_dflt(dcx: &dcx::Dcx, data: &[u8]) -> Result<Vec<u8>, PackError> {
+    let level = dcx.params.unk0C as u32;  // Unsure if it really is compression level.
+    let half_size = data.len() / 2;  // Quicker allocation.
+    let encoder = ZlibEncoder::new(Vec::with_capacity(half_size), Compression::new(level));
+    Ok(encoder.finish()?)
+}

src/repackers/errors.rs

@@ -0,0 +1,14 @@
+use std::io;
+
+#[derive(Debug)]
+pub enum PackError {
+    Io(io::Error),
+    Compression(String),
+    Unknown(String),
+}
+
+impl From<io::Error> for PackError {
+    fn from(e: io::Error) -> Self {
+        PackError::Io(e)
+    }
+}

src/unpackers/dcx.rs

@@ -36,10 +36,7 @@ fn decompress_dcx(dcx: &dcx::Dcx, comp_data: &[u8]) -> Result<Vec<u8>, UnpackErr
     if method == b"DFLT" {
         decompress_dcx_dflt(dcx, comp_data)
     } else {
-        let method_string = match std::str::from_utf8(method) {
-            Ok(s) => { String::from(s) }
-            Err(_) => { format!("{:?}", method) }
-        };
+        let method_string = String::from_utf8_lossy(method).to_string();
         Err(UnpackError::Compression(format!("Unknown method: {}", method_string)))
     }
 }