Mercurial > touhou

comparison formats/src/th06/pbg3.rs @ 757:21b186be2590

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
Split the Rust version into multiple crates.
author Emmanuel Gil Peyrot <linkmauve@linkmauve.fr>
date Tue, 05 Jan 2021 02:16:32 +0100
parents src/th06/pbg3.rs@0ebf6467e4ff
children eba9a3d0c484
comparison
equal deleted inserted replaced
756:4d91790cf8ab 757:21b186be2590
1 //! PBG3 archive files handling.
2 //!
3 //! This module provides classes for handling the PBG3 file format.
4 //! The PBG3 format is the archive format used by Touhou 6: EoSD.
5 //!
6 //! PBG3 files are merely a bitstream composed of a header, a file
7 //! table, and LZSS-compressed files.
8
9 use touhou_utils::bitstream::BitStream;
10 use touhou_utils::lzss;
11 use std::fs::File;
12 use std::io;
13 use std::collections::hash_map::{self, HashMap};
14 use std::path::Path;
15
16 /// Helper struct to handle strings and integers in PBG3 bitstreams.
17 pub struct PBG3BitStream<R: io::Read + io::Seek> {
18 bitstream: BitStream<R>,
19 }
20
21 impl<R: io::Read + io::Seek> PBG3BitStream<R> {
22 /// Create a bitstream capable of reading u32 and strings.
23 pub fn new(bitstream: BitStream<R>) -> PBG3BitStream<R> {
24 PBG3BitStream {
25 bitstream,
26 }
27 }
28
29 /// Seek inside the bitstream, ditching any unused data read.
30 pub fn seek(&mut self, seek_from: io::SeekFrom) -> io::Result<u64> {
31 self.bitstream.seek(seek_from)
32 }
33
34 /// Return the current position in the stream.
35 pub fn tell(&mut self) -> io::Result<u64> {
36 self.bitstream.seek(io::SeekFrom::Current(0))
37 }
38
39 /// Read a given amount of bits.
40 pub fn read(&mut self, nb_bits: usize) -> io::Result<usize> {
41 self.bitstream.read(nb_bits)
42 }
43
44 /// Read a given amount of bytes.
45 pub fn read_bytes(&mut self, nb_bytes: usize) -> io::Result<Vec<u8>> {
46 self.bitstream.read_bytes(nb_bytes)
47 }
48
49 /// Read an integer from the bitstream.
50 ///
51 /// Integers have variable sizes. They begin with a two-bit value indicating
52 /// the number of (non-aligned) bytes to read.
53 pub fn read_u32(&mut self) -> io::Result<u32> {
54 let size = self.read(2)?;
55 Ok(self.read((size + 1) * 8)? as u32)
56 }
57
58 /// Read a string from the bitstream.
59 ///
60 /// Strings are stored as NULL-terminated sequences of bytes.
61 /// The only catch is that they are not byte-aligned.
62 pub fn read_string(&mut self, mut max_size: usize) -> io::Result<Vec<u8>> {
63 let mut buf = Vec::new();
64 while max_size > 0 {
65 let byte = self.read(8)? as u8;
66 if byte == 0 {
67 break;
68 }
69 buf.push(byte);
70 max_size -= 1;
71 }
72 Ok(buf)
73 }
74 }
75
76 type Entry = (u32, u32, u32, u32, u32);
77
78 /// Handle PBG3 archive files.
79 ///
80 /// PBG3 is a file archive format used in Touhou 6: EoSD.
81 /// This class provides a representation of such files, as well as functions to
82 /// read and extract files from a PBG3 archive.
83 pub struct PBG3<R: io::Read + io::Seek> {
84 /// List of PBG3Entry objects describing files present in the archive.
85 entries: HashMap<String, Entry>,
86
87 /// PBG3BitStream struct.
88 bitstream: PBG3BitStream<R>,
89 }
90
91 impl<R: io::Read + io::Seek> PBG3<R> {
92 /// Create a PBG3 archive.
93 fn new(entries: HashMap<String, Entry>, bitstream: PBG3BitStream<R>) -> PBG3<R> {
94 PBG3 {
95 entries,
96 bitstream,
97 }
98 }
99
100 /// Open a PBG3 archive.
101 pub fn from_file(mut file: R) -> io::Result<PBG3<R>> {
102 let mut magic = [0; 4];
103 file.read(&mut magic)?;
104 if &magic != b"PBG3" {
105 return Err(io::Error::new(io::ErrorKind::Other, "Wrong magic!"));
106 }
107
108 let bitstream = BitStream::new(file);
109 let mut bitstream = PBG3BitStream::new(bitstream);
110 let mut entries = HashMap::new();
111
112 let nb_entries = bitstream.read_u32()?;
113 let offset = bitstream.read_u32()?;
114 bitstream.seek(io::SeekFrom::Start(offset as u64))?;
115
116 for _ in 0..nb_entries {
117 let unknown_1 = bitstream.read_u32()?;
118 let unknown_2 = bitstream.read_u32()?;
119 let checksum = bitstream.read_u32()?; // Checksum of *compressed data*
120 let offset = bitstream.read_u32()?;
121 let size = bitstream.read_u32()?;
122 let name = bitstream.read_string(255)?;
123 // XXX: no unwrap!
124 let name = String::from_utf8(name).unwrap();
125 entries.insert(name, (unknown_1, unknown_2, checksum, offset, size));
126 }
127
128 Ok(PBG3::new(entries, bitstream))
129 }
130
131 /// List all file entries in this PBG3 archive.
132 pub fn list_files(&self) -> hash_map::Keys<String, Entry> {
133 self.entries.keys()
134 }
135
136 /// Read a single file from this PBG3 archive.
137 pub fn get_file(&mut self, filename: &str, check: bool) -> io::Result<Vec<u8>> {
138 // XXX: no unwrap!
139 let (_unknown_1, _unknown_2, checksum, offset, size) = self.entries.get(filename).unwrap();
140 self.bitstream.seek(io::SeekFrom::Start(*offset as u64))?;
141 let data = lzss::decompress(&mut self.bitstream.bitstream, *size as usize, 0x2000, 13, 4, 3)?;
142 if check {
143 // Verify the checksum.
144 let compressed_size = self.bitstream.tell()? as u32 - *offset;
145 self.bitstream.seek(io::SeekFrom::Start(*offset as u64))?;
146 let mut value: u32 = 0;
147 for c in self.bitstream.read_bytes(compressed_size as usize)? {
148 value += c as u32;
149 value &= 0xffffffff;
150 }
151 if value != *checksum {
152 return Err(io::Error::new(io::ErrorKind::Other, "Corrupted data!"));
153 }
154 }
155 Ok(data)
156 }
157 }
158
159 /// Open a PBG3 archive from its path.
160 pub fn from_path_buffered<P: AsRef<Path>>(path: P) -> io::Result<PBG3<io::BufReader<File>>> {
161 let file = File::open(path)?;
162 let buf_file = io::BufReader::new(file);
163 PBG3::from_file(buf_file)
164 }
165
166 #[cfg(test)]
167 mod tests {
168 use super::*;
169 use crate::util::SeekableSlice;
170 use std::fs::File;
171
172 #[test]
173 fn bitstream() {
174 let data = SeekableSlice::new(b"Hello world!\0");
175 let bitstream = BitStream::new(data);
176 let mut pbg3 = PBG3BitStream::new(bitstream);
177 assert_eq!(pbg3.read_string(42).unwrap(), b"Hello world!");
178 }
179
180 #[test]
181 fn file_present() {
182 let file = File::open("EoSD/MD.DAT").unwrap();
183 let file = io::BufReader::new(file);
184 let pbg3 = PBG3::from_file(file).unwrap();
185 let files = pbg3.list_files().cloned().collect::<Vec<String>>();
186 assert!(files.contains(&String::from("th06_01.pos")));
187 }
188
189 #[test]
190 fn check_all_files() {
191 let file = File::open("EoSD/MD.DAT").unwrap();
192 let file = io::BufReader::new(file);
193 let mut pbg3 = PBG3::from_file(file).unwrap();
194 let files = pbg3.list_files().cloned().collect::<Vec<String>>();
195 for filename in files {
196 pbg3.get_file(filename, true).unwrap();
197 }
198 }
199 }