diff options
Diffstat (limited to 'crates/nres/src/lib.rs')
| -rw-r--r-- | crates/nres/src/lib.rs | 702 |
1 files changed, 702 insertions, 0 deletions
diff --git a/crates/nres/src/lib.rs b/crates/nres/src/lib.rs new file mode 100644 index 0000000..e0631e3 --- /dev/null +++ b/crates/nres/src/lib.rs @@ -0,0 +1,702 @@ +pub mod error; + +use crate::error::Error; +use common::{OutputBuffer, ResourceData}; +use core::ops::Range; +use std::cmp::Ordering; +use std::fs::{self, OpenOptions as FsOpenOptions}; +use std::io::Write; +use std::path::{Path, PathBuf}; +use std::sync::Arc; +use std::time::{SystemTime, UNIX_EPOCH}; + +pub type Result<T> = core::result::Result<T, Error>; + +#[derive(Clone, Debug, Default)] +pub struct OpenOptions { + pub raw_mode: bool, + pub sequential_hint: bool, + pub prefetch_pages: bool, +} + +#[derive(Clone, Debug, Default)] +pub enum OpenMode { + #[default] + ReadOnly, + ReadWrite, +} + +#[derive(Debug)] +pub struct Archive { + bytes: Arc<[u8]>, + entries: Vec<EntryRecord>, + raw_mode: bool, +} + +#[derive(Copy, Clone, Debug, PartialEq, Eq, Hash)] +pub struct EntryId(pub u32); + +#[derive(Clone, Debug)] +pub struct EntryMeta { + pub kind: u32, + pub attr1: u32, + pub attr2: u32, + pub attr3: u32, + pub name: String, + pub data_offset: u64, + pub data_size: u32, + pub sort_index: u32, +} + +#[derive(Copy, Clone, Debug)] +pub struct EntryRef<'a> { + pub id: EntryId, + pub meta: &'a EntryMeta, +} + +#[derive(Clone, Debug)] +struct EntryRecord { + meta: EntryMeta, + name_raw: [u8; 36], +} + +impl Archive { + pub fn open_path(path: impl AsRef<Path>) -> Result<Self> { + Self::open_path_with(path, OpenMode::ReadOnly, OpenOptions::default()) + } + + pub fn open_path_with( + path: impl AsRef<Path>, + _mode: OpenMode, + opts: OpenOptions, + ) -> Result<Self> { + let bytes = fs::read(path.as_ref())?; + let arc: Arc<[u8]> = Arc::from(bytes.into_boxed_slice()); + Self::open_bytes(arc, opts) + } + + pub fn open_bytes(bytes: Arc<[u8]>, opts: OpenOptions) -> Result<Self> { + let (entries, _) = parse_archive(&bytes, opts.raw_mode)?; + if opts.prefetch_pages { + prefetch_pages(&bytes); + } + Ok(Self { + bytes, + entries, + raw_mode: opts.raw_mode, + }) + } + + pub fn entry_count(&self) -> usize { + self.entries.len() + } + + pub fn entries(&self) -> impl Iterator<Item = EntryRef<'_>> { + self.entries + .iter() + .enumerate() + .map(|(idx, entry)| EntryRef { + id: EntryId(u32::try_from(idx).expect("entry count validated at parse")), + meta: &entry.meta, + }) + } + + pub fn find(&self, name: &str) -> Option<EntryId> { + if self.entries.is_empty() { + return None; + } + + if !self.raw_mode { + let mut low = 0usize; + let mut high = self.entries.len(); + while low < high { + let mid = low + (high - low) / 2; + let Ok(target_idx) = usize::try_from(self.entries[mid].meta.sort_index) else { + break; + }; + if target_idx >= self.entries.len() { + break; + } + let cmp = cmp_name_case_insensitive( + name.as_bytes(), + entry_name_bytes(&self.entries[target_idx].name_raw), + ); + match cmp { + Ordering::Less => high = mid, + Ordering::Greater => low = mid + 1, + Ordering::Equal => { + return Some(EntryId( + u32::try_from(target_idx).expect("entry count validated at parse"), + )) + } + } + } + } + + self.entries.iter().enumerate().find_map(|(idx, entry)| { + if cmp_name_case_insensitive(name.as_bytes(), entry_name_bytes(&entry.name_raw)) + == Ordering::Equal + { + Some(EntryId( + u32::try_from(idx).expect("entry count validated at parse"), + )) + } else { + None + } + }) + } + + pub fn get(&self, id: EntryId) -> Option<EntryRef<'_>> { + let idx = usize::try_from(id.0).ok()?; + let entry = self.entries.get(idx)?; + Some(EntryRef { + id, + meta: &entry.meta, + }) + } + + pub fn read(&self, id: EntryId) -> Result<ResourceData<'_>> { + let range = self.entry_range(id)?; + Ok(ResourceData::Borrowed(&self.bytes[range])) + } + + pub fn read_into(&self, id: EntryId, out: &mut dyn OutputBuffer) -> Result<usize> { + let range = self.entry_range(id)?; + out.write_exact(&self.bytes[range.clone()])?; + Ok(range.len()) + } + + pub fn raw_slice(&self, id: EntryId) -> Result<Option<&[u8]>> { + let range = self.entry_range(id)?; + Ok(Some(&self.bytes[range])) + } + + pub fn edit_path(path: impl AsRef<Path>) -> Result<Editor> { + let path_buf = path.as_ref().to_path_buf(); + let bytes = fs::read(&path_buf)?; + let arc: Arc<[u8]> = Arc::from(bytes.into_boxed_slice()); + let (entries, _) = parse_archive(&arc, false)?; + let mut editable = Vec::with_capacity(entries.len()); + for entry in &entries { + let range = checked_range(entry.meta.data_offset, entry.meta.data_size, arc.len())?; + editable.push(EditableEntry { + meta: entry.meta.clone(), + name_raw: entry.name_raw, + data: EntryData::Borrowed(range), // Copy-on-write: only store range + }); + } + Ok(Editor { + path: path_buf, + source: arc, + entries: editable, + }) + } + + fn entry_range(&self, id: EntryId) -> Result<Range<usize>> { + let idx = usize::try_from(id.0).map_err(|_| Error::IntegerOverflow)?; + let Some(entry) = self.entries.get(idx) else { + return Err(Error::EntryIdOutOfRange { + id: id.0, + entry_count: self.entries.len().try_into().unwrap_or(u32::MAX), + }); + }; + checked_range( + entry.meta.data_offset, + entry.meta.data_size, + self.bytes.len(), + ) + } +} + +pub struct Editor { + path: PathBuf, + source: Arc<[u8]>, + entries: Vec<EditableEntry>, +} + +#[derive(Clone, Debug)] +enum EntryData { + Borrowed(Range<usize>), + Modified(Vec<u8>), +} + +#[derive(Clone, Debug)] +struct EditableEntry { + meta: EntryMeta, + name_raw: [u8; 36], + data: EntryData, +} + +impl EditableEntry { + fn data_slice<'a>(&'a self, source: &'a Arc<[u8]>) -> &'a [u8] { + match &self.data { + EntryData::Borrowed(range) => &source[range.clone()], + EntryData::Modified(vec) => vec.as_slice(), + } + } +} + +#[derive(Clone, Debug)] +pub struct NewEntry<'a> { + pub kind: u32, + pub attr1: u32, + pub attr2: u32, + pub attr3: u32, + pub name: &'a str, + pub data: &'a [u8], +} + +impl Editor { + pub fn entries(&self) -> impl Iterator<Item = EntryRef<'_>> { + self.entries + .iter() + .enumerate() + .map(|(idx, entry)| EntryRef { + id: EntryId(u32::try_from(idx).expect("entry count validated at add")), + meta: &entry.meta, + }) + } + + pub fn add(&mut self, entry: NewEntry<'_>) -> Result<EntryId> { + let name_raw = encode_name_field(entry.name)?; + let id_u32 = u32::try_from(self.entries.len()).map_err(|_| Error::IntegerOverflow)?; + let data_size = u32::try_from(entry.data.len()).map_err(|_| Error::IntegerOverflow)?; + self.entries.push(EditableEntry { + meta: EntryMeta { + kind: entry.kind, + attr1: entry.attr1, + attr2: entry.attr2, + attr3: entry.attr3, + name: decode_name(entry_name_bytes(&name_raw)), + data_offset: 0, + data_size, + sort_index: 0, + }, + name_raw, + data: EntryData::Modified(entry.data.to_vec()), + }); + Ok(EntryId(id_u32)) + } + + pub fn replace_data(&mut self, id: EntryId, data: &[u8]) -> Result<()> { + let idx = usize::try_from(id.0).map_err(|_| Error::IntegerOverflow)?; + let Some(entry) = self.entries.get_mut(idx) else { + return Err(Error::EntryIdOutOfRange { + id: id.0, + entry_count: self.entries.len().try_into().unwrap_or(u32::MAX), + }); + }; + entry.meta.data_size = u32::try_from(data.len()).map_err(|_| Error::IntegerOverflow)?; + // Replace with new data (triggers copy-on-write if borrowed) + entry.data = EntryData::Modified(data.to_vec()); + Ok(()) + } + + pub fn remove(&mut self, id: EntryId) -> Result<()> { + let idx = usize::try_from(id.0).map_err(|_| Error::IntegerOverflow)?; + if idx >= self.entries.len() { + return Err(Error::EntryIdOutOfRange { + id: id.0, + entry_count: self.entries.len().try_into().unwrap_or(u32::MAX), + }); + } + self.entries.remove(idx); + Ok(()) + } + + pub fn commit(mut self) -> Result<()> { + let count_u32 = u32::try_from(self.entries.len()).map_err(|_| Error::IntegerOverflow)?; + + // Pre-calculate capacity to avoid reallocations + let total_data_size: usize = self + .entries + .iter() + .map(|e| e.data_slice(&self.source).len()) + .sum(); + let padding_estimate = self.entries.len() * 8; // Max 8 bytes padding per entry + let directory_size = self.entries.len() * 64; // 64 bytes per entry + let capacity = 16 + total_data_size + padding_estimate + directory_size; + + let mut out = Vec::with_capacity(capacity); + out.resize(16, 0); // Header + + // Keep reference to source for copy-on-write + let source = &self.source; + + for entry in &mut self.entries { + entry.meta.data_offset = + u64::try_from(out.len()).map_err(|_| Error::IntegerOverflow)?; + + // Calculate size and get slice separately to avoid borrow conflicts + let data_len = entry.data_slice(source).len(); + entry.meta.data_size = u32::try_from(data_len).map_err(|_| Error::IntegerOverflow)?; + + // Now get the slice again for writing + let data_slice = entry.data_slice(source); + out.extend_from_slice(data_slice); + + let padding = (8 - (out.len() % 8)) % 8; + if padding > 0 { + out.resize(out.len() + padding, 0); + } + } + + let mut sort_order: Vec<usize> = (0..self.entries.len()).collect(); + sort_order.sort_by(|a, b| { + cmp_name_case_insensitive( + entry_name_bytes(&self.entries[*a].name_raw), + entry_name_bytes(&self.entries[*b].name_raw), + ) + }); + + for (idx, entry) in self.entries.iter_mut().enumerate() { + entry.meta.sort_index = + u32::try_from(sort_order[idx]).map_err(|_| Error::IntegerOverflow)?; + } + + for entry in &self.entries { + let data_offset_u32 = + u32::try_from(entry.meta.data_offset).map_err(|_| Error::IntegerOverflow)?; + push_u32(&mut out, entry.meta.kind); + push_u32(&mut out, entry.meta.attr1); + push_u32(&mut out, entry.meta.attr2); + push_u32(&mut out, entry.meta.data_size); + push_u32(&mut out, entry.meta.attr3); + out.extend_from_slice(&entry.name_raw); + push_u32(&mut out, data_offset_u32); + push_u32(&mut out, entry.meta.sort_index); + } + + let total_size_u32 = u32::try_from(out.len()).map_err(|_| Error::IntegerOverflow)?; + out[0..4].copy_from_slice(b"NRes"); + out[4..8].copy_from_slice(&0x100_u32.to_le_bytes()); + out[8..12].copy_from_slice(&count_u32.to_le_bytes()); + out[12..16].copy_from_slice(&total_size_u32.to_le_bytes()); + + write_atomic(&self.path, &out) + } +} + +fn parse_archive(bytes: &[u8], raw_mode: bool) -> Result<(Vec<EntryRecord>, u64)> { + if raw_mode { + let data_size = u32::try_from(bytes.len()).map_err(|_| Error::IntegerOverflow)?; + let entry = EntryRecord { + meta: EntryMeta { + kind: 0, + attr1: 0, + attr2: 0, + attr3: 0, + name: String::from("RAW"), + data_offset: 0, + data_size, + sort_index: 0, + }, + name_raw: { + let mut name = [0u8; 36]; + let bytes_name = b"RAW"; + name[..bytes_name.len()].copy_from_slice(bytes_name); + name + }, + }; + return Ok(( + vec![entry], + u64::try_from(bytes.len()).map_err(|_| Error::IntegerOverflow)?, + )); + } + + if bytes.len() < 16 { + let mut got = [0u8; 4]; + let copy_len = bytes.len().min(4); + got[..copy_len].copy_from_slice(&bytes[..copy_len]); + return Err(Error::InvalidMagic { got }); + } + + let mut magic = [0u8; 4]; + magic.copy_from_slice(&bytes[0..4]); + if &magic != b"NRes" { + return Err(Error::InvalidMagic { got: magic }); + } + + let version = read_u32(bytes, 4)?; + if version != 0x100 { + return Err(Error::UnsupportedVersion { got: version }); + } + + let entry_count_i32 = i32::from_le_bytes( + bytes[8..12] + .try_into() + .map_err(|_| Error::IntegerOverflow)?, + ); + if entry_count_i32 < 0 { + return Err(Error::InvalidEntryCount { + got: entry_count_i32, + }); + } + let entry_count = usize::try_from(entry_count_i32).map_err(|_| Error::IntegerOverflow)?; + + // Validate entry_count fits in u32 (required for EntryId) + if entry_count > u32::MAX as usize { + return Err(Error::TooManyEntries { got: entry_count }); + } + + let total_size = read_u32(bytes, 12)?; + let actual_size = u64::try_from(bytes.len()).map_err(|_| Error::IntegerOverflow)?; + if u64::from(total_size) != actual_size { + return Err(Error::TotalSizeMismatch { + header: total_size, + actual: actual_size, + }); + } + + let directory_len = u64::try_from(entry_count) + .map_err(|_| Error::IntegerOverflow)? + .checked_mul(64) + .ok_or(Error::IntegerOverflow)?; + let directory_offset = + u64::from(total_size) + .checked_sub(directory_len) + .ok_or(Error::DirectoryOutOfBounds { + directory_offset: 0, + directory_len, + file_len: actual_size, + })?; + + if directory_offset < 16 || directory_offset + directory_len > actual_size { + return Err(Error::DirectoryOutOfBounds { + directory_offset, + directory_len, + file_len: actual_size, + }); + } + + let mut entries = Vec::with_capacity(entry_count); + for index in 0..entry_count { + let base = usize::try_from(directory_offset) + .map_err(|_| Error::IntegerOverflow)? + .checked_add(index.checked_mul(64).ok_or(Error::IntegerOverflow)?) + .ok_or(Error::IntegerOverflow)?; + + let kind = read_u32(bytes, base)?; + let attr1 = read_u32(bytes, base + 4)?; + let attr2 = read_u32(bytes, base + 8)?; + let data_size = read_u32(bytes, base + 12)?; + let attr3 = read_u32(bytes, base + 16)?; + + let mut name_raw = [0u8; 36]; + let name_slice = bytes + .get(base + 20..base + 56) + .ok_or(Error::IntegerOverflow)?; + name_raw.copy_from_slice(name_slice); + + let name_bytes = entry_name_bytes(&name_raw); + if name_bytes.len() > 35 { + return Err(Error::NameTooLong { + got: name_bytes.len(), + max: 35, + }); + } + + let data_offset = u64::from(read_u32(bytes, base + 56)?); + let sort_index = read_u32(bytes, base + 60)?; + + let end = data_offset + .checked_add(u64::from(data_size)) + .ok_or(Error::IntegerOverflow)?; + if data_offset < 16 || end > directory_offset { + return Err(Error::EntryDataOutOfBounds { + id: u32::try_from(index).map_err(|_| Error::IntegerOverflow)?, + offset: data_offset, + size: data_size, + directory_offset, + }); + } + + entries.push(EntryRecord { + meta: EntryMeta { + kind, + attr1, + attr2, + attr3, + name: decode_name(name_bytes), + data_offset, + data_size, + sort_index, + }, + name_raw, + }); + } + + Ok((entries, directory_offset)) +} + +fn checked_range(offset: u64, size: u32, bytes_len: usize) -> Result<Range<usize>> { + let start = usize::try_from(offset).map_err(|_| Error::IntegerOverflow)?; + let len = usize::try_from(size).map_err(|_| Error::IntegerOverflow)?; + let end = start.checked_add(len).ok_or(Error::IntegerOverflow)?; + if end > bytes_len { + return Err(Error::IntegerOverflow); + } + Ok(start..end) +} + +fn read_u32(bytes: &[u8], offset: usize) -> Result<u32> { + let data = bytes + .get(offset..offset + 4) + .ok_or(Error::IntegerOverflow)?; + let arr: [u8; 4] = data.try_into().map_err(|_| Error::IntegerOverflow)?; + Ok(u32::from_le_bytes(arr)) +} + +fn push_u32(out: &mut Vec<u8>, value: u32) { + out.extend_from_slice(&value.to_le_bytes()); +} + +fn encode_name_field(name: &str) -> Result<[u8; 36]> { + let bytes = name.as_bytes(); + if bytes.contains(&0) { + return Err(Error::NameContainsNul); + } + if bytes.len() > 35 { + return Err(Error::NameTooLong { + got: bytes.len(), + max: 35, + }); + } + + let mut out = [0u8; 36]; + out[..bytes.len()].copy_from_slice(bytes); + Ok(out) +} + +fn entry_name_bytes(raw: &[u8; 36]) -> &[u8] { + let len = raw.iter().position(|&b| b == 0).unwrap_or(raw.len()); + &raw[..len] +} + +fn decode_name(name: &[u8]) -> String { + name.iter().map(|b| char::from(*b)).collect() +} + +fn cmp_name_case_insensitive(a: &[u8], b: &[u8]) -> Ordering { + let mut idx = 0usize; + let min_len = a.len().min(b.len()); + while idx < min_len { + let left = ascii_lower(a[idx]); + let right = ascii_lower(b[idx]); + if left != right { + return left.cmp(&right); + } + idx += 1; + } + a.len().cmp(&b.len()) +} + +fn ascii_lower(value: u8) -> u8 { + if value.is_ascii_uppercase() { + value + 32 + } else { + value + } +} + +fn prefetch_pages(bytes: &[u8]) { + use std::sync::atomic::{compiler_fence, Ordering}; + + let mut cursor = 0usize; + let mut sink = 0u8; + while cursor < bytes.len() { + sink ^= bytes[cursor]; + cursor = cursor.saturating_add(4096); + } + compiler_fence(Ordering::SeqCst); + let _ = sink; +} + +fn write_atomic(path: &Path, content: &[u8]) -> Result<()> { + let file_name = path + .file_name() + .and_then(|name| name.to_str()) + .unwrap_or("archive"); + let parent = path.parent().unwrap_or_else(|| Path::new(".")); + + let mut temp_path = None; + for attempt in 0..128u32 { + let name = format!( + ".{}.tmp.{}.{}.{}", + file_name, + std::process::id(), + unix_time_nanos(), + attempt + ); + let candidate = parent.join(name); + let opened = FsOpenOptions::new() + .create_new(true) + .write(true) + .open(&candidate); + if let Ok(mut file) = opened { + file.write_all(content)?; + file.sync_all()?; + temp_path = Some((candidate, file)); + break; + } + } + + let Some((tmp_path, mut file)) = temp_path else { + return Err(Error::Io(std::io::Error::new( + std::io::ErrorKind::AlreadyExists, + "failed to create temporary file for atomic write", + ))); + }; + + file.flush()?; + drop(file); + + if let Err(err) = replace_file_atomically(&tmp_path, path) { + let _ = fs::remove_file(&tmp_path); + return Err(Error::Io(err)); + } + + Ok(()) +} + +#[cfg(not(windows))] +fn replace_file_atomically(src: &Path, dst: &Path) -> std::io::Result<()> { + fs::rename(src, dst) +} + +#[cfg(windows)] +fn replace_file_atomically(src: &Path, dst: &Path) -> std::io::Result<()> { + use std::iter; + use std::os::windows::ffi::OsStrExt; + use windows_sys::Win32::Storage::FileSystem::{ + MoveFileExW, MOVEFILE_REPLACE_EXISTING, MOVEFILE_WRITE_THROUGH, + }; + + let src_wide: Vec<u16> = src.as_os_str().encode_wide().chain(iter::once(0)).collect(); + let dst_wide: Vec<u16> = dst.as_os_str().encode_wide().chain(iter::once(0)).collect(); + + // Replace destination in one OS call, avoiding remove+rename gaps on Windows. + let ok = unsafe { + MoveFileExW( + src_wide.as_ptr(), + dst_wide.as_ptr(), + MOVEFILE_REPLACE_EXISTING | MOVEFILE_WRITE_THROUGH, + ) + }; + + if ok == 0 { + Err(std::io::Error::last_os_error()) + } else { + Ok(()) + } +} + +fn unix_time_nanos() -> u128 { + match SystemTime::now().duration_since(UNIX_EPOCH) { + Ok(duration) => duration.as_nanos(), + Err(_) => 0, + } +} + +#[cfg(test)] +mod tests; |
