Add S3 encryption

crates/pile-io/Cargo.toml

@@ -0,0 +1,15 @@
+[package]
+name = "pile-io"
+version = { workspace = true }
+rust-version = { workspace = true }
+edition = { workspace = true }
+
+[lints]
+workspace = true
+
+[dependencies]
+tokio = { workspace = true }
+smartstring = { workspace = true }
+aws-sdk-s3 = { workspace = true }
+chacha20poly1305 = { workspace = true }
+binrw = { workspace = true }

crates/pile-io/src/asyncreader.rs

@@ -0,0 +1,75 @@
+use std::io::{Read, Seek, SeekFrom};
+use tokio::runtime::Handle;
+
+//
+// MARK: asyncreader
+//
+
+/// An `async` equivalent of [std::io::Read].
+pub trait AsyncReader: Send {
+	/// Read a chunk of bytes.
+	fn read(
+		&mut self,
+		buf: &mut [u8],
+	) -> impl Future<Output = Result<usize, std::io::Error>> + Send;
+
+	/// Read all remaining bytes into a `Vec`.
+	fn read_to_end(&mut self) -> impl Future<Output = Result<Vec<u8>, std::io::Error>> + Send {
+		async {
+			let mut buf = Vec::new();
+			let mut chunk = vec![0u8; 65536];
+			loop {
+				let n = self.read(&mut chunk).await?;
+				if n == 0 {
+					break;
+				}
+				buf.extend_from_slice(&chunk[..n]);
+			}
+			Ok(buf)
+		}
+	}
+}
+
+/// An `async` equivalent of [std::io::Read] + [std::io::Seek].
+pub trait AsyncSeekReader: AsyncReader {
+	fn seek(&mut self, pos: SeekFrom) -> impl Future<Output = Result<u64, std::io::Error>> + Send;
+}
+
+//
+// MARK: sync bridge
+//
+
+/// Turn an async [Reader] into a sync [Read] + [Seek].
+///
+/// Never use this outside of [tokio::task::spawn_blocking],
+/// the async runtime will deadlock if this struct blocks
+/// the runtime.
+pub struct SyncReadBridge<R: AsyncReader> {
+	inner: R,
+	handle: Handle,
+}
+
+impl<R: AsyncReader> SyncReadBridge<R> {
+	/// Creates a new adapter using a handle to the current runtime.
+	/// Panics if called outside of a tokio context.
+	pub fn new_current(inner: R) -> Self {
+		Self::new(inner, Handle::current())
+	}
+
+	/// Creates a new adapter using a handle to an existing runtime.
+	pub fn new(inner: R, handle: Handle) -> Self {
+		Self { inner, handle }
+	}
+}
+
+impl<R: AsyncReader> Read for SyncReadBridge<R> {
+	fn read(&mut self, buf: &mut [u8]) -> Result<usize, std::io::Error> {
+		self.handle.block_on(self.inner.read(buf))
+	}
+}
+
+impl<R: AsyncReader + AsyncSeekReader> Seek for SyncReadBridge<R> {
+	fn seek(&mut self, pos: SeekFrom) -> Result<u64, std::io::Error> {
+		self.handle.block_on(self.inner.seek(pos))
+	}
+}

crates/pile-io/src/chachareader.rs

@@ -0,0 +1,205 @@
+use std::io::{Read, Seek, SeekFrom};
+
+use binrw::binrw;
+
+use crate::{AsyncReader, AsyncSeekReader};
+
+//
+// MARK: header
+//
+
+/// Serialized size of [`ChaChaHeader`] in bytes: 12 magic + 3×8 config + 8 plaintext_size.
+pub const HEADER_SIZE: usize = 44;
+
+#[binrw]
+#[brw(little, magic = b"PileChaChav1")]
+#[derive(Debug, Clone, Copy)]
+pub struct ChaChaHeader {
+	pub chunk_size: u64,
+	pub nonce_size: u64,
+	pub tag_size: u64,
+	pub plaintext_size: u64,
+}
+
+//
+// MARK: config
+//
+
+#[derive(Debug, Clone, Copy)]
+pub struct ChaChaReaderConfig {
+	pub chunk_size: u64,
+	pub nonce_size: u64,
+	pub tag_size: u64,
+}
+
+impl Default for ChaChaReaderConfig {
+	fn default() -> Self {
+		Self {
+			chunk_size: 1_048_576, // 1MiB
+			nonce_size: 24,
+			tag_size: 16,
+		}
+	}
+}
+
+impl ChaChaReaderConfig {
+	pub(crate) fn enc_chunk_size(&self) -> u64 {
+		self.chunk_size + self.nonce_size + self.tag_size
+	}
+}
+
+impl From<ChaChaHeader> for ChaChaReaderConfig {
+	fn from(h: ChaChaHeader) -> Self {
+		Self {
+			chunk_size: h.chunk_size,
+			nonce_size: h.nonce_size,
+			tag_size: h.tag_size,
+		}
+	}
+}
+
+//
+// MARK: reader
+//
+
+pub struct ChaChaReader<R: Read + Seek> {
+	inner: R,
+	config: ChaChaReaderConfig,
+	data_offset: u64,
+	encryption_key: [u8; 32],
+	cursor: u64,
+	plaintext_size: u64,
+	cached_chunk: Option<(u64, Vec<u8>)>,
+}
+
+impl<R: Read + Seek> ChaChaReader<R> {
+	pub fn new(mut inner: R, encryption_key: [u8; 32]) -> Result<Self, std::io::Error> {
+		use binrw::BinReaderExt;
+
+		inner.seek(SeekFrom::Start(0))?;
+		let header: ChaChaHeader = inner.read_le().map_err(std::io::Error::other)?;
+		let data_offset = inner.stream_position()?;
+
+		Ok(Self {
+			inner,
+			config: header.into(),
+			data_offset,
+			encryption_key,
+			cursor: 0,
+			plaintext_size: header.plaintext_size,
+			cached_chunk: None,
+		})
+	}
+
+	fn fetch_chunk(&mut self, chunk_index: u64) -> Result<(), std::io::Error> {
+		use chacha20poly1305::{KeyInit, XChaCha20Poly1305, XNonce, aead::Aead};
+
+		let enc_start = self.data_offset + chunk_index * self.config.enc_chunk_size();
+		self.inner.seek(SeekFrom::Start(enc_start))?;
+
+		let mut encrypted = vec![0u8; self.config.enc_chunk_size() as usize];
+		let n = self.read_exact_or_eof(&mut encrypted)?;
+		encrypted.truncate(n);
+
+		if encrypted.len() < (self.config.nonce_size + self.config.tag_size) as usize {
+			return Err(std::io::Error::new(
+				std::io::ErrorKind::InvalidData,
+				"encrypted chunk too short",
+			));
+		}
+
+		let (nonce_bytes, ciphertext) = encrypted.split_at(self.config.nonce_size as usize);
+		let nonce = XNonce::from_slice(nonce_bytes);
+		let key = chacha20poly1305::Key::from_slice(&self.encryption_key);
+		let cipher = XChaCha20Poly1305::new(key);
+		let plaintext = cipher.decrypt(nonce, ciphertext).map_err(|_| {
+			std::io::Error::new(std::io::ErrorKind::InvalidData, "decryption failed")
+		})?;
+
+		self.cached_chunk = Some((chunk_index, plaintext));
+		Ok(())
+	}
+
+	fn read_exact_or_eof(&mut self, buf: &mut [u8]) -> Result<usize, std::io::Error> {
+		let mut total = 0;
+		while total < buf.len() {
+			match self.inner.read(&mut buf[total..])? {
+				0 => break,
+				n => total += n,
+			}
+		}
+		Ok(total)
+	}
+}
+
+impl<R: Read + Seek + Send> AsyncReader for ChaChaReader<R> {
+	async fn read(&mut self, buf: &mut [u8]) -> Result<usize, std::io::Error> {
+		let remaining = self.plaintext_size.saturating_sub(self.cursor);
+		if remaining == 0 || buf.is_empty() {
+			return Ok(0);
+		}
+
+		let chunk_index = self.cursor / self.config.chunk_size;
+
+		let need_fetch = match &self.cached_chunk {
+			None => true,
+			Some((idx, _)) => *idx != chunk_index,
+		};
+
+		if need_fetch {
+			self.fetch_chunk(chunk_index)?;
+		}
+
+		#[expect(clippy::unwrap_used)]
+		let (_, chunk_data) = self.cached_chunk.as_ref().unwrap();
+
+		let offset_in_chunk = (self.cursor % self.config.chunk_size) as usize;
+		let available = chunk_data.len() - offset_in_chunk;
+		let to_copy = available.min(buf.len());
+
+		buf[..to_copy].copy_from_slice(&chunk_data[offset_in_chunk..offset_in_chunk + to_copy]);
+		self.cursor += to_copy as u64;
+		Ok(to_copy)
+	}
+}
+
+impl<R: Read + Seek + Send> AsyncSeekReader for ChaChaReader<R> {
+	async fn seek(&mut self, pos: SeekFrom) -> Result<u64, std::io::Error> {
+		match pos {
+			SeekFrom::Start(x) => self.cursor = x.min(self.plaintext_size),
+
+			SeekFrom::Current(x) => {
+				if x < 0 {
+					let abs = x.unsigned_abs();
+					if abs > self.cursor {
+						return Err(std::io::Error::new(
+							std::io::ErrorKind::InvalidInput,
+							"cannot seek past start",
+						));
+					}
+					self.cursor -= abs;
+				} else {
+					self.cursor += x as u64;
+				}
+			}
+
+			SeekFrom::End(x) => {
+				if x < 0 {
+					let abs = x.unsigned_abs();
+					if abs > self.plaintext_size {
+						return Err(std::io::Error::new(
+							std::io::ErrorKind::InvalidInput,
+							"cannot seek past start",
+						));
+					}
+					self.cursor = self.plaintext_size - abs;
+				} else {
+					self.cursor = self.plaintext_size + x as u64;
+				}
+			}
+		}
+
+		self.cursor = self.cursor.min(self.plaintext_size);
+		Ok(self.cursor)
+	}
+}

crates/pile-io/src/chachareader_async.rs

@@ -0,0 +1,163 @@
+use std::io::SeekFrom;
+
+use crate::{AsyncReader, AsyncSeekReader, ChaChaHeader, ChaChaReaderConfig, HEADER_SIZE};
+
+pub struct ChaChaReaderAsync<R: AsyncSeekReader> {
+	inner: R,
+	config: ChaChaReaderConfig,
+	data_offset: u64,
+	encryption_key: [u8; 32],
+	cursor: u64,
+	plaintext_size: u64,
+	cached_chunk: Option<(u64, Vec<u8>)>,
+}
+
+impl<R: AsyncSeekReader> ChaChaReaderAsync<R> {
+	pub async fn new(mut inner: R, encryption_key: [u8; 32]) -> Result<Self, std::io::Error> {
+		use binrw::BinReaderExt;
+		use std::io::Cursor;
+
+		inner.seek(SeekFrom::Start(0)).await?;
+		let mut buf = [0u8; HEADER_SIZE];
+		read_exact(&mut inner, &mut buf).await?;
+		let header: ChaChaHeader = Cursor::new(&buf[..])
+			.read_le()
+			.map_err(std::io::Error::other)?;
+
+		Ok(Self {
+			inner,
+			config: header.into(),
+			data_offset: HEADER_SIZE as u64,
+			encryption_key,
+			cursor: 0,
+			plaintext_size: header.plaintext_size,
+			cached_chunk: None,
+		})
+	}
+
+	async fn fetch_chunk(&mut self, chunk_index: u64) -> Result<(), std::io::Error> {
+		use chacha20poly1305::{KeyInit, XChaCha20Poly1305, XNonce, aead::Aead};
+
+		let enc_start = self.data_offset + chunk_index * self.config.enc_chunk_size();
+		self.inner.seek(SeekFrom::Start(enc_start)).await?;
+
+		let mut encrypted = vec![0u8; self.config.enc_chunk_size() as usize];
+		let n = read_exact_or_eof(&mut self.inner, &mut encrypted).await?;
+		encrypted.truncate(n);
+
+		if encrypted.len() < (self.config.nonce_size + self.config.tag_size) as usize {
+			return Err(std::io::Error::new(
+				std::io::ErrorKind::InvalidData,
+				"encrypted chunk too short",
+			));
+		}
+
+		let (nonce_bytes, ciphertext) = encrypted.split_at(self.config.nonce_size as usize);
+		let nonce = XNonce::from_slice(nonce_bytes);
+		let key = chacha20poly1305::Key::from_slice(&self.encryption_key);
+		let cipher = XChaCha20Poly1305::new(key);
+		let plaintext = cipher.decrypt(nonce, ciphertext).map_err(|_| {
+			std::io::Error::new(std::io::ErrorKind::InvalidData, "decryption failed")
+		})?;
+
+		self.cached_chunk = Some((chunk_index, plaintext));
+		Ok(())
+	}
+}
+
+async fn read_exact<R: AsyncReader>(inner: &mut R, buf: &mut [u8]) -> Result<(), std::io::Error> {
+	let n = read_exact_or_eof(inner, buf).await?;
+	if n < buf.len() {
+		return Err(std::io::Error::new(
+			std::io::ErrorKind::UnexpectedEof,
+			"unexpected EOF reading header",
+		));
+	}
+	Ok(())
+}
+
+async fn read_exact_or_eof<R: AsyncReader>(
+	inner: &mut R,
+	buf: &mut [u8],
+) -> Result<usize, std::io::Error> {
+	let mut total = 0;
+	while total < buf.len() {
+		match inner.read(&mut buf[total..]).await? {
+			0 => break,
+			n => total += n,
+		}
+	}
+	Ok(total)
+}
+
+impl<R: AsyncSeekReader> AsyncReader for ChaChaReaderAsync<R> {
+	async fn read(&mut self, buf: &mut [u8]) -> Result<usize, std::io::Error> {
+		let remaining = self.plaintext_size.saturating_sub(self.cursor);
+		if remaining == 0 || buf.is_empty() {
+			return Ok(0);
+		}
+
+		let chunk_index = self.cursor / self.config.chunk_size;
+
+		let need_fetch = match &self.cached_chunk {
+			None => true,
+			Some((idx, _)) => *idx != chunk_index,
+		};
+
+		if need_fetch {
+			self.fetch_chunk(chunk_index).await?;
+		}
+
+		#[expect(clippy::unwrap_used)]
+		let (_, chunk_data) = self.cached_chunk.as_ref().unwrap();
+
+		let offset_in_chunk = (self.cursor % self.config.chunk_size) as usize;
+		let available = chunk_data.len() - offset_in_chunk;
+		let to_copy = available.min(buf.len());
+
+		buf[..to_copy].copy_from_slice(&chunk_data[offset_in_chunk..offset_in_chunk + to_copy]);
+		self.cursor += to_copy as u64;
+		Ok(to_copy)
+	}
+}
+
+impl<R: AsyncSeekReader> AsyncSeekReader for ChaChaReaderAsync<R> {
+	async fn seek(&mut self, pos: SeekFrom) -> Result<u64, std::io::Error> {
+		match pos {
+			SeekFrom::Start(x) => self.cursor = x.min(self.plaintext_size),
+
+			SeekFrom::Current(x) => {
+				if x < 0 {
+					let abs = x.unsigned_abs();
+					if abs > self.cursor {
+						return Err(std::io::Error::new(
+							std::io::ErrorKind::InvalidInput,
+							"cannot seek past start",
+						));
+					}
+					self.cursor -= abs;
+				} else {
+					self.cursor += x as u64;
+				}
+			}
+
+			SeekFrom::End(x) => {
+				if x < 0 {
+					let abs = x.unsigned_abs();
+					if abs > self.plaintext_size {
+						return Err(std::io::Error::new(
+							std::io::ErrorKind::InvalidInput,
+							"cannot seek past start",
+						));
+					}
+					self.cursor = self.plaintext_size - abs;
+				} else {
+					self.cursor = self.plaintext_size + x as u64;
+				}
+			}
+		}
+
+		self.cursor = self.cursor.min(self.plaintext_size);
+		Ok(self.cursor)
+	}
+}

crates/pile-io/src/chachawriter.rs

@@ -0,0 +1,264 @@
+use std::io::{Seek, SeekFrom, Write};
+
+use crate::{ChaChaHeader, ChaChaReaderConfig};
+
+/// Generate a random 32-byte encryption key suitable for use with [`ChaChaWriter`].
+pub fn generate_key() -> [u8; 32] {
+	use chacha20poly1305::aead::OsRng;
+	use chacha20poly1305::{KeyInit, XChaCha20Poly1305};
+	XChaCha20Poly1305::generate_key(&mut OsRng).into()
+}
+
+pub struct ChaChaWriter<W: Write + Seek> {
+	inner: W,
+	config: ChaChaReaderConfig,
+	encryption_key: [u8; 32],
+	buffer: Vec<u8>,
+	plaintext_bytes_written: u64,
+}
+
+impl<W: Write + Seek> ChaChaWriter<W> {
+	pub fn new(mut inner: W, encryption_key: [u8; 32]) -> Result<Self, std::io::Error> {
+		use binrw::BinWriterExt;
+
+		let config = ChaChaReaderConfig::default();
+		let header = ChaChaHeader {
+			chunk_size: config.chunk_size,
+			nonce_size: config.nonce_size,
+			tag_size: config.tag_size,
+			plaintext_size: 0,
+		};
+		inner.write_le(&header).map_err(std::io::Error::other)?;
+
+		Ok(Self {
+			inner,
+			config,
+			encryption_key,
+			buffer: Vec::new(),
+			plaintext_bytes_written: 0,
+		})
+	}
+
+	/// Encrypt and write any buffered plaintext, patch the header with the
+	/// final `plaintext_size`, then return the inner writer.
+	pub fn finish(mut self) -> Result<W, std::io::Error> {
+		use binrw::BinWriterExt;
+
+		self.flush_buffer()?;
+
+		self.inner.seek(SeekFrom::Start(0))?;
+		let header = ChaChaHeader {
+			chunk_size: self.config.chunk_size,
+			nonce_size: self.config.nonce_size,
+			tag_size: self.config.tag_size,
+			plaintext_size: self.plaintext_bytes_written,
+		};
+		self.inner
+			.write_le(&header)
+			.map_err(std::io::Error::other)?;
+
+		Ok(self.inner)
+	}
+
+	fn encrypt_chunk(&self, plaintext: &[u8]) -> Result<Vec<u8>, std::io::Error> {
+		use chacha20poly1305::{
+			XChaCha20Poly1305,
+			aead::{Aead, AeadCore, KeyInit, OsRng},
+		};
+
+		let nonce = XChaCha20Poly1305::generate_nonce(&mut OsRng);
+		let key = chacha20poly1305::Key::from_slice(&self.encryption_key);
+		let cipher = XChaCha20Poly1305::new(key);
+		let ciphertext = cipher
+			.encrypt(&nonce, plaintext)
+			.map_err(|_| std::io::Error::other("encryption failed"))?;
+
+		let mut output = Vec::with_capacity(nonce.len() + ciphertext.len());
+		output.extend_from_slice(&nonce);
+		output.extend_from_slice(&ciphertext);
+		Ok(output)
+	}
+
+	fn flush_buffer(&mut self) -> Result<(), std::io::Error> {
+		if !self.buffer.is_empty() {
+			let encrypted = self.encrypt_chunk(&self.buffer)?;
+			self.inner.write_all(&encrypted)?;
+			self.buffer.clear();
+		}
+		Ok(())
+	}
+}
+
+impl<W: Write + Seek> Write for ChaChaWriter<W> {
+	fn write(&mut self, buf: &[u8]) -> Result<usize, std::io::Error> {
+		self.buffer.extend_from_slice(buf);
+		self.plaintext_bytes_written += buf.len() as u64;
+
+		let chunk_size = self.config.chunk_size as usize;
+		while self.buffer.len() >= chunk_size {
+			let encrypted = self.encrypt_chunk(&self.buffer[..chunk_size])?;
+			self.inner.write_all(&encrypted)?;
+			self.buffer.drain(..chunk_size);
+		}
+
+		Ok(buf.len())
+	}
+
+	/// Encrypts and flushes any buffered plaintext as a partial chunk.
+	///
+	/// Prefer [`finish`](Self::finish) to retrieve the inner writer after
+	/// all data has been written. Calling `flush` multiple times will produce
+	/// multiple small encrypted chunks for the same partial data.
+	fn flush(&mut self) -> Result<(), std::io::Error> {
+		self.flush_buffer()?;
+		self.inner.flush()
+	}
+}
+
+#[cfg(test)]
+#[expect(clippy::unwrap_used)]
+mod tests {
+	use std::io::{Cursor, SeekFrom, Write};
+
+	use super::ChaChaWriter;
+	use crate::{AsyncReader, AsyncSeekReader, ChaChaReader};
+
+	const KEY: [u8; 32] = [42u8; 32];
+
+	fn encrypt(data: &[u8]) -> Cursor<Vec<u8>> {
+		let mut writer = ChaChaWriter::new(Cursor::new(Vec::new()), KEY).unwrap();
+		writer.write_all(data).unwrap();
+		let mut buf = writer.finish().unwrap();
+		buf.set_position(0);
+		buf
+	}
+
+	async fn decrypt_all(buf: Cursor<Vec<u8>>) -> Vec<u8> {
+		let mut reader = ChaChaReader::new(buf, KEY).unwrap();
+		reader.read_to_end().await.unwrap()
+	}
+
+	#[tokio::test]
+	async fn roundtrip_empty() {
+		let buf = encrypt(&[]);
+		// Header present but no chunks
+		assert!(!buf.get_ref().is_empty());
+		assert!(decrypt_all(buf).await.is_empty());
+	}
+
+	#[tokio::test]
+	async fn roundtrip_small() {
+		let data = b"hello, world!";
+		assert_eq!(decrypt_all(encrypt(data)).await, data);
+	}
+
+	#[tokio::test]
+	async fn roundtrip_exact_chunk() {
+		let data = vec![0xABu8; 65536];
+		assert_eq!(decrypt_all(encrypt(&data)).await, data);
+	}
+
+	#[tokio::test]
+	async fn roundtrip_multi_chunk() {
+		// 2.5 chunks
+		let data: Vec<u8> = (0u8..=255).cycle().take(65536 * 2 + 1000).collect();
+		assert_eq!(decrypt_all(encrypt(&data)).await, data);
+	}
+
+	#[tokio::test]
+	async fn roundtrip_incremental_writes() {
+		// Write one byte at a time
+		let data: Vec<u8> = (0u8..200).collect();
+		let mut writer = ChaChaWriter::new(Cursor::new(Vec::new()), KEY).unwrap();
+		for byte in &data {
+			writer.write_all(&[*byte]).unwrap();
+		}
+		let mut buf = writer.finish().unwrap();
+		buf.set_position(0);
+		assert_eq!(decrypt_all(buf).await, data);
+	}
+
+	#[tokio::test]
+	async fn wrong_key_fails() {
+		let buf = encrypt(b"secret data");
+		let mut reader = ChaChaReader::new(buf, [0u8; 32]).unwrap();
+		assert!(reader.read_to_end().await.is_err());
+	}
+
+	#[tokio::test]
+	async fn header_magic_checked() {
+		// Corrupt the magic bytes — reader should fail
+		let mut buf = encrypt(b"data");
+		buf.get_mut()[0] = 0xFF;
+		buf.set_position(0);
+		assert!(ChaChaReader::new(buf, KEY).is_err());
+	}
+
+	#[tokio::test]
+	async fn seek_from_start() {
+		let data: Vec<u8> = (0u8..100).collect();
+		let mut reader = ChaChaReader::new(encrypt(&data), KEY).unwrap();
+
+		reader.seek(SeekFrom::Start(50)).await.unwrap();
+		let mut buf = [0u8; 10];
+		let mut read = 0;
+		while read < buf.len() {
+			read += reader.read(&mut buf[read..]).await.unwrap();
+		}
+		assert_eq!(buf, data[50..60]);
+	}
+
+	#[tokio::test]
+	async fn seek_from_end() {
+		let data: Vec<u8> = (0u8..100).collect();
+		let mut reader = ChaChaReader::new(encrypt(&data), KEY).unwrap();
+
+		reader.seek(SeekFrom::End(-10)).await.unwrap();
+		assert_eq!(reader.read_to_end().await.unwrap(), &data[90..]);
+	}
+
+	#[tokio::test]
+	async fn seek_across_chunk_boundary() {
+		// Seek to 6 bytes before the end of chunk 0, read 12 bytes spanning into chunk 1
+		let data: Vec<u8> = (0u8..=255).cycle().take(65536 + 500).collect();
+		let mut reader = ChaChaReader::new(encrypt(&data), KEY).unwrap();
+
+		reader.seek(SeekFrom::Start(65530)).await.unwrap();
+		let mut buf = vec![0u8; 12];
+		let mut read = 0;
+		while read < buf.len() {
+			read += reader.read(&mut buf[read..]).await.unwrap();
+		}
+		assert_eq!(buf, data[65530..65542]);
+	}
+
+	#[tokio::test]
+	async fn seek_current() {
+		let data: Vec<u8> = (0u8..=255).cycle().take(200).collect();
+		let mut reader = ChaChaReader::new(encrypt(&data), KEY).unwrap();
+
+		// Read 10, seek back 5, read 5 — should get bytes 5..10
+		let mut first = [0u8; 10];
+		let mut n = 0;
+		while n < first.len() {
+			n += reader.read(&mut first[n..]).await.unwrap();
+		}
+		reader.seek(SeekFrom::Current(-5)).await.unwrap();
+		let mut second = [0u8; 5];
+		n = 0;
+		while n < second.len() {
+			n += reader.read(&mut second[n..]).await.unwrap();
+		}
+		assert_eq!(second, data[5..10]);
+	}
+
+	#[tokio::test]
+	async fn seek_past_end_clamps() {
+		let data = b"hello";
+		let mut reader = ChaChaReader::new(encrypt(data), KEY).unwrap();
+
+		let pos = reader.seek(SeekFrom::Start(9999)).await.unwrap();
+		assert_eq!(pos, data.len() as u64);
+		assert_eq!(reader.read_to_end().await.unwrap(), b"");
+	}
+}

crates/pile-io/src/chachawriter_async.rs

@@ -0,0 +1,95 @@
+use std::io::SeekFrom;
+
+use tokio::io::{AsyncSeek, AsyncSeekExt, AsyncWrite, AsyncWriteExt};
+
+use crate::{ChaChaHeader, ChaChaReaderConfig};
+
+pub struct ChaChaWriterAsync<W: AsyncWrite + AsyncSeek + Unpin + Send> {
+	inner: W,
+	config: ChaChaReaderConfig,
+	encryption_key: [u8; 32],
+	buffer: Vec<u8>,
+	plaintext_bytes_written: u64,
+}
+
+impl<W: AsyncWrite + AsyncSeek + Unpin + Send> ChaChaWriterAsync<W> {
+	pub async fn new(mut inner: W, encryption_key: [u8; 32]) -> Result<Self, std::io::Error> {
+		let config = ChaChaReaderConfig::default();
+		let header_bytes = serialize_header(ChaChaHeader {
+			chunk_size: config.chunk_size,
+			nonce_size: config.nonce_size,
+			tag_size: config.tag_size,
+			plaintext_size: 0,
+		})?;
+		inner.write_all(&header_bytes).await?;
+
+		Ok(Self {
+			inner,
+			config,
+			encryption_key,
+			buffer: Vec::new(),
+			plaintext_bytes_written: 0,
+		})
+	}
+
+	pub async fn write(&mut self, buf: &[u8]) -> Result<(), std::io::Error> {
+		self.buffer.extend_from_slice(buf);
+		self.plaintext_bytes_written += buf.len() as u64;
+
+		let chunk_size = self.config.chunk_size as usize;
+		while self.buffer.len() >= chunk_size {
+			let encrypted = encrypt_chunk(&self.encryption_key, &self.buffer[..chunk_size])?;
+			self.inner.write_all(&encrypted).await?;
+			self.buffer.drain(..chunk_size);
+		}
+
+		Ok(())
+	}
+
+	/// Encrypt and write any buffered plaintext, patch the header with the
+	/// final `plaintext_size`, then return the inner writer.
+	pub async fn finish(mut self) -> Result<W, std::io::Error> {
+		if !self.buffer.is_empty() {
+			let encrypted = encrypt_chunk(&self.encryption_key, &self.buffer)?;
+			self.inner.write_all(&encrypted).await?;
+		}
+
+		self.inner.seek(SeekFrom::Start(0)).await?;
+		let header_bytes = serialize_header(ChaChaHeader {
+			chunk_size: self.config.chunk_size,
+			nonce_size: self.config.nonce_size,
+			tag_size: self.config.tag_size,
+			plaintext_size: self.plaintext_bytes_written,
+		})?;
+		self.inner.write_all(&header_bytes).await?;
+
+		Ok(self.inner)
+	}
+}
+
+fn encrypt_chunk(key: &[u8; 32], plaintext: &[u8]) -> Result<Vec<u8>, std::io::Error> {
+	use chacha20poly1305::{
+		XChaCha20Poly1305,
+		aead::{Aead, AeadCore, KeyInit, OsRng},
+	};
+
+	let nonce = XChaCha20Poly1305::generate_nonce(&mut OsRng);
+	let cipher = XChaCha20Poly1305::new(chacha20poly1305::Key::from_slice(key));
+	let ciphertext = cipher
+		.encrypt(&nonce, plaintext)
+		.map_err(|_| std::io::Error::other("encryption failed"))?;
+
+	let mut output = Vec::with_capacity(nonce.len() + ciphertext.len());
+	output.extend_from_slice(&nonce);
+	output.extend_from_slice(&ciphertext);
+	Ok(output)
+}
+
+fn serialize_header(header: ChaChaHeader) -> Result<Vec<u8>, std::io::Error> {
+	use binrw::BinWriterExt;
+	use std::io::Cursor;
+
+	let mut buf = Cursor::new(Vec::new());
+	buf.write_le(&header).map_err(std::io::Error::other)?;
+	Ok(buf.into_inner())
+}

crates/pile-io/src/lib.rs

@@ -0,0 +1,17 @@
+mod asyncreader;
+pub use asyncreader::*;
+
+mod s3reader;
+pub use s3reader::*;
+
+mod chachareader;
+pub use chachareader::*;
+
+mod chachawriter;
+pub use chachawriter::*;
+
+mod chachareader_async;
+pub use chachareader_async::*;
+
+mod chachawriter_async;
+pub use chachawriter_async::*;

crates/pile-io/src/s3reader.rs

@@ -0,0 +1,88 @@
+use smartstring::{LazyCompact, SmartString};
+use std::{io::SeekFrom, sync::Arc};
+
+use crate::{AsyncReader, AsyncSeekReader};
+
+pub struct S3Reader {
+	pub client: Arc<aws_sdk_s3::Client>,
+	pub bucket: SmartString<LazyCompact>,
+	pub key: SmartString<LazyCompact>,
+	pub cursor: u64,
+	pub size: u64,
+}
+
+impl AsyncReader for S3Reader {
+	async fn read(&mut self, buf: &mut [u8]) -> Result<usize, std::io::Error> {
+		let len_left = self.size.saturating_sub(self.cursor);
+		if len_left == 0 || buf.is_empty() {
+			return Ok(0);
+		}
+
+		let start_byte = self.cursor;
+		let len_to_read = (buf.len() as u64).min(len_left);
+		let end_byte = start_byte + len_to_read - 1;
+
+		let resp = self
+			.client
+			.get_object()
+			.bucket(self.bucket.as_str())
+			.key(self.key.as_str())
+			.range(format!("bytes={start_byte}-{end_byte}"))
+			.send()
+			.await
+			.map_err(std::io::Error::other)?;
+
+		let bytes = resp
+			.body
+			.collect()
+			.await
+			.map(|x| x.into_bytes())
+			.map_err(std::io::Error::other)?;
+
+		let n = bytes.len().min(buf.len());
+		buf[..n].copy_from_slice(&bytes[..n]);
+		self.cursor += n as u64;
+		Ok(n)
+	}
+}
+
+impl AsyncSeekReader for S3Reader {
+	async fn seek(&mut self, pos: SeekFrom) -> Result<u64, std::io::Error> {
+		match pos {
+			SeekFrom::Start(x) => self.cursor = x.min(self.size),
+
+			SeekFrom::Current(x) => {
+				if x < 0 {
+					let abs = x.unsigned_abs();
+					if abs > self.cursor {
+						return Err(std::io::Error::new(
+							std::io::ErrorKind::InvalidInput,
+							"cannot seek past start",
+						));
+					}
+					self.cursor -= abs;
+				} else {
+					self.cursor += x as u64;
+				}
+			}
+
+			std::io::SeekFrom::End(x) => {
+				if x < 0 {
+					let abs = x.unsigned_abs();
+					if abs > self.size {
+						return Err(std::io::Error::new(
+							std::io::ErrorKind::InvalidInput,
+							"cannot seek past start",
+						));
+					}
+					self.cursor = self.size - abs;
+				} else {
+					self.cursor = self.size + x as u64;
+				}
+			}
+		}
+
+		self.cursor = self.cursor.min(self.size);
+		Ok(self.cursor)
+	}
+}