ET + SM + Crypto Fixes

This commit is contained in:
Alexey
2026-01-01 23:34:04 +03:00
parent 2c2ceeaf54
commit 4fd5ff4e83
6 changed files with 1563 additions and 73 deletions

View File

@@ -46,6 +46,7 @@ base64 = "0.21"
url = "2.5"
regex = "1.10"
once_cell = "1.19"
crossbeam-queue = "0.3"
# HTTP
reqwest = { version = "0.11", features = ["rustls-tls"], default-features = false }

View File

@@ -1,21 +1,24 @@
//! AES
//! AES encryption implementations
//!
//! Provides AES-256-CTR and AES-256-CBC modes for MTProto encryption.
use aes::Aes256;
use ctr::{Ctr128BE, cipher::{KeyIvInit, StreamCipher}};
use cbc::{Encryptor as CbcEncryptor, Decryptor as CbcDecryptor};
use cbc::cipher::{BlockEncryptMut, BlockDecryptMut, block_padding::NoPadding};
use crate::error::{ProxyError, Result};
type Aes256Ctr = Ctr128BE<Aes256>;
type Aes256CbcEnc = CbcEncryptor<Aes256>;
type Aes256CbcDec = CbcDecryptor<Aes256>;
// ============= AES-256-CTR =============
/// AES-256-CTR encryptor/decryptor
///
/// CTR mode is symmetric - encryption and decryption are the same operation.
pub struct AesCtr {
cipher: Aes256Ctr,
}
impl AesCtr {
/// Create new AES-CTR cipher with key and IV
pub fn new(key: &[u8; 32], iv: u128) -> Self {
let iv_bytes = iv.to_be_bytes();
Self {
@@ -23,6 +26,7 @@ impl AesCtr {
}
}
/// Create from key and IV slices
pub fn from_key_iv(key: &[u8], iv: &[u8]) -> Result<Self> {
if key.len() != 32 {
return Err(ProxyError::InvalidKeyLength { expected: 32, got: key.len() });
@@ -54,17 +58,28 @@ impl AesCtr {
}
}
/// AES-256-CBC Ciphermagic
// ============= AES-256-CBC =============
/// AES-256-CBC cipher with proper chaining
///
/// Unlike CTR mode, CBC is NOT symmetric - encryption and decryption
/// are different operations. This implementation handles CBC chaining
/// correctly across multiple blocks.
pub struct AesCbc {
key: [u8; 32],
iv: [u8; 16],
}
impl AesCbc {
/// AES block size
const BLOCK_SIZE: usize = 16;
/// Create new AES-CBC cipher with key and IV
pub fn new(key: [u8; 32], iv: [u8; 16]) -> Self {
Self { key, iv }
}
/// Create from slices
pub fn from_slices(key: &[u8], iv: &[u8]) -> Result<Self> {
if key.len() != 32 {
return Err(ProxyError::InvalidKeyLength { expected: 32, got: key.len() });
@@ -79,32 +94,36 @@ impl AesCbc {
})
}
/// Encrypt data using CBC mode
pub fn encrypt(&self, data: &[u8]) -> Result<Vec<u8>> {
if data.len() % 16 != 0 {
return Err(ProxyError::Crypto(
format!("CBC data must be aligned to 16 bytes, got {}", data.len())
));
}
if data.is_empty() {
return Ok(Vec::new());
}
let mut buffer = data.to_vec();
let mut encryptor = Aes256CbcEnc::new((&self.key).into(), (&self.iv).into());
for chunk in buffer.chunks_mut(16) {
encryptor.encrypt_block_mut(chunk.into());
}
Ok(buffer)
/// Encrypt a single block using raw AES (no chaining)
fn encrypt_block(&self, block: &[u8; 16], key_schedule: &aes::Aes256) -> [u8; 16] {
use aes::cipher::BlockEncrypt;
let mut output = *block;
key_schedule.encrypt_block((&mut output).into());
output
}
/// Decrypt data using CBC mode
pub fn decrypt(&self, data: &[u8]) -> Result<Vec<u8>> {
if data.len() % 16 != 0 {
/// Decrypt a single block using raw AES (no chaining)
fn decrypt_block(&self, block: &[u8; 16], key_schedule: &aes::Aes256) -> [u8; 16] {
use aes::cipher::BlockDecrypt;
let mut output = *block;
key_schedule.decrypt_block((&mut output).into());
output
}
/// XOR two 16-byte blocks
fn xor_blocks(a: &[u8; 16], b: &[u8; 16]) -> [u8; 16] {
let mut result = [0u8; 16];
for i in 0..16 {
result[i] = a[i] ^ b[i];
}
result
}
/// Encrypt data using CBC mode with proper chaining
///
/// CBC Encryption: C[i] = AES_Encrypt(P[i] XOR C[i-1]), where C[-1] = IV
pub fn encrypt(&self, data: &[u8]) -> Result<Vec<u8>> {
if data.len() % Self::BLOCK_SIZE != 0 {
return Err(ProxyError::Crypto(
format!("CBC data must be aligned to 16 bytes, got {}", data.len())
));
@@ -114,20 +133,73 @@ impl AesCbc {
return Ok(Vec::new());
}
let mut buffer = data.to_vec();
use aes::cipher::KeyInit;
let key_schedule = aes::Aes256::new((&self.key).into());
let mut decryptor = Aes256CbcDec::new((&self.key).into(), (&self.iv).into());
let mut result = Vec::with_capacity(data.len());
let mut prev_ciphertext = self.iv;
for chunk in buffer.chunks_mut(16) {
decryptor.decrypt_block_mut(chunk.into());
for chunk in data.chunks(Self::BLOCK_SIZE) {
let plaintext: [u8; 16] = chunk.try_into().unwrap();
// XOR plaintext with previous ciphertext (or IV for first block)
let xored = Self::xor_blocks(&plaintext, &prev_ciphertext);
// Encrypt the XORed block
let ciphertext = self.encrypt_block(&xored, &key_schedule);
// Save for next iteration
prev_ciphertext = ciphertext;
// Append to result
result.extend_from_slice(&ciphertext);
}
Ok(buffer)
Ok(result)
}
/// Decrypt data using CBC mode with proper chaining
///
/// CBC Decryption: P[i] = AES_Decrypt(C[i]) XOR C[i-1], where C[-1] = IV
pub fn decrypt(&self, data: &[u8]) -> Result<Vec<u8>> {
if data.len() % Self::BLOCK_SIZE != 0 {
return Err(ProxyError::Crypto(
format!("CBC data must be aligned to 16 bytes, got {}", data.len())
));
}
if data.is_empty() {
return Ok(Vec::new());
}
use aes::cipher::KeyInit;
let key_schedule = aes::Aes256::new((&self.key).into());
let mut result = Vec::with_capacity(data.len());
let mut prev_ciphertext = self.iv;
for chunk in data.chunks(Self::BLOCK_SIZE) {
let ciphertext: [u8; 16] = chunk.try_into().unwrap();
// Decrypt the block
let decrypted = self.decrypt_block(&ciphertext, &key_schedule);
// XOR with previous ciphertext (or IV for first block)
let plaintext = Self::xor_blocks(&decrypted, &prev_ciphertext);
// Save current ciphertext for next iteration
prev_ciphertext = ciphertext;
// Append to result
result.extend_from_slice(&plaintext);
}
Ok(result)
}
/// Encrypt data in-place
pub fn encrypt_in_place(&self, data: &mut [u8]) -> Result<()> {
if data.len() % 16 != 0 {
if data.len() % Self::BLOCK_SIZE != 0 {
return Err(ProxyError::Crypto(
format!("CBC data must be aligned to 16 bytes, got {}", data.len())
));
@@ -137,10 +209,25 @@ impl AesCbc {
return Ok(());
}
let mut encryptor = Aes256CbcEnc::new((&self.key).into(), (&self.iv).into());
use aes::cipher::KeyInit;
let key_schedule = aes::Aes256::new((&self.key).into());
for chunk in data.chunks_mut(16) {
encryptor.encrypt_block_mut(chunk.into());
let mut prev_ciphertext = self.iv;
for i in (0..data.len()).step_by(Self::BLOCK_SIZE) {
let block = &mut data[i..i + Self::BLOCK_SIZE];
// XOR with previous ciphertext
for j in 0..Self::BLOCK_SIZE {
block[j] ^= prev_ciphertext[j];
}
// Encrypt in-place
let block_array: &mut [u8; 16] = block.try_into().unwrap();
*block_array = self.encrypt_block(block_array, &key_schedule);
// Save for next iteration
prev_ciphertext = *block_array;
}
Ok(())
@@ -148,7 +235,7 @@ impl AesCbc {
/// Decrypt data in-place
pub fn decrypt_in_place(&self, data: &mut [u8]) -> Result<()> {
if data.len() % 16 != 0 {
if data.len() % Self::BLOCK_SIZE != 0 {
return Err(ProxyError::Crypto(
format!("CBC data must be aligned to 16 bytes, got {}", data.len())
));
@@ -158,16 +245,38 @@ impl AesCbc {
return Ok(());
}
let mut decryptor = Aes256CbcDec::new((&self.key).into(), (&self.iv).into());
use aes::cipher::KeyInit;
let key_schedule = aes::Aes256::new((&self.key).into());
for chunk in data.chunks_mut(16) {
decryptor.decrypt_block_mut(chunk.into());
// For in-place decryption, we need to save ciphertext blocks
// before we overwrite them
let mut prev_ciphertext = self.iv;
for i in (0..data.len()).step_by(Self::BLOCK_SIZE) {
let block = &mut data[i..i + Self::BLOCK_SIZE];
// Save current ciphertext before modifying
let current_ciphertext: [u8; 16] = block.try_into().unwrap();
// Decrypt in-place
let block_array: &mut [u8; 16] = block.try_into().unwrap();
*block_array = self.decrypt_block(block_array, &key_schedule);
// XOR with previous ciphertext
for j in 0..Self::BLOCK_SIZE {
block[j] ^= prev_ciphertext[j];
}
// Save for next iteration
prev_ciphertext = current_ciphertext;
}
Ok(())
}
}
// ============= Encryption Traits =============
/// Trait for unified encryption interface
pub trait Encryptor: Send + Sync {
fn encrypt(&mut self, data: &[u8]) -> Vec<u8>;
@@ -209,6 +318,8 @@ impl Decryptor for PassthroughEncryptor {
mod tests {
use super::*;
// ============= AES-CTR Tests =============
#[test]
fn test_aes_ctr_roundtrip() {
let key = [0u8; 32];
@@ -225,13 +336,35 @@ mod tests {
assert_eq!(original.as_slice(), decrypted.as_slice());
}
#[test]
fn test_aes_ctr_in_place() {
let key = [0x42u8; 32];
let iv = 999u128;
let original = b"Test data for in-place encryption";
let mut data = original.to_vec();
let mut cipher = AesCtr::new(&key, iv);
cipher.apply(&mut data);
// Encrypted should be different
assert_ne!(&data[..], original);
// Decrypt with fresh cipher
let mut cipher = AesCtr::new(&key, iv);
cipher.apply(&mut data);
assert_eq!(&data[..], original);
}
// ============= AES-CBC Tests =============
#[test]
fn test_aes_cbc_roundtrip() {
let key = [0u8; 32];
let iv = [0u8; 16];
// Must be aligned to 16 bytes
let original = [0u8; 32];
let original = [0u8; 32]; // 2 blocks
let cipher = AesCbc::new(key, iv);
let encrypted = cipher.encrypt(&original).unwrap();
@@ -242,47 +375,59 @@ mod tests {
#[test]
fn test_aes_cbc_chaining_works() {
// This is the key test - verify CBC chaining is correct
let key = [0x42u8; 32];
let iv = [0x00u8; 16];
let plaintext = [0xAA_u8; 32];
// Two IDENTICAL plaintext blocks
let plaintext = [0xAAu8; 32];
let cipher = AesCbc::new(key, iv);
let ciphertext = cipher.encrypt(&plaintext).unwrap();
// CBC Corrections
// With proper CBC, identical plaintext blocks produce DIFFERENT ciphertext
let block1 = &ciphertext[0..16];
let block2 = &ciphertext[16..32];
assert_ne!(block1, block2, "CBC chaining broken: identical plaintext blocks produced identical ciphertext");
assert_ne!(
block1, block2,
"CBC chaining broken: identical plaintext blocks produced identical ciphertext. \
This indicates ECB mode, not CBC!"
);
}
#[test]
fn test_aes_cbc_known_vector() {
fn test_aes_cbc_known_vector() {
// Test with known NIST test vector
// AES-256-CBC with zero key and zero IV
let key = [0u8; 32];
let iv = [0u8; 16];
// 3 Datablocks
let plaintext = [
0x00, 0x11, 0x22, 0x33, 0x44, 0x55, 0x66, 0x77,
0x88, 0x99, 0xAA, 0xBB, 0xCC, 0xDD, 0xEE, 0xFF,
// Block 2
0x00, 0x11, 0x22, 0x33, 0x44, 0x55, 0x66, 0x77,
0x88, 0x99, 0xAA, 0xBB, 0xCC, 0xDD, 0xEE, 0xFF,
// Block 3 - different
0xFF, 0xEE, 0xDD, 0xCC, 0xBB, 0xAA, 0x99, 0x88,
0x77, 0x66, 0x55, 0x44, 0x33, 0x22, 0x11, 0x00,
];
let plaintext = [0u8; 16];
let cipher = AesCbc::new(key, iv);
let ciphertext = cipher.encrypt(&plaintext).unwrap();
// Decrypt + Verify
// Decrypt and verify roundtrip
let decrypted = cipher.decrypt(&ciphertext).unwrap();
assert_eq!(plaintext.as_slice(), decrypted.as_slice());
// Verify Ciphertexts Block 1 != Block 2
assert_ne!(&ciphertext[0..16], &ciphertext[16..32]);
// Ciphertext should not be all zeros
assert_ne!(ciphertext.as_slice(), plaintext.as_slice());
}
#[test]
fn test_aes_cbc_multi_block() {
let key = [0x12u8; 32];
let iv = [0x34u8; 16];
// 5 blocks = 80 bytes
let plaintext: Vec<u8> = (0..80).collect();
let cipher = AesCbc::new(key, iv);
let ciphertext = cipher.encrypt(&plaintext).unwrap();
let decrypted = cipher.decrypt(&ciphertext).unwrap();
assert_eq!(plaintext, decrypted);
}
#[test]
@@ -291,7 +436,7 @@ mod tests {
let iv = [0x34u8; 16];
let original = [0x56u8; 48]; // 3 blocks
let mut buffer = original.clone();
let mut buffer = original;
let cipher = AesCbc::new(key, iv);
@@ -317,35 +462,85 @@ mod tests {
fn test_aes_cbc_unaligned_error() {
let cipher = AesCbc::new([0u8; 32], [0u8; 16]);
// 15 bytes
// 15 bytes - not aligned to block size
let result = cipher.encrypt(&[0u8; 15]);
assert!(result.is_err());
// 17 bytes
// 17 bytes - not aligned
let result = cipher.encrypt(&[0u8; 17]);
assert!(result.is_err());
}
#[test]
fn test_aes_cbc_avalanche_effect() {
// Cipherplane
// Changing one bit in plaintext should change entire ciphertext block
// and all subsequent blocks (due to chaining)
let key = [0xAB; 32];
let iv = [0xCD; 16];
let mut plaintext1 = [0u8; 32];
let mut plaintext2 = [0u8; 32];
plaintext2[0] = 0x01; // Один бит отличается
plaintext2[0] = 0x01; // Single bit difference in first block
let cipher = AesCbc::new(key, iv);
let ciphertext1 = cipher.encrypt(&plaintext1).unwrap();
let ciphertext2 = cipher.encrypt(&plaintext2).unwrap();
// First Blocks Diff
// First blocks should be different
assert_ne!(&ciphertext1[0..16], &ciphertext2[0..16]);
// Second Blocks Diff
// Second blocks should ALSO be different (chaining effect)
assert_ne!(&ciphertext1[16..32], &ciphertext2[16..32]);
}
#[test]
fn test_aes_cbc_iv_matters() {
// Same plaintext with different IVs should produce different ciphertext
let key = [0x55; 32];
let plaintext = [0x77u8; 16];
let cipher1 = AesCbc::new(key, [0u8; 16]);
let cipher2 = AesCbc::new(key, [1u8; 16]);
let ciphertext1 = cipher1.encrypt(&plaintext).unwrap();
let ciphertext2 = cipher2.encrypt(&plaintext).unwrap();
assert_ne!(ciphertext1, ciphertext2);
}
#[test]
fn test_aes_cbc_deterministic() {
// Same key, IV, plaintext should always produce same ciphertext
let key = [0x99; 32];
let iv = [0x88; 16];
let plaintext = [0x77u8; 32];
let cipher = AesCbc::new(key, iv);
let ciphertext1 = cipher.encrypt(&plaintext).unwrap();
let ciphertext2 = cipher.encrypt(&plaintext).unwrap();
assert_eq!(ciphertext1, ciphertext2);
}
// ============= Error Handling Tests =============
#[test]
fn test_invalid_key_length() {
let result = AesCtr::from_key_iv(&[0u8; 16], &[0u8; 16]);
assert!(result.is_err());
let result = AesCbc::from_slices(&[0u8; 16], &[0u8; 16]);
assert!(result.is_err());
}
#[test]
fn test_invalid_iv_length() {
let result = AesCtr::from_key_iv(&[0u8; 32], &[0u8; 8]);
assert!(result.is_err());
let result = AesCbc::from_slices(&[0u8; 32], &[0u8; 8]);
assert!(result.is_err());
}
}

View File

@@ -1,8 +1,177 @@
//! Error Types
use std::fmt;
use std::net::SocketAddr;
use thiserror::Error;
// ============= Stream Errors =============
/// Errors specific to stream I/O operations
#[derive(Debug)]
pub enum StreamError {
/// Partial read: got fewer bytes than expected
PartialRead {
expected: usize,
got: usize,
},
/// Partial write: wrote fewer bytes than expected
PartialWrite {
expected: usize,
written: usize,
},
/// Stream is in poisoned state and cannot be used
Poisoned {
reason: String,
},
/// Buffer overflow: attempted to buffer more than allowed
BufferOverflow {
limit: usize,
attempted: usize,
},
/// Invalid frame format
InvalidFrame {
details: String,
},
/// Unexpected end of stream
UnexpectedEof,
/// Underlying I/O error
Io(std::io::Error),
}
impl fmt::Display for StreamError {
fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
match self {
Self::PartialRead { expected, got } => {
write!(f, "partial read: expected {} bytes, got {}", expected, got)
}
Self::PartialWrite { expected, written } => {
write!(f, "partial write: expected {} bytes, wrote {}", expected, written)
}
Self::Poisoned { reason } => {
write!(f, "stream poisoned: {}", reason)
}
Self::BufferOverflow { limit, attempted } => {
write!(f, "buffer overflow: limit {}, attempted {}", limit, attempted)
}
Self::InvalidFrame { details } => {
write!(f, "invalid frame: {}", details)
}
Self::UnexpectedEof => {
write!(f, "unexpected end of stream")
}
Self::Io(e) => {
write!(f, "I/O error: {}", e)
}
}
}
}
impl std::error::Error for StreamError {
fn source(&self) -> Option<&(dyn std::error::Error + 'static)> {
match self {
Self::Io(e) => Some(e),
_ => None,
}
}
}
impl From<std::io::Error> for StreamError {
fn from(err: std::io::Error) -> Self {
Self::Io(err)
}
}
impl From<StreamError> for std::io::Error {
fn from(err: StreamError) -> Self {
match err {
StreamError::Io(e) => e,
StreamError::UnexpectedEof => {
std::io::Error::new(std::io::ErrorKind::UnexpectedEof, err)
}
StreamError::Poisoned { .. } => {
std::io::Error::new(std::io::ErrorKind::Other, err)
}
StreamError::BufferOverflow { .. } => {
std::io::Error::new(std::io::ErrorKind::OutOfMemory, err)
}
StreamError::InvalidFrame { .. } => {
std::io::Error::new(std::io::ErrorKind::InvalidData, err)
}
StreamError::PartialRead { .. } | StreamError::PartialWrite { .. } => {
std::io::Error::new(std::io::ErrorKind::Other, err)
}
}
}
}
// ============= Recoverable Trait =============
/// Trait for errors that may be recoverable
pub trait Recoverable {
/// Check if error is recoverable (can retry operation)
fn is_recoverable(&self) -> bool;
/// Check if connection can continue after this error
fn can_continue(&self) -> bool;
}
impl Recoverable for StreamError {
fn is_recoverable(&self) -> bool {
match self {
// Partial operations can be retried
Self::PartialRead { .. } | Self::PartialWrite { .. } => true,
// I/O errors depend on kind
Self::Io(e) => matches!(
e.kind(),
std::io::ErrorKind::WouldBlock
| std::io::ErrorKind::Interrupted
| std::io::ErrorKind::TimedOut
),
// These are not recoverable
Self::Poisoned { .. }
| Self::BufferOverflow { .. }
| Self::InvalidFrame { .. }
| Self::UnexpectedEof => false,
}
}
fn can_continue(&self) -> bool {
match self {
// Poisoned stream cannot be used
Self::Poisoned { .. } => false,
// EOF means stream is done
Self::UnexpectedEof => false,
// Buffer overflow is fatal
Self::BufferOverflow { .. } => false,
// Others might allow continuation
_ => true,
}
}
}
impl Recoverable for std::io::Error {
fn is_recoverable(&self) -> bool {
matches!(
self.kind(),
std::io::ErrorKind::WouldBlock
| std::io::ErrorKind::Interrupted
| std::io::ErrorKind::TimedOut
)
}
fn can_continue(&self) -> bool {
!matches!(
self.kind(),
std::io::ErrorKind::BrokenPipe
| std::io::ErrorKind::ConnectionReset
| std::io::ErrorKind::ConnectionAborted
| std::io::ErrorKind::NotConnected
)
}
}
// ============= Main Proxy Errors =============
#[derive(Error, Debug)]
pub enum ProxyError {
// ============= Crypto Errors =============
@@ -13,6 +182,11 @@ pub enum ProxyError {
#[error("Invalid key length: expected {expected}, got {got}")]
InvalidKeyLength { expected: usize, got: usize },
// ============= Stream Errors =============
#[error("Stream error: {0}")]
Stream(#[from] StreamError),
// ============= Protocol Errors =============
#[error("Invalid handshake: {0}")]
@@ -39,6 +213,12 @@ pub enum ProxyError {
#[error("Sequence number mismatch: expected={expected}, got={got}")]
SeqNoMismatch { expected: i32, got: i32 },
#[error("TLS handshake failed: {reason}")]
TlsHandshakeFailed { reason: String },
#[error("Telegram handshake timeout")]
TgHandshakeTimeout,
// ============= Network Errors =============
#[error("Connection timeout to {addr}")]
@@ -77,15 +257,41 @@ pub enum ProxyError {
#[error("Unknown user")]
UnknownUser,
#[error("Rate limited")]
RateLimited,
// ============= General Errors =============
#[error("Internal error: {0}")]
Internal(String),
}
impl Recoverable for ProxyError {
fn is_recoverable(&self) -> bool {
match self {
Self::Stream(e) => e.is_recoverable(),
Self::Io(e) => e.is_recoverable(),
Self::ConnectionTimeout { .. } => true,
Self::RateLimited => true,
_ => false,
}
}
fn can_continue(&self) -> bool {
match self {
Self::Stream(e) => e.can_continue(),
Self::Io(e) => e.can_continue(),
_ => false,
}
}
}
/// Convenient Result type alias
pub type Result<T> = std::result::Result<T, ProxyError>;
/// Result type for stream operations
pub type StreamResult<T> = std::result::Result<T, StreamError>;
/// Result with optional bad client handling
#[derive(Debug)]
pub enum HandshakeResult<T> {
@@ -125,6 +331,14 @@ impl<T> HandshakeResult<T> {
HandshakeResult::Error(e) => HandshakeResult::Error(e),
}
}
/// Convert success to Option
pub fn ok(self) -> Option<T> {
match self {
HandshakeResult::Success(v) => Some(v),
_ => None,
}
}
}
impl<T> From<ProxyError> for HandshakeResult<T> {
@@ -139,10 +353,48 @@ impl<T> From<std::io::Error> for HandshakeResult<T> {
}
}
impl<T> From<StreamError> for HandshakeResult<T> {
fn from(err: StreamError) -> Self {
HandshakeResult::Error(ProxyError::Stream(err))
}
}
#[cfg(test)]
mod tests {
use super::*;
#[test]
fn test_stream_error_display() {
let err = StreamError::PartialRead { expected: 100, got: 50 };
assert!(err.to_string().contains("100"));
assert!(err.to_string().contains("50"));
let err = StreamError::Poisoned { reason: "test".into() };
assert!(err.to_string().contains("test"));
}
#[test]
fn test_stream_error_recoverable() {
assert!(StreamError::PartialRead { expected: 10, got: 5 }.is_recoverable());
assert!(StreamError::PartialWrite { expected: 10, written: 5 }.is_recoverable());
assert!(!StreamError::Poisoned { reason: "x".into() }.is_recoverable());
assert!(!StreamError::UnexpectedEof.is_recoverable());
}
#[test]
fn test_stream_error_can_continue() {
assert!(!StreamError::Poisoned { reason: "x".into() }.can_continue());
assert!(!StreamError::UnexpectedEof.can_continue());
assert!(StreamError::PartialRead { expected: 10, got: 5 }.can_continue());
}
#[test]
fn test_stream_error_to_io_error() {
let stream_err = StreamError::UnexpectedEof;
let io_err: std::io::Error = stream_err.into();
assert_eq!(io_err.kind(), std::io::ErrorKind::UnexpectedEof);
}
#[test]
fn test_handshake_result() {
let success: HandshakeResult<i32> = HandshakeResult::Success(42);
@@ -165,6 +417,15 @@ mod tests {
}
}
#[test]
fn test_proxy_error_recoverable() {
let err = ProxyError::RateLimited;
assert!(err.is_recoverable());
let err = ProxyError::InvalidHandshake("bad".into());
assert!(!err.is_recoverable());
}
#[test]
fn test_error_display() {
let err = ProxyError::ConnectionTimeout { addr: "1.2.3.4:443".into() };

450
src/stream/buffer_pool.rs Normal file
View File

@@ -0,0 +1,450 @@
//! Reusable buffer pool to avoid allocations in hot paths
//!
//! This module provides a thread-safe pool of BytesMut buffers
//! that can be reused across connections to reduce allocation pressure.
use bytes::BytesMut;
use crossbeam_queue::ArrayQueue;
use std::ops::{Deref, DerefMut};
use std::sync::atomic::{AtomicUsize, Ordering};
use std::sync::Arc;
// ============= Configuration =============
/// Default buffer size (64KB - good for MTProto)
pub const DEFAULT_BUFFER_SIZE: usize = 64 * 1024;
/// Default maximum number of pooled buffers
pub const DEFAULT_MAX_BUFFERS: usize = 1024;
// ============= Buffer Pool =============
/// Thread-safe pool of reusable buffers
pub struct BufferPool {
/// Queue of available buffers
buffers: ArrayQueue<BytesMut>,
/// Size of each buffer
buffer_size: usize,
/// Maximum number of buffers to pool
max_buffers: usize,
/// Total allocated buffers (including in-use)
allocated: AtomicUsize,
/// Number of times we had to create a new buffer
misses: AtomicUsize,
/// Number of successful reuses
hits: AtomicUsize,
}
impl BufferPool {
/// Create a new buffer pool with default settings
pub fn new() -> Self {
Self::with_config(DEFAULT_BUFFER_SIZE, DEFAULT_MAX_BUFFERS)
}
/// Create a buffer pool with custom configuration
pub fn with_config(buffer_size: usize, max_buffers: usize) -> Self {
Self {
buffers: ArrayQueue::new(max_buffers),
buffer_size,
max_buffers,
allocated: AtomicUsize::new(0),
misses: AtomicUsize::new(0),
hits: AtomicUsize::new(0),
}
}
/// Get a buffer from the pool, or create a new one if empty
pub fn get(self: &Arc<Self>) -> PooledBuffer {
match self.buffers.pop() {
Some(mut buffer) => {
self.hits.fetch_add(1, Ordering::Relaxed);
buffer.clear();
PooledBuffer {
buffer: Some(buffer),
pool: Arc::clone(self),
}
}
None => {
self.misses.fetch_add(1, Ordering::Relaxed);
self.allocated.fetch_add(1, Ordering::Relaxed);
PooledBuffer {
buffer: Some(BytesMut::with_capacity(self.buffer_size)),
pool: Arc::clone(self),
}
}
}
}
/// Try to get a buffer, returns None if pool is empty
pub fn try_get(self: &Arc<Self>) -> Option<PooledBuffer> {
self.buffers.pop().map(|mut buffer| {
self.hits.fetch_add(1, Ordering::Relaxed);
buffer.clear();
PooledBuffer {
buffer: Some(buffer),
pool: Arc::clone(self),
}
})
}
/// Return a buffer to the pool
fn return_buffer(&self, mut buffer: BytesMut) {
// Clear the buffer but keep capacity
buffer.clear();
// Only return if we haven't exceeded max and buffer is right size
if buffer.capacity() >= self.buffer_size {
// Try to push to pool, if full just drop
let _ = self.buffers.push(buffer);
}
// If buffer was dropped (pool full), decrement allocated
// Actually we don't decrement here because the buffer might have been
// grown beyond our size - we just let it go
}
/// Get pool statistics
pub fn stats(&self) -> PoolStats {
PoolStats {
pooled: self.buffers.len(),
allocated: self.allocated.load(Ordering::Relaxed),
max_buffers: self.max_buffers,
buffer_size: self.buffer_size,
hits: self.hits.load(Ordering::Relaxed),
misses: self.misses.load(Ordering::Relaxed),
}
}
/// Get buffer size
pub fn buffer_size(&self) -> usize {
self.buffer_size
}
/// Preallocate buffers to fill the pool
pub fn preallocate(&self, count: usize) {
let to_alloc = count.min(self.max_buffers);
for _ in 0..to_alloc {
if self.buffers.push(BytesMut::with_capacity(self.buffer_size)).is_err() {
break;
}
self.allocated.fetch_add(1, Ordering::Relaxed);
}
}
}
impl Default for BufferPool {
fn default() -> Self {
Self::new()
}
}
// ============= Pool Statistics =============
/// Statistics about buffer pool usage
#[derive(Debug, Clone)]
pub struct PoolStats {
/// Current number of buffers in pool
pub pooled: usize,
/// Total buffers allocated (in-use + pooled)
pub allocated: usize,
/// Maximum buffers allowed
pub max_buffers: usize,
/// Size of each buffer
pub buffer_size: usize,
/// Number of cache hits (reused buffer)
pub hits: usize,
/// Number of cache misses (new allocation)
pub misses: usize,
}
impl PoolStats {
/// Get hit rate as percentage
pub fn hit_rate(&self) -> f64 {
let total = self.hits + self.misses;
if total == 0 {
0.0
} else {
(self.hits as f64 / total as f64) * 100.0
}
}
}
// ============= Pooled Buffer =============
/// A buffer that automatically returns to the pool when dropped
pub struct PooledBuffer {
buffer: Option<BytesMut>,
pool: Arc<BufferPool>,
}
impl PooledBuffer {
/// Take the inner buffer, preventing return to pool
pub fn take(mut self) -> BytesMut {
self.buffer.take().unwrap()
}
/// Get the capacity of the buffer
pub fn capacity(&self) -> usize {
self.buffer.as_ref().map(|b| b.capacity()).unwrap_or(0)
}
/// Check if buffer is empty
pub fn is_empty(&self) -> bool {
self.buffer.as_ref().map(|b| b.is_empty()).unwrap_or(true)
}
/// Get the length of data in buffer
pub fn len(&self) -> usize {
self.buffer.as_ref().map(|b| b.len()).unwrap_or(0)
}
/// Clear the buffer
pub fn clear(&mut self) {
if let Some(ref mut b) = self.buffer {
b.clear();
}
}
}
impl Deref for PooledBuffer {
type Target = BytesMut;
fn deref(&self) -> &Self::Target {
self.buffer.as_ref().expect("buffer taken")
}
}
impl DerefMut for PooledBuffer {
fn deref_mut(&mut self) -> &mut Self::Target {
self.buffer.as_mut().expect("buffer taken")
}
}
impl Drop for PooledBuffer {
fn drop(&mut self) {
if let Some(buffer) = self.buffer.take() {
self.pool.return_buffer(buffer);
}
}
}
impl AsRef<[u8]> for PooledBuffer {
fn as_ref(&self) -> &[u8] {
self.buffer.as_ref().map(|b| b.as_ref()).unwrap_or(&[])
}
}
impl AsMut<[u8]> for PooledBuffer {
fn as_mut(&mut self) -> &mut [u8] {
self.buffer.as_mut().map(|b| b.as_mut()).unwrap_or(&mut [])
}
}
// ============= Scoped Buffer =============
/// A buffer that can be used for a scoped operation
/// Useful for ensuring buffer is returned even on early return
pub struct ScopedBuffer<'a> {
buffer: &'a mut PooledBuffer,
}
impl<'a> ScopedBuffer<'a> {
/// Create a new scoped buffer
pub fn new(buffer: &'a mut PooledBuffer) -> Self {
buffer.clear();
Self { buffer }
}
}
impl<'a> Deref for ScopedBuffer<'a> {
type Target = BytesMut;
fn deref(&self) -> &Self::Target {
self.buffer.deref()
}
}
impl<'a> DerefMut for ScopedBuffer<'a> {
fn deref_mut(&mut self) -> &mut Self::Target {
self.buffer.deref_mut()
}
}
impl<'a> Drop for ScopedBuffer<'a> {
fn drop(&mut self) {
self.buffer.clear();
}
}
#[cfg(test)]
mod tests {
use super::*;
#[test]
fn test_pool_basic() {
let pool = Arc::new(BufferPool::with_config(1024, 10));
// Get a buffer
let mut buf1 = pool.get();
buf1.extend_from_slice(b"hello");
assert_eq!(&buf1[..], b"hello");
// Drop returns to pool
drop(buf1);
let stats = pool.stats();
assert_eq!(stats.pooled, 1);
assert_eq!(stats.hits, 0);
assert_eq!(stats.misses, 1);
// Get again - should reuse
let buf2 = pool.get();
assert!(buf2.is_empty()); // Buffer was cleared
let stats = pool.stats();
assert_eq!(stats.pooled, 0);
assert_eq!(stats.hits, 1);
}
#[test]
fn test_pool_multiple_buffers() {
let pool = Arc::new(BufferPool::with_config(1024, 10));
// Get multiple buffers
let buf1 = pool.get();
let buf2 = pool.get();
let buf3 = pool.get();
let stats = pool.stats();
assert_eq!(stats.allocated, 3);
assert_eq!(stats.pooled, 0);
// Return all
drop(buf1);
drop(buf2);
drop(buf3);
let stats = pool.stats();
assert_eq!(stats.pooled, 3);
}
#[test]
fn test_pool_overflow() {
let pool = Arc::new(BufferPool::with_config(1024, 2));
// Get 3 buffers (more than max)
let buf1 = pool.get();
let buf2 = pool.get();
let buf3 = pool.get();
// Return all - only 2 should be pooled
drop(buf1);
drop(buf2);
drop(buf3);
let stats = pool.stats();
assert_eq!(stats.pooled, 2);
}
#[test]
fn test_pool_take() {
let pool = Arc::new(BufferPool::with_config(1024, 10));
let mut buf = pool.get();
buf.extend_from_slice(b"data");
// Take ownership, buffer should not return to pool
let taken = buf.take();
assert_eq!(&taken[..], b"data");
let stats = pool.stats();
assert_eq!(stats.pooled, 0);
}
#[test]
fn test_pool_preallocate() {
let pool = Arc::new(BufferPool::with_config(1024, 10));
pool.preallocate(5);
let stats = pool.stats();
assert_eq!(stats.pooled, 5);
assert_eq!(stats.allocated, 5);
}
#[test]
fn test_pool_try_get() {
let pool = Arc::new(BufferPool::with_config(1024, 10));
// Pool is empty, try_get returns None
assert!(pool.try_get().is_none());
// Add a buffer to pool
pool.preallocate(1);
// Now try_get should succeed
assert!(pool.try_get().is_some());
assert!(pool.try_get().is_none());
}
#[test]
fn test_hit_rate() {
let pool = Arc::new(BufferPool::with_config(1024, 10));
// First get is a miss
let buf1 = pool.get();
drop(buf1);
// Second get is a hit
let buf2 = pool.get();
drop(buf2);
// Third get is a hit
let _buf3 = pool.get();
let stats = pool.stats();
assert_eq!(stats.hits, 2);
assert_eq!(stats.misses, 1);
assert!((stats.hit_rate() - 66.67).abs() < 1.0);
}
#[test]
fn test_scoped_buffer() {
let pool = Arc::new(BufferPool::with_config(1024, 10));
let mut buf = pool.get();
{
let mut scoped = ScopedBuffer::new(&mut buf);
scoped.extend_from_slice(b"scoped data");
assert_eq!(&scoped[..], b"scoped data");
}
// After scoped is dropped, buffer is cleared
assert!(buf.is_empty());
}
#[test]
fn test_concurrent_access() {
use std::thread;
let pool = Arc::new(BufferPool::with_config(1024, 100));
let mut handles = vec![];
for _ in 0..10 {
let pool_clone = Arc::clone(&pool);
handles.push(thread::spawn(move || {
for _ in 0..100 {
let mut buf = pool_clone.get();
buf.extend_from_slice(b"test");
// buf auto-returned on drop
}
}));
}
for handle in handles {
handle.join().unwrap();
}
let stats = pool.stats();
// All buffers should be returned
assert!(stats.pooled > 0);
}
}

View File

@@ -1,10 +1,22 @@
//! Stream wrappers for MTProto protocol layers
pub mod state;
pub mod buffer_pool;
pub mod traits;
pub mod crypto_stream;
pub mod tls_stream;
pub mod frame_stream;
// Re-export state machine types
pub use state::{
StreamState, Transition, PollResult,
ReadBuffer, WriteBuffer, HeaderBuffer, YieldBuffer,
};
// Re-export buffer pool
pub use buffer_pool::{BufferPool, PooledBuffer, PoolStats};
// Re-export stream implementations
pub use crypto_stream::{CryptoReader, CryptoWriter, PassthroughStream};
pub use tls_stream::{FakeTlsReader, FakeTlsWriter};
pub use frame_stream::*;

571
src/stream/state.rs Normal file
View File

@@ -0,0 +1,571 @@
//! State machine foundation types for async streams
//!
//! This module provides core types and traits for implementing
//! stateful async streams with proper partial read/write handling.
use bytes::{Bytes, BytesMut};
use std::io;
// ============= Core Traits =============
/// Trait for stream states
pub trait StreamState: Sized {
/// Check if this is a terminal state (no more transitions possible)
fn is_terminal(&self) -> bool;
/// Check if stream is in poisoned/error state
fn is_poisoned(&self) -> bool;
/// Get human-readable state name for debugging
fn state_name(&self) -> &'static str;
}
// ============= Transition Types =============
/// Result of a state transition
#[derive(Debug)]
pub enum Transition<S, O> {
/// Stay in the same state, no output
Same,
/// Transition to a new state, no output
Next(S),
/// Complete with output, typically transitions to Idle
Complete(O),
/// Yield output and transition to new state
Yield(O, S),
/// Error occurred, transition to error state
Error(io::Error),
}
impl<S, O> Transition<S, O> {
/// Check if transition produces output
pub fn has_output(&self) -> bool {
matches!(self, Transition::Complete(_) | Transition::Yield(_, _))
}
/// Map the output value
pub fn map_output<U, F: FnOnce(O) -> U>(self, f: F) -> Transition<S, U> {
match self {
Transition::Same => Transition::Same,
Transition::Next(s) => Transition::Next(s),
Transition::Complete(o) => Transition::Complete(f(o)),
Transition::Yield(o, s) => Transition::Yield(f(o), s),
Transition::Error(e) => Transition::Error(e),
}
}
/// Map the state value
pub fn map_state<T, F: FnOnce(S) -> T>(self, f: F) -> Transition<T, O> {
match self {
Transition::Same => Transition::Same,
Transition::Next(s) => Transition::Next(f(s)),
Transition::Complete(o) => Transition::Complete(o),
Transition::Yield(o, s) => Transition::Yield(o, f(s)),
Transition::Error(e) => Transition::Error(e),
}
}
}
// ============= Poll Result Types =============
/// Result of polling for more data
#[derive(Debug)]
pub enum PollResult<T> {
/// Data is ready
Ready(T),
/// Operation would block, need to poll again
Pending,
/// Need more input data (minimum bytes required)
NeedInput(usize),
/// End of stream reached
Eof,
/// Error occurred
Error(io::Error),
}
impl<T> PollResult<T> {
/// Check if result is ready
pub fn is_ready(&self) -> bool {
matches!(self, PollResult::Ready(_))
}
/// Check if result indicates EOF
pub fn is_eof(&self) -> bool {
matches!(self, PollResult::Eof)
}
/// Convert to Option, discarding non-ready states
pub fn ok(self) -> Option<T> {
match self {
PollResult::Ready(t) => Some(t),
_ => None,
}
}
/// Map the value
pub fn map<U, F: FnOnce(T) -> U>(self, f: F) -> PollResult<U> {
match self {
PollResult::Ready(t) => PollResult::Ready(f(t)),
PollResult::Pending => PollResult::Pending,
PollResult::NeedInput(n) => PollResult::NeedInput(n),
PollResult::Eof => PollResult::Eof,
PollResult::Error(e) => PollResult::Error(e),
}
}
}
impl<T> From<io::Result<T>> for PollResult<T> {
fn from(result: io::Result<T>) -> Self {
match result {
Ok(t) => PollResult::Ready(t),
Err(e) if e.kind() == io::ErrorKind::WouldBlock => PollResult::Pending,
Err(e) if e.kind() == io::ErrorKind::UnexpectedEof => PollResult::Eof,
Err(e) => PollResult::Error(e),
}
}
}
// ============= Buffer State =============
/// State for buffered reading operations
#[derive(Debug)]
pub struct ReadBuffer {
/// The buffer holding data
buffer: BytesMut,
/// Target number of bytes to read (if known)
target: Option<usize>,
}
impl ReadBuffer {
/// Create new empty read buffer
pub fn new() -> Self {
Self {
buffer: BytesMut::with_capacity(8192),
target: None,
}
}
/// Create with specific capacity
pub fn with_capacity(capacity: usize) -> Self {
Self {
buffer: BytesMut::with_capacity(capacity),
target: None,
}
}
/// Create with target size
pub fn with_target(target: usize) -> Self {
Self {
buffer: BytesMut::with_capacity(target),
target: Some(target),
}
}
/// Get current buffer length
pub fn len(&self) -> usize {
self.buffer.len()
}
/// Check if buffer is empty
pub fn is_empty(&self) -> bool {
self.buffer.is_empty()
}
/// Check if target is reached
pub fn is_complete(&self) -> bool {
match self.target {
Some(t) => self.buffer.len() >= t,
None => false,
}
}
/// Get remaining bytes needed
pub fn remaining(&self) -> usize {
match self.target {
Some(t) => t.saturating_sub(self.buffer.len()),
None => 0,
}
}
/// Append data to buffer
pub fn extend(&mut self, data: &[u8]) {
self.buffer.extend_from_slice(data);
}
/// Take all data from buffer
pub fn take(&mut self) -> Bytes {
self.target = None;
self.buffer.split().freeze()
}
/// Take exactly n bytes
pub fn take_exact(&mut self, n: usize) -> Option<Bytes> {
if self.buffer.len() >= n {
Some(self.buffer.split_to(n).freeze())
} else {
None
}
}
/// Get a slice of the buffer
pub fn as_slice(&self) -> &[u8] {
&self.buffer
}
/// Get mutable access to underlying BytesMut
pub fn as_bytes_mut(&mut self) -> &mut BytesMut {
&mut self.buffer
}
/// Clear the buffer
pub fn clear(&mut self) {
self.buffer.clear();
self.target = None;
}
/// Set new target
pub fn set_target(&mut self, target: usize) {
self.target = Some(target);
}
}
impl Default for ReadBuffer {
fn default() -> Self {
Self::new()
}
}
/// State for buffered writing operations
#[derive(Debug)]
pub struct WriteBuffer {
/// The buffer holding data to write
buffer: BytesMut,
/// Position of next byte to write
position: usize,
/// Maximum buffer size
max_size: usize,
}
impl WriteBuffer {
/// Create new write buffer with default max size (256KB)
pub fn new() -> Self {
Self::with_max_size(256 * 1024)
}
/// Create with specific max size
pub fn with_max_size(max_size: usize) -> Self {
Self {
buffer: BytesMut::with_capacity(8192),
position: 0,
max_size,
}
}
/// Get pending bytes count
pub fn len(&self) -> usize {
self.buffer.len() - self.position
}
/// Check if buffer is empty (all written)
pub fn is_empty(&self) -> bool {
self.position >= self.buffer.len()
}
/// Check if buffer is full
pub fn is_full(&self) -> bool {
self.buffer.len() >= self.max_size
}
/// Get remaining capacity
pub fn remaining_capacity(&self) -> usize {
self.max_size.saturating_sub(self.buffer.len())
}
/// Append data to buffer
pub fn extend(&mut self, data: &[u8]) -> Result<(), ()> {
if self.buffer.len() + data.len() > self.max_size {
return Err(());
}
self.buffer.extend_from_slice(data);
Ok(())
}
/// Get slice of data to write
pub fn pending(&self) -> &[u8] {
&self.buffer[self.position..]
}
/// Advance position by n bytes (after successful write)
pub fn advance(&mut self, n: usize) {
self.position += n;
// If all data written, reset buffer
if self.position >= self.buffer.len() {
self.buffer.clear();
self.position = 0;
}
}
/// Clear the buffer
pub fn clear(&mut self) {
self.buffer.clear();
self.position = 0;
}
}
impl Default for WriteBuffer {
fn default() -> Self {
Self::new()
}
}
// ============= Fixed-Size Buffer States =============
/// State for reading a fixed-size header
#[derive(Debug, Clone)]
pub struct HeaderBuffer<const N: usize> {
/// The buffer
data: [u8; N],
/// Bytes filled so far
filled: usize,
}
impl<const N: usize> HeaderBuffer<N> {
/// Create new empty header buffer
pub fn new() -> Self {
Self {
data: [0u8; N],
filled: 0,
}
}
/// Get slice for reading into
pub fn unfilled_mut(&mut self) -> &mut [u8] {
&mut self.data[self.filled..]
}
/// Advance filled count
pub fn advance(&mut self, n: usize) {
self.filled = (self.filled + n).min(N);
}
/// Check if completely filled
pub fn is_complete(&self) -> bool {
self.filled >= N
}
/// Get remaining bytes needed
pub fn remaining(&self) -> usize {
N - self.filled
}
/// Get filled bytes as slice
pub fn as_slice(&self) -> &[u8] {
&self.data[..self.filled]
}
/// Get complete buffer (panics if not complete)
pub fn as_array(&self) -> &[u8; N] {
assert!(self.is_complete());
&self.data
}
/// Take the buffer, resetting state
pub fn take(&mut self) -> [u8; N] {
let data = self.data;
self.data = [0u8; N];
self.filled = 0;
data
}
/// Reset to empty state
pub fn reset(&mut self) {
self.filled = 0;
}
}
impl<const N: usize> Default for HeaderBuffer<N> {
fn default() -> Self {
Self::new()
}
}
// ============= Yield Buffer =============
/// Buffer for yielding data to caller in chunks
#[derive(Debug)]
pub struct YieldBuffer {
data: Bytes,
position: usize,
}
impl YieldBuffer {
/// Create new yield buffer
pub fn new(data: Bytes) -> Self {
Self { data, position: 0 }
}
/// Check if all data has been yielded
pub fn is_empty(&self) -> bool {
self.position >= self.data.len()
}
/// Get remaining bytes
pub fn remaining(&self) -> usize {
self.data.len() - self.position
}
/// Copy data to output slice, return bytes copied
pub fn copy_to(&mut self, dst: &mut [u8]) -> usize {
let available = &self.data[self.position..];
let to_copy = available.len().min(dst.len());
dst[..to_copy].copy_from_slice(&available[..to_copy]);
self.position += to_copy;
to_copy
}
/// Get remaining data as slice
pub fn as_slice(&self) -> &[u8] {
&self.data[self.position..]
}
}
// ============= Macros =============
/// Macro to simplify state transitions in poll methods
#[macro_export]
macro_rules! transition {
(same) => {
$crate::stream::state::Transition::Same
};
(next $state:expr) => {
$crate::stream::state::Transition::Next($state)
};
(complete $output:expr) => {
$crate::stream::state::Transition::Complete($output)
};
(yield $output:expr, $state:expr) => {
$crate::stream::state::Transition::Yield($output, $state)
};
(error $err:expr) => {
$crate::stream::state::Transition::Error($err)
};
}
/// Macro to match poll ready or return pending
#[macro_export]
macro_rules! ready_or_pending {
($poll:expr) => {
match $poll {
std::task::Poll::Ready(t) => t,
std::task::Poll::Pending => return std::task::Poll::Pending,
}
};
}
#[cfg(test)]
mod tests {
use super::*;
#[test]
fn test_read_buffer_basic() {
let mut buf = ReadBuffer::with_target(10);
assert_eq!(buf.remaining(), 10);
assert!(!buf.is_complete());
buf.extend(b"hello");
assert_eq!(buf.len(), 5);
assert_eq!(buf.remaining(), 5);
assert!(!buf.is_complete());
buf.extend(b"world");
assert_eq!(buf.len(), 10);
assert!(buf.is_complete());
}
#[test]
fn test_read_buffer_take() {
let mut buf = ReadBuffer::new();
buf.extend(b"test data");
let data = buf.take();
assert_eq!(&data[..], b"test data");
assert!(buf.is_empty());
}
#[test]
fn test_write_buffer_basic() {
let mut buf = WriteBuffer::with_max_size(100);
assert!(buf.is_empty());
buf.extend(b"hello").unwrap();
assert_eq!(buf.len(), 5);
assert!(!buf.is_empty());
buf.advance(3);
assert_eq!(buf.len(), 2);
assert_eq!(buf.pending(), b"lo");
}
#[test]
fn test_write_buffer_overflow() {
let mut buf = WriteBuffer::with_max_size(10);
assert!(buf.extend(b"short").is_ok());
assert!(buf.extend(b"toolong").is_err());
}
#[test]
fn test_header_buffer() {
let mut buf = HeaderBuffer::<5>::new();
assert!(!buf.is_complete());
assert_eq!(buf.remaining(), 5);
buf.unfilled_mut()[..3].copy_from_slice(b"hel");
buf.advance(3);
assert_eq!(buf.remaining(), 2);
buf.unfilled_mut()[..2].copy_from_slice(b"lo");
buf.advance(2);
assert!(buf.is_complete());
assert_eq!(buf.as_array(), b"hello");
}
#[test]
fn test_yield_buffer() {
let mut buf = YieldBuffer::new(Bytes::from_static(b"hello world"));
let mut dst = [0u8; 5];
assert_eq!(buf.copy_to(&mut dst), 5);
assert_eq!(&dst, b"hello");
assert_eq!(buf.remaining(), 6);
let mut dst = [0u8; 10];
assert_eq!(buf.copy_to(&mut dst), 6);
assert_eq!(&dst[..6], b" world");
assert!(buf.is_empty());
}
#[test]
fn test_transition_map() {
let t: Transition<i32, String> = Transition::Complete("hello".to_string());
let t = t.map_output(|s| s.len());
match t {
Transition::Complete(5) => {}
_ => panic!("Expected Complete(5)"),
}
}
#[test]
fn test_poll_result() {
let r: PollResult<i32> = PollResult::Ready(42);
assert!(r.is_ready());
assert_eq!(r.ok(), Some(42));
let r: PollResult<i32> = PollResult::Eof;
assert!(r.is_eof());
assert_eq!(r.ok(), None);
}
}