mirror-github-speiceio/marketdata-shootout
1
0
Fork 0
mirror of https://github.com/speice-io/marketdata-shootout synced 2024-12-26 08:38:21 -05:00
Code Issues Releases Wiki Activity

Add the SBE runner

Browse Source 
Works for the test dataset, now time to make sure everyone agrees.
This commit is contained in:
Bradlee Speice 2019-09-04 23:47:10 -04:00
parent df5198993b
commit e58861458e
7 changed files with 525 additions and 429 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

29
marketdata.xml
View File

@ -59,23 +59,26 @@
-->
<field name="timestamp" id="3" type="int64"/>
<!--
SBE specifically doesn't have "union" types, so we include both a `trade` and `quote`
here and the tag to identify.
SBE specifically doesn't have "union" types, so we include a type tag, and both
`trade` and `quote` as "optional". This style was chosen to approximate how
Cap'n Proto and Flatbuffers do it:
https://github.com/real-logic/simple-binary-encoding/issues/232
In the future, there are a couple options:
1. Create a "payload header" type and promote "trade" and "quote" to <sbe:message>,
since the SBE message header is already able to distinguish between message types
2. Create a "group" for each message type, and just include no elements of that type.
Uses a `u16` for each message type, instead of padding bytes for the message size.
3. Split up the message components and use "session types" to chain things together,
as in https://polysync.io/blog/session-types-for-hearty-codecs
For now, this message format was chosen to approximate the schemas of Cap'n Proto and Flatbuffers
However, space is actually reserved for *both* `trade` and `quote` in the message;
that is, the payload size is the same if neither, one, or both are filled.
Other ways you can try to emulate unions:
1. Use a "payload header" composite type and promote "trade" and "quote" to <sbe:message>;
SBE can distinguish message types based on the SBE header.
2. Create a "group" for each message type; adds an extra `u16` per type, but overall payload
size goes down because we no longer reserve space that is potentially unused.
3. Split up the message components into individual <composite/> blocks, and write
a state machine (by hand) to chain the blocks. For a better explanation,
see "session types" in:
https://polysync.io/blog/session-types-for-hearty-codecs
-->
<field name="msg_type" id="4" type="MsgType"/>
<field name="trade" id="5" type="Trade"/>
<field name="quote" id="6" type="Quote"/>
<field name="trade" id="5" type="Trade" presence="optional"/>
<field name="quote" id="6" type="Quote" presence="optional"/>
<data name="symbol" id="100" type="varAsciiEncoding"/>
</group>
</sbe:message>

2
src/lib.rs
View File

@ -1,2 +0,0 @@
// This file is needed for tests outside the main source tree to find the project files
pub mod marketdata_capnp;

17
src/main.rs
View File

@ -17,9 +17,11 @@ use crate::iex::IexParser;
pub mod marketdata_capnp;
#[allow(unused_imports)]
pub mod marketdata_generated; // Flatbuffers
pub mod marketdata_sbe;
mod capnp_runner;
mod flatbuffers_runner;
mod sbe_runner;
mod iex;
mod parsers;
@ -81,6 +83,7 @@ fn main() {
}
*/
/*
let mut capnp_writer = capnp_runner::CapnpWriter::new();
for iex_payload in parser {
//let iex_payload = parser.next().unwrap();
@ -93,6 +96,20 @@ fn main() {
while let Ok(_) = capnp_reader.deserialize_unpacked(&mut read_buf, &mut summarizer) {
parsed_msgs += 1;
}
*/
let mut sbe_writer = sbe_runner::SBEWriter::new();
for iex_payload in parser {
//let iex_payload = parser.next().unwrap();
sbe_writer.serialize(&iex_payload, &mut output_buf);
}
let sbe_reader = sbe_runner::SBEReader::new();
let mut read_buf = StreamVec::new(output_buf);
let mut parsed_msgs: u64 = 0;
while let Ok(_) = sbe_reader.deserialize(&mut read_buf, &mut summarizer) {
parsed_msgs += 1;
}
dbg!(parsed_msgs);
dbg!(summarizer);

20
src/marketdata_capnp.rs
View File

@ -144,17 +144,17 @@ pub mod multi_message {
impl Pipeline {
}
mod _private {
use capnp::private::layout;
use capnp::private::layout;
pub const STRUCT_SIZE: layout::StructSize = layout::StructSize { data: 1, pointers: 1 };
pub const STRUCT_SIZE: layout::StructSize = layout::StructSize { data: 1, pointers: 1 };
pub const TYPE_ID: u64 = 0xd13b_1bd4_36e1_ca9f;
}
}
pub mod message {
pub use self::Which::{Quote, Trade};
pub use self::Which::{Quote, Trade};
#[derive(Copy, Clone)]
#[derive(Copy, Clone)]
pub struct Owned;
impl <'a> ::capnp::traits::Owned<'a> for Owned { type Reader = Reader<'a>; type Builder = Builder<'a>; }
impl <'a> ::capnp::traits::OwnedStruct<'a> for Owned { type Reader = Reader<'a>; type Builder = Builder<'a>; }
@ -362,9 +362,9 @@ pub mod message {
impl Pipeline {
}
mod _private {
use capnp::private::layout;
use capnp::private::layout;
pub const STRUCT_SIZE: layout::StructSize = layout::StructSize { data: 2, pointers: 2 };
pub const STRUCT_SIZE: layout::StructSize = layout::StructSize { data: 2, pointers: 2 };
pub const TYPE_ID: u64 = 0x91d7_2965_3a3d_4be4;
}
pub enum Which<A0,A1> {
@ -506,9 +506,9 @@ pub mod trade {
impl Pipeline {
}
mod _private {
use capnp::private::layout;
use capnp::private::layout;
pub const STRUCT_SIZE: layout::StructSize = layout::StructSize { data: 2, pointers: 0 };
pub const STRUCT_SIZE: layout::StructSize = layout::StructSize { data: 2, pointers: 0 };
pub const TYPE_ID: u64 = 0xd29e_10bd_4e5f_c241;
}
}
@ -668,9 +668,9 @@ pub mod level_update {
impl Pipeline {
}
mod _private {
use capnp::private::layout;
use capnp::private::layout;
pub const STRUCT_SIZE: layout::StructSize = layout::StructSize { data: 2, pointers: 0 };
pub const STRUCT_SIZE: layout::StructSize = layout::StructSize { data: 2, pointers: 0 };
pub const TYPE_ID: u64 = 0xe664_c3b5_6628_c453;
}
}

692
src/marketdata_sbe.rs
View File

@ -9,13 +9,13 @@ extern crate core;
/// Errors that may occur during the course of encoding or decoding.
#[derive(Debug)]
pub enum CodecErr {
/// Too few bytes in the byte-slice to read or write the data structure relevant
/// to the current state of the codec
NotEnoughBytes,
/// Too few bytes in the byte-slice to read or write the data structure relevant
/// to the current state of the codec
NotEnoughBytes,
/// Groups and vardata are constrained by the numeric type chosen to represent their
/// length as well as optional maxima imposed by the schema
SliceIsLongerThanAllowedBySchema,
/// Groups and vardata are constrained by the numeric type chosen to represent their
/// length as well as optional maxima imposed by the schema
SliceIsLongerThanAllowedBySchema,
}
pub type CodecResult<T> = core::result::Result<T, CodecErr>;
@ -23,208 +23,206 @@ pub type CodecResult<T> = core::result::Result<T, CodecErr>;
/// Scratch Decoder Data Wrapper - codec internal use only
#[derive(Debug)]
pub struct ScratchDecoderData<'d> {
data: &'d [u8],
pos: usize,
data: &'d [u8],
pos: usize,
}
impl<'d> ScratchDecoderData<'d> {
/// Create a struct reference overlaid atop the data buffer
/// such that the struct's contents directly reflect the buffer.
/// Advances the `pos` index by the size of the struct in bytes.
#[inline]
fn read_type<T>(&mut self, num_bytes: usize) -> CodecResult<&'d T> {
let end = self.pos + num_bytes;
if end <= self.data.len() {
let s = self.data[self.pos..end].as_ptr() as *mut T;
let v: &'d T = unsafe { &*s };
self.pos = end;
Ok(v)
} else {
Err(CodecErr::NotEnoughBytes)
}
/// Create a struct reference overlaid atop the data buffer
/// such that the struct's contents directly reflect the buffer.
/// Advances the `pos` index by the size of the struct in bytes.
#[inline]
fn read_type<T>(&mut self, num_bytes: usize) -> CodecResult<&'d T> {
let end = self.pos + num_bytes;
if end <= self.data.len() {
let s = self.data[self.pos..end].as_ptr() as *mut T;
let v: &'d T = unsafe { &*s };
self.pos = end;
Ok(v)
} else {
Err(CodecErr::NotEnoughBytes)
}
}
/// Advances the `pos` index by a set number of bytes.
#[inline]
fn skip_bytes(&mut self, num_bytes: usize) -> CodecResult<()> {
let end = self.pos + num_bytes;
if end <= self.data.len() {
self.pos = end;
Ok(())
} else {
Err(CodecErr::NotEnoughBytes)
}
/// Advances the `pos` index by a set number of bytes.
#[inline]
fn skip_bytes(&mut self, num_bytes: usize) -> CodecResult<()> {
let end = self.pos + num_bytes;
if end <= self.data.len() {
self.pos = end;
Ok(())
} else {
Err(CodecErr::NotEnoughBytes)
}
}
/// Create a slice reference overlaid atop the data buffer
/// such that the slice's members' contents directly reflect the buffer.
/// Advances the `pos` index by the size of the slice contents in bytes.
#[inline]
fn read_slice<T>(&mut self, count: usize, bytes_per_item: usize) -> CodecResult<&'d [T]> {
let num_bytes = bytes_per_item * count;
let end = self.pos + num_bytes;
if end <= self.data.len() {
let v: &'d [T] = unsafe {
core::slice::from_raw_parts(self.data[self.pos..end].as_ptr() as *const T, count)
};
self.pos = end;
Ok(v)
} else {
Err(CodecErr::NotEnoughBytes)
}
/// Create a slice reference overlaid atop the data buffer
/// such that the slice's members' contents directly reflect the buffer.
/// Advances the `pos` index by the size of the slice contents in bytes.
#[inline]
fn read_slice<T>(&mut self, count: usize, bytes_per_item: usize) -> CodecResult<&'d [T]> {
let num_bytes = bytes_per_item * count;
let end = self.pos + num_bytes;
if end <= self.data.len() {
let v: &'d [T] = unsafe {
core::slice::from_raw_parts(self.data[self.pos..end].as_ptr() as *const T, count)
};
self.pos = end;
Ok(v)
} else {
Err(CodecErr::NotEnoughBytes)
}
}
}
/// Scratch Encoder Data Wrapper - codec internal use only
#[derive(Debug)]
pub struct ScratchEncoderData<'d> {
data: &'d mut [u8],
pos: usize,
data: &'d mut [u8],
pos: usize,
}
impl<'d> ScratchEncoderData<'d> {
/// Copy the bytes of a value into the data buffer
/// Advances the `pos` index to after the newly-written bytes.
#[inline]
fn write_type<T>(&mut self, t: &T, num_bytes: usize) -> CodecResult<()> {
let end = self.pos + num_bytes;
if end <= self.data.len() {
let source_bytes: &[u8] = unsafe {
core::slice::from_raw_parts(t as *const T as *const u8, num_bytes)
};
(&mut self.data[self.pos..end]).copy_from_slice(source_bytes);
self.pos = end;
Ok(())
} else {
Err(CodecErr::NotEnoughBytes)
}
/// Copy the bytes of a value into the data buffer
/// Advances the `pos` index to after the newly-written bytes.
#[inline]
fn write_type<T>(&mut self, t: &T, num_bytes: usize) -> CodecResult<()> {
let end = self.pos + num_bytes;
if end <= self.data.len() {
let source_bytes: &[u8] = unsafe {
core::slice::from_raw_parts(t as *const T as *const u8, num_bytes)
};
(&mut self.data[self.pos..end]).copy_from_slice(source_bytes);
self.pos = end;
Ok(())
} else {
Err(CodecErr::NotEnoughBytes)
}
}
/// Advances the `pos` index by a set number of bytes.
#[inline]
fn skip_bytes(&mut self, num_bytes: usize) -> CodecResult<()> {
let end = self.pos + num_bytes;
if end <= self.data.len() {
self.pos = end;
Ok(())
} else {
Err(CodecErr::NotEnoughBytes)
}
/// Advances the `pos` index by a set number of bytes.
#[inline]
fn skip_bytes(&mut self, num_bytes: usize) -> CodecResult<()> {
let end = self.pos + num_bytes;
if end <= self.data.len() {
self.pos = end;
Ok(())
} else {
Err(CodecErr::NotEnoughBytes)
}
}
/// Create a struct reference overlaid atop the data buffer
/// such that changes to the struct directly edit the buffer.
/// Note that the initial content of the struct's fields may be garbage.
/// Advances the `pos` index to after the newly-written bytes.
#[inline]
fn writable_overlay<T>(&mut self, num_bytes: usize) -> CodecResult<&'d mut T> {
let end = self.pos + num_bytes;
if end <= self.data.len() {
let v: &'d mut T = unsafe {
let s = self.data.as_ptr().offset(self.pos as isize) as *mut T;
&mut *s
};
self.pos = end;
Ok(v)
} else {
Err(CodecErr::NotEnoughBytes)
}
/// Create a struct reference overlaid atop the data buffer
/// such that changes to the struct directly edit the buffer.
/// Note that the initial content of the struct's fields may be garbage.
/// Advances the `pos` index to after the newly-written bytes.
#[inline]
fn writable_overlay<T>(&mut self, num_bytes: usize) -> CodecResult<&'d mut T> {
let end = self.pos + num_bytes;
if end <= self.data.len() {
let v: &'d mut T = unsafe {
let s = self.data.as_ptr().offset(self.pos as isize) as *mut T;
&mut *s
};
self.pos = end;
Ok(v)
} else {
Err(CodecErr::NotEnoughBytes)
}
}
/// Copy the bytes of a value into the data buffer at a specific position
/// Does **not** alter the `pos` index.
#[inline]
fn write_at_position<T>(&mut self, position: usize, t: &T, num_bytes: usize) -> CodecResult<()> {
let end = position + num_bytes;
if end <= self.data.len() {
let source_bytes: &[u8] = unsafe {
core::slice::from_raw_parts(t as *const T as *const u8, num_bytes)
};
(&mut self.data[position..end]).copy_from_slice(source_bytes);
Ok(())
} else {
Err(CodecErr::NotEnoughBytes)
}
/// Copy the bytes of a value into the data buffer at a specific position
/// Does **not** alter the `pos` index.
#[inline]
fn write_at_position<T>(&mut self, position: usize, t: &T, num_bytes: usize) -> CodecResult<()> {
let end = position + num_bytes;
if end <= self.data.len() {
let source_bytes: &[u8] = unsafe {
core::slice::from_raw_parts(t as *const T as *const u8, num_bytes)
};
(&mut self.data[position..end]).copy_from_slice(source_bytes);
Ok(())
} else {
Err(CodecErr::NotEnoughBytes)
}
/// Create a mutable slice overlaid atop the data buffer directly
/// such that changes to the slice contents directly edit the buffer
/// Note that the initial content of the slice's members' fields may be garbage.
/// Advances the `pos` index to after the region representing the slice.
#[inline]
fn writable_slice<T>(&mut self, count: usize, bytes_per_item: usize) -> CodecResult<&'d mut [T]> {
let end = self.pos + (count * bytes_per_item);
if end <= self.data.len() {
let v: &'d mut [T] = unsafe {
core::slice::from_raw_parts_mut(self.data[self.pos..end].as_mut_ptr() as *mut T, count)
};
self.pos = end;
Ok(v)
} else {
Err(CodecErr::NotEnoughBytes)
}
}
/// Create a mutable slice overlaid atop the data buffer directly
/// such that changes to the slice contents directly edit the buffer
/// Note that the initial content of the slice's members' fields may be garbage.
/// Advances the `pos` index to after the region representing the slice.
#[inline]
fn writable_slice<T>(&mut self, count: usize, bytes_per_item: usize) -> CodecResult<&'d mut [T]> {
let end = self.pos + (count * bytes_per_item);
if end <= self.data.len() {
let v: &'d mut [T] = unsafe {
core::slice::from_raw_parts_mut(self.data[self.pos..end].as_mut_ptr() as *mut T, count)
};
self.pos = end;
Ok(v)
} else {
Err(CodecErr::NotEnoughBytes)
}
}
/// Copy the raw bytes of a slice's contents into the data buffer
/// Does **not** encode the length of the slice explicitly into the buffer.
/// Advances the `pos` index to after the newly-written slice bytes.
#[inline]
fn write_slice_without_count<T>(&mut self, t: &[T], bytes_per_item: usize) -> CodecResult<()> {
let content_bytes_size = bytes_per_item * t.len();
let end = self.pos + content_bytes_size;
if end <= self.data.len() {
let source_bytes: &[u8] = unsafe {
core::slice::from_raw_parts(t.as_ptr() as *const u8, content_bytes_size)
};
(&mut self.data[self.pos..end]).copy_from_slice(source_bytes);
self.pos = end;
Ok(())
} else {
Err(CodecErr::NotEnoughBytes)
}
/// Copy the raw bytes of a slice's contents into the data buffer
/// Does **not** encode the length of the slice explicitly into the buffer.
/// Advances the `pos` index to after the newly-written slice bytes.
#[inline]
fn write_slice_without_count<T>(&mut self, t: &[T], bytes_per_item: usize) -> CodecResult<()> {
let content_bytes_size = bytes_per_item * t.len();
let end = self.pos + content_bytes_size;
if end <= self.data.len() {
let source_bytes: &[u8] = unsafe {
core::slice::from_raw_parts(t.as_ptr() as *const u8, content_bytes_size)
};
(&mut self.data[self.pos..end]).copy_from_slice(source_bytes);
self.pos = end;
Ok(())
} else {
Err(CodecErr::NotEnoughBytes)
}
}
}
/// Convenience Either enum
#[derive(Copy, Clone, PartialEq, Eq, PartialOrd, Ord, Hash, Debug)]
pub enum Either<L, R> {
Left(L),
Right(R),
Left(L),
Right(R)
}
/// Enum Side
#[derive(Clone, Copy, Debug, PartialEq, Eq, PartialOrd, Ord, Hash)]
#[repr(u8)]
pub enum Side {
Buy = 0u8,
Sell = 1u8,
NullVal = 255u8,
Buy = 0u8,
Sell = 1u8,
NullVal = 255u8,
}
impl Default for Side {
fn default() -> Self { Side::NullVal }
fn default() -> Self { Side::NullVal }
}
/// Enum MsgType
#[derive(Clone, Copy, Debug, PartialEq, Eq, PartialOrd, Ord, Hash)]
#[repr(u8)]
pub enum MsgType {
Trade = 0u8,
Quote = 1u8,
NullVal = 255u8,
Trade = 0u8,
Quote = 1u8,
NullVal = 255u8,
}
impl Default for MsgType {
fn default() -> Self { MsgType::NullVal }
fn default() -> Self { MsgType::NullVal }
}
/// Quote
#[repr(C, packed)]
#[derive(Default)]
pub struct Quote {
pub price: u64,
pub size: u32,
pub flags: u8,
pub side: Side,
pub price: u64,
pub size: u32,
pub flags: u8,
pub side: Side,
}
impl Quote {}
@ -233,8 +231,8 @@ impl Quote {}
#[repr(C, packed)]
#[derive(Default)]
pub struct Trade {
pub price: u64,
pub size: u32,
pub price: u64,
pub size: u32,
}
impl Trade {}
@ -243,10 +241,10 @@ impl Trade {}
#[repr(C, packed)]
#[derive(Default)]
pub struct MessageHeader {
pub block_length: u16,
pub template_id: u16,
pub schema_id: u16,
pub version: u16,
pub block_length: u16,
pub template_id: u16,
pub schema_id: u16,
pub version: u16,
}
impl MessageHeader {}
@ -255,8 +253,8 @@ impl MessageHeader {}
#[repr(C, packed)]
#[derive(Default)]
pub struct GroupSizeEncoding {
pub block_length: u16,
pub num_in_group: u16,
pub block_length: u16,
pub num_in_group: u16,
}
impl GroupSizeEncoding {}
@ -265,24 +263,24 @@ impl GroupSizeEncoding {}
#[repr(C, packed)]
#[derive(Default)]
pub struct VarAsciiEncoding {
pub length: u32,
pub var_data: u8,
pub length: u32,
pub var_data: u8,
}
impl VarAsciiEncoding {}
/// MessageHeader Decoder entry point
pub fn start_decoding_message_header<'d>(data: &'d [u8]) -> CodecResult<(&'d MessageHeader, ScratchDecoderData<'d>)> {
let mut scratch = ScratchDecoderData { data: data, pos: 0 };
let v = scratch.read_type::<MessageHeader>(8)?;
Ok((v, scratch))
let mut scratch = ScratchDecoderData { data: data, pos: 0 };
let v = scratch.read_type::<MessageHeader>(8)?;
Ok((v, scratch))
}
/// MultiMessage Fixed-size Fields (8 bytes)
#[repr(C, packed)]
#[derive(Default)]
pub struct MultiMessageFields {
pub sequence_number: u64,
pub sequence_number: u64,
}
@ -291,293 +289,279 @@ impl MultiMessageFields {}
/// MultiMessage specific Message Header
#[repr(C, packed)]
pub struct MultiMessageMessageHeader {
pub message_header: MessageHeader
pub message_header: MessageHeader
}
impl MultiMessageMessageHeader {
pub const BLOCK_LENGTH: u16 = 8;
pub const TEMPLATE_ID: u16 = 1;
pub const SCHEMA_ID: u16 = 1;
pub const VERSION: u16 = 0;
pub const BLOCK_LENGTH: u16 = 8;
pub const TEMPLATE_ID: u16 = 1;
pub const SCHEMA_ID: u16 = 1;
pub const VERSION: u16 = 0;
}
impl Default for MultiMessageMessageHeader {
fn default() -> MultiMessageMessageHeader {
MultiMessageMessageHeader {
message_header: MessageHeader {
block_length: 8u16,
template_id: 1u16,
schema_id: 1u16,
version: 0u16,
}
}
fn default() -> MultiMessageMessageHeader {
MultiMessageMessageHeader {
message_header: MessageHeader {
block_length: 8u16,
template_id: 1u16,
schema_id: 1u16,
version: 0u16,
}
}
}
}
/// Group fixed-field member representations
#[repr(C, packed)]
#[derive(Default)]
pub struct MultiMessageMessagesMember {
pub timestamp: i64,
pub msg_type: MsgType,
pub trade: Trade,
pub quote: Quote,
pub timestamp: i64,
pub msg_type: MsgType,
pub trade: Trade,
pub quote: Quote,
}
impl MultiMessageMessagesMember {}
/// MultiMessageDecoderDone
pub struct MultiMessageDecoderDone<'d> {
scratch: ScratchDecoderData<'d>,
scratch: ScratchDecoderData<'d>,
}
impl<'d> MultiMessageDecoderDone<'d> {
/// Returns the number of bytes decoded
pub fn unwrap(self) -> usize {
self.scratch.pos
}
/// Returns the number of bytes decoded
pub fn unwrap(self) -> usize {
self.scratch.pos
}
pub fn wrap(scratch: ScratchDecoderData<'d>) -> MultiMessageDecoderDone<'d> {
MultiMessageDecoderDone { scratch: scratch }
}
pub fn wrap(scratch: ScratchDecoderData<'d>) -> MultiMessageDecoderDone<'d> {
MultiMessageDecoderDone { scratch: scratch }
}
}
/// symbol variable-length data
pub struct MultiMessageMessagesSymbolDecoder<'d> {
parent: MultiMessageMessagesMemberDecoder<'d>,
parent: MultiMessageMessagesMemberDecoder<'d>,
}
impl<'d> MultiMessageMessagesSymbolDecoder<'d> {
fn wrap(parent: MultiMessageMessagesMemberDecoder<'d>) -> Self {
MultiMessageMessagesSymbolDecoder { parent: parent }
}
pub fn symbol(mut self) -> CodecResult<(&'d [u8], Either<MultiMessageMessagesMemberDecoder<'d>, MultiMessageDecoderDone<'d>>)> {
let count = *self.parent.scratch.read_type::<u32>(4)?;
Ok((self.parent.scratch.read_slice::<u8>(count as usize, 1)?, self.parent.after_member()))
}
fn wrap(parent: MultiMessageMessagesMemberDecoder<'d>) -> Self {
MultiMessageMessagesSymbolDecoder { parent: parent }
}
pub fn symbol(mut self) -> CodecResult<(&'d [u8], Either<MultiMessageMessagesMemberDecoder<'d>, MultiMessageDecoderDone<'d>>)> {
let count = *self.parent.scratch.read_type::<u32>(4)?;
Ok((self.parent.scratch.read_slice::<u8>(count as usize, 1)?, self.parent.after_member()))
}
}
/// MultiMessageMessages Decoder for fields and header
pub struct MultiMessageMessagesMemberDecoder<'d> {
scratch: ScratchDecoderData<'d>,
max_index: u16,
index: u16,
scratch: ScratchDecoderData<'d>,
max_index: u16,
index: u16,
}
impl<'d> MultiMessageMessagesMemberDecoder<'d> {
fn new(scratch: ScratchDecoderData<'d>, count: u16) -> Self {
assert!(count > 0u16);
MultiMessageMessagesMemberDecoder {
scratch: scratch,
max_index: count - 1,
index: 0,
}
fn new(scratch: ScratchDecoderData<'d>, count: u16) -> Self {
assert!(count > 0u16);
MultiMessageMessagesMemberDecoder {
scratch: scratch,
max_index: count - 1,
index: 0,
}
}
pub fn next_messages_member(mut self) -> CodecResult<(&'d MultiMessageMessagesMember, MultiMessageMessagesSymbolDecoder<'d>)> {
let v = self.scratch.read_type::<MultiMessageMessagesMember>(35)?;
self.index += 1;
Ok((v, MultiMessageMessagesSymbolDecoder::wrap(self)))
}
#[inline]
fn after_member(self) -> Either<MultiMessageMessagesMemberDecoder<'d>, MultiMessageDecoderDone<'d>> {
if self.index <= self.max_index {
Either::Left(self)
} else {
Either::Right(MultiMessageDecoderDone::wrap(self.scratch))
}
pub fn next_messages_member(mut self) -> CodecResult<(&'d MultiMessageMessagesMember, MultiMessageMessagesSymbolDecoder<'d>)> {
let v = self.scratch.read_type::<MultiMessageMessagesMember>(35)?;
self.index += 1;
Ok((v, MultiMessageMessagesSymbolDecoder::wrap(self)))
}
#[inline]
fn after_member(self) -> Either<MultiMessageMessagesMemberDecoder<'d>, MultiMessageDecoderDone<'d>> {
if self.index <= self.max_index {
Either::Left(self)
} else {
Either::Right(MultiMessageDecoderDone::wrap(self.scratch))
}
}
}
pub struct MultiMessageMessagesHeaderDecoder<'d> {
scratch: ScratchDecoderData<'d>,
scratch: ScratchDecoderData<'d>,
}
impl<'d> MultiMessageMessagesHeaderDecoder<'d> {
fn wrap(scratch: ScratchDecoderData<'d>) -> Self {
MultiMessageMessagesHeaderDecoder { scratch: scratch }
}
pub fn messages_individually(mut self) -> CodecResult<Either<MultiMessageMessagesMemberDecoder<'d>, MultiMessageDecoderDone<'d>>> {
let dim = self.scratch.read_type::<GroupSizeEncoding>(4)?;
if dim.num_in_group > 0 {
Ok(Either::Left(MultiMessageMessagesMemberDecoder::new(self.scratch, dim.num_in_group)))
} else {
Ok(Either::Right(MultiMessageDecoderDone::wrap(self.scratch)))
}
fn wrap(scratch: ScratchDecoderData<'d>) -> Self {
MultiMessageMessagesHeaderDecoder { scratch: scratch }
}
pub fn messages_individually(mut self) -> CodecResult<Either<MultiMessageMessagesMemberDecoder<'d>, MultiMessageDecoderDone<'d>>> {
let dim = self.scratch.read_type::<GroupSizeEncoding>(4)?;
if dim.num_in_group > 0 {
Ok(Either::Left(MultiMessageMessagesMemberDecoder::new(self.scratch, dim.num_in_group)))
} else {
Ok(Either::Right(MultiMessageDecoderDone::wrap(self.scratch)))
}
}
}
/// MultiMessage Fixed fields Decoder
pub struct MultiMessageFieldsDecoder<'d> {
scratch: ScratchDecoderData<'d>,
scratch: ScratchDecoderData<'d>,
}
impl<'d> MultiMessageFieldsDecoder<'d> {
pub fn wrap(scratch: ScratchDecoderData<'d>) -> MultiMessageFieldsDecoder<'d> {
MultiMessageFieldsDecoder { scratch: scratch }
}
pub fn multi_message_fields(mut self) -> CodecResult<(&'d MultiMessageFields, MultiMessageMessagesHeaderDecoder<'d>)> {
let v = self.scratch.read_type::<MultiMessageFields>(8)?;
Ok((v, MultiMessageMessagesHeaderDecoder::wrap(self.scratch)))
}
pub fn wrap(scratch: ScratchDecoderData<'d>) -> MultiMessageFieldsDecoder<'d> {
MultiMessageFieldsDecoder { scratch: scratch }
}
pub fn multi_message_fields(mut self) -> CodecResult<(&'d MultiMessageFields, MultiMessageMessagesHeaderDecoder<'d>)> {
let v = self.scratch.read_type::<MultiMessageFields>(8)?;
Ok((v, MultiMessageMessagesHeaderDecoder::wrap(self.scratch)))
}
}
/// MultiMessageMessageHeaderDecoder
pub struct MultiMessageMessageHeaderDecoder<'d> {
scratch: ScratchDecoderData<'d>,
scratch: ScratchDecoderData<'d>,
}
impl<'d> MultiMessageMessageHeaderDecoder<'d> {
pub fn wrap(scratch: ScratchDecoderData<'d>) -> MultiMessageMessageHeaderDecoder<'d> {
MultiMessageMessageHeaderDecoder { scratch: scratch }
}
pub fn header(mut self) -> CodecResult<(&'d MessageHeader, MultiMessageFieldsDecoder<'d>)> {
let v = self.scratch.read_type::<MessageHeader>(8)?;
Ok((v, MultiMessageFieldsDecoder::wrap(self.scratch)))
}
pub fn wrap(scratch: ScratchDecoderData<'d>) -> MultiMessageMessageHeaderDecoder<'d> {
MultiMessageMessageHeaderDecoder { scratch: scratch }
}
pub fn header(mut self) -> CodecResult<(&'d MessageHeader, MultiMessageFieldsDecoder<'d>)> {
let v = self.scratch.read_type::<MessageHeader>(8)?;
Ok((v, MultiMessageFieldsDecoder::wrap(self.scratch)))
}
}
/// MultiMessage Decoder entry point
pub fn start_decoding_multi_message<'d>(data: &'d [u8]) -> MultiMessageMessageHeaderDecoder<'d> {
MultiMessageMessageHeaderDecoder::wrap(ScratchDecoderData { data: data, pos: 0 })
MultiMessageMessageHeaderDecoder::wrap(ScratchDecoderData { data: data, pos: 0 })
}
/// MultiMessageEncoderDone
pub struct MultiMessageEncoderDone<'d> {
scratch: ScratchEncoderData<'d>,
scratch: ScratchEncoderData<'d>,
}
impl<'d> MultiMessageEncoderDone<'d> {
/// Returns the number of bytes encoded
pub fn unwrap(self) -> usize {
self.scratch.pos
}
/// Returns the number of bytes encoded
pub fn unwrap(self) -> usize {
self.scratch.pos
}
pub fn wrap(scratch: ScratchEncoderData<'d>) -> MultiMessageEncoderDone<'d> {
MultiMessageEncoderDone { scratch: scratch }
}
pub fn wrap(scratch: ScratchEncoderData<'d>) -> MultiMessageEncoderDone<'d> {
MultiMessageEncoderDone { scratch: scratch }
}
}
/// symbol variable-length data
pub struct MultiMessageMessagesSymbolEncoder<'d> {
parent: MultiMessageMessagesMemberEncoder<'d>,
parent: MultiMessageMessagesMemberEncoder<'d>,
}
impl<'d> MultiMessageMessagesSymbolEncoder<'d> {
fn wrap(parent: MultiMessageMessagesMemberEncoder<'d>) -> Self {
MultiMessageMessagesSymbolEncoder { parent: parent }
}
pub fn symbol(mut self, s: &'d [u8]) -> CodecResult<MultiMessageMessagesMemberEncoder> {
let l = s.len();
if l > 4294967294 {
return Err(CodecErr::SliceIsLongerThanAllowedBySchema);
}
// Write data length
self.parent.scratch.write_type::<u32>(&(l as u32), 4)?; // group length
self.parent.scratch.write_slice_without_count::<u8>(s, 1)?;
Ok(self.parent)
fn wrap(parent: MultiMessageMessagesMemberEncoder<'d>) -> Self {
MultiMessageMessagesSymbolEncoder { parent: parent }
}
pub fn symbol(mut self, s: &'d [u8]) -> CodecResult<MultiMessageMessagesMemberEncoder> {
let l = s.len();
if l > 4294967294 {
return Err(CodecErr::SliceIsLongerThanAllowedBySchema)
}
// Write data length
self.parent.scratch.write_type::<u32>(&(l as u32), 4)?; // group length
self.parent.scratch.write_slice_without_count::<u8>(s, 1)?;
Ok(self.parent)
}
}
/// MultiMessageMessages Encoder for fields and header
pub struct MultiMessageMessagesMemberEncoder<'d> {
scratch: ScratchEncoderData<'d>,
count_write_pos: usize,
count: u16,
scratch: ScratchEncoderData<'d>,
count_write_pos: usize,
count: u16,
}
impl<'d> MultiMessageMessagesMemberEncoder<'d> {
#[inline]
fn new(scratch: ScratchEncoderData<'d>, count_write_pos: usize) -> Self {
MultiMessageMessagesMemberEncoder {
scratch: scratch,
count_write_pos: count_write_pos,
count: 0,
}
#[inline]
fn new(scratch: ScratchEncoderData<'d>, count_write_pos: usize) -> Self {
MultiMessageMessagesMemberEncoder {
scratch: scratch,
count_write_pos: count_write_pos,
count: 0,
}
}
#[inline]
pub fn next_messages_member(mut self, fields: &MultiMessageMessagesMember) -> CodecResult<MultiMessageMessagesSymbolEncoder<'d>> {
self.scratch.write_type::<MultiMessageMessagesMember>(fields, 35)?; // block length
self.count += 1;
Ok(MultiMessageMessagesSymbolEncoder::wrap(self))
}
#[inline]
pub fn done_with_messages(mut self) -> CodecResult<MultiMessageEncoderDone<'d>> {
self.scratch.write_at_position::<u16>(self.count_write_pos, &self.count, 2)?;
Ok(MultiMessageEncoderDone::wrap(self.scratch))
}
#[inline]
pub fn next_messages_member(mut self, fields: &MultiMessageMessagesMember) -> CodecResult<MultiMessageMessagesSymbolEncoder<'d>> {
self.scratch.write_type::<MultiMessageMessagesMember>(fields, 35)?; // block length
self.count += 1;
Ok(MultiMessageMessagesSymbolEncoder::wrap(self))
}
#[inline]
pub fn done_with_messages(mut self) -> CodecResult<MultiMessageEncoderDone<'d>> {
self.scratch.write_at_position::<u16>(self.count_write_pos, &self.count, 2)?;
Ok(MultiMessageEncoderDone::wrap(self.scratch))
}
}
pub struct MultiMessageMessagesHeaderEncoder<'d> {
scratch: ScratchEncoderData<'d>,
scratch: ScratchEncoderData<'d>,
}
impl<'d> MultiMessageMessagesHeaderEncoder<'d> {
#[inline]
fn wrap(scratch: ScratchEncoderData<'d>) -> Self {
MultiMessageMessagesHeaderEncoder { scratch: scratch }
}
#[inline]
pub fn messages_individually(mut self) -> CodecResult<MultiMessageMessagesMemberEncoder<'d>> {
self.scratch.write_type::<u16>(&35u16, 2)?; // block length
let count_pos = self.scratch.pos;
self.scratch.write_type::<u16>(&0, 2)?; // preliminary group member count
Ok(MultiMessageMessagesMemberEncoder::new(self.scratch, count_pos))
}
#[inline]
fn wrap(scratch: ScratchEncoderData<'d>) -> Self {
MultiMessageMessagesHeaderEncoder { scratch: scratch }
}
#[inline]
pub fn messages_individually(mut self) -> CodecResult<MultiMessageMessagesMemberEncoder<'d>> {
self.scratch.write_type::<u16>(&35u16, 2)?; // block length
let count_pos = self.scratch.pos;
self.scratch.write_type::<u16>(&0, 2)?; // preliminary group member count
Ok(MultiMessageMessagesMemberEncoder::new(self.scratch, count_pos))
}
}
/// MultiMessage Fixed fields Encoder
pub struct MultiMessageFieldsEncoder<'d> {
scratch: ScratchEncoderData<'d>,
scratch: ScratchEncoderData<'d>,
}
impl<'d> MultiMessageFieldsEncoder<'d> {
pub fn wrap(scratch: ScratchEncoderData<'d>) -> MultiMessageFieldsEncoder<'d> {
MultiMessageFieldsEncoder { scratch: scratch }
}
pub fn wrap(scratch: ScratchEncoderData<'d>) -> MultiMessageFieldsEncoder<'d> {
MultiMessageFieldsEncoder { scratch: scratch }
}
/// Create a mutable struct reference overlaid atop the data buffer
/// such that changes to the struct directly edit the buffer.
/// Note that the initial content of the struct's fields may be garbage.
pub fn multi_message_fields(mut self) -> CodecResult<(&'d mut MultiMessageFields, MultiMessageMessagesHeaderEncoder<'d>)> {
let v = self.scratch.writable_overlay::<MultiMessageFields>(8 + 0)?;
Ok((v, MultiMessageMessagesHeaderEncoder::wrap(self.scratch)))
}
/// Create a mutable struct reference overlaid atop the data buffer
/// such that changes to the struct directly edit the buffer.
/// Note that the initial content of the struct's fields may be garbage.
pub fn multi_message_fields(mut self) -> CodecResult<(&'d mut MultiMessageFields, MultiMessageMessagesHeaderEncoder<'d>)> {
let v = self.scratch.writable_overlay::<MultiMessageFields>(8 + 0)?;
Ok((v, MultiMessageMessagesHeaderEncoder::wrap(self.scratch)))
}
/// Copy the bytes of a value into the data buffer
pub fn multi_message_fields_copy(mut self, t: &MultiMessageFields) -> CodecResult<MultiMessageMessagesHeaderEncoder<'d>> {
self.scratch.write_type::<MultiMessageFields>(t, 8)?;
Ok(MultiMessageMessagesHeaderEncoder::wrap(self.scratch))
}
/// Copy the bytes of a value into the data buffer
pub fn multi_message_fields_copy(mut self, t: &MultiMessageFields) -> CodecResult<MultiMessageMessagesHeaderEncoder<'d>> {
self.scratch.write_type::<MultiMessageFields>(t, 8)?;
Ok(MultiMessageMessagesHeaderEncoder::wrap(self.scratch))
}
}
/// MultiMessageMessageHeaderEncoder
pub struct MultiMessageMessageHeaderEncoder<'d> {
scratch: ScratchEncoderData<'d>,
scratch: ScratchEncoderData<'d>,
}
impl<'d> MultiMessageMessageHeaderEncoder<'d> {
pub fn wrap(scratch: ScratchEncoderData<'d>) -> MultiMessageMessageHeaderEncoder<'d> {
MultiMessageMessageHeaderEncoder { scratch: scratch }
}
pub fn wrap(scratch: ScratchEncoderData<'d>) -> MultiMessageMessageHeaderEncoder<'d> {
MultiMessageMessageHeaderEncoder { scratch: scratch }
}
/// Create a mutable struct reference overlaid atop the data buffer
/// such that changes to the struct directly edit the buffer.
/// Note that the initial content of the struct's fields may be garbage.
pub fn header(mut self) -> CodecResult<(&'d mut MessageHeader, MultiMessageFieldsEncoder<'d>)> {
let v = self.scratch.writable_overlay::<MessageHeader>(8 + 0)?;
Ok((v, MultiMessageFieldsEncoder::wrap(self.scratch)))
}
/// Create a mutable struct reference overlaid atop the data buffer
/// such that changes to the struct directly edit the buffer.
/// Note that the initial content of the struct's fields may be garbage.
pub fn header(mut self) -> CodecResult<(&'d mut MessageHeader, MultiMessageFieldsEncoder<'d>)> {
let v = self.scratch.writable_overlay::<MessageHeader>(8 + 0)?;
Ok((v, MultiMessageFieldsEncoder::wrap(self.scratch)))
}
/// Copy the bytes of a value into the data buffer
pub fn header_copy(mut self, t: &MessageHeader) -> CodecResult<MultiMessageFieldsEncoder<'d>> {
self.scratch.write_type::<MessageHeader>(t, 8)?;
Ok(MultiMessageFieldsEncoder::wrap(self.scratch))
}
/// Copy the bytes of a value into the data buffer
pub fn header_copy(mut self, t: &MessageHeader) -> CodecResult<MultiMessageFieldsEncoder<'d>> {
self.scratch.write_type::<MessageHeader>(t, 8)?;
Ok(MultiMessageFieldsEncoder::wrap(self.scratch))
}
}
/// MultiMessage Encoder entry point
pub fn start_encoding_multi_message<'d>(data: &'d mut [u8]) -> MultiMessageMessageHeaderEncoder<'d> {
MultiMessageMessageHeaderEncoder::wrap(ScratchEncoderData { data: data, pos: 0 })
MultiMessageMessageHeaderEncoder::wrap(ScratchEncoderData { data: data, pos: 0 })
}

144
src/sbe_runner.rs Normal file
View File

@ -0,0 +1,144 @@
use std::io::{BufRead, Write};
use std::str::from_utf8_unchecked;
use nom::bytes::complete::take_until;
use nom::IResult;
use crate::{marketdata_sbe, StreamVec, Summarizer};
use crate::iex::{IexMessage, IexPayload};
use crate::marketdata_sbe::{Either, MultiMessageFields, MultiMessageMessageHeader, MultiMessageMessagesMember, MultiMessageMessagesMemberEncoder, MultiMessageMessagesSymbolEncoder, Side, start_decoding_multi_message, start_encoding_multi_message};
fn __take_until<'a>(tag: &'static str, input: &'a [u8]) -> IResult<&'a [u8], &'a [u8]> {
take_until(tag)(input)
}
fn parse_symbol(sym: &[u8; 8]) -> &str {
// TODO: Use the `jetscii` library for all that SIMD goodness
// IEX guarantees ASCII, so we're fine using an unsafe conversion
let (_, sym_bytes) = __take_until(" ", &sym[..]).unwrap();
unsafe { from_utf8_unchecked(sym_bytes) }
}
pub struct SBEWriter {
/// Buffer to construct messages before copying. While SBE benefits
/// from easily being able to create messages directly in output buffer,
/// we'll construct in a scratch buffer and then copy to more fairly
/// benchmark against Cap'n Proto and Flatbuffers.
scratch_buffer: Vec<u8>,
default_header: MultiMessageMessageHeader,
}
impl SBEWriter {
pub fn new() -> SBEWriter {
SBEWriter {
// 8K scratch buffer is *way* more than necessary,
// but we don't want to run into issues with not enough
// data to encode messages
scratch_buffer: vec![0; 1024 * 8],
default_header: MultiMessageMessageHeader::default(),
}
}
pub fn serialize(&mut self, payload: &IexPayload, output: &mut Vec<u8>) {
let (fields, encoder) = start_encoding_multi_message(&mut self.scratch_buffer[..])
.header_copy(&self.default_header.message_header).unwrap()
.multi_message_fields().unwrap();
fields.sequence_number = payload.first_seq_no;
let mut encoder = encoder.messages_individually().unwrap();
let mut encoder: MultiMessageMessagesMemberEncoder = payload.messages.iter().fold(encoder, |enc, m| {
match m {
IexMessage::TradeReport(tr) => {
let fields = MultiMessageMessagesMember {
msg_type: marketdata_sbe::MsgType::Trade,
timestamp: tr.timestamp,
trade: marketdata_sbe::Trade {
size: tr.size,
price: tr.price,
},
..Default::default()
};
let sym_enc: MultiMessageMessagesSymbolEncoder = enc.next_messages_member(&fields).unwrap();
sym_enc.symbol(parse_symbol(&tr.symbol).as_bytes()).unwrap()
}
IexMessage::PriceLevelUpdate(plu) => {
let fields = MultiMessageMessagesMember {
msg_type: marketdata_sbe::MsgType::Quote,
timestamp: plu.timestamp,
quote: marketdata_sbe::Quote {
price: plu.price,
size: plu.size,
flags: plu.event_flags,
side: if plu.msg_type == 0x38 { Side::Buy } else { Side::Sell },
},
..Default::default()
};
let sym_enc: MultiMessageMessagesSymbolEncoder = enc.next_messages_member(&fields).unwrap();
sym_enc.symbol(parse_symbol(&plu.symbol).as_bytes()).unwrap()
}
_ => enc
}
});
let finished = encoder.done_with_messages().unwrap();
let data_len = finished.unwrap();
output.write(&self.scratch_buffer[..data_len]).unwrap();
}
}
pub struct SBEReader;
impl SBEReader {
pub fn new() -> SBEReader {
SBEReader {}
}
pub fn deserialize<'a>(&self, buf: &'a mut StreamVec, stats: &mut Summarizer) -> Result<(), ()> {
let data = buf.fill_buf().unwrap();
if data.len() == 0 {
return Err(());
}
let (header, decoder) = start_decoding_multi_message(data)
.header().unwrap();
let (fields, decoder) = decoder.multi_message_fields().unwrap();
let mut msg_decoder = decoder.messages_individually().unwrap();
while let Either::Left(msg) = msg_decoder {
let (member, sym_dec) = msg.next_messages_member().unwrap();
let (sym, next_msg_dec) = sym_dec.symbol().unwrap();
match member.msg_type {
marketdata_sbe::MsgType::Trade => stats.append_trade_volume(
unsafe { from_utf8_unchecked(sym) },
member.trade.size as u64,
),
marketdata_sbe::MsgType::Quote => stats.update_quote_prices(
unsafe { from_utf8_unchecked(sym) },
member.quote.price,
match member.quote.side {
Side::Buy => true,
_ => false
},
),
_ => ()
}
msg_decoder = next_msg_dec;
}
// We now have a `Right`, which is a finished messages block
let msg_decoder = match msg_decoder {
Either::Right(r) => r,
_ => panic!("Didn't parse all messages")
};
// Interestingly enough, `buf.consume(msg_decoder.unwrap())` isn't OK,
// presumably something to do with when *precisely* the drop of `self`
// happens for `msg_decoder`. Leave it as two statments so that
// Rust is able to prove our immutable borrow of `data` ends in time
// to consume the buffer
let msg_len = msg_decoder.unwrap();
buf.consume(msg_len);
Ok(())
}
}

50
tests/capnp.rs
View File

@ -1,50 +0,0 @@
use alloc_counter::{AllocCounterSystem, count_alloc, deny_alloc};
use md_shootout::marketdata_capnp::multi_message;
#[global_allocator]
static A: AllocCounterSystem = AllocCounterSystem;
#[test]
fn reinit_memory_check() {
// Setting up the builder doesn't reserve any heap memory
let mut msg_block = deny_alloc(|| {
capnp::message::Builder::new_default()
});
// Setting up the root object, however, does reserve a first segment
let (stats, result) = count_alloc(|| {
let multimsg = msg_block.init_root::<multi_message::Builder>();
multimsg.init_messages(32);
});
assert_eq!(stats.0, 4);
assert_eq!(stats.1, 0);
assert_eq!(stats.2, 0);
// If we reinitialize an object on that original builder, we re-use memory
deny_alloc(|| {
let multimsg = msg_block.init_root::<multi_message::Builder>();
multimsg.init_messages(32);
// Even if we down-size and up-size the message list size, we don't need
// to re-allocate
let multimsg = msg_block.init_root::<multi_message::Builder>();
multimsg.init_messages(16);
let multimsg = msg_block.init_root::<multi_message::Builder>();
multimsg.init_messages(32);
});
// It's only when we init a larger message count that a fresh allocation occurs
let (stats, _) = count_alloc(|| {
let multimsg = msg_block.init_root::<multi_message::Builder>();
// Note: calling `init_messages(33)` doesn't force allocation because
// the Capnproto builder reserved extra memory the first time around
multimsg.init_messages(256);
});
assert_eq!(stats.0, 1);
assert_eq!(stats.1, 3);
assert_eq!(stats.2, 0);
}