Brummel/Aura
1
0
Fork 0
Code Issues 31 Pull Requests Actions Packages Projects Releases Wiki Activity

plan: 0011 data-server M1 ingestion boundary

Browse Source 
Placeholder-free 6-task plan for spec 0011 (refs #7). T1 scaffolds the
aura-ingest crate + workspace member + unix_ms_to_epoch_ns (proves the
data-server git dep resolves); T2 M1Columns + transpose_m1; T3 close_stream;
T4 load_m1_window (data-server adapter, compile-gated against the real API);
T5 the gated real-bars integration test (cycle-0007 sample harness over real
AAPL.US close bars, finite + deterministic, skips where data absent);
T6 full-workspace gates. Mirrors report.rs build_two_sink_harness with the
shipped aura_std::Recorder. External data-server signatures verified against
its source in the spec commit.
This commit is contained in:
Brummel
2026-06-04 21:47:06 +02:00
parent df0574a75d
commit 44aa90be00
1 changed files with 567 additions and 0 deletions
Download Patch File Download Diff File Expand all files Collapse all files
docs/plans/0011-ingest-data-server-m1.md
+567
View File
@@ -0,0 +1,567 @@
# data-server M1 ingestion boundary — Implementation Plan
> **Parent spec:** `docs/specs/0011-ingest-data-server-m1.md`
>
> **For agentic workers:** REQUIRED SUB-SKILL: use the `implement` skill to run
> this plan. Steps use `- [ ]` checkboxes for tracking.
**Goal:** Ship a new `aura-ingest` crate that transposes data-server's AoS M1
records into SoA columns, normalizes Unix-ms to epoch-ns at the one ingestion
boundary, and feeds the engine a real close-price stream for a deterministic
backtest.
**Architecture:** A new workspace member `aura-ingest` (prod deps `aura-core` +
the `data-server` git crate; dev deps `aura-engine` + `aura-std`) holds the pure
C3/C7 boundary (`transpose_m1`, `unix_ms_to_epoch_ns`, `M1Columns::close_stream`)
plus the thin `data-server` adapter (`load_m1_window`). The boundary is the
external-dependency firewall: only this crate links `data-server` (and its
transitive `chrono`/`regex`/`zip`), keeping core/std/engine zero-external-dep.
**Tech Stack:** Rust 2024; `aura-core` (`Scalar`/`Timestamp`); `data-server`
(`M1Parsed`/`DataServer`); `aura-engine` + `aura-std` (sample harness, in the
integration test only).
---
## Files this plan creates or modifies
- Create: `crates/aura-ingest/Cargo.toml` — new member manifest; deps as above.
- Create: `crates/aura-ingest/src/lib.rs` — the C3/C7 boundary + hermetic unit tests.
- Create: `crates/aura-ingest/tests/real_bars.rs` — gated real-bars integration test.
- Modify: `Cargo.toml:11-16` — add `"crates/aura-ingest"` to `members`.
Verified external `data-server` API (from its actual source; used verbatim):
- `data_server::DataServer::new(impl AsRef<Path>) -> Self`
- `DataServer::has_symbol(&self, &str) -> bool`
- `DataServer::stream_m1_windowed(self: &Arc<Self>, symbol: &str, from_ms: Option<i64>, to_ms: Option<i64>) -> Option<SymbolChunkIter<M1Parsed>>`
- `SymbolChunkIter::<M1Parsed>::next_chunk(&mut self) -> Option<Arc<[M1Parsed]>>`
- `data_server::DEFAULT_DATA_PATH: &str` (crate root)
- `data_server::records::M1Parsed { time_ms: i64, open: f64, high: f64, low: f64, close: f64, spread: f64, volume: i64 }` derives `Copy + Clone` (plain struct, not packed — direct field access is sound).
---
## Task 1: Crate scaffold + workspace member + `unix_ms_to_epoch_ns`
Establishes the new member, proves the `data-server` git dependency resolves and
the crate compiles, and lands the first boundary function with its test.
**Files:**
- Modify: `Cargo.toml:11-16`
- Create: `crates/aura-ingest/Cargo.toml`
- Create: `crates/aura-ingest/src/lib.rs`
- [ ] **Step 1: Add the workspace member**
In `Cargo.toml`, replace the `members` array (lines 11-16):
```toml
members = [
"crates/aura-core",
"crates/aura-std",
"crates/aura-engine",
"crates/aura-cli",
"crates/aura-ingest",
]
```
- [ ] **Step 2: Write the crate manifest**
Create `crates/aura-ingest/Cargo.toml`:
```toml
[package]
name = "aura-ingest"
edition.workspace = true
version.workspace = true
license.workspace = true
publish.workspace = true
[dependencies]
aura-core = { path = "../aura-core" }
# data-server: aura's first real data source AND its one external dependency
# (transitively chrono + regex + zip). Isolated in this crate — the firewall
# that keeps aura-core/std/engine zero-external-dep (spec §Architecture).
data-server = { git = "http://192.168.178.103:3000/Brummel/data-server.git", branch = "main" }
[dev-dependencies]
# the integration test bootstraps a sample harness + folds a RunReport
aura-engine = { path = "../aura-engine" }
aura-std = { path = "../aura-std" }
```
- [ ] **Step 3: Write `lib.rs` module doc + `unix_ms_to_epoch_ns` + its test**
Create `crates/aura-ingest/src/lib.rs`:
```rust
//! aura's first real data source: the C3/C7 ingestion boundary.
//!
//! Transposes [`data_server`]'s Array-of-Structs `M1Parsed` records into aura's
//! Structure-of-Arrays base columns (C7) and normalizes data-server's
//! Unix-millisecond time to aura's canonical epoch-nanosecond [`Timestamp`] at
//! this one boundary (C3). A transposed [`M1Columns`] exposes its close column
//! as the engine source-stream shape (`Vec<(Timestamp, Scalar)>`) the SMA-cross
//! sample strategy consumes.
//!
//! This crate is the workspace's **external-dependency firewall**: it is the
//! only crate that links `data-server` (and its transitive `chrono`/`regex`/
//! `zip`), so `aura-core`/`aura-std`/`aura-engine` stay zero-external-dependency.
use aura_core::{Scalar, Timestamp};
use data_server::records::M1Parsed;
use data_server::DataServer;
use std::sync::Arc;
/// Normalize data-server's Unix-millisecond time to aura's canonical epoch-ns
/// [`Timestamp`] — the single unit normalization of C3, performed at the one
/// ingestion boundary and nowhere else. The `i64` epoch-ns range covers all
/// market data through year ~2262, well beyond any real file.
pub fn unix_ms_to_epoch_ns(time_ms: i64) -> Timestamp {
Timestamp(time_ms * 1_000_000)
}
#[cfg(test)]
mod tests {
use super::*;
#[test]
fn unix_ms_to_epoch_ns_scales_ms_to_ns() {
// data-server's own epoch fixture: 2017-03-01 00:00 UTC.
assert_eq!(
unix_ms_to_epoch_ns(1_488_326_400_000),
Timestamp(1_488_326_400_000_000_000)
);
// epoch maps to epoch.
assert_eq!(unix_ms_to_epoch_ns(0), Timestamp(0));
}
}
```
- [ ] **Step 4: Build the crate (proves the git dep resolves) and run the test**
Run: `cargo test -p aura-ingest`
Expected: PASS — compiles (fetching `data-server` from Gitea on first build, then
cached) and `unix_ms_to_epoch_ns_scales_ms_to_ns` is green. (If the Gitea fetch
fails, that is an infra/network problem to resolve, not a code failure.)
---
## Task 2: `M1Columns` + `transpose_m1`
The pure AoS→SoA transpose, the heart of the C3/C7 boundary, with hermetic tests
on hand-built records.
**Files:**
- Modify: `crates/aura-ingest/src/lib.rs`
- [ ] **Step 1: Add `M1Columns` + `with_capacity` + `transpose_m1`**
In `crates/aura-ingest/src/lib.rs`, after `unix_ms_to_epoch_ns` (before the
`#[cfg(test)]` module), insert:
```rust
/// One M1 window transposed Array-of-Structs → Structure-of-Arrays (C7): the
/// OHLCV bar as a bundle of base columns, time already normalized to epoch-ns.
/// `volume` is the one `i64` column; the price/spread columns are `f64`. All
/// columns share an index: `ts[i]` is the timestamp of `close[i]`, etc.
#[derive(Clone, Debug, PartialEq)]
pub struct M1Columns {
pub ts: Vec<Timestamp>,
pub open: Vec<f64>,
pub high: Vec<f64>,
pub low: Vec<f64>,
pub close: Vec<f64>,
pub spread: Vec<f64>,
pub volume: Vec<i64>,
}
impl M1Columns {
/// Pre-size all columns for `n` bars.
fn with_capacity(n: usize) -> Self {
Self {
ts: Vec::with_capacity(n),
open: Vec::with_capacity(n),
high: Vec::with_capacity(n),
low: Vec::with_capacity(n),
close: Vec::with_capacity(n),
spread: Vec::with_capacity(n),
volume: Vec::with_capacity(n),
}
}
}
/// Transpose data-server's AoS M1 records into aura's SoA columns (C7),
/// normalizing time at the boundary (C3). Pure: identical bars yield identical
/// columns (C1) — it reads no clock and no external state.
pub fn transpose_m1(bars: &[M1Parsed]) -> M1Columns {
let mut c = M1Columns::with_capacity(bars.len());
for b in bars {
c.ts.push(unix_ms_to_epoch_ns(b.time_ms));
c.open.push(b.open);
c.high.push(b.high);
c.low.push(b.low);
c.close.push(b.close);
c.spread.push(b.spread);
c.volume.push(b.volume);
}
c
}
```
- [ ] **Step 2: Add the transpose tests**
In the `#[cfg(test)] mod tests` block in `crates/aura-ingest/src/lib.rs`, add a
record-builder helper and the transpose tests (inside the `mod tests`, after the
existing test):
```rust
/// A hand-built M1 bar with distinct per-field values so a transpose test
/// can tell the columns apart.
fn full_bar(time_ms: i64) -> M1Parsed {
M1Parsed {
time_ms,
open: 1.0,
high: 2.0,
low: 0.5,
close: 1.5,
spread: 0.1,
volume: 100,
}
}
#[test]
fn transpose_m1_maps_every_field_to_its_column() {
let bars = [full_bar(1_000), full_bar(2_000)];
let c = transpose_m1(&bars);
// ts normalized ms -> ns; each column equal length to the input.
assert_eq!(c.ts, vec![Timestamp(1_000_000_000), Timestamp(2_000_000_000)]);
assert_eq!(c.open, vec![1.0, 1.0]);
assert_eq!(c.high, vec![2.0, 2.0]);
assert_eq!(c.low, vec![0.5, 0.5]);
assert_eq!(c.close, vec![1.5, 1.5]);
assert_eq!(c.spread, vec![0.1, 0.1]);
// volume is the one i64 column.
assert_eq!(c.volume, vec![100_i64, 100_i64]);
}
#[test]
fn transpose_m1_is_pure() {
let bars = [full_bar(1_000), full_bar(2_000), full_bar(3_000)];
assert_eq!(transpose_m1(&bars), transpose_m1(&bars));
}
#[test]
fn transpose_m1_empty_input_yields_empty_columns() {
let c = transpose_m1(&[]);
assert!(c.ts.is_empty());
assert!(c.close.is_empty());
assert!(c.volume.is_empty());
}
```
- [ ] **Step 3: Run the tests**
Run: `cargo test -p aura-ingest`
Expected: PASS — `transpose_m1_maps_every_field_to_its_column`,
`transpose_m1_is_pure`, `transpose_m1_empty_input_yields_empty_columns` green
alongside Task 1's test.
---
## Task 3: `M1Columns::close_stream`
The adapter from a transposed column to the engine's source-stream shape.
**Files:**
- Modify: `crates/aura-ingest/src/lib.rs`
- [ ] **Step 1: Add `close_stream`**
In `crates/aura-ingest/src/lib.rs`, add an `impl M1Columns` block (after the
existing `impl M1Columns { fn with_capacity … }`, or extend it) with:
```rust
impl M1Columns {
/// The close column as an engine source stream: each normalized timestamp
/// zipped with its close as a [`Scalar::F64`]. Ascending in timestamp iff
/// the bars were (the C3 ingestion precondition `Harness::run` relies on).
/// This is the price input the SMA-cross sample strategy consumes; a project
/// that wants another field reads the public columns and zips its own.
pub fn close_stream(&self) -> Vec<(Timestamp, Scalar)> {
self.ts
.iter()
.zip(&self.close)
.map(|(&t, &v)| (t, Scalar::F64(v)))
.collect()
}
}
```
- [ ] **Step 2: Add the close_stream tests**
In `#[cfg(test)] mod tests`, add:
```rust
#[test]
fn close_stream_zips_ts_with_close_in_order() {
// distinct close per bar so order is observable; ts ascending.
let bars = [
M1Parsed { time_ms: 1, open: 0.0, high: 0.0, low: 0.0, close: 10.0, spread: 0.0, volume: 0 },
M1Parsed { time_ms: 2, open: 0.0, high: 0.0, low: 0.0, close: 20.0, spread: 0.0, volume: 0 },
M1Parsed { time_ms: 3, open: 0.0, high: 0.0, low: 0.0, close: 30.0, spread: 0.0, volume: 0 },
];
let stream = transpose_m1(&bars).close_stream();
assert_eq!(
stream,
vec![
(Timestamp(1_000_000), Scalar::F64(10.0)),
(Timestamp(2_000_000), Scalar::F64(20.0)),
(Timestamp(3_000_000), Scalar::F64(30.0)),
]
);
// ascending in timestamp (the merge precondition).
assert!(stream.windows(2).all(|w| w[0].0 < w[1].0));
}
#[test]
fn close_stream_empty_on_empty_columns() {
assert!(transpose_m1(&[]).close_stream().is_empty());
}
```
- [ ] **Step 3: Run the tests**
Run: `cargo test -p aura-ingest`
Expected: PASS — `close_stream_zips_ts_with_close_in_order` and
`close_stream_empty_on_empty_columns` green alongside the prior tests.
---
## Task 4: `load_m1_window` — the data-server adapter
Drains a data-server M1 window into transposed columns. Not hermetically
unit-testable (it touches files/the cache); its correctness is exercised by the
Task 5 integration test. This task is impl + compile gate.
**Files:**
- Modify: `crates/aura-ingest/src/lib.rs`
- [ ] **Step 1: Add `load_m1_window`**
In `crates/aura-ingest/src/lib.rs`, after `transpose_m1` (outside the test
module), insert:
```rust
/// Drain a data-server M1 window into transposed SoA columns. Reads the
/// chronological chunk iterator to exhaustion into one AoS buffer, then
/// transposes once at the boundary (C3 — one merge point, not per chunk).
/// `None` if the symbol has no data in `[from_ms, to_ms]` (data-server's own
/// `None`); the inclusive Unix-ms window bounds are data-server's contract.
pub fn load_m1_window(
server: &Arc<DataServer>,
symbol: &str,
from_ms: i64,
to_ms: i64,
) -> Option<M1Columns> {
let mut it = server.stream_m1_windowed(symbol, Some(from_ms), Some(to_ms))?;
let mut bars: Vec<M1Parsed> = Vec::new();
// M1Parsed is Copy; &Arc<[M1Parsed]> derefs to &[M1Parsed] in arg position.
while let Some(chunk) = it.next_chunk() {
bars.extend_from_slice(&chunk);
}
Some(transpose_m1(&bars))
}
```
- [ ] **Step 2: Build (compiles against the real data-server API) + lint**
Run: `cargo build -p aura-ingest && cargo clippy -p aura-ingest --all-targets -- -D warnings`
Expected: PASS — `load_m1_window` compiles against data-server's real
`stream_m1_windowed`/`next_chunk` signatures with no warnings. (A signature
mismatch here surfaces as a compile error — the load-bearing external-API check.)
---
## Task 5: Gated real-bars integration test
Runs the cycle-0007 sample harness over a real AAPL.US close stream, asserting a
finite, deterministic `RunReport`. Skips cleanly where `/mnt/tickdata` is absent.
**Files:**
- Create: `crates/aura-ingest/tests/real_bars.rs`
- [ ] **Step 1: Write the integration test**
Create `crates/aura-ingest/tests/real_bars.rs`:
```rust
//! Gated integration test: a real data-server M1 close stream driven through
//! the cycle-0007 signal-quality sample harness (SMA-cross → Exposure →
//! SimBroker), folded into a `RunReport`. Skips with a note where the local
//! Pepperstone data directory is absent, so `cargo test --workspace` stays green
//! anywhere; it exercises the real ingestion path where data exists.
use std::sync::mpsc;
use std::sync::Arc;
use aura_core::{Firing, Scalar, ScalarKind, Timestamp};
use aura_engine::{
f64_field, summarize, Edge, Harness, RunManifest, RunReport, SourceSpec, Target,
};
use aura_ingest::load_m1_window;
use aura_std::{Exposure, Recorder, SimBroker, Sma, Sub};
use data_server::{DataServer, DEFAULT_DATA_PATH};
/// Bootstrap the cycle-0007 two-sink signal-quality harness (mirrors
/// `aura-engine`'s `report::tests::build_two_sink_harness`, with the shipped
/// `aura_std::Recorder`), run it on `prices`, and fold the recorded equity +
/// exposure into a `RunReport` whose window is the first/last real bar ts.
fn run_sample_over(prices: Vec<(Timestamp, Scalar)>) -> RunReport {
let (tx_eq, rx_eq) = mpsc::channel();
let (tx_ex, rx_ex) = mpsc::channel();
let mut h = Harness::bootstrap(
vec![
Box::new(Sma::new(2)), // 0
Box::new(Sma::new(4)), // 1
Box::new(Sub::new()), // 2
Box::new(Exposure::new(0.5)), // 3
Box::new(SimBroker::new(0.0001)), // 4
Box::new(Recorder::new(&[ScalarKind::F64], Firing::Any, tx_eq)), // 5 equity sink
Box::new(Recorder::new(&[ScalarKind::F64], Firing::Any, tx_ex)), // 6 exposure sink
],
vec![SourceSpec {
kind: ScalarKind::F64,
targets: vec![
Target { node: 0, slot: 0 },
Target { node: 1, slot: 0 },
Target { node: 4, slot: 1 }, // price into the broker
],
}],
vec![
Edge { from: 0, to: 2, slot: 0, from_field: 0 },
Edge { from: 1, to: 2, slot: 1, from_field: 0 },
Edge { from: 2, to: 3, slot: 0, from_field: 0 },
Edge { from: 3, to: 4, slot: 0, from_field: 0 },
Edge { from: 4, to: 5, slot: 0, from_field: 0 }, // equity -> sink 5
Edge { from: 3, to: 6, slot: 0, from_field: 0 }, // exposure -> sink 6
],
)
.expect("valid signal-quality DAG");
let window = (
prices.first().map(|&(t, _)| t).unwrap_or(Timestamp(0)),
prices.last().map(|&(t, _)| t).unwrap_or(Timestamp(0)),
);
h.run(vec![prices]);
let eq_rows: Vec<(Timestamp, Vec<Scalar>)> = rx_eq.try_iter().collect();
let ex_rows: Vec<(Timestamp, Vec<Scalar>)> = rx_ex.try_iter().collect();
let equity = f64_field(&eq_rows, 0);
let exposure = f64_field(&ex_rows, 0);
let metrics = summarize(&equity, &exposure);
RunReport {
manifest: RunManifest {
commit: "real-bars-test".to_string(),
params: vec![
("sma_fast".to_string(), 2.0),
("sma_slow".to_string(), 4.0),
("exposure_scale".to_string(), 0.5),
],
window,
seed: 0,
broker: "sim-optimal(pip_size=0.0001)".to_string(),
},
metrics,
}
}
#[test]
fn sample_strategy_runs_over_real_m1_bars_deterministically() {
let server = Arc::new(DataServer::new(DEFAULT_DATA_PATH));
if !server.has_symbol("AAPL.US") {
eprintln!("skip: no local data at {DEFAULT_DATA_PATH} (symbol AAPL.US absent)");
return; // hermetic elsewhere; exercises the real path where files exist
}
// 2006-08 in inclusive Unix-ms (data-server skips files outside the window).
let (from_ms, to_ms) = (1_154_390_400_000_i64, 1_157_068_799_999_i64);
let load = || {
load_m1_window(&server, "AAPL.US", from_ms, to_ms)
.expect("AAPL.US has data in the 2006-08 window")
.close_stream()
};
let prices = load();
assert!(!prices.is_empty(), "window resolved to zero bars");
let r1 = run_sample_over(prices);
assert!(r1.metrics.total_pips.is_finite(), "a backtest ran over real bars");
// same window -> bit-identical report (C1).
let r2 = run_sample_over(load());
assert_eq!(r1.to_json(), r2.to_json());
}
```
- [ ] **Step 2: Run the integration test**
Run: `cargo test -p aura-ingest --test real_bars`
Expected: PASS — on this machine `/mnt/tickdata/Pepperstone` exists, so the test
runs the real path (loads AAPL.US 2006-08, runs the sample harness, asserts
finite + deterministic). On a machine without the data it prints the skip note
and passes.
---
## Task 6: Full-workspace gates
Confirm the additive crate leaves the whole workspace green under the project's
gate commands.
**Files:** none (verification only).
- [ ] **Step 1: Workspace test suite**
Run: `cargo test --workspace`
Expected: PASS — all existing crate tests plus the new `aura-ingest` unit tests
and the gated integration test. (Builds `aura-ingest` + its `data-server` tree;
the cargo cache must be populated — Task 1 already fetched it.)
- [ ] **Step 2: Lint**
Run: `cargo clippy --workspace --all-targets -- -D warnings`
Expected: PASS — no warnings across the workspace including `aura-ingest`.
- [ ] **Step 3: Docs**
Run: `RUSTDOCFLAGS="-D warnings" cargo doc --workspace --no-deps`
Expected: PASS — `aura-ingest`'s rustdoc (module doc + public-item docs) builds
with no warnings.
---
## Self-review (planner Step 5)
1. **Spec coverage:** boundary (`unix_ms_to_epoch_ns` T1, `transpose_m1`/`M1Columns`
T2, `close_stream` T3, `load_m1_window` T4), hermetic tests (T2/T3), gated
integration test (T5), workspace gates (T6), crate/firewall + offline-build
(T1 manifest + members). All spec sections covered.
2. **Placeholder scan:** no TBD/TODO/"implement later"/"similar to"/"add appropriate".
3. **Type consistency:** `M1Columns`, `transpose_m1`, `unix_ms_to_epoch_ns`,
`close_stream`, `load_m1_window`, node/edge wiring identical across tasks and to
the recon-anchored `report.rs` fixture; `data-server` signatures match the
verified source.
4. **Step granularity:** each step is one edit or one command.
5. **No commit steps:** none present.
6. **Pin/replacement contiguity:** no presence-pin paired with a verbatim text
edit (the `members` edit is a whole-array replacement, self-contained).
7. **Compile-gate vs deferred-caller:** purely additive new crate; Task 1 adds the
member AND manifest AND a compiling lib.rs together, so its build gate is
satisfiable with no deferred caller. No existing signature changes.
8. **Verification-command filters resolve:** per-task gates use `-p aura-ingest`
(whole-crate, named expected tests) and `--test real_bars` (the file created in
T5); no empty-filter "0 ran" masquerade.
9. **Parse-the-bytes gate:** profile declares no `spec_validation` parser → no-op;
inlined bodies are Rust (the source language), caught by the implement compile
gate, not the planner parse gate.