Crate wickra_strategy_ci_core 

Data-driven core of Wickra Strategy-CI.

A StrategyTest carries an opaque StrategySpec sub-JSON that is forwarded verbatim to the wickra-backtest engine. The resulting BacktestReport is golden-pinned and asserted against per-field Tolerances, checked for invariant Propertys, and fuzzed with seeded data Perturbations. The Session command_json boundary is the single JSON entry point every language binding drives, so results are byte-identical across languages and between the parallel and sequential paths.

Structs

Candle

The OHLCV candle type, re-exported from the backtest engine so tests, data and bindings all speak the exact same JSON. One OHLCV bar. time is the bar’s open time (engine-defined epoch unit; it is passed straight through to indicators that need a timestamp).

Config

Per-session configuration. The runner is stateless with respect to tests and data, so the session holds only configuration; this is the extension point for future runtime knobs (report-hash policy, output formatting, …).

FieldDiff

One field that differs between the golden and the actual report.

FuzzFailure

A property that failed on a specific fuzz run.

FuzzSpec

A fuzz axis: perturb the data runs times from seed and re-check properties.

PropertyResult

The outcome of one property check.

Session

A stateless handle over the test runner. It holds only configuration; tests and data are passed in with each command.

StrategyTest

A single strategy test: a strategy, the dataset it runs on, an optional golden report with tolerances, invariant properties, and an optional fuzz axis.

SuiteResult

The result of running a whole suite, sorted by test id.

TestResult

The full result of running one test.

Enums

DiffKind

The nature of a field-level difference between two reports.

Error

Everything that can go wrong while running a strategy test.

Perturbation

How the input data is perturbed for a fuzz run.

Property

An invariant that must hold for a run, independent of any golden value.

Tolerance

A per-field tolerance for the golden diff.

Constants

VERSION

The crate version, surfaced through the version command and crate::Session::version.

Functions

bless

Re-run a test and return a copy with its expected golden set to the freshly produced (rounded) report.

check

Check a single property against a flattened report.

check_all

Check every property against report, returning one result each (in the order the properties were given).

diff_reports

Diff two reports into the list of fields that differ. A perfect match yields an empty vector. Values are rounded to eight decimals before comparison, and the BTreeMap iteration order makes the output deterministic.

flatten_report

Flatten a report into a sorted map of numeric leaves. A scalar x maps to its key; equity_curve[i] maps to "equity_curve[i]"; a nested object a.b maps to "a.b". Non-numeric leaves (strings, bools, null) are ignored — only f64 values are compared. The BTreeMap makes iteration order deterministic.

round8

Round to eight decimals — the canonical numeric precision for diffs and for the values stored in a blessed golden.

run_suite

Run a suite of tests, sorted by id. Tests are independent, so with the parallel feature they run concurrently; the sorted output is identical either way.

run_test

Run one strategy test to a full TestResult.

Type Aliases

ReportJson

A BacktestReport as produced by the engine, held schema-agnostically so new report fields flow through without a code change.

Result

A Result specialized to [Error].

StrategyJson

An opaque backtest strategy definition. It is not parsed here; it is forwarded verbatim to the wickra-backtest engine, whose schema owns it.