Relicenses Wickra from PolyForm Noncommercial 1.0.0 to the dual, OSI-approved **MIT OR Apache-2.0** (the de-facto Rust convention). Wickra becomes permissive, commercial-use-permitted open source; users may choose either license.
## Changes
- Replace `LICENSE` (PolyForm) with `LICENSE-MIT` + `LICENSE-APACHE` (full texts).
- Cargo: workspace `license = "MIT OR Apache-2.0"` (SPDX) + all 7 sub-crates switched from `license-file.workspace` to `license.workspace`.
- `deny.toml`: drop PolyForm from the allowlist.
- Python: `pyproject.toml` PEP 639 SPDX expression; remove the non-commercial classifier (verified: sdist metadata emits `License-Expression: MIT OR Apache-2.0`).
- Node: `package.json`, the 6 platform manifests and both lockfiles.
- README + Python/Node/WASM binding READMEs, CONTRIBUTING, CITATION.cff, PR template, and the WASM `pkg.license` step in `release.yml`.
- SECURITY.md: refresh supported versions 0.1.x -> 0.4.x.
- CHANGELOG: note the relicense under [Unreleased].
## Notes
- No code changes; metadata/text only. `cargo build` and `cargo deny check licenses` pass locally.
- GitHub will auto-detect "MIT, Apache-2.0" once this lands (currently NOASSERTION).
- Matching downstream changes (org `.github` profile, webpage, docs) are in separate PRs; merge those together with the relicense release so the live sites and org profile do not claim MIT before the packages do.
The LICENSE now carries an Additional Permissions section on top of
PolyForm Noncommercial 1.0.0, so the bare SPDX id no longer describes
it exactly. Update the package manifests to reference the actual file
instead of claiming the unmodified standard:
- Cargo (workspace + all crates): license -> license-file = "LICENSE"
- npm (main + 6 platform packages): LicenseRef-Wickra-Noncommercial-1.0.0
- PyPI: license text notes the additional personal-account permissions
A full `cargo doc --workspace` (and docs.rs, which builds with
`-D rustdoc::all`) emitted five broken intra-doc links. docs.rs treats
these as hard errors, so any 0.2.x doc build was at risk of aborting.
- mama.rs: `[`Fama`]` -> `[`crate::Fama`]` (Fama lives in another module)
- standard_error.rs: `[`crate::Bollinger`]` -> `[`crate::BollingerBands`]`
(the public type is `BollingerBands`, not `Bollinger`)
- aggregator.rs: drop the link to the private `OpenBar::into_candle`,
keep it as plain code text
- resample.rs: drop the link to the private `RolledBar::into_candle`
- csv.rs: `CandleReader::with_timestamp_parser` never existed; reword to
state plainly that ISO/RFC timestamps must be converted to integers
Verified with `RUSTDOCFLAGS="-D rustdoc::broken_intra_doc_links \
-D rustdoc::private_intra_doc_links" cargo doc --workspace --no-deps`:
clean, zero warnings.
* fix(docs-rs): rename `doc_auto_cfg` to `doc_cfg` after Rust 1.92 merge
`doc_auto_cfg` was removed in Rust 1.92.0 and folded back into
`doc_cfg` (rust-lang/rust#138907). docs.rs builds with the latest
nightly and sets `--cfg docsrs`, so the previous
#![cfg_attr(docsrs, feature(doc_auto_cfg))]
aborts compilation with E0557 on every published 0.2.x. GitHub CI
never tripped this — stable rustc ignores the line because nothing
sets the `docsrs` cfg there.
Switch all three published library crates (`wickra`, `wickra-core`,
`wickra-data`) to the merged-into `doc_cfg` gate. Same intent, same
on-docs.rs output, builds again on nightly.
* docs(readme): float Wickra to the top of the comparison tables
Reorders the "Why Wickra exists" library-comparison table and the two
benchmark headers so Wickra is the first row (with a ★ marker) instead
of the last. The previous order placed Wickra at the bottom, which
buries the only row a reader landing on the README is here to compare
against. Same column data, same ★/winner annotations, just the row
order flipped and a ★ prefix on the Wickra label.
Mirrored across the umbrella README and every binding README so the
crates.io / PyPI / npm landing pages stay in sync.
* release: bump workspace + bindings to 0.2.6
Workspace, every binding (Python, Node, Node platform stubs), the
release.yml comment and the CHANGELOG all move together to 0.2.6 so
the next tagged release lines every artefact up.
0.2.6 carries two changes from the [0.2.6] CHANGELOG entry:
- fix(docs-rs): swap the now-removed `doc_auto_cfg` feature gate for
the merged-into `doc_cfg` so docs.rs nightly builds resume.
- docs(readme): float ★ Wickra to the top of every comparison table
across the umbrella + binding READMEs.
wickra-win32-arm64-msvc stays excluded for this release with the same
npm spam-filter rationale that held for 0.2.5.
* refactor(binance): introduce BinanceConfig for endpoint + timing knobs
Replaces the file-private READ_TIMEOUT / MAX_RECONNECT_ATTEMPTS / size
limit constants with a Default-equipped BinanceConfig the caller can hand
to a new connect_with_config(). connect() forwards to it with the
defaults, so the public API stays backwards-compatible.
Behaviour-preserving: every default matches the value of the constant it
replaces, and the WebSocketConfig is built the same way. The change
unlocks two real use-cases — pointing at Binance Testnet
(wss://testnet.binance.vision) and pointing at a local mock server with
millisecond-scale reconnect timing in tests.
* test(binance): cover the Interval table and the empty-symbol guard
Three quick wins that don't need a live or mock socket:
- interval_as_str_covers_every_variant pins every wire-format mapping in
one table so a typo on any of the 14 variants is caught.
- binance_config_default_matches_production_endpoint guards the default
base URL and timing knobs against an accidental drift.
- connect_rejects_an_empty_symbol_list exercises the guard before the
WebSocket handshake — the one async path we can hit without a server.
* test(binance): cover the async / reconnect / control-frame paths
Adds a small mock-WebSocket scaffold built on a `127.0.0.1:0` listener
and tokio-tungstenite's `accept_async`, plus nine integration tests that
drive `BinanceKlineStream::next_event` through every branch:
- text + binary kline frames decode to a KlineEvent
- inbound Ping is answered with a Pong, then the kline arrives
- inbound Pong / Frame variants are silently skipped
- a server-side Close triggers a transparent reconnect that then
serves the kline
- a stalled connection trips read_timeout and reconnects on its own
- close() flips the closed flag and next_event() yields None forever
- when every reconnect attempt is refused, next_event surfaces an Err
- a "kline" envelope whose numbers are unparseable bubbles up as
Error::Malformed rather than being silently skipped
`one_shot_server` drops the listener as soon as the first accept is
done, so a follow-up reconnect lands on a refused port — that is what
lets the exhaustion test hit the final `last_err.expect(...)`.
The whole suite runs in ~4 s with millisecond-scale reconnect timings
supplied via the new test-only [`test_config`].
* test(binance): drop defensive cold-paths in the mock-WS scaffolding
Codecov's patch report on PR #34 flagged seven uncovered lines, all of
them in the test scaffolding rather than in production code:
- the `let Ok((stream, _)) = … else { return }` shortcut and the
`if let Ok(ws) = accept_async(stream).await { … }` branch in the
mock-server helpers — both error arms never fire on a passing test
- the closing braces of the spawned-task bodies in the close-frame and
read-timeout reconnect tests — the spawned async blocks were getting
killed mid-drain when the test asserted and returned
Refactor the helpers to `.unwrap()` every Result (a failure here is a
bug in the scaffold, not in production) and have `multi_shot_server`
accept a fixed `n_accepts`, await every spawned inner task, and hand
the outer JoinHandle back to the caller.
Refactor the two affected tests to capture that JoinHandle, collapse
the per-index `if/else` so both arms reach the same trailing
expression, swap the read-timeout drain for a bounded sleep, and
await `server_done` at the end. Every handler now reaches its closing
brace before the runtime is torn down, so coverage on the patch should
collapse from 97.89 % to 100 %.
* test(binance): cover the non-kline-skip path and simplify the Ping arm
After the scaffolding fix landed three lines on binance.rs were still
uncovered:
- L305 / L313: the Text- and Binary-arm "frame was not a kline, keep
reading" fall-throughs. No existing test drove the loop through a
non-kline frame followed by a kline; the new
`next_event_skips_non_kline_frames_and_returns_the_next_kline` does
exactly that (Text ack, Binary id frame, then a real kline).
- L317: the Ping-Err defensive arm that forced a reconnect when the
Pong reply itself failed to write. A failed Pong reply means the
socket is already dead, so the very next read will surface the error
and reconnect through the existing timeout/err branch — one tokio
scheduling iteration later. Drop the defensive arm and write the
Pong reply best-effort. Same observable behaviour, no test back
door, no dead-line guard.
Repo coverage on `cov/binance-mock-ws` now sits at 100 %.
* test(vertical_horizontal_filter): cover period accessor + name metadata
Codecov flagged 6 lines (file at 94.44%): period (61-63) + name (119-121).
* test(z_score): cover period accessor + name metadata
Codecov flagged 6 lines (file at 93.75%): period (59-61) + name (106-108).
* test(vpt): cover value() Some branch, name, zero-prev fallback
Codecov flagged 5 lines (file at 94.38%): value() Some branch (57),
prev==0.0 ROC fallback (77), and Indicator-impl name (100-102).
Add accessors_and_metadata covering value()/name and zero_previous_
close_contributes_zero — feeding a 0.0 baseline + non-zero candle
proves the divide-by-zero guard yields a 0 contribution rather than NaN.
* test(csv): cover from_csv_reader + kill rejects_header dead panic arm
Codecov flagged 5 lines in csv.rs (file at 96.98%): from_csv_reader
(201-204) — never called by existing tests which use from_reader /
open — and the cold arm in
rejects_header_missing_a_column (279). Add from_csv_reader_accepts_a_
prebuilt_reader (demonstrates the API by building a custom-delimited
csv::Reader and passing it in), and refactor the header-missing test
to use a single matches!() assertion so the panic arm is gone.
* test(adl): cover name metadata
Codecov flagged 3 lines (file at 96.84%): Indicator-impl name body (94-96).
Codecov flagged 15 uncovered lines in crates/wickra-data/src/aggregator.rs
(file at 95.11%):
- Timeframe::millis / Timeframe::seconds / Timeframe::one_minute_ms
convenience constructors (40-52) — every existing test built
Timeframes via new / minutes / hours / days, never via these three
- the cold `?` Err arm on `Candle::new(...)?` for the flat gap-fill
candle (line 334) — `prev.close` is already finite (came from a
closed bar), volume is exactly 0.0, OHLC are trivially equal, so
Candle::new's error path is unreachable here
- the cold `ok_or_else` overflow closure on `t.checked_add(step)`
inside the gap-fill loop (336-337) — bucket alignment guarantees
start + (gap_count-1)*step ≤ next_bucket - step < i64::MAX, so
every aligned-bucket layout reaches t == next_bucket cleanly and
exits without ever invoking the overflow path
- TickAggregator::timeframe accessor (353-355) — never queried
Add two new tests:
- timeframe_convenience_constructors exercises millis/seconds/
one_minute_ms with both happy-path and rejection cases
- aggregator_timeframe_getter asserts timeframe().bucket() round-trips
Refactor fill_between to use Candle::new_unchecked for the flat-candle
push (the OHLCV invariants hold by construction) and iterate via
`0..gap_count` with `saturating_add(step)` instead of `while t <
next_bucket` with `checked_add(...).ok_or_else(...)?`. gap_count
already controls iteration count and saturating_add cannot panic,
preserving observable behaviour on every reachable input while
removing the unreachable overflow-error branch.
aggregator.rs is now at 307/307 lines, no observable behaviour change
on aligned-bucket inputs (which is every input fill_between can be
called with given the call site's preconditions).
Codecov flagged a single uncovered line in crates/wickra-data/src/resample.rs:
line 46, the `self.low = c.low;` assignment inside RolledBar::absorb.
None of the existing resampler tests fed a follow-up candle with a strictly
lower low than the first candle in the bucket, so the `c.low < self.low`
branch never fired. Coverage stayed at 122/123.
Add a small dedicated test that pushes a 10.0-low candle into bucket 0, then
a 8.0-low candle into the same bucket, and asserts the rolled bar's low
reflects the dip. Resample file is now at 123/123 lines, no behavioural change.
Dependabot opened #13 to bump tokio-tungstenite from 0.24 to 0.29 but
the bare-version bump fails to compile: WebSocketConfig became
#[non_exhaustive] starting with 0.27, so the existing struct-literal
construction
let ws_config = WebSocketConfig {
max_message_size: Some(MAX_MESSAGE_SIZE),
max_frame_size: Some(MAX_FRAME_SIZE),
..WebSocketConfig::default()
};
produces
error[E0639]: cannot create non-exhaustive struct using struct expression
Switch to the builder-style setters that 0.29 exposes on the default
value. Semantics are unchanged; both fields still carry the
MAX_MESSAGE_SIZE / MAX_FRAME_SIZE caps from the original config and the
rest of the WebSocketConfig defaults are preserved by starting from
WebSocketConfig::default().
This supersedes #13 — same target version, plus the code change
Dependabot can't make on its own.
Verified locally:
cargo check -p wickra-data --features live-binance # clean
cargo test --workspace --all-features # 630 passed / 0 failed
cargo clippy --workspace --all-targets --all-features -- -D warnings
The tick_aggregator fuzz target found that TickAggregator::fill_between
allocates one placeholder Candle per skipped bucket without bounding the
gap size. An adversarial input (a clock-glitch tick years in the future)
produced an OOM on libFuzzer after malloc(~3 GB):
SUMMARY: libFuzzer: out-of-memory (malloc(3221225472))
A real-world failure mode too: a single bad timestamp from a flaky feed
could OOM the host process even though every individual tick passed
Tick::new validation.
Fix:
- Compute the gap size up-front via saturating arithmetic, before any
allocation, and refuse with Error::Malformed when it exceeds the new
MAX_GAP_FILL_CANDLES = 1_000_000 cap (≈ 1.9 years of contiguous
one-minute bars, well above any realistic missing-data window).
- Reserve the right Vec capacity in advance once we know the gap fits,
avoiding intermediate reallocations.
- Add two regression tests: gap_fill_rejects_runaway_timestamp_jump
(the fuzz scenario) and gap_fill_at_the_cap_succeeds (exact-cap input
still works).
The cap is exposed as pub const so callers can pre-validate their input
without relying on the error string.
The three Rust examples (backtest, fetch_btcusdt, live_binance) used to
live each in their own crate's examples/ dir, splitting the example set
across crates and burying it inside the source tree. Move them into a new
workspace member crate at `examples/rust/` (package `wickra-examples`,
`publish = false`) so all language examples sit under one top-level
`examples/<lang>/` tree.
* `examples/rust/Cargo.toml` declares the per-binary deps (wickra,
wickra-data with the `live-binance` feature always on, serde_json, tokio
for the macro and current-thread runtime).
* `examples/rust/src/bin/{backtest,fetch_btcusdt,live_binance}.rs` are the
three migrated binaries; their doc-comments and the fetch_btcusdt output
path are updated for the new location and run command
(`cargo run -p wickra-examples --bin <name>`).
* Workspace `Cargo.toml` lists the new member; the now-empty
`[dev-dependencies]` extras (`wickra`, `tokio` in wickra-data and
`serde_json` in wickra) that existed only for these examples are dropped.
* The `[[example]] live_binance` table is removed from wickra-data's
manifest since the file moved out.
* README "Languages" + project-layout, examples/README.md, Quickstart-Rust
and Data-Layer are pointed at the new paths and commands.
`cargo build -p wickra-examples` and `cargo run --release -p wickra-examples
--bin backtest -- examples/data/btcusdt-1d.csv` both succeed; the rest of
the workspace (core, data, wickra) builds, clippies (`--all-targets -D
warnings`) and tests (508 core + 28 data + 1 integration + 74+3+1
doctests) all stay green.
Timeframe gained new/millis/seconds/one_minute_ms; add minutes, hours and
days alongside them. Each builds on seconds (minutes(5) -> a 300-second
bucket), consistent with Timeframe::seconds, and guards the multiplication
with checked_mul so an oversized n yields Error::InvalidTimeframe instead
of an overflow panic. A non-positive n is rejected by Timeframe::new.
Each method carries a runnable doctest, and unit tests cover the known
bucket sizes, non-positive rejection and overflow rejection.
The three published crates had no documentation link and no docs.rs
configuration, so wickra-data's feature-gated live-binance module would
not render on docs.rs.
Add documentation = "https://docs.rs/<crate>" and a
[package.metadata.docs.rs] section with all-features = true to
wickra-core, wickra, and wickra-data. No `exclude` is added: each crate
directory contains only src/ (plus benches/examples that are useful
source), so there is nothing irrelevant to drop from the .crate.
OpenBar::into_candle and RolledBar::into_candle built their result with
Candle::new_unchecked, skipping the finiteness check. volume is summed
across every absorbed tick/candle, so a long or large run can drift it
to +inf — and an inf-volume candle would silently poison every
downstream indicator.
Switch both to Candle::new, which validates volume finiteness, and
return Result<Candle>. The OHLC fields are finite and correctly ordered
by construction, so the only invariant Candle::new can reject here is a
non-finite volume. push propagates the error with `?`; both flush
methods now return Result<Option<Candle>> and resample_all pulls the
result through.
push rejected ticks that went backwards across buckets but absorbed any
tick whose timestamp fell inside the open bucket — including one older
than the last tick already absorbed. Such a stale tick silently
overwrote the bar's close with an outdated price.
Track last_ts on OpenBar (set in from_tick, advanced in absorb) and, on
the same-bucket path, reject a tick whose timestamp predates it with
Error::Malformed, leaving the open bar untouched. Ticks that share a
timestamp are still accepted, since several trades can land in the same
millisecond.
Timeframe::floor computed `ts - ts.rem_euclid(bucket)`. For a timestamp
within one bucket of i64::MIN the subtrahend is a positive remainder
and the true boundary lies below i64::MIN, so the subtraction overflowed
and panicked in debug builds.
Switch to saturating_sub: the result clamps to i64::MIN in that
practically unreachable case and stays exact everywhere else. floor
keeps its infallible `-> i64` signature, so neither push path changes.
The CSV reader set has_headers(true) with no trimming and no header
check, so three real-world inputs failed silently or opaquely:
- A file with no header row had its first data row consumed as the
header and silently dropped.
- A leading UTF-8 BOM (Excel exports it) became part of the first
header name, breaking the `timestamp` column mapping.
- Leading/trailing whitespace around values broke serde parsing.
Add a BomStripReader<R> Read adapter that discards a leading EF BB BF,
set csv::Trim::All on the builder, and validate after opening that the
header names every required OHLCV column — a missing column now yields
a clear Error::Malformed instead of a silent misread. open/from_reader
route through a shared build() helper; from_reader and from_csv_reader
now return Result because header validation can fail.
A tick that jumped across one or more empty buckets previously opened
the next non-empty bar directly, so the candle series silently grew
time holes — downstream indicators (EMA, ATR, ...) computed over such a
series drift from one computed over an unbroken series.
Add an opt-in gap-fill mode: with_gap_fill(true) makes push emit a flat
placeholder candle (open == high == low == close = the pre-gap close,
volume = 0) for every skipped bucket. push now returns Result<Vec<Candle>>
so a single tick can yield the closed bar plus its trailing fillers;
the empty vector replaces the former Ok(None). Timestamp overflow while
filling is reported as Error::Malformed. Default behaviour is unchanged
(gaps skipped) and is now documented on the type and on push.
Resampler::push previously closed the open bar and opened a new one for
any candle whose bucket differed from the open bar, including buckets
strictly before it — silently corrupting the output for out-of-order
input. TickAggregator::push already rejects this case with an error.
Change push to return Result<Option<Candle>>: candles in an earlier
bucket than the open bar now yield Error::Malformed, matching the
aggregator. resample_all propagates the error via `?`. The doc comment
keeps the input/output multiple relationship as a documented caller
responsibility, since Resampler does not know the input timeframe.
A 24-hour forced disconnect or a network blip permanently killed the
feed: next_event returned Ok(None)/Err and the stream was dead. The
struct now retains the subscribed symbols, an open() helper rebuilds the
socket, and reconnect() retries with exponential backoff (1s..30s, up to
MAX_RECONNECT_ATTEMPTS). next_event transparently reconnects on a
protocol error, a server close or a read stall, and only reports Ok(None)
after the caller has closed the stream. close() now takes &mut self.
connect() used connect_async with no WebSocketConfig and next_event
awaited the socket with no deadline, so a stalled server hung the feed
forever and an oversized message could force an unbounded allocation.
connect() now passes a WebSocketConfig capping message/frame size, and
next_event wraps the read in a 300s tokio timeout (well above Binance's
~3-minute ping), surfacing a stall as the new Error::Timeout.
BinanceKlineStream had no closed-state flag, so after the server closed
the connection (Ok(None)) a caller could keep calling next_event and
poll a dead socket. A closed: bool is now set when the server closes or
sends a Close frame; next_event short-circuits to Ok(None) once set, and
is_closed() exposes the state.
next_event deserialized every text frame straight into RawWsEnvelope, so
a subscription acknowledgement, heartbeat or error object propagated a
decode Err and killed the feed. Frames are now routed through
parse_frame, which inspects data.e: kline frames yield an event,
everything else is skipped, and an Err is raised only when a genuine
kline frame fails to decode. Adds tests for skipped acks/errors.
The previous `[[example]] path = "../../examples/rust/..."` entries
pointed at files outside the crate root. cargo only packages files
inside the crate directory, so the examples would have been silently
dropped on publish. Moving them to crates/wickra/examples/ and
crates/wickra-data/examples/ keeps them as part of the published
package and lets cargo discover them automatically (no [[example]]
override needed). README and project-layout section updated.
A multi-language technical analysis library: 25 indicators across trend,
momentum, volatility, and volume families, every one a state machine with
O(1) per-tick updates. Batch evaluation is provided by a blanket extension
trait over the streaming primitive, so live trading bots and historical
backtests run the same code path.
What ships in this initial drop:
crates/wickra-core - 25 indicators, Indicator/BatchExt/Chain traits,
OHLCV types with validation; 171 unit tests,
property tests, Wilder/Bollinger textbook tests.
crates/wickra - top-level facade + criterion benches for every
indicator at 1K/10K/100K series sizes.
crates/wickra-data - streaming CSV reader, tick-to-candle aggregator,
multi-timeframe resampler, Binance Spot kline
WebSocket adapter behind feature live-binance;
11 unit + 1 doctest.
bindings/python - PyO3 + maturin, NumPy I/O, type stubs (.pyi),
56 pytest tests including streaming==batch
equivalence, Wilder reference values, lifecycle.
bindings/node - napi-rs native module, TypeScript .d.ts
auto-generated, 7 node --test cases.
bindings/wasm - wasm-bindgen ES module for browser/bundler/Node;
interactive HTML demo at examples/index.html.
examples/ - Python and Rust scripts: backtest, live trading,
parallel multi-asset, multi-timeframe, Binance.
benchmarks/ - cross-library comparison against TA-Lib,
pandas-ta, finta, talipp; Wickra wins every
category by 11-1030x (batch) and 17x+ streaming.
.github/workflows/ - CI matrix (Rust + Python + Node + WASM on
Linux/macOS/Windows), release pipeline for
PyPI wheels and npm.
Indicators (25):
Trend SMA EMA WMA DEMA TEMA HMA KAMA
Momentum RSI MACD Stochastic CCI ROC WilliamsR ADX MFI TRIX
AwesomeOscillator Aroon
Volatility BollingerBands ATR Keltner Donchian PSAR
Volume OBV VWAP (cumulative + rolling)
cargo clippy --workspace --all-targets -D warnings is clean. License: Apache-2.0.
kingchenc