tesser-ledger

tesser-ledger contains the accounting primitives that keep Tesser's balances, realized PnL, and adjustments auditable. It defines the canonical LedgerEntry structure plus repositories that write to SQLite or Apache Parquet, so the same APIs cover low-latency disk storage and analytics exports.

Components

• LedgerEntry / LedgerType – Strongly typed rows that satisfy the accounting identity and capture metadata (exchange, asset, reference id, optional JSON payloads).
• LedgerRepository – Trait for storage engines; the crate ships SqliteLedgerRepository for on-disk durability and ParquetLedgerRepository for analytical pipelines.
• LedgerSequencer – Monotonic sequence allocator ensuring deterministic replay.
• Journal helpers – entries_from_fill converts tesser_core::Fill events into the correct ledger lines for spot or perpetual instruments, including realized PnL and fees.

Quick start

use rust_decimal::Decimal;
use tesser_core::{AssetId, ExchangeId};
use tesser_ledger::{LedgerEntry, LedgerRepository, LedgerType, LedgerSequencer, SqliteLedgerRepository};

let repo = SqliteLedgerRepository::new("ledger.db").expect("open sqlite");
let mut sequencer = LedgerSequencer::new(repo.latest_sequence()?.unwrap_or(0));

let entry = LedgerEntry::new(
    ExchangeId::from("paper"),
    AssetId::from("paper:USDT"),
    Decimal::new(1_000, 0),
    LedgerType::TransferIn,
    "initial_funding",
)
.with_sequence(sequencer.next());

repo.append(&entry).expect("persist ledger line");

To export the same data for downstream analytics, instantiate ParquetLedgerRepository and call append_batch with the same LedgerEntry values. The schema aligns with Arrow so you can load the files directly into Python/Polars.

Querying

LedgerQuery lets you filter the repository by exchange, asset, ledger type, or sequence range. This makes it easy to:

• Reconstruct balance history for a specific settlement currency.
• Audit fees or funding across sessions.
• Generate statements for sub-accounts.

use tesser_core::AssetId;
use tesser_ledger::LedgerQuery;

let entries = repo.query(LedgerQuery::default().with_asset(AssetId::from("binance:USDT")))?;

Integration notes

• Transactions are batched so parents can use append_batch to commit all adjustments derived from a fill atomically.
• entries_from_fill and FillLedgerContext ensure derivative vs. spot products produce the correct cash movements and metadata.
• The crate is no_std-friendly aside from the storage backends; you can depend on just the types if you disable the repositories or gate them behind features in your own crate.