2.6 KiB
2.6 KiB
Usage: finance-data-adapter
Normalize provider-specific market bars into one reproducible OHLCV schema so LawSynth can discover math worlds from price series.
Install
pip install -e plugins/finance-data-adapter
No third-party dependencies are required.
Constructing the adapter
from finance_data_adapter.plugin import FinanceDataAdapter
adapter = FinanceDataAdapter(max_rows=1_000_000)
Request
| key | type | notes |
|---|---|---|
symbol |
str |
1..32 chars; upper-cased in the output |
records |
sequence of mappings | vendor bars |
mapping |
mapping | overrides field names for any OHLCV field |
The default mapping is the identity map over
timestamp, open, high, low, close, volume. Supply overrides only for the
fields your vendor names differently.
Output schema
Each row: symbol, timestamp, open, high, low, close, volume.
timestamp: timezone-aware UTC ISO-8601. Epoch seconds (int/float) or ISO strings (trailingZaccepted) are supported; naive timestamps are rejected.- OHLCV numeric fields are coerced to
float. - Rows are sorted by timestamp; duplicate timestamps are rejected.
- OHLC integrity is enforced:
low <= min(open, close) <= max(open, close) <= high.
Response
{"records": [{...}, ...], "row_count": int, "symbol": str}
Example
adapter.invoke({
"symbol": "aapl",
"mapping": {"timestamp": "t", "open": "o", "high": "h",
"low": "l", "close": "c", "volume": "v"},
"records": [
{"t": 1_704_153_600, "o": 100, "h": 102, "l": 99, "c": 101, "v": 1000},
],
})
# {'records': [{'symbol': 'AAPL', 'timestamp': '2024-01-02T00:00:00+00:00',
# 'open': 100.0, 'high': 102.0, 'low': 99.0, 'close': 101.0,
# 'volume': 1000.0}], 'row_count': 1, 'symbol': 'AAPL'}
Feeding discovery
A normalized close series maps onto a lawsynth.dataset.Dataset column:
rows = adapter.invoke({...})["records"]
time = tuple(float(i) for i, _ in enumerate(rows)) # or convert timestamps to epoch
columns = {"close": tuple(r["close"] for r in rows)}
# Dataset.from_columns(time, columns)
Errors
ValueError for: an invalid symbol, missing OHLC data, OHLC-bound violations,
duplicate timestamps, naive timestamps, or exceeding max_rows. TypeError
when a record is not a mapping.