Function align_calendar

python/openquant/data.py:249–283

(
    df: pl.DataFrame,
    *,
    interval: str = "1d",
)

def align_calendar(
    df: pl.DataFrame,
    *,
    interval: str = "1d",
) -> pl.DataFrame:
    interval_seconds = _interval_to_seconds(interval)
    if interval_seconds <= 0:
        raise ValueError("interval_seconds must be > 0")

    clean = clean_ohlcv(df).lazy()
    bounds = clean.group_by("symbol").agg(
        pl.col("ts").min().alias("ts_min"),
        pl.col("ts").max().alias("ts_max"),
    )
    calendar = (
        bounds.with_columns(
            pl.datetime_ranges(
                "ts_min",
                "ts_max",
                interval=f"{interval_seconds}s",
                closed="both",
            ).alias("ts")
        )
        .explode("ts")
        .select(["symbol", "ts"])
    )

    out = (
        calendar.join(clean, on=["symbol", "ts"], how="left")
        .with_columns(pl.col("open").is_null().alias("is_missing_bar"))
        .select(CANONICAL_OHLCV_COLUMNS + ["is_missing_bar"])
        .sort(["symbol", "ts"])
        .collect()
    )
    return out