1"""Derive 5-minute (or other) bars from validated 1-minute bars, per session_date (delta layout, supersede)."""2from __future__ import annotations34from datetime import date5from typing import Any67import pyarrow as pa89from findata_core.db import connection, transaction10from findata_ingest.lake import Lake, live_files11from findata_ingest.reader import read_bars12from findata_ingest.tasks.common import live_delta_paths_for_dates, upsert_coverage13from findata_quality import aggregate_bars14from findata_schema import Interval151617def run(job: dict[str, Any], lake: Lake, log) -> dict[str, Any]:18 p = job.get("params") or {}19 src = p["source_dataset"]20 dst = job.get("dataset_id") or src.replace(".1m.", ".5m.")21 target = Interval(p.get("target_interval", "5m"))22 dates = [date.fromisoformat(d) for d in p.get("dates", [])]23 symbols = p.get("symbols")24 run_id = job.get("run_id")25 with connection() as conn:26 if not dates:27 # all live source dates for requested symbols28 fl = live_files(conn, src, symbols)29 ds = set()30 for f in fl:31 d = f["date_min"]32 while d <= f["date_max"]:33 ds.add(d)34 d = d.fromordinal(d.toordinal() + 1)35 dates = sorted(ds)36 all_syms = symbols or [r["symbol"] for r in conn.execute("SELECT DISTINCT symbol FROM coverage WHERE dataset_id=%s AND session_date = ANY(%s)", (src, dates)).fetchall()]37 written, rows = [], 038 tables = {}39 for d in dates:40 t, _ = read_bars(src, all_syms, d, d, lake=lake)41 if t.num_rows == 0:42 continue43 agg = aggregate_bars(t, target, Interval.m1)44 tables[d] = agg45 rows += agg.num_rows46 written.append(lake.write_table(dst, agg, "delta", {"session_date": d.isoformat()}))47 if not written:48 return {"rows": 0, "dates": len(dates)}49 with connection() as conn:50 retire = live_delta_paths_for_dates(conn, dst, list(tables)) if not symbols else []51 pub = lake.publish(dst, written, retire, run_id, note=f"derive {target.value} from {src}")52 with transaction() as conn:53 for t in tables.values():54 upsert_coverage(conn, dst, t, target)55 log.info("derive.done", rows=rows, files=len(written), **pub)56 return {"rows": rows, "files": len(written), "dates": len(tables), **pub}57