"""`findata` operator CLI.""" from __future__ import annotations import json from pathlib import Path import typer app = typer.Typer(no_args_is_help=True, help="FinData OS operator CLI") @app.callback() def _main() -> None: import atexit from findata_core.db import close_pool atexit.register(close_pool) ingest_app = typer.Typer(no_args_is_help=True) app.add_typer(ingest_app, name="ingest") @app.command() def migrate() -> None: """Apply database migrations.""" from findata_core.db import migrate as _m applied = _m() typer.echo(f"applied: {applied or 'nothing'}") @app.command() def discover(source: str = "all") -> None: """Sanitized live probe of vendor capabilities (no bulk download).""" out = {} if source in ("all", "firstrate"): from findata_market_firstrate import FirstRateMarketConnector out["firstrate"] = FirstRateMarketConnector().discover() if source in ("all", "firstrate_fundamentals"): from findata_fundamentals import FirstRateFundamentalsConnector out["firstrate_fundamentals"] = FirstRateFundamentalsConnector().discover() if source in ("all", "sec_edgar"): from findata_fundamentals import SecEdgarConnector out["sec_edgar"] = SecEdgarConnector().status() typer.echo(json.dumps(out, indent=2, default=str)) @app.command() def worker(once: bool = False, no_scheduler: bool = False, kinds: str | None = None) -> None: """Run the single ingest worker (+ scheduler).""" from findata_ingest.worker import main main(once=once, kinds=kinds.split(",") if kinds else None, with_scheduler=not no_scheduler) @ingest_app.command("metadata") def ingest_metadata(types: str = "stock,etf") -> None: from findata_ingest import queue typer.echo(queue.enqueue("metadata", None, {"types": types.split(",")}, priority=10, created_by="cli")) @ingest_app.command("incremental") def ingest_incremental(type_: str = typer.Option("stock", "--type"), period: str = "day", timeframe: str = "1min", adjustment: str = "UNADJUSTED", force: bool = False) -> None: from findata_ingest import queue from findata_ingest.tasks.common import DATASET_ASSET typer.echo(queue.enqueue("incremental", DATASET_ASSET[type_], {"type": type_, "period": period, "timeframe": timeframe, "adjustment": adjustment, "force": force}, priority=20, created_by="cli")) @ingest_app.command("backfill") def ingest_backfill(ticker_range: str, type_: str = typer.Option("stock", "--type"), timeframe: str = "1min", adjustment: str = "UNADJUSTED", symbols: str | None = None, max_symbols: int = 0) -> None: """Bounded backfill of one ticker_range letter (full archive). Disk budget enforced by the lake.""" from findata_ingest import queue from findata_ingest.tasks.common import DATASET_ASSET for letter in ticker_range.upper(): jid = queue.enqueue("backfill", DATASET_ASSET[type_], {"type": type_, "ticker_range": letter, "timeframe": timeframe, "adjustment": adjustment, "symbols": symbols.split(",") if symbols else None, "max_symbols": max_symbols}, priority=35, created_by="cli", dedupe_key=f"backfill:{type_}:{letter}:{timeframe}:{adjustment}") typer.echo(f"{letter}: {jid or 'already queued'}") @ingest_app.command("fundamentals") def ingest_fundamentals(tickers: str) -> None: from findata_ingest import queue typer.echo(queue.enqueue("fundamentals", "fundamentals.financials", {"tickers": tickers.split(",")}, priority=30, created_by="cli")) @ingest_app.command("derive") def ingest_derive(dates: str | None = None, source: str = "market_bars.stock.1m.raw") -> None: from findata_ingest import queue typer.echo(queue.enqueue("derive", source.replace(".1m.", ".5m."), {"source_dataset": source, "target_interval": "5m", "dates": dates.split(",") if dates else []}, created_by="cli")) @ingest_app.command("compaction") def ingest_compaction(dataset: str = "market_bars.stock.1m.raw", min_age_days: int = 35) -> None: from findata_ingest import queue typer.echo(queue.enqueue("compaction", dataset, {"min_age_days": min_age_days}, priority=60, created_by="cli")) @ingest_app.command("jobs") def ingest_jobs(limit: int = 20) -> None: from findata_core.db import fetch_all for r in fetch_all("SELECT id, kind, dataset_id, status, attempts, created_at, finished_at, left(last_error, 80) err FROM ingest_jobs ORDER BY created_at DESC LIMIT %s", (limit,)): typer.echo(f"{str(r['id'])[:8]} {r['kind']:<12} {r['dataset_id']!s:<28} {r['status']:<9} a={r['attempts']} {r['created_at']:%m-%d %H:%M} {r['err'] or ''}") @app.command() def backup(now: bool = True) -> None: from findata_ingest import queue typer.echo(queue.enqueue("backup", None, {}, priority=5, created_by="cli")) @app.command("restore-drill") def restore_drill(dest: Path = Path("/tmp/findata-restore-drill"), dataset: str = "market_bars.stock.1m.raw") -> None: from findata_core.config import get_settings from findata_ingest.tasks.backup import restore_drill as rd typer.echo(json.dumps(rd(get_settings().backup_target, dest, dataset), indent=2, default=str)) @app.command("create-owner") def create_owner(email: str, password: str | None = None) -> None: """Create (or reset the password of) the owner account; prints the TOTP provisioning URI.""" from findata_api.auth import create_owner as _co typer.echo(json.dumps(_co(email, password), indent=2)) @app.command("create-key") def create_key(email: str, name: str, scopes: str = "read:market,read:fundamentals,query,export") -> None: from findata_api.auth import create_api_key_for_email typer.echo(json.dumps(create_api_key_for_email(email, name, scopes.split(",")), indent=2)) @app.command() def status() -> None: from findata_core.db import fetch_all, fetch_one from findata_ingest.lake import Lake lake = Lake() typer.echo(f"lake {lake.root} free {lake.free_gb():.1f} GB curated {lake.used_bytes('curated')/1e9:.2f} GB raw {lake.used_bytes('raw')/1e9:.2f} GB") for r in fetch_all("SELECT d.id, d.status, d.current_version, v.file_count, v.row_count, v.bytes FROM datasets d LEFT JOIN dataset_versions v ON v.dataset_id=d.id AND v.version=d.current_version ORDER BY d.id"): typer.echo(f" {r['id']:<36} {r['status']:<22} v{r['current_version']:<4} files={r['file_count'] or 0:<6} rows={r['row_count'] or 0:<12} {(r['bytes'] or 0)/1e6:.1f} MB") q = fetch_one("SELECT count(*) FILTER (WHERE status='queued') q, count(*) FILTER (WHERE status='running') r, count(*) FILTER (WHERE status='dead') d FROM ingest_jobs") typer.echo(f"jobs queued={q['q']} running={q['r']} dead={q['d']}") if __name__ == "__main__": app()