Lyra/worker/lyra_worker/main.py

import time

import psycopg

from lyra_worker.claim import claim_next
from lyra_worker.config import get_config
from lyra_worker.db import wait_for_db
from lyra_worker.discovery import DiscoveryConfig, run_discovery
from lyra_worker.library import clear_staging_root
from lyra_worker.monitor import MonitorConfig, fulfill_owned_releases, reconcile, sweep
from lyra_worker.pipeline import run_pipeline
from lyra_worker.registry import (
    build_adapters, build_browser, build_probe, build_resolver,
    build_similarity_sources, build_tagger,
)
from lyra_worker.scan import scan_chunk

IDLE_SLEEP = 2.0
MONITOR_TICK_SECONDS = 60.0
DISCOVER_TICK_SECONDS = 300.0  # check the discovery interval every 5 min (interval itself is hours)
DEST_ROOT = "/music"  # must match run_pipeline's default library root
DEFAULT_SCAN_CHUNK = 25  # albums resolved per loop iteration (~25-50s at MB ~1 req/s)
_TRUE = {"1", "true", "yes", "on"}


def _set_config(conn, key: str, value: str) -> None:
    with conn.cursor() as cur:
        cur.execute(
            'INSERT INTO "Config"(key, value, secret, "updatedAt") VALUES (%s, %s, false, now()) '
            'ON CONFLICT (key) DO UPDATE SET value = EXCLUDED.value, "updatedAt" = now()',
            (key, value),
        )
    conn.commit()


def _scan_chunk_size(config) -> int:
    try:
        return max(1, int(config.get("scan.chunkSize", "")))
    except (TypeError, ValueError):
        return DEFAULT_SCAN_CHUNK


def _parse_progress(s: str) -> tuple[int, int]:
    """Parse the internal 'imported/skipped' running-total marker; ('', junk) -> (0, 0)."""
    try:
        i, _, k = (s or "").partition("/")
        return int(i), int(k)
    except (ValueError, AttributeError):
        return 0, 0


def maybe_run_scan(conn, resolver, probe, browser, config, dest_root: str = DEST_ROOT) -> None:
    """Advance a chunked library scan by at most one chunk per call, so the worker loop
    keeps claiming jobs and ticking the monitor between chunks. `scan.requested` starts a
    fresh scan; `scan.inProgress`/`scan.cursor`/`scan.progress` persist state until the walk
    is exhausted, at which point `scan.result` is written (matching the old one-shot output)."""
    requested = str(config.get("scan.requested", "")).strip().lower() in _TRUE
    in_progress = str(config.get("scan.inProgress", "")).strip().lower() in _TRUE
    if not requested and not in_progress:
        return
    if requested and not in_progress:  # start a fresh scan
        _set_config(conn, "scan.inProgress", "true")
        _set_config(conn, "scan.requested", "false")
        _set_config(conn, "scan.cursor", "")
        _set_config(conn, "scan.progress", "0/0")
        cursor, imported_so_far, skipped_so_far = "", 0, 0
    else:  # resume the in-progress scan from its persisted cursor
        cursor = config.get("scan.cursor", "") or ""
        imported_so_far, skipped_so_far = _parse_progress(config.get("scan.progress", ""))
    try:
        new_cursor, imp, skp, done = scan_chunk(
            conn, resolver, probe, browser, dest_root, cursor, _scan_chunk_size(config)
        )
    except Exception as e:  # a scan error must never kill the worker
        print(f"worker: library scan chunk failed: {e}", flush=True)
        conn.rollback()
        _set_config(conn, "scan.inProgress", "false")  # abandon the scan; a new request restarts it
        _set_config(conn, "scan.requested", "false")
        return
    imported_total = imported_so_far + imp
    skipped_total = skipped_so_far + skp
    if done:
        _set_config(conn, "scan.result", f"imported {imported_total}, skipped {skipped_total}")
        _set_config(conn, "scan.progress", "")
        _set_config(conn, "scan.cursor", "")
        _set_config(conn, "scan.inProgress", "false")
        print(f"worker: library scan done — {imported_total} imported, {skipped_total} skipped", flush=True)
    else:
        _set_config(conn, "scan.cursor", new_cursor)
        _set_config(conn, "scan.progress", f"{imported_total}/{skipped_total}")


def _run_discovery(conn, sources, browser, config=None) -> None:
    cfg = DiscoveryConfig.from_config(config if config is not None else get_config(conn))
    result = run_discovery(conn, sources, browser, cfg)
    _set_config(conn, "discover.result", f"artists {result.artists}, albums {result.albums}")
    _set_config(conn, "discover.requested", "false")
    print(f"worker: discovery done — {result.artists} artists, {result.albums} albums", flush=True)


def finish_job(conn, job_id, mcfg) -> None:
    """Feed a finished job's outcome back into its MonitoredRelease. Runs for EVERY job,
    monitor enabled or not: a manually-wanted release must still transition to fulfilled so
    it drops off the Wanted list. A no-op for plain (unlinked) requests."""
    try:
        reconcile(conn, job_id, mcfg)
    except Exception as e:  # reconcile must never take down the worker loop
        print(f"worker: reconcile failed for job {job_id}: {e}", flush=True)
        conn.rollback()


def maybe_run_discovery(conn, sources, browser, config, dcfg, now, last_discover_tick,
                        tick_seconds: float = DISCOVER_TICK_SECONDS) -> float:
    """Run discovery at most once per loop iteration, whether triggered by the schedule
    (enabled + interval elapsed) or a one-shot ``discover.requested`` flag. A single run
    satisfies both, so the two triggers must not fire it twice off the same stale config
    snapshot. Returns the (possibly advanced) last_discover_tick."""
    requested = str(config.get("discover.requested", "")).strip().lower() in _TRUE
    due = dcfg.enabled and now - last_discover_tick >= tick_seconds
    if not (requested or due):
        return last_discover_tick
    try:
        _run_discovery(conn, sources, browser, config)  # also clears discover.requested
    except Exception as e:  # a discovery error must never kill the worker
        print(f"worker: discovery run failed: {e}", flush=True)
        conn.rollback()
        if requested:
            _set_config(conn, "discover.requested", "false")
    return now if due else last_discover_tick


def run_forever() -> None:
    conn = wait_for_db()
    clear_staging_root(DEST_ROOT)  # sweep any staging dirs orphaned by a prior crash
    adapters = build_adapters(get_config(conn))
    resolver = build_resolver()
    tagger = build_tagger()
    browser = build_browser()
    probe = build_probe()
    similarity_sources = build_similarity_sources(get_config(conn))
    # Backfill: releases pressed while the monitor was off never got reconciled, so they linger
    # on the wanted list. Fix any already-owned ones once at startup.
    fixed = fulfill_owned_releases(conn, MonitorConfig.from_config(get_config(conn)))
    if fixed:
        print(f"worker: reconciled {fixed} already-owned release(s) to fulfilled", flush=True)
    print(f"worker: {len(adapters)} adapter(s) enabled: {[a.name for a in adapters]}", flush=True)
    print("worker: waiting for jobs", flush=True)
    last_tick = 0.0
    last_discover_tick = 0.0
    try:
        while True:
            # A transient DB connection loss (e.g. AdminShutdown when Postgres restarts) raises
            # OperationalError from any DB op below; reconnect in-process rather than crashing and
            # relying on the container restart policy. Inner rollbacks on a dead connection also
            # raise OperationalError and compose up to here.
            try:
                config = get_config(conn)
                mcfg = MonitorConfig.from_config(config)
                now = time.monotonic()
                if mcfg.enabled and now - last_tick >= MONITOR_TICK_SECONDS:
                    try:
                        sweep(conn, browser, mcfg)
                    except Exception as e:  # a monitor error must never kill the worker
                        print(f"worker: monitor sweep failed: {e}", flush=True)
                        conn.rollback()
                    last_tick = now

                dcfg = DiscoveryConfig.from_config(config)
                last_discover_tick = maybe_run_discovery(
                    conn, similarity_sources, browser, config, dcfg, now, last_discover_tick
                )

                maybe_run_scan(conn, resolver, probe, browser, config)

                job_id = claim_next(conn)
                if job_id is None:
                    time.sleep(IDLE_SLEEP)
                    continue
                print(f"worker: claimed job {job_id}", flush=True)
                try:
                    run_pipeline(conn, job_id, adapters, resolver=resolver, tagger=tagger,
                                 upgrade_cutoff=mcfg.quality_cutoff)
                except Exception as e:  # one bad job must not take down the worker loop
                    print(f"worker: pipeline failed for job {job_id}: {e}", flush=True)
                    conn.rollback()
                finish_job(conn, job_id, mcfg)
                print(f"worker: finished job {job_id}", flush=True)
            except psycopg.OperationalError as e:
                print(f"worker: db connection lost ({e}); reconnecting...", flush=True)
                try:
                    conn.close()
                except Exception:
                    pass
                conn = wait_for_db()
    finally:
        conn.close()


if __name__ == "__main__":
    run_forever()