prop/lib/microwaveprop/pskr/aggregator.ex
Graham McIntire d67186176f
fix: resolve all dialyzer type errors across project (46→0 project errors)
Type spec fixes:
- duct_usable_* return boolean→float (delegate to duct_usable_for_band)
- sanitize/1 spec includes :unicode error tuples
- match_delete/1 broadened from :ets.match_spec()
- telemetry_event local type replaces :telemetry.event/0
- wgrib2 parse_lon_val_segment corrected to tuple spec
- preloaded Ecto assoc types in get_mission/get_contact! specs

Unmatched returns:
- _ = prefix on Task.start, Oban.insert, Repo.query!, PubSub.subscribe,
  :ets.new, and if-expression returns across 18 files

Pattern match fixes:
- markdown: restructure acc!=[] guard as direct pattern match
- path_compute/pskr/skewt_location_resolver: remove dead clauses
- calibrate.aprs_144: remove unreachable format_float catch-all
- unused.ex: suppress MapSet.union no_opaque

Also: remove unused unicode_util_compat from mix.lock
2026-06-08 17:51:13 -05:00

136 lines
4.7 KiB
Elixir
Raw Blame History

This file contains ambiguous Unicode characters

This file contains Unicode characters that might be confused with other characters. If you think that this is intentional, you can safely ignore this warning. Use the Escape button to reveal them.

defmodule Microwaveprop.Pskr.Aggregator do
@moduledoc """
In-memory accumulator for incoming PSK Reporter spots.
The MQTT firehose can push 500+ messages/second when subscribed
broadly. Writing each spot as a row would (a) saturate the
Postgres write path and (b) bloat the dataset 50× because every
QSO is heard by many reporters. This GenServer keeps a small
state map keyed by `Pskr.path_key/1`, folds each spot into the
matching row with `Pskr.merge_spot/2`, and periodically batches
the whole map into a single `Repo.insert_all/3` upsert.
Conflict handling is additive: when a row with the same
`(hour_utc, band, sender_grid, receiver_grid)` already exists in
the DB, the upsert merges counts, expands the SNR envelope, and
unions the mode set rather than overwriting. That makes flushes
idempotent in the face of overlap (e.g. a flush firing a few
seconds before the hour boundary).
Memory profile: each pending row is small (~200 bytes), and the
unique key space is bounded by `bands × paths_per_hour`. With our
CONUS-anchored topic filter that's typically a few thousand keys
per hour — flushed every minute, the working set never exceeds a
handful of MiB.
"""
use GenServer
import Ecto.Query
alias Microwaveprop.Pskr
alias Microwaveprop.Pskr.SpotHourly
alias Microwaveprop.Repo
require Logger
@default_flush_ms 60_000
@spec start_link(keyword()) :: GenServer.on_start()
def start_link(opts \\ []) do
{name, opts} = Keyword.pop(opts, :name, __MODULE__)
GenServer.start_link(__MODULE__, opts, name: name)
end
@doc "Push one raw MQTT payload (binary JSON) into the accumulator."
@spec ingest(GenServer.server(), binary()) :: :ok
def ingest(server \\ __MODULE__, payload) when is_binary(payload) do
GenServer.cast(server, {:ingest, payload})
end
@doc "Force a synchronous flush. Returns the number of rows written."
@spec flush(GenServer.server()) :: non_neg_integer()
def flush(server \\ __MODULE__) do
GenServer.call(server, :flush, 30_000)
end
@doc "Inspect the current pending row count without flushing."
@spec pending_count(GenServer.server()) :: non_neg_integer()
def pending_count(server \\ __MODULE__), do: GenServer.call(server, :pending_count)
@impl true
def init(opts) do
flush_ms = Keyword.get(opts, :flush_ms, @default_flush_ms)
_ = if flush_ms > 0, do: schedule_flush(flush_ms)
{:ok, %{rows: %{}, parsed: 0, dropped: 0, flush_ms: flush_ms}}
end
@impl true
def handle_cast({:ingest, payload}, state) do
case Pskr.parse_spot(payload) do
{:ok, spot} ->
key = Pskr.path_key(spot)
rows = Map.update(state.rows, key, Pskr.merge_spot(nil, spot), &Pskr.merge_spot(&1, spot))
{:noreply, %{state | rows: rows, parsed: state.parsed + 1}}
{:error, _reason} ->
{:noreply, %{state | dropped: state.dropped + 1}}
end
end
@impl true
def handle_call(:flush, _from, state) do
written = do_flush(state.rows)
{:reply, written, %{state | rows: %{}}}
end
def handle_call(:pending_count, _from, state) do
{:reply, map_size(state.rows), state}
end
@impl true
def handle_info(:flush_tick, state) do
written = do_flush(state.rows)
if written > 0 or state.dropped > 0 do
Logger.info("Pskr.Aggregator flushed #{written} rows (parsed #{state.parsed}, dropped #{state.dropped})")
end
_ = schedule_flush(state.flush_ms)
{:noreply, %{state | rows: %{}, parsed: 0, dropped: 0}}
end
defp schedule_flush(ms) when ms > 0, do: Process.send_after(self(), :flush_tick, ms)
defp schedule_flush(_), do: :ok
defp do_flush(rows) when map_size(rows) == 0, do: 0
defp do_flush(rows) do
now = DateTime.truncate(DateTime.utc_now(), :second)
entries =
rows
|> Map.values()
|> Enum.map(&Map.merge(&1, %{inserted_at: now, updated_at: now}))
{count, _} =
Repo.insert_all(SpotHourly, entries,
on_conflict:
from(s in SpotHourly,
update: [
set: [
spot_count: fragment("? + EXCLUDED.spot_count", s.spot_count),
max_snr_db: fragment("GREATEST(?, EXCLUDED.max_snr_db)", s.max_snr_db),
min_snr_db: fragment("LEAST(?, EXCLUDED.min_snr_db)", s.min_snr_db),
modes: fragment("ARRAY(SELECT DISTINCT unnest(? || EXCLUDED.modes))", s.modes),
first_spot_at: fragment("LEAST(?, EXCLUDED.first_spot_at)", s.first_spot_at),
last_spot_at: fragment("GREATEST(?, EXCLUDED.last_spot_at)", s.last_spot_at),
updated_at: fragment("EXCLUDED.updated_at")
]
]
),
conflict_target: [:hour_utc, :band, :sender_grid, :receiver_grid]
)
count
end
end