Merge pull request #534 from BigBodyCobain/fix/aprs-is-resource-safety-533

fix(aprs): stop global APRS-IS load and bound receive lifecycle

.github/workflows/ci.yml

@@ -84,4 +84,5 @@ jobs:
             tests/test_liveuamap_provider.py \
             tests/test_liveuamap_docker_contract.py \
             tests/test_xquik_news.py \
+            tests/test_aprs_is_resource_safety.py \
             -v --tb=short

backend/services/aprs_is_bridge.py

@@ -0,0 +1,430 @@
+"""Resource-safe APRS-IS receive bridge.
+
+The public APRS-IS network is community infrastructure. Shadowbroker therefore
+uses it only when the operator explicitly opts in and supplies a bounded
+geographic range. Public APRS-IS hosts never receive an unbounded/full-feed
+subscription from this client.
+"""
+
+from __future__ import annotations
+
+import logging
+import os
+import random
+import re
+import socket
+import threading
+from collections import deque
+from dataclasses import dataclass
+from datetime import datetime, timezone
+
+from services.sigint_bridge import _decode_aprs_symbol, _parse_aprs_comment, _scan_emergency
+
+logger = logging.getLogger("services.sigint.aprs")
+
+_TRUE_VALUES = {"1", "true", "yes", "on"}
+_PUBLIC_APRS_SUFFIXES = (".aprs2.net", ".aprs-is.net", ".aprs.net")
+_PUBLIC_APRS_HOSTS = {"rotate.aprs2.net", "srvr.aprs-is.net", "rotate.aprs.net"}
+_DEFAULT_HOST = "rotate.aprs2.net"
+_DEFAULT_PORT = 14580
+_DEFAULT_RADIUS_KM = 100.0
+_MAX_PUBLIC_RADIUS_KM = 500.0
+_DEFAULT_MAX_SIGNALS = 5000
+_MAX_SIGNAL_CAP = 20000
+_SIGNAL_MAX_AGE_S = 600.0
+_RECONNECT_BASE_S = 30.0
+_RECONNECT_MAX_S = 900.0
+_CALLSIGN_RE = re.compile(r"^[A-Z0-9-]{1,9}$")
+
+
+@dataclass(frozen=True)
+class APRSISConfig:
+    host: str
+    port: int
+    login: str
+    filter_expr: str
+    private_server: bool
+
+
+def _env_enabled(name: str, default: bool = False) -> bool:
+    raw = os.getenv(name)
+    if raw is None:
+        return default
+    return raw.strip().lower() in _TRUE_VALUES
+
+
+def _bounded_int(name: str, default: int, low: int, high: int) -> int:
+    raw = (os.getenv(name) or "").strip()
+    if not raw:
+        return default
+    try:
+        value = int(raw)
+    except ValueError:
+        return default
+    return max(low, min(value, high))
+
+
+def aprs_is_enabled() -> bool:
+    """Return whether APRS-IS networking is explicitly enabled."""
+    return _env_enabled("APRS_IS_ENABLED", False)
+
+
+def _is_public_aprs_host(host: str) -> bool:
+    host = host.strip().lower().rstrip(".")
+    return host in _PUBLIC_APRS_HOSTS or host.endswith(_PUBLIC_APRS_SUFFIXES)
+
+
+def _parse_public_range() -> tuple[float, float, float]:
+    lat_raw = (os.getenv("APRS_IS_LAT") or "").strip()
+    lon_raw = (os.getenv("APRS_IS_LON") or "").strip()
+    if not lat_raw or not lon_raw:
+        raise ValueError(
+            "APRS_IS_LAT and APRS_IS_LON are required for public APRS-IS; "
+            "global/unbounded subscriptions are intentionally unsupported"
+        )
+    try:
+        lat = float(lat_raw)
+        lon = float(lon_raw)
+    except ValueError as exc:
+        raise ValueError("APRS_IS_LAT/APRS_IS_LON must be decimal degrees") from exc
+    if not -90.0 <= lat <= 90.0 or not -180.0 <= lon <= 180.0:
+        raise ValueError("APRS_IS_LAT/APRS_IS_LON are outside valid coordinate bounds")
+
+    radius_raw = (os.getenv("APRS_IS_RADIUS_KM") or str(_DEFAULT_RADIUS_KM)).strip()
+    try:
+        radius = float(radius_raw)
+    except ValueError as exc:
+        raise ValueError("APRS_IS_RADIUS_KM must be numeric") from exc
+    if not 1.0 <= radius <= _MAX_PUBLIC_RADIUS_KM:
+        raise ValueError(
+            f"APRS_IS_RADIUS_KM must be between 1 and {_MAX_PUBLIC_RADIUS_KM:g} km "
+            "when using public APRS-IS"
+        )
+    return lat, lon, radius
+
+
+def aprs_connection_config() -> APRSISConfig | None:
+    """Build the receive configuration, failing closed on unsafe public filters.
+
+    Public APRS-IS requires an explicit geographic center and enforces a 500 km
+    maximum range. Operators running their own APRS-IS server may set
+    APRS_IS_PRIVATE_SERVER=true; only then may APRS_IS_FILTER be arbitrary or
+    blank. The private-server override is rejected for known public APRS hosts.
+    """
+    if not aprs_is_enabled():
+        return None
+
+    host = (os.getenv("APRS_IS_HOST") or _DEFAULT_HOST).strip()
+    if not host or any(ch.isspace() for ch in host):
+        raise ValueError("APRS_IS_HOST is invalid")
+    port = _bounded_int("APRS_IS_PORT", _DEFAULT_PORT, 1, 65535)
+    private_server = _env_enabled("APRS_IS_PRIVATE_SERVER", False)
+
+    if private_server:
+        if _is_public_aprs_host(host):
+            raise ValueError("APRS_IS_PRIVATE_SERVER cannot be used with a public APRS-IS host")
+        filter_expr = (os.getenv("APRS_IS_FILTER") or "").strip()
+        if "\r" in filter_expr or "\n" in filter_expr:
+            raise ValueError("APRS_IS_FILTER cannot contain line breaks")
+    else:
+        lat, lon, radius = _parse_public_range()
+        filter_expr = f"r/{lat:.5f}/{lon:.5f}/{radius:g}"
+
+    callsign = (os.getenv("APRS_IS_CALLSIGN") or "N0CALL").strip().upper()
+    if not _CALLSIGN_RE.fullmatch(callsign):
+        raise ValueError("APRS_IS_CALLSIGN must be 1-9 APRS-safe characters")
+
+    login = f"user {callsign} pass -1 vers ShadowBroker 1.0"
+    if filter_expr:
+        login += f" filter {filter_expr}"
+    login += "\r\n"
+    return APRSISConfig(
+        host=host,
+        port=port,
+        login=login,
+        filter_expr=filter_expr,
+        private_server=private_server,
+    )
+
+
+class APRSISBridge:
+    """Long-lived, bounded, opt-in APRS-IS receive client."""
+
+    CONFIDENCE = 0.7
+
+    def __init__(self) -> None:
+        max_signals = _bounded_int(
+            "APRS_IS_MAX_SIGNALS",
+            _DEFAULT_MAX_SIGNALS,
+            100,
+            _MAX_SIGNAL_CAP,
+        )
+        self.signals: deque[dict] = deque(maxlen=max_signals)
+        self._thread: threading.Thread | None = None
+        self._stop = threading.Event()
+        self._socket_lock = threading.Lock()
+        self._socket: socket.socket | None = None
+        self._config: APRSISConfig | None = None
+        self._connected = False
+        self._last_error = ""
+
+    def is_running(self) -> bool:
+        return bool(self._thread and self._thread.is_alive() and not self._stop.is_set())
+
+    def status(self) -> dict[str, object]:
+        return {
+            "enabled": aprs_is_enabled(),
+            "running": self.is_running(),
+            "connected": self._connected,
+            "buffered": len(self.signals),
+            "last_error": self._last_error,
+            "host": self._config.host if self._config else "",
+            "filter": self._config.filter_expr if self._config else "",
+        }
+
+    def reconcile(self, layer_enabled: bool) -> None:
+        """Match the network bridge to current operator layer/config state."""
+        if not layer_enabled:
+            self.stop()
+            return
+        try:
+            config = aprs_connection_config()
+        except ValueError as exc:
+            self._last_error = str(exc)
+            self.stop()
+            logger.warning("APRS-IS disabled by safety validation: %s", exc)
+            return
+        if config is None:
+            self._last_error = "APRS_IS_ENABLED is false"
+            self.stop()
+            return
+
+        if self.is_running() and self._config == config:
+            return
+        if self.is_running() or self._config != config:
+            self.stop()
+        self._config = config
+        self._last_error = ""
+        self.start()
+
+    def start(self) -> None:
+        if self._config is None:
+            try:
+                self._config = aprs_connection_config()
+            except ValueError as exc:
+                self._last_error = str(exc)
+                logger.warning("APRS-IS not started: %s", exc)
+                return
+        if self._config is None:
+            return
+        if self._thread and self._thread.is_alive():
+            if not self._stop.is_set():
+                return
+            self._thread.join(timeout=2.0)
+            if self._thread.is_alive():
+                logger.warning("APRS-IS bridge is still stopping; start deferred")
+                return
+        self._stop.clear()
+        self._thread = threading.Thread(target=self._run, daemon=True, name="aprs-safe-bridge")
+        self._thread.start()
+        logger.info(
+            "APRS-IS bridge started on %s:%s with bounded filter %s",
+            self._config.host,
+            self._config.port,
+            self._config.filter_expr or "server-default/private",
+        )
+
+    def stop(self) -> None:
+        self._stop.set()
+        self._connected = False
+        with self._socket_lock:
+            sock = self._socket
+            self._socket = None
+        if sock is not None:
+            try:
+                sock.shutdown(socket.SHUT_RDWR)
+            except OSError:
+                pass
+            try:
+                sock.close()
+            except OSError:
+                pass
+        thread = self._thread
+        if thread and thread.is_alive() and thread is not threading.current_thread():
+            thread.join(timeout=2.0)
+
+    @staticmethod
+    def _reconnect_delay(failures: int) -> float:
+        exponent = max(0, min(failures - 1, 5))
+        base = min(_RECONNECT_BASE_S * (2**exponent), _RECONNECT_MAX_S)
+        return base * random.uniform(0.85, 1.15)
+
+    def _run(self) -> None:
+        failures = 0
+        while not self._stop.is_set():
+            config = self._config
+            if config is None:
+                return
+            try:
+                self._connect_and_read(config)
+                if self._stop.is_set():
+                    return
+                failures += 1
+                self._last_error = "connection_closed"
+            except Exception as exc:
+                if self._stop.is_set():
+                    return
+                failures += 1
+                self._last_error = f"{type(exc).__name__}: {exc}"
+                logger.warning("APRS-IS connection error: %s", exc)
+            delay = self._reconnect_delay(failures)
+            logger.info("APRS-IS reconnect backoff %.0fs after %d failure(s)", delay, failures)
+            self._stop.wait(delay)
+
+    def _connect_and_read(self, config: APRSISConfig) -> None:
+        sock = socket.create_connection((config.host, config.port), timeout=30)
+        with self._socket_lock:
+            if self._stop.is_set():
+                sock.close()
+                return
+            self._socket = sock
+        try:
+            sock.settimeout(30)
+            banner = sock.recv(512).decode("utf-8", errors="replace")
+            logger.info("APRS-IS: %s", banner.strip())
+            sock.sendall(config.login.encode("ascii"))
+            self._connected = True
+            self._last_error = ""
+            buf = b""
+            while not self._stop.is_set():
+                try:
+                    chunk = sock.recv(4096)
+                except socket.timeout:
+                    sock.sendall(b"#keepalive\r\n")
+                    continue
+                if not chunk:
+                    break
+                buf += chunk
+                while b"\n" in buf:
+                    line_bytes, buf = buf.split(b"\n", 1)
+                    line_bytes = line_bytes.strip()
+                    if not line_bytes or line_bytes.startswith(b"#"):
+                        continue
+                    self._parse_packet(self._decode_line(line_bytes))
+        finally:
+            self._connected = False
+            with self._socket_lock:
+                if self._socket is sock:
+                    self._socket = None
+            try:
+                sock.close()
+            except OSError:
+                pass
+
+    @staticmethod
+    def _decode_line(raw_bytes: bytes) -> str:
+        try:
+            return raw_bytes.decode("utf-8")
+        except UnicodeDecodeError:
+            pass
+        try:
+            return raw_bytes.decode("gbk")
+        except UnicodeDecodeError:
+            pass
+        return raw_bytes.decode("latin-1")
+
+    def _parse_packet(self, raw: str) -> None:
+        try:
+            if ":" not in raw:
+                return
+            header, payload = raw.split(":", 1)
+            callsign = header.split(">")[0].strip()
+            if not callsign or callsign == "N0CALL" or not payload or payload[0] not in "!@/=":
+                return
+
+            pos = payload[1:]
+            lat = self._parse_lat(pos[:8])
+            lng = self._parse_lng(pos[9:18])
+            if lat is None or lng is None:
+                return
+            symbol = pos[8] + pos[18] if len(pos) > 18 else ""
+            comment = pos[19:].strip() if len(pos) > 19 else ""
+            meta = _parse_aprs_comment(comment)
+            signal = {
+                "callsign": callsign,
+                "lat": lat,
+                "lng": lng,
+                "source": "aprs",
+                "confidence": self.CONFIDENCE,
+                "timestamp": datetime.now(timezone.utc).isoformat(),
+                "raw_message": raw[:200],
+                "symbol": symbol,
+                "station_type": _decode_aprs_symbol(symbol),
+                "comment": comment[:100],
+            }
+            for key in (
+                "frequency",
+                "altitude_ft",
+                "speed_knots",
+                "battery_v",
+                "power_watts",
+                "status",
+            ):
+                if meta.get(key) is not None:
+                    signal[key] = meta[key]
+            if meta.get("speed_knots"):
+                signal["course"] = meta.get("course", 0)
+            emergency_kw = _scan_emergency(comment) or _scan_emergency(signal.get("status", ""))
+            if emergency_kw:
+                signal["emergency"] = True
+                signal["emergency_keyword"] = emergency_kw
+            self.signals.append(signal)
+        except (ValueError, IndexError):
+            return
+
+    def get_signals(self) -> list[dict]:
+        """Return only recent APRS signals while keeping memory hard-bounded."""
+        now = datetime.now(timezone.utc)
+        result: list[dict] = []
+        for signal in list(self.signals):
+            try:
+                timestamp = datetime.fromisoformat(str(signal["timestamp"]).replace("Z", "+00:00"))
+                if (now - timestamp).total_seconds() > _SIGNAL_MAX_AGE_S:
+                    continue
+            except (KeyError, TypeError, ValueError):
+                continue
+            result.append(dict(signal))
+        return result
+
+    @staticmethod
+    def _parse_lat(value: str) -> float | None:
+        try:
+            if len(value) < 8:
+                return None
+            degrees = int(value[:2])
+            minutes = float(value[2:7])
+            direction = value[7].upper()
+            lat = degrees + minutes / 60.0
+            if direction == "S":
+                lat = -lat
+            return round(lat, 5) if -90 <= lat <= 90 else None
+        except (ValueError, IndexError):
+            return None
+
+    @staticmethod
+    def _parse_lng(value: str) -> float | None:
+        try:
+            if len(value) < 9:
+                return None
+            degrees = int(value[:3])
+            minutes = float(value[3:8])
+            direction = value[8].upper()
+            lng = degrees + minutes / 60.0
+            if direction == "W":
+                lng = -lng
+            return round(lng, 5) if -180 <= lng <= 180 else None
+        except (ValueError, IndexError):
+            return None
+
+
+aprs_is_bridge = APRSISBridge()

backend/services/fetchers/sigint.py

@@ -1,12 +1,15 @@
 """SIGINT fetcher — pulls latest signals from the SIGINT Grid into latest_data.
 
-Merges live MQTT signals with cached Meshtastic map API nodes.
+Merges live APRS/MQTT/JS8Call signals with cached Meshtastic map API nodes.
-Live MQTT signals always take priority (fresher) — API nodes fill in the gaps
+Each external bridge is reconciled independently so enabling one transport does
-for the thousands of nodes our MQTT listener hasn't heard yet.
+not implicitly start another.
 """
 
 import logging
+
+from services.aprs_is_bridge import aprs_is_bridge
 from services.fetchers._store import latest_data, _data_lock, _mark_fresh
+from services.meshtastic_mqtt_settings import mqtt_bridge_enabled
 
 logger = logging.getLogger("services.data_fetcher")
 
@@ -23,7 +26,7 @@ def _merge_sigint_snapshot(
 
     # Shallow-copy every entry so the published list owns its own dicts. The
     # inputs alias objects that other threads keep mutating in place: live
-    # signals are the SIGINT bridge's own dicts (updated as packets arrive),
+    # signals are the SIGINT bridges' own dicts (updated as packets arrive),
     # and api_nodes are the same objects published under latest_data
     # ["meshtastic_map_nodes"]. Publishing those references into
     # latest_data["sigint"] lets a concurrent mutation race the lock-free
@@ -69,7 +72,11 @@ def build_sigint_snapshot() -> tuple[list[dict], dict[str, object], dict[str, in
 
     from services.sigint_bridge import sigint_grid
 
+    # The legacy APRS member in SIGINTGrid is intentionally never started by
+    # the production fetch path after #533. Safe APRS-IS receive traffic comes
+    # from aprs_is_bridge, which requires explicit bounded configuration.
     live_signals = sigint_grid.get_all_signals()
+    live_signals.extend(aprs_is_bridge.get_signals())
     with _data_lock:
         api_nodes = list(latest_data.get("meshtastic_map_nodes", []))
     merged = _merge_sigint_snapshot(live_signals, api_nodes)
@@ -90,22 +97,54 @@ def refresh_sigint_snapshot() -> tuple[list[dict], dict[str, object], dict[str,
     return signals, channel_stats, totals
 
 
-def fetch_sigint():
+def _reconcile_sigint_bridges(aprs_requested: bool, mesh_requested: bool) -> None:
-    """Fetch all signals from the SIGINT Grid, merge with Meshtastic map nodes."""
+    """Start/stop each bridge independently from current operator state."""
-    from services.fetchers._store import is_any_active
-
-    if not is_any_active("sigint_meshtastic", "sigint_aprs"):
-        return
     from services.sigint_bridge import sigint_grid
 
-    # Start bridges on first call (idempotent)
+    # Defense-in-depth: the old SIGINTGrid APRS client used an effectively
-    sigint_grid.start()
+    # global range subscription. It is no longer part of the production fetch
+    # path; force it stopped even if another caller started it accidentally.
+    sigint_grid.aprs.stop()
+
+    aprs_is_bridge.reconcile(aprs_requested)
+
+    try:
+        mesh_network_enabled = mqtt_bridge_enabled()
+    except Exception:
+        mesh_network_enabled = False
+    if mesh_requested and mesh_network_enabled:
+        sigint_grid.mesh.start()
+    else:
+        sigint_grid.mesh.stop()
+
+    # JS8Call is localhost-only and historically accompanies the SIGINT view.
+    # Keep that behavior without coupling it to either public network bridge.
+    if aprs_requested or mesh_requested:
+        sigint_grid.js8.start()
+    else:
+        sigint_grid.js8.stop()
+
+
+def fetch_sigint():
+    """Refresh SIGINT while matching bridge lifecycles to operator settings."""
+    from services.fetchers._store import effective_layers
+
+    layers = effective_layers()
+    aprs_requested = bool(layers.get("sigint_aprs", False))
+    mesh_requested = bool(layers.get("sigint_meshtastic", False))
+    _reconcile_sigint_bridges(aprs_requested, mesh_requested)
+
+    if not aprs_requested and not mesh_requested:
+        return
 
     signals, channel_stats, totals = refresh_sigint_snapshot()
+    from services.sigint_bridge import sigint_grid
 
-    status = sigint_grid.status
     logger.info(
-        f"SIGINT: {len(signals)} signals "
+        "SIGINT: %d signals (APRS:%d MESH:%d JS8:%d MAP:%d)",
-        f"(APRS:{status['aprs']} MESH:{status['meshtastic']} "
+        len(signals),
-        f"JS8:{status['js8call']} MAP:{totals['meshtastic_map']})"
+        totals["aprs"],
+        totals["meshtastic_live"],
+        len(sigint_grid.js8.signals),
+        totals["meshtastic_map"],
     )

backend/tests/test_aprs_is_resource_safety.py

@@ -0,0 +1,201 @@
+"""Regression coverage for APRS-IS resource safety (#533)."""
+
+from __future__ import annotations
+
+from unittest.mock import MagicMock
+
+import pytest
+
+from services.aprs_is_bridge import (
+    APRSISBridge,
+    aprs_connection_config,
+)
+
+
+_APRS_ENV = (
+    "APRS_IS_ENABLED",
+    "APRS_IS_HOST",
+    "APRS_IS_PORT",
+    "APRS_IS_PRIVATE_SERVER",
+    "APRS_IS_FILTER",
+    "APRS_IS_LAT",
+    "APRS_IS_LON",
+    "APRS_IS_RADIUS_KM",
+    "APRS_IS_CALLSIGN",
+    "APRS_IS_MAX_SIGNALS",
+)
+
+
+def _clear_aprs_env(monkeypatch: pytest.MonkeyPatch) -> None:
+    for name in _APRS_ENV:
+        monkeypatch.delenv(name, raising=False)
+
+
+def _enable_public(monkeypatch: pytest.MonkeyPatch) -> None:
+    _clear_aprs_env(monkeypatch)
+    monkeypatch.setenv("APRS_IS_ENABLED", "true")
+    monkeypatch.setenv("APRS_IS_LAT", "40.7128")
+    monkeypatch.setenv("APRS_IS_LON", "-74.0060")
+
+
+def test_aprs_is_is_network_off_by_default(monkeypatch: pytest.MonkeyPatch) -> None:
+    _clear_aprs_env(monkeypatch)
+    assert aprs_connection_config() is None
+
+
+def test_public_aprs_uses_bounded_geographic_filter(monkeypatch: pytest.MonkeyPatch) -> None:
+    _enable_public(monkeypatch)
+    monkeypatch.setenv("APRS_IS_RADIUS_KM", "250")
+
+    config = aprs_connection_config()
+
+    assert config is not None
+    assert config.host == "rotate.aprs2.net"
+    assert config.port == 14580
+    assert config.filter_expr == "r/40.71280/-74.00600/250"
+    assert "filter r/40.71280/-74.00600/250" in config.login
+    assert "25000" not in config.login
+
+
+def test_public_aprs_rejects_global_scale_radius(monkeypatch: pytest.MonkeyPatch) -> None:
+    _enable_public(monkeypatch)
+    monkeypatch.setenv("APRS_IS_RADIUS_KM", "25000")
+
+    with pytest.raises(ValueError, match="between 1 and 500 km"):
+        aprs_connection_config()
+
+
+def test_public_aprs_requires_explicit_center(monkeypatch: pytest.MonkeyPatch) -> None:
+    _clear_aprs_env(monkeypatch)
+    monkeypatch.setenv("APRS_IS_ENABLED", "true")
+
+    with pytest.raises(ValueError, match="APRS_IS_LAT and APRS_IS_LON"):
+        aprs_connection_config()
+
+
+def test_private_full_feed_override_cannot_target_public_aprs(monkeypatch: pytest.MonkeyPatch) -> None:
+    _clear_aprs_env(monkeypatch)
+    monkeypatch.setenv("APRS_IS_ENABLED", "true")
+    monkeypatch.setenv("APRS_IS_PRIVATE_SERVER", "true")
+    monkeypatch.setenv("APRS_IS_HOST", "rotate.aprs2.net")
+
+    with pytest.raises(ValueError, match="cannot be used with a public APRS-IS host"):
+        aprs_connection_config()
+
+
+def test_private_operator_server_may_use_its_own_filter(monkeypatch: pytest.MonkeyPatch) -> None:
+    _clear_aprs_env(monkeypatch)
+    monkeypatch.setenv("APRS_IS_ENABLED", "true")
+    monkeypatch.setenv("APRS_IS_PRIVATE_SERVER", "true")
+    monkeypatch.setenv("APRS_IS_HOST", "aprs.internal.example")
+    monkeypatch.setenv("APRS_IS_FILTER", "t/p")
+
+    config = aprs_connection_config()
+
+    assert config is not None
+    assert config.private_server is True
+    assert config.filter_expr == "t/p"
+    assert config.login.endswith("filter t/p\r\n")
+
+
+def test_aprs_buffer_is_hard_bounded(monkeypatch: pytest.MonkeyPatch) -> None:
+    _clear_aprs_env(monkeypatch)
+    monkeypatch.setenv("APRS_IS_MAX_SIGNALS", "100")
+    bridge = APRSISBridge()
+
+    for index in range(150):
+        bridge.signals.append({"id": index})
+
+    assert bridge.signals.maxlen == 100
+    assert len(bridge.signals) == 100
+    assert bridge.signals[0]["id"] == 50
+
+
+def test_reconnect_backoff_is_exponential_and_capped(monkeypatch: pytest.MonkeyPatch) -> None:
+    monkeypatch.setattr("services.aprs_is_bridge.random.uniform", lambda _a, _b: 1.0)
+
+    assert APRSISBridge._reconnect_delay(1) == 30.0
+    assert APRSISBridge._reconnect_delay(2) == 60.0
+    assert APRSISBridge._reconnect_delay(3) == 120.0
+    assert APRSISBridge._reconnect_delay(4) == 240.0
+    assert APRSISBridge._reconnect_delay(5) == 480.0
+    assert APRSISBridge._reconnect_delay(6) == 900.0
+    assert APRSISBridge._reconnect_delay(99) == 900.0
+
+
+def test_invalid_public_config_fails_closed_without_starting_thread(
+    monkeypatch: pytest.MonkeyPatch,
+) -> None:
+    _clear_aprs_env(monkeypatch)
+    monkeypatch.setenv("APRS_IS_ENABLED", "true")
+    bridge = APRSISBridge()
+
+    bridge.reconcile(True)
+
+    assert bridge.is_running() is False
+    assert "APRS_IS_LAT" in bridge.status()["last_error"]
+
+
+def test_sigint_lifecycle_does_not_start_legacy_aprs_when_only_mesh_is_requested(
+    monkeypatch: pytest.MonkeyPatch,
+) -> None:
+    from services.fetchers import sigint as sigint_fetcher
+    from services.sigint_bridge import sigint_grid
+
+    legacy_stop = MagicMock()
+    mesh_start = MagicMock()
+    mesh_stop = MagicMock()
+    js8_start = MagicMock()
+    safe_reconcile = MagicMock()
+
+    monkeypatch.setattr(sigint_grid.aprs, "stop", legacy_stop)
+    monkeypatch.setattr(sigint_grid.mesh, "start", mesh_start)
+    monkeypatch.setattr(sigint_grid.mesh, "stop", mesh_stop)
+    monkeypatch.setattr(sigint_grid.js8, "start", js8_start)
+    monkeypatch.setattr(sigint_fetcher.aprs_is_bridge, "reconcile", safe_reconcile)
+    monkeypatch.setattr(sigint_fetcher, "mqtt_bridge_enabled", lambda: True)
+
+    sigint_fetcher._reconcile_sigint_bridges(aprs_requested=False, mesh_requested=True)
+
+    legacy_stop.assert_called_once_with()
+    safe_reconcile.assert_called_once_with(False)
+    mesh_start.assert_called_once_with()
+    mesh_stop.assert_not_called()
+    js8_start.assert_called_once_with()
+
+
+def test_turning_aprs_layer_off_stops_aprs_independently(
+    monkeypatch: pytest.MonkeyPatch,
+) -> None:
+    from services.fetchers import sigint as sigint_fetcher
+    from services.sigint_bridge import sigint_grid
+
+    safe_reconcile = MagicMock()
+    monkeypatch.setattr(sigint_fetcher.aprs_is_bridge, "reconcile", safe_reconcile)
+    monkeypatch.setattr(sigint_fetcher, "mqtt_bridge_enabled", lambda: False)
+    monkeypatch.setattr(sigint_grid.aprs, "stop", MagicMock())
+    monkeypatch.setattr(sigint_grid.mesh, "stop", MagicMock())
+    monkeypatch.setattr(sigint_grid.js8, "start", MagicMock())
+
+    sigint_fetcher._reconcile_sigint_bridges(aprs_requested=False, mesh_requested=True)
+
+    safe_reconcile.assert_called_once_with(False)
+
+
+def test_fetch_path_never_calls_legacy_grid_start(monkeypatch: pytest.MonkeyPatch) -> None:
+    from services.fetchers import sigint as sigint_fetcher
+    from services.fetchers import _store
+    from services.sigint_bridge import sigint_grid
+
+    monkeypatch.setattr(
+        _store,
+        "effective_layers",
+        lambda: {"sigint_aprs": False, "sigint_meshtastic": False},
+    )
+    legacy_start = MagicMock(side_effect=AssertionError("legacy grid start must not run"))
+    monkeypatch.setattr(sigint_grid, "start", legacy_start)
+    monkeypatch.setattr(sigint_fetcher, "_reconcile_sigint_bridges", MagicMock())
+
+    sigint_fetcher.fetch_sigint()
+
+    legacy_start.assert_not_called()

docker-compose.yml

@@ -80,6 +80,18 @@ services:
       - MESH_MQTT_EXTRA_ROOTS=${MESH_MQTT_EXTRA_ROOTS:-}
       - MESH_MQTT_EXTRA_TOPICS=${MESH_MQTT_EXTRA_TOPICS:-}
       - MESHTASTIC_OPERATOR_CALLSIGN=${MESHTASTIC_OPERATOR_CALLSIGN:-}
+      # APRS-IS receive is opt-in and geographically bounded on public servers.
+      # Configure a center before enabling; public radius is hard-capped at 500 km.
+      - APRS_IS_ENABLED=${APRS_IS_ENABLED:-false}
+      - APRS_IS_HOST=${APRS_IS_HOST:-rotate.aprs2.net}
+      - APRS_IS_PORT=${APRS_IS_PORT:-14580}
+      - APRS_IS_LAT=${APRS_IS_LAT:-}
+      - APRS_IS_LON=${APRS_IS_LON:-}
+      - APRS_IS_RADIUS_KM=${APRS_IS_RADIUS_KM:-100}
+      - APRS_IS_CALLSIGN=${APRS_IS_CALLSIGN:-N0CALL}
+      - APRS_IS_MAX_SIGNALS=${APRS_IS_MAX_SIGNALS:-5000}
+      - APRS_IS_PRIVATE_SERVER=${APRS_IS_PRIVATE_SERVER:-false}
+      - APRS_IS_FILTER=${APRS_IS_FILTER:-}
       # The bundled Docker UI talks to the backend across Docker's private bridge.
       # Treat that bridge as local operator access while ports remain bound to 127.0.0.1 by default.
       - SHADOWBROKER_TRUST_DOCKER_BRIDGE_LOCAL_OPERATOR=${SHADOWBROKER_TRUST_DOCKER_BRIDGE_LOCAL_OPERATOR:-1}

docs/APRS_IS.md

@@ -0,0 +1,59 @@
+# APRS-IS receive configuration
+
+Shadowbroker treats public APRS-IS as community infrastructure. Receive access is therefore **off by default** and public-server subscriptions must be geographically bounded.
+
+## Public APRS-IS (recommended client mode)
+
+Set a geographic center and a reasonable radius in `.env`:
+
+```env
+APRS_IS_ENABLED=true
+APRS_IS_LAT=40.7128
+APRS_IS_LON=-74.0060
+APRS_IS_RADIUS_KM=100
+```
+
+Docker Compose passes these values to the backend automatically. The default host is `rotate.aprs2.net:14580`, the APRS-IS user-defined filtered client port.
+
+Shadowbroker constructs a server-side range filter of the form:
+
+```text
+r/<latitude>/<longitude>/<radius-km>
+```
+
+For public APRS-IS hosts, `APRS_IS_RADIUS_KM` must be between **1 and 500 km**. Missing coordinates, invalid coordinates, or a larger radius fail closed: the APRS receive bridge does not connect.
+
+Optional public-client settings:
+
+```env
+APRS_IS_HOST=rotate.aprs2.net
+APRS_IS_PORT=14580
+APRS_IS_CALLSIGN=N0CALL
+APRS_IS_MAX_SIGNALS=5000
+```
+
+`APRS_IS_MAX_SIGNALS` is hard-capped at 20,000 and only recent observations are published.
+
+## Dedicated/private APRS-IS server
+
+Operators who run or control their own APRS-IS aggregation server can opt into private-server mode:
+
+```env
+APRS_IS_ENABLED=true
+APRS_IS_PRIVATE_SERVER=true
+APRS_IS_HOST=aprs.internal.example
+APRS_IS_PORT=14580
+APRS_IS_FILTER=t/p
+```
+
+In private-server mode, `APRS_IS_FILTER` may be any filter accepted by that server, or blank to use the server's default. Shadowbroker refuses to enable private/full-feed mode when the configured host is a known public APRS-IS domain such as `*.aprs2.net`, `*.aprs-is.net`, or `*.aprs.net`.
+
+Use dedicated infrastructure for genuinely global/full-feed APRS collection. Do not point an unbounded client at public Tier-2 filtered servers.
+
+## Runtime behavior
+
+- APRS and Meshtastic now have independent connection lifecycles.
+- Turning the APRS layer off stops the APRS receive bridge; leaving Meshtastic on does not restart it.
+- Failed APRS connections use exponential reconnect backoff with jitter (30 seconds up to 15 minutes) instead of reconnecting every 15 seconds forever.
+- APRS receive memory is bounded, and only observations from the recent retention window are published.
+- APRS transmit remains a separate operator-initiated path and is not enabled by `APRS_IS_ENABLED`.