osint-dashboard/app/camera_config.py

"""Camera discovery configuration (12-factor, env-driven).

All knobs read from the environment with container-friendly defaults.
"""

from __future__ import annotations

import os

# ── Camera scraper ─────────────────────────────────────────────────────────
# Comma-separated public directory/list/API URLs to scrape.
# Formats: Insecam-style HTML, plain-text URL lists, ALERTWest JSON.
# NOTE: docker-compose always defines CAMERA_SOURCE_URLS (empty when no .env),
# so os.getenv()'s default would never apply — use `or` semantics instead.
CALTRANS_CCTV_URLS = tuple(
    f"https://cwwp2.dot.ca.gov/data/d{n}/cctv/cctvStatusD{n:02d}.json"
    for n in range(1, 13)
)
_DEFAULT_SOURCE_URL = ",".join((
    # Publicly published open-camera list (markdown bullets of stream URLs).
    "https://raw.githubusercontent.com/fury999io/public-ip-cams/main/README.md",
    # ALERTCalifornia / ALERTWest official public JPEG API (wildfire + DOT + FAA).
    "https://api.cdn.prod.alertwest.com/api/getCameraDataByLoc",
    # Curated global outdoor streams (HLS / YouTube / JPEG) — VDOT, MDSHA, etc.
    "https://raw.githubusercontent.com/willytop8/Live-Environment-Streams/main/streams.geojson",
    # Official Caltrans CWWP2 JPEG + HLS CCTV (districts 1–12).
    *CALTRANS_CCTV_URLS,
))
CAMERA_SOURCE_URLS = [
    u.strip()
    for u in (os.getenv("CAMERA_SOURCE_URLS") or _DEFAULT_SOURCE_URL).split(",")
    if u.strip()  # empty env var → fall back to the default above
]

# Seconds between full scrape cycles of every source.
CAMERA_SCRAPE_INTERVAL = int(os.getenv("CAMERA_SCRAPE_INTERVAL", "3600"))

# Politeness: minimum seconds between consecutive requests to the SAME host.
CAMERA_REQUEST_DELAY = float(os.getenv("CAMERA_REQUEST_DELAY", "2.0"))

# Max cameras accepted per scrape cycle per source (safety cap).
CAMERA_MAX_PER_SOURCE = int(os.getenv("CAMERA_MAX_PER_SOURCE", "20000"))

# ── Geocoding (Nominatim — free, 1 req/s hard politeness limit) ────────────
NOMINATIM_URL = os.getenv("NOMINATIM_URL", "https://nominatim.openstreetmap.org")
NOMINATIM_MIN_INTERVAL = float(os.getenv("NOMINATIM_MIN_INTERVAL", "1.0"))
USER_AGENT = os.getenv(
    "OSINT_USER_AGENT", "osint-dashboard-camera-scraper/1.0 (self-hosted)"
)

# ── Snapshot cache ─────────────────────────────────────────────────────────
SNAPSHOT_CACHE_DIR = os.getenv("SNAPSHOT_CACHE_DIR", "/data/snapshots")
SNAPSHOT_TTL_SECONDS = int(os.getenv("SNAPSHOT_TTL_SECONDS", "300"))
SNAPSHOT_TIMEOUT = float(os.getenv("SNAPSHOT_TIMEOUT", "8.0"))

# NATS subject cameras are published on (consumed by the shared ingester).
CAMERA_NATS_SUBJECT = os.getenv("CAMERA_NATS_SUBJECT", "events.camera")


# ── UDOT IBI 511 traffic cameras ──────────────────────────────────────────
# DataTables endpoint (POST form-encoded; server caps at 100 rows/page no
# matter what `length` is sent). No API key. Snapshot stills live at a stable
# /map/Cctv/{id} URL — same URL always serves the latest frame, so we store
# the URL and never scrape every frame ourselves.
UDOT_IBI_URL = "https://prod-ut.ibi511.com/List/GetData/Cameras"
UDOT_IBI_BASE = "https://prod-ut.ibi511.com"
UDOT_IBI_PAGE_SIZE = 100
# Safety cap on pages per cycle so a runaway recordsTotal cannot fan out.
UDOT_IBI_MAX_PAGES = int(os.getenv("UDOT_IBI_MAX_PAGES", "40"))