we_monitor/collect.py

#!/usr/bin/env python3
"""
WeConnect vehicle data collector.

Periodically fetches selected domains from a VW WeConnect vehicle and
appends timestamped records to a JSON file.  The JSON structure mirrors
the WeConnect domain hierarchy: domain → status-object → field.

Usage examples
--------------
# Credentials file with all settings:
  python collect.py -c credentials/alex.json

# Override username/password via env vars:
  export WC_USER=me@example.com WC_PASS=secret
  python collect.py -c credentials/alex.json

# Override domains on the CLI:
  python collect.py -c credentials/alex.json -d charging,measurements

Available domains: charging, climatisation, measurements, readiness, parking, access

Credentials file format (JSON)
-------------------------------
{
  "credentials": {
    "username": "me@example.com",
    "password": "secret"
  },
  "vin":        "WVWZZZE1ZMP123456",
  "domains":    ["charging", "measurements", "readiness"],
  "interval_s": 300,
  "host":       "0.0.0.0",
  "port":       9999,
  "log_dir":    "./logs"
}

Push server
-----------
The collector listens on host/port from the credentials file (default 0.0.0.0:9999).
Each connected TCP client receives every new snapshot as a single
line of JSON (newline-delimited) the moment it is collected.
Connect with:  nc localhost 9999  or any TCP client that reads lines.
"""

import argparse
import json
import logging
import os
import sys
import time
from datetime import datetime, timezone
from pathlib import Path

from server import log_config, network, we_connect
from server.data_model import ALL_DOMAINS, collect_snapshot
from server.storage_helpers import auto_out_path, load_last_24h_records, load_store, save_store
from utils.jay_diff import jay_diff_full

log = logging.getLogger(__name__)


# ── main loop ────────────────────────────────────────────────────────────────

def run(args: argparse.Namespace) -> None:
    log_config.setup(args.verbose)

    raw = [d.strip() for d in args.domains.split(",")]
    domains = list(ALL_DOMAINS) if raw == ["all"] else raw
    unknown = [d for d in domains if d not in ALL_DOMAINS]
    if unknown:
        log.error("Unknown domain(s): %s.  Available: %s", unknown, list(ALL_DOMAINS))
        sys.exit(1)

    if args.dry:
        log.info("Dry-run mode — skipping WeConnect login")
        wc = None
        vehicle = None
        vin = args.vin or "UNKNOWN"
    else:
        log.info("Connecting to WeConnect…")
        wc = we_connect.connect(args.username, args.password)
        vehicle, err = we_connect.select_vehicle(wc, args.vin)
        if err:
            log.error(err)
            sys.exit(1)
        vin = vehicle.vin.value

    log.info("Vehicle: %s", vin)

    logs_dir = Path(args.log_dir)
    logs_dir.mkdir(parents=True, exist_ok=True)
    out = auto_out_path(logs_dir, vin)

    push_clients, push_lock, push_store_ref = network.start_push_server(args.host, args.port)
    push_store_ref["diff_mode"] = args.diff
    push_store_ref["preloaded"] = load_last_24h_records(logs_dir, vin)
    if push_store_ref["preloaded"]:
        log.info("Pre-loaded %d record(s) from the last 24 h", len(push_store_ref["preloaded"]))

    current_day = datetime.now(timezone.utc).date()
    store = load_store(out, vin, args.interval)
    push_store_ref["current"] = store
    last_broadcast: dict = {}

    if args.dry:
        log.info("Push server running on %s:%d — no live collection  (Ctrl-C to stop)", args.host, args.port)
    else:
        log.info(
            "Collecting [%s] every %ds → %s  (Ctrl-C to stop)",
            ", ".join(domains),
            args.interval,
            out,
        )

    try:
        while True:
            if not args.dry:
                today = datetime.now(timezone.utc).date()
                if today != current_day:
                    out = auto_out_path(logs_dir, vin)
                    store = load_store(out, vin, args.interval)
                    push_store_ref["current"] = store
                    current_day = today
                    log.info("Midnight UTC rotation → %s", out)

                try:
                    wc.update()
                    snapshot = collect_snapshot(vehicle, domains)
                    store["records"].append(snapshot)
                    save_store(out, store, args.max_records)
                    if args.diff and last_broadcast:
                        payload = jay_diff_full(last_broadcast, snapshot, combine_upd_add=True)
                        payload["_diff"] = True
                    else:
                        payload = snapshot
                    network.broadcast(payload, push_clients, push_lock)
                    last_broadcast = snapshot
                    log.info("Record #%d saved at %s", len(store["records"]), snapshot["ts"])
                except KeyboardInterrupt:
                    raise
                except Exception:
                    log.exception("Collection failed — will retry at next interval")

            time.sleep(args.interval)

    except KeyboardInterrupt:
        log.info("Stopped by user.  %d records in %s", len(store["records"]), out)


# ── CLI ──────────────────────────────────────────────────────────────────────

def main() -> None:
    p = argparse.ArgumentParser(
        description=__doc__,
        formatter_class=argparse.RawDescriptionHelpFormatter,
    )

    creds = p.add_argument_group(
        "credentials (also accepted via WC_USER / WC_PASS env vars or -c FILE)"
    )
    creds.add_argument(
        "-c", "--credentials",
        metavar="FILE",
        help="JSON file with connection settings (see format above).",
    )
    creds.add_argument(
        "-u", "--username",
        default=os.environ.get("WC_USER"),
        help="WeConnect / MyVolkswagen email (overrides credentials file)",
    )
    creds.add_argument(
        "-p", "--password",
        default=os.environ.get("WC_PASS"),
        help="WeConnect password (overrides credentials file)",
    )

    p.add_argument(
        "-d", "--domains",
        default=None,
        metavar="DOMAIN[,DOMAIN…]",
        help=(
            "Comma-separated domains to collect, or 'all' (overrides credentials file).  "
            f"Available: {', '.join(ALL_DOMAINS)}.  "
            "Default: charging,measurements,readiness"
        ),
    )
    p.add_argument(
        "--max-records",
        type=int,
        default=None,
        metavar="N",
        help="Keep only the last N records per file (default: unlimited)",
    )
    p.add_argument("-v", "--verbose", action="store_true", help="Enable debug logging")
    p.add_argument(
        "--diff",
        action="store_true",
        help="Send diffs instead of full snapshots over the push server (default: off)",
    )
    p.add_argument(
        "--dry",
        action="store_true",
        help="Skip WeConnect login and data collection; run push server only (for GUI testing)",
    )
    p.add_argument(
        "--list-domains",
        action="store_true",
        help="Print available domains and exit",
    )

    args = p.parse_args()

    if args.list_domains:
        print("\n".join(ALL_DOMAINS))
        sys.exit(0)

    creds_data: dict = {}
    if args.credentials:
        creds_path = Path(args.credentials)
        if not creds_path.exists():
            p.error(f"Credentials file not found: {args.credentials}")
        try:
            creds_data = json.loads(creds_path.read_text())
        except json.JSONDecodeError as exc:
            p.error(f"Invalid JSON in credentials file: {exc}")

    # username / password: env / CLI flag > credentials.username > credentials file root
    nested = creds_data.get("credentials", {})
    if args.username is None:
        args.username = nested.get("username") or creds_data.get("username")
    if args.password is None:
        args.password = nested.get("password") or creds_data.get("password")

    args.vin = creds_data.get("vin")

    if args.domains is None:
        raw_domains = creds_data.get("domains")
        if isinstance(raw_domains, list):
            args.domains = ",".join(raw_domains)
        else:
            args.domains = raw_domains or "charging,measurements,readiness"

    args.interval = int(creds_data.get("interval_s", 300))
    args.host     = creds_data.get("host", "0.0.0.0")
    args.port     = int(creds_data.get("port", 9999))
    args.log_dir  = creds_data.get("log_dir", "./logs")

    if not args.dry and (not args.username or not args.password):
        p.error(
            "Username and password are required.  "
            "Provide them in the credentials file or via -u/-p / WC_USER / WC_PASS."
        )

    run(args)


if __name__ == "__main__":
    main()