"""Project-wide constants and paths (UQO WP36)."""
from __future__ import annotations

import os
from pathlib import Path

ROOT = Path(os.environ.get("UCML_ROOT", Path(__file__).resolve().parents[2]))
DATA = ROOT / "data"
RAW = DATA / "raw"
INTERIM = DATA / "interim"
PROCESSED = DATA / "processed"
SYNTHETIC = DATA / "synthetic"
RESULTS = ROOT / "results"
FIGURES = ROOT / "figures"
TABLES = ROOT / "paper" / "tables"

SNAPSHOT = Path(os.environ.get("UCML_SNAPSHOT", RAW / "autoka-snapshot-20261101.db"))
SNAPSHOT_DATE = "2026-11-01"
# last sync in the snapshot (UTC) — used as the right-censoring point
def _manifest_ts() -> tuple[float, float]:
    """(first sync ts, last sync ts) from data/interim/manifest.json when available."""
    import json
    m = INTERIM / "manifest.json"
    if m.exists():
        d = json.loads(m.read_text())
        return float(d["sync_ts_min"]), float(d["sync_ts_max"])
    return 1786537835.64518, 1793515782.732466  # 2026-08-12 12:30:35 UTC, 2026-11-01 06:49:42 UTC


WINDOW_START_TS, SNAPSHOT_TS = _manifest_ts()

# --- Source classification (DECISIONS.md D-003) ---------------------------------
SAMPLED_PORTALS = {"autotrader", "cargurus"}          # capped sliding windows -> no exit information
PRIVATE_MARKETPLACE = {"kijiji"}                       # private sellers, no VIN
FULL_PORTALS = {"otogo", "autousagee"}                 # full-inventory portals (dealer-identified)
BANNERS = {"hgregoire", "automobileendirect", "megacentredeliquidation", "clubautozone", "leprixdugros"}
SMALL_PORTALS = {"okaze", "classeauto", "ototr"}
FLICKER_OR_FAILED = {"leprixdugros", "albioccasion", "baiecomeaugm", "septilesgm",
                     "stjeromeautodepot", "excelmoto"}
UID_COLLISIONS = {"classeauto:-98"}

# --- Cleaning thresholds (analysis_plan.md §1-2) --------------------------------
PRICE_MIN, PRICE_MAX = 2_000.0, 250_000.0
KM_MIN, KM_MAX = 1_000.0, 400_000.0
YEAR_MIN, YEAR_MAX = 2005, 2026
TRANSIENT_SECONDS = 48 * 3600          # A->B->A level shorter than this is removed
OSCILLATION_MAX_TRANSIENTS = 3         # >= 3 transients -> oscillating uid
MARKDOWN_MIN_PCT, MARKDOWN_MIN_ABS = 0.01, 100.0
EXIT_CONFIRM_DAYS = 7
LEFT_CENSOR_CUTOFF_TS = 1786579200.0   # 2026-08-13 00:00:00 UTC: entries before this are the day-1 stock
SOURCE_STARTUP_SECONDS = 24 * 3600     # entries within 24 h of a source's first non-empty crawl = start-up stock (D-014)

RESTRICTED_COLUMNS = ["description", "images", "url", "carfax_url", "features"]