scripts/dev_seed.py

"""Dev-only: seed a LOCAL dashboard with clickable batch + booking leads (overlay image, outline
geometry, pricing, prior edit) so the address popup / drag-to-edit outline / measurement history
can be browser-tested WITHOUT running the real (model-heavy) pipeline.

    ./.venv/Scripts/python.exe scripts/dev_seed.py
    # then open  http://127.0.0.1:8000/dashboard?token=devtoken   (Leads screen → click an address)

Standalone — nothing here ships to prod. It runs the app in-process (so the in-memory viz cache
is seeded too) with a dev-token login (no Clerk) and an in-memory SQLite DB.
"""

from __future__ import annotations

import json
import os
import uuid

# Dev env must be set BEFORE importing the app (config is read at import).
os.environ.setdefault("WARM_MODEL_ON_STARTUP", "0")   # skip the 3-model load
os.environ.setdefault("DASHBOARD_DEV_TOKEN", "devtoken")
os.environ.setdefault("DASHBOARD_DEV_TENANT", "thelawnstandard")
os.environ.setdefault("ALLOWED_API_KEYS", "dev:devkey")
os.environ.setdefault("WIDGET_API_KEYS", "thelawnstandard:pub-dev")   # tenant:publishable → /widget?key=pub-dev
os.environ.setdefault("ADMIN_DEV_TOKEN", "admintoken")               # /admin?token=admintoken
os.environ.setdefault("PLATFORM_ADMIN_EMAILS", "admin@localhost")
os.environ.setdefault("LAWN_DB_PATH", ":memory:")
for _k in ("DATABASE_URL", "CLERK_SECRET_KEY", "CLERK_PUBLISHABLE_KEY"):
    os.environ.pop(_k, None)   # force the dev-token gate, local SQLite

import matplotlib

matplotlib.use("Agg")
import matplotlib.pyplot as plt

from lawn_estimator import api

TENANT = os.environ["DASHBOARD_DEV_TENANT"]
TOKEN = os.environ["DASHBOARD_DEV_TOKEN"]


def _overlay_png(path, poly):
    """A faux 'lawn overlay' (aerial-ish background + the measured lawn) so the popup image +
    drag editor render. The frontend draws its own editable SVG polygon on top of this."""
    fig = plt.figure(figsize=(6.4, 6.4), dpi=200)   # 1280x1280
    ax = fig.add_axes([0, 0, 1, 1])
    ax.set_xlim(0, 1280)
    ax.set_ylim(1280, 0)
    ax.axis("off")
    ax.add_patch(plt.Rectangle((0, 0), 1280, 1280, color="#5c6b4f"))
    xs = [p[0] for p in poly] + [poly[0][0]]
    ys = [p[1] for p in poly] + [poly[0][1]]
    ax.fill(xs, ys, color="#38c860", alpha=0.35)
    ax.plot(xs, ys, color="#2e7d32", lw=3)
    fig.savefig(path)
    plt.close(fig)


def _seed_batch(address, zip_code, sqft, poly):
    rid = uuid.uuid4().hex[:12]
    pricing = {"currency": "USD", "line_items": [
        {"service": "mow", "label": "Mowing", "price": round(sqft / 1000 * 10, 2)},
        {"service": "fert", "label": "Fertilizer", "price": round(sqft / 1000 * 7, 2)}]}
    price_total = round(sum(li["price"] for li in pricing["line_items"]), 2)
    api.METER.record(TENANT, address, api.BATCH, lawn_sqft=sqft, result_id=rid, zip=zip_code,
                     method="lidar+rgb", confidence="high")
    detail = json.dumps({"lawn_polygon_px": poly, "pixel_area_sqft": 0.35, "image_size_px": [1280, 1280],
                         "pricing": pricing, "original_sqft": sqft, "original_pricing": pricing})
    api.LEADS.record(TENANT, "batch", address, email="ops@thelawnstandard.com", zip=zip_code,
                     lawn_sqft=sqft, price_total=price_total, confidence="high", result_id=rid, detail=detail)
    api.OUTPUT_DIR.mkdir(parents=True, exist_ok=True)
    overlay = api.OUTPUT_DIR / f"dev_{rid}.png"
    _overlay_png(overlay, poly)
    api._remember_viz(rid, overlay, overlay, TENANT)   # viz + overlay = the placeholder; owner=tenant
    return rid


def seed():
    if not api.USERS.by_clerk_id(api.DASHBOARD_DEV_CLERK_ID):
        api.USERS.create(TENANT, api.DASHBOARD_DEV_CLERK_ID, "dev@localhost", "owner", status="active")

    r1 = _seed_batch("1423 N 75th St, Omaha, NE 68134", "68134", 5200.0,
                     [[300, 320], [980, 300], [1010, 980], [280, 1000]])
    # a batch lead that already carries a prior tenant edit, so its popup shows Original + Edit + Revert
    r2 = _seed_batch("7863 N 144th Ave, Bennington, NE 68007", "68007", 8100.0,
                     [[260, 260], [1020, 300], [980, 1010], [300, 980]])
    api.MEDITS.record(TENANT, r2, 8600.0, kind="tenant", original_sqft=8100.0,
                      address="7863 N 144th Ave, Bennington, NE 68007", user_id=1)
    api._persist_measurement_edit(TENANT, r2, 8600.0,
                                  {"currency": "USD", "line_items": [{"service": "mow", "label": "Mowing", "price": 86.0}]})

    # A team member (Team screen) + a pending invite.
    if not api.USERS.by_clerk_id("clerk_staff"):
        api.USERS.create(TENANT, "clerk_staff", "staff@thelawnstandard.com", "staff", status="active")
    if not api.USERS.pending_by_email("invitee@thelawnstandard.com"):
        api.USERS.invite(TENANT, "invitee@thelawnstandard.com", "staff", invited_by=1)

    # A widget-origin quote → lead → booking with long-format line items (Leads + Overview + popup).
    rw = uuid.uuid4().hex[:12]
    api.METER.record(TENANT, "5534 Mayberry St, Omaha, NE 68106", api.SINGLE, lawn_sqft=4200.0,
                     result_id=rw, zip="68106", method="lidar+rgb", confidence="high")
    api.LEADS.record(TENANT, "widget", "5534 Mayberry St, Omaha, NE 68106", name="Sam Homeowner",
                     email="sam@example.com", phone="402-555-0100", zip="68106", result_id=rw)
    api.LEADS.record(TENANT, "booking", "5534 Mayberry St, Omaha, NE 68106", name="Sam Homeowner",
                     email="sam@example.com", zip="68106", lawn_sqft=4200.0, price_total=84.0,
                     result_id=rw, detail=json.dumps({"original_sqft": 4200.0,
                         "bundle": {"id": "spring", "name": "Spring Package", "cadence": "seasonal"},
                         "pricing": {"line_items": [{"service": "mow", "label": "Mowing", "price": 42.0},
                                                     {"service": "fert", "label": "Fertilizer", "price": 42.0}]}}))
    api.BOOKING_ITEMS.record_many(TENANT, rw, [
        {"service": "mow", "label": "Mowing", "price": 42.0, "purchased": True},
        {"service": "fert", "label": "Fertilizer", "price": 42.0, "purchased": True},
        {"service": "aerate", "label": "Aeration", "price": 120.0, "purchased": False}])

    # Backdated runs so the Overview volume sparkline shows a trend (raw insert — record() can't backdate).
    import random
    from datetime import date, timedelta
    zips = ["68106", "68134", "68164", "68007", "68135"]
    for i in range(1, 15):
        d = (date.today() - timedelta(days=i)).isoformat() + "T12:00:00.000Z"
        for _ in range(random.randint(0, 4)):
            oc = random.choices(["ok", "rejected", "error"], weights=[85, 10, 5])[0]
            api.METER._db.execute(
                "INSERT INTO runs (tenant_id, surface, address, address_norm, zip, billable, "
                "method, confidence, lawn_sqft, duration_s, outcome, created_at) "
                "VALUES (?, ?, ?, ?, ?, ?, 'lidar+rgb', 'high', ?, ?, ?, ?)",
                (TENANT, random.choice(["single", "batch"]), f"seed {i}", f"seed {i}",
                 random.choice(zips), 1, random.uniform(2500, 9000), random.uniform(18, 55), oc, d))
    api.METER._db.commit()

    print(f"\n  Seeded → http://127.0.0.1:8000/dashboard?token={TOKEN}   (widget: /widget?key=pub-dev)")
    print("  Leads → click an address (batch: view+drag-edit+history/revert; booking: view).")
    print(f"  results: {r1} (fresh), {r2} (has a tenant edit), {rw} (widget booking)\n")


if __name__ == "__main__":
    seed()
    import uvicorn

    uvicorn.run(api.app, host="127.0.0.1", port=8000, log_level="warning")