pull_sales.py

"""Pull Elko County vacant-land sale records via the assessor advanced search export.

Two forms:
    uv run python pull_sales.py --months 36
    uv run python pull_sales.py 2023-08-01 2026-08-01 V data/raw/sales_window_20260802.json

The rolling form is what refresh.py calls: it computes the window backwards from today
and writes a date-stamped file into data/raw/, which publish_snapshot.py then redacts and
compresses into the committed set the notebook reads. A rerun always lands on current data
without anyone editing a date.
"""
import sys
from datetime import date

from dateutil.relativedelta import relativedelta

import elko_client


def query(dos_min, dos_max, acreage_min="40", vi_cd="V"):
    q = {
        "type": "s",
        "query[parcel][acreage_min]": acreage_min,
        "query[sale][dos_min]": dos_min,
        "query[sale][dos_max]": dos_max,
        "sort": "last_sale_date",
    }
    if vi_cd:
        q["query[sale][vi_cd]"] = vi_cd
    return q


def rolling(months, out_dir="data/raw"):
    end = date.today()
    start = end - relativedelta(months=months)
    out = f"{out_dir}/sales_window_{end:%Y%m%d}.json"
    return start.isoformat(), end.isoformat(), out


if __name__ == "__main__":
    if sys.argv[1:2] == ["--months"]:
        dos_min, dos_max, out = rolling(int(sys.argv[2]))
        vi = "V"
    else:
        dos_min, dos_max, vi, out = sys.argv[1], sys.argv[2], sys.argv[3], sys.argv[4]
        vi = "" if vi == "any" else vi
    print(f"pulling {dos_min}..{dos_max} vi={vi or 'ANY'}", file=sys.stderr)
    rows = elko_client.pull("/search/adv", query(dos_min, dos_max, vi_cd=vi),
                            "/export/adv/s", out)
    print(f"wrote {out}: {len(rows)} records", file=sys.stderr)