gimme-job/tests/proactive/test_repo.py

"""Tests for the proactive lead ledger status transitions."""
from datetime import date

import pytest
from sqlalchemy import create_engine
from sqlalchemy.orm import sessionmaker
from sqlalchemy.pool import StaticPool

from gimme_job.db.proactive_repo import ProactiveLeadRepo
from gimme_job.models.db import Base
from gimme_job.models.dto import ProactiveLeadCandidate


@pytest.fixture()
def session():
    engine = create_engine(
        "sqlite:///:memory:", connect_args={"check_same_thread": False}, poolclass=StaticPool
    )
    Base.metadata.create_all(engine)
    factory = sessionmaker(bind=engine)
    s = factory()
    yield s
    s.close()


def _cand(status: str, url: str, title: str = "Orthodontist") -> ProactiveLeadCandidate:
    return ProactiveLeadCandidate(
        status=status,
        title=title,
        official_url=url,
        source_engine="duckduckgo",
        source_query="orthodontist FQHC Alabama",
        fingerprint=f"fp-{url}",
    )


def test_new_active_lead(session):
    repo = ProactiveLeadRepo()
    changed, status = repo.upsert(
        session, _cand("ACTIVE", "https://a.org/1"), date(2026, 9, 8)
    )
    assert changed is True
    assert status == "NEW"
    lead = ProactiveLeadRepo().get_by_fingerprint(session, "fp-https://a.org/1")
    assert lead.is_active is True
    assert lead.ever_active is True
    assert lead.is_new is True


def test_active_twice_is_still_open(session):
    repo = ProactiveLeadRepo()
    repo.upsert(session, _cand("ACTIVE", "https://a.org/1"), date(2026, 9, 8))
    changed, status = repo.upsert(session, _cand("ACTIVE", "https://a.org/1"), date(2026, 9, 9))
    assert changed is True
    assert status == "STILL_OPEN"
    lead = repo.get_by_fingerprint(session, "fp-https://a.org/1")
    assert lead.is_new is False
    assert lead.is_active is True


def test_closed_then_active_is_reopened(session):
    repo = ProactiveLeadRepo()
    repo.upsert(session, _cand("ACTIVE", "https://a.org/1"), date(2026, 9, 8))
    repo.upsert(session, _cand("CLOSED", "https://a.org/1"), date(2026, 9, 10))
    lead = repo.get_by_fingerprint(session, "fp-https://a.org/1")
    assert lead.status == "CLOSED"
    assert lead.is_active is False
    changed, status = repo.upsert(session, _cand("ACTIVE", "https://a.org/1"), date(2026, 9, 12))
    assert status == "REOPENED"
    assert changed is True
    lead = repo.get_by_fingerprint(session, "fp-https://a.org/1")
    assert lead.is_new is True
    assert lead.is_active is True


def test_verify_lead_stored_as_verify(session):
    repo = ProactiveLeadRepo()
    changed, status = repo.upsert(
        session, _cand("VERIFY", "https://indeed.com/jobs/1"), date(2026, 9, 8)
    )
    assert changed is True
    assert status == "VERIFY"
    lead = repo.get_by_fingerprint(session, "fp-https://indeed.com/jobs/1")
    assert lead.is_active is False
    assert lead.ever_active is False


def test_verify_upgrade_to_active_is_new(session):
    repo = ProactiveLeadRepo()
    repo.upsert(session, _cand("VERIFY", "https://indeed.com/jobs/1"), date(2026, 9, 8))
    changed, status = repo.upsert(
        session, _cand("ACTIVE", "https://indeed.com/jobs/1"), date(2026, 9, 9)
    )
    assert status == "NEW"
    lead = repo.get_by_fingerprint(session, "fp-https://indeed.com/jobs/1")
    assert lead.ever_active is True
    assert lead.is_active is True


def test_closed_first_seen_not_stored(session):
    changed, status = ProactiveLeadRepo().upsert(
        session, _cand("CLOSED", "https://x.org/9"), date(2026, 9, 8)
    )
    assert changed is False
    assert status == "CLOSED"
    assert ProactiveLeadRepo().get_by_fingerprint(session, "fp-https://x.org/9") is None


def test_stale_transition_from_active(session):
    repo = ProactiveLeadRepo()
    repo.upsert(session, _cand("ACTIVE", "https://a.org/2"), date(2026, 9, 8))
    changed, status = repo.upsert(session, _cand("STALE", "https://a.org/2"), date(2026, 9, 9))
    assert status == "STALE"
    lead = repo.get_by_fingerprint(session, "fp-https://a.org/2")
    assert lead.is_active is False


def test_fields_refreshed_on_upsert(session):
    repo = ProactiveLeadRepo()
    repo.upsert(session, _cand("ACTIVE", "https://a.org/3", title="Orthodontist"), date(2026, 9, 8))
    cand = _cand("ACTIVE", "https://a.org/3", title="Staff Orthodontist")
    cand.salary_text = "$200k-$250k"
    cand.evidence = {"apply_found": True}
    repo.upsert(session, cand, date(2026, 9, 9))
    lead = repo.get_by_fingerprint(session, "fp-https://a.org/3")
    assert lead.title == "Staff Orthodontist"
    assert lead.salary_text == "$200k-$250k"
    assert lead.evidence is not None


def test_get_inactive_returns_closed_and_stale(session):
    repo = ProactiveLeadRepo()
    repo.upsert(session, _cand("ACTIVE", "https://a.org/10"), date(2026, 9, 8))
    repo.upsert(session, _cand("ACTIVE", "https://a.org/11"), date(2026, 9, 8))
    repo.upsert(session, _cand("ACTIVE", "https://a.org/12"), date(2026, 9, 8))
    # close two of them
    repo.upsert(session, _cand("CLOSED", "https://a.org/10"), date(2026, 9, 9))
    repo.upsert(session, _cand("STALE", "https://a.org/11"), date(2026, 9, 9))
    inactive = repo.get_inactive(session)
    urls = {rec.official_url for rec in inactive}
    assert "https://a.org/10" in urls
    assert "https://a.org/11" in urls
    assert "https://a.org/12" not in urls


def test_get_inactive_respects_limit(session):
    repo = ProactiveLeadRepo()
    for i in range(5):
        repo.upsert(session, _cand("ACTIVE", f"https://a.org/l{i}"), date(2026, 9, 8))
        repo.upsert(session, _cand("CLOSED", f"https://a.org/l{i}"), date(2026, 9, 9))
    assert len(repo.get_inactive(session, limit=2)) == 2