gimme-job/tests/proactive/test_boards.py

"""Tests for ATS board crawling (mock HTTP — no network)."""
from gimme_job.proactive.boards import (
    BoardConfig,
    _parse_icims_jobs,
    fetch_board,
    fetch_greenhouse,
    fetch_lever,
    fetch_smartrecruiters,
    fetch_usajobs,
    fetch_workday,
)


class _FakeClient:
    """Stand-in for httpx.Client: returns a configured response for each call."""

    def __init__(self, responses):
        self._responses = list(responses)
        self._calls = 0
        self.bodies = []

    def __enter__(self):
        return self

    def __exit__(self, *a):
        return False

    def _next(self):
        self._calls += 1
        return self._responses[self._calls - 1]

    def get(self, url):
        return self._next()

    def post(self, url, json=None):
        self.bodies.append(json or {})
        return self._next()


def _patch(monkeypatch, client):
    monkeypatch.setattr("httpx.Client", lambda *a, **k: client)


class _Resp:
    def __init__(self, **kw):
        self._kw = kw

    def raise_for_status(self):
        pass

    def json(self):
        return self._kw.get("json", {})

    @property
    def content(self):
        return self._kw.get("content", b"")


def test_greenhouse_filters_dental(monkeypatch):
    client = _FakeClient(
        [
            _Resp(
                json={
                    "jobs": [
                        {
                            "title": "Orthodontist",
                            "absolute_url": "https://gh/j/1",
                            "content": "braces",
                        },
                        {
                            "title": "Office Manager",
                            "absolute_url": "https://gh/j/2",
                            "content": "admin",
                        },
                    ]
                }
            )
        ]
    )
    _patch(monkeypatch, client)
    res = fetch_greenhouse(BoardConfig(name="Test", org="testorg", source="greenhouse"))
    assert len(res) == 1
    assert res[0].title == "Orthodontist"
    assert res[0].url == "https://gh/j/1"
    assert res[0].source == "greenhouse"


def test_greenhouse_ignores_employer_name_in_match(monkeypatch):
    payload = {
        "jobs": [
            {
                "title": "Account Payable Specialist at Specialty Dental Brands",
                "absolute_url": "https://gh/j/1",
                "content": "finance role",
            },
            {
                "title": "Dentist at Anthem Pediatric Dentistry",
                "absolute_url": "https://gh/j/2",
                "content": "braces support",
            },
        ]
    }
    client = _FakeClient([_Resp(json=payload)])
    _patch(monkeypatch, client)
    res = fetch_greenhouse(
        BoardConfig(name="Specialty Dental Brands", org="sdb", source="greenhouse")
    )
    assert len(res) == 1
    assert res[0].title.startswith("Dentist")


def test_boards_role_gate_drops_support_roles(monkeypatch):
    payload = {
        "jobs": [
            {
                "title": "Dental Assistant",
                "absolute_url": "https://gh/j/1",
                "content": "assist the orthodontic team",
            },
            {
                "title": "Orthodontist",
                "absolute_url": "https://gh/j/2",
                "content": "braces",
            },
        ]
    }
    client = _FakeClient([_Resp(json=payload)])
    _patch(monkeypatch, client)
    res = fetch_greenhouse(BoardConfig(name="Acme", org="acme", source="greenhouse"))
    assert [r.title for r in res] == ["Orthodontist"]


def test_boards_drop_non_ortho_specialties(monkeypatch):
    payload = {
        "jobs": [
            {
                "title": "Pediatric Dentist",
                "absolute_url": "https://gh/j/1",
                "content": "orthodontics referrals",
            },
            {
                "title": "Orthodontist",
                "absolute_url": "https://gh/j/2",
                "content": "braces",
            },
        ]
    }
    client = _FakeClient([_Resp(json=payload)])
    _patch(monkeypatch, client)
    res = fetch_greenhouse(BoardConfig(name="Acme", org="acme", source="greenhouse"))
    assert [r.title for r in res] == ["Orthodontist"]


def test_boards_pass_employer_type(monkeypatch):
    payload = {
        "jobs": [
            {
                "title": "Orthodontist",
                "absolute_url": "https://gh/j/1",
                "content": "braces",
            }
        ]
    }
    client = _FakeClient([_Resp(json=payload)])
    _patch(monkeypatch, client)
    res = fetch_greenhouse(
        BoardConfig(
            name="Navajo Nation",
            org="navajo",
            source="greenhouse",
            employer_type="tribal",
        )
    )
    assert res[0].employer_type == "tribal"


def test_lever_filters_dental(monkeypatch):
    client = _FakeClient(
        [
            _Resp(
                json=[
                    {
                        "text": "Dentist II (Orthodontics)",
                        "hostedUrl": "https://lever/p/1",
                        "categories": {"location": "Phoenix, AZ"},
                        "descriptionPlain": "full-time orthodontics",
                    },
                    {
                        "text": "Nurse",
                        "hostedUrl": "https://lever/p/2",
                        "categories": {},
                        "descriptionPlain": "nursing",
                    },
                ]
            )
        ]
    )
    _patch(monkeypatch, client)
    res = fetch_lever(BoardConfig(name="Test", org="testorg", source="lever"))
    assert len(res) == 1
    assert res[0].title == "Dentist II (Orthodontics)"


def test_smartrecruiters_builds_job_url(monkeypatch):
    payload = {
        "content": [
            {
                "name": "Dentist",
                "id": "12345",
                "location": {"city": "Phoenix"},
            },
        ]
    }
    # fetcher queries the API once per keyword (orthodont/dentist/dental)
    client = _FakeClient([_Resp(json=payload)] * 3)
    _patch(monkeypatch, client)
    res = fetch_smartrecruiters(
        BoardConfig(name="NH", org="NATIVEHEALTH", source="smartrecruiters")
    )
    assert len(res) == 1
    assert res[0].url == "https://jobs.smartrecruiters.com/NATIVEHEALTH/12345"


def test_workday_filters(monkeypatch):
    payload = {
        "jobPostings": [
            {
                "title": "Orthodontist",
                "externalPath": "/en-US/org/job/R1",
                "locationsText": "OH",
            },
            {
                "title": "Receptionist",
                "externalPath": "/en-US/org/job/R2",
                "locationsText": "OH",
            },
        ]
    }
    # fetcher queries the tenant once per keyword (orthodontist/dentist/dental)
    client = _FakeClient([_Resp(json=payload)] * 3)
    _patch(monkeypatch, client)
    res = fetch_workday(
        BoardConfig(
            name="Test",
            org="org",
            tenant="tenant",
            domain="wd5.myworkdayjobs.com",
            source="workday",
        )
    )
    assert len(res) == 1
    assert "myworkdayjobs.com/en-US/org/job/R1" in res[0].url
    # limit=50 is rejected by some tenants (HTTP 400) — always request 20
    assert client.bodies and all(b["limit"] == 20 for b in client.bodies)


def test_icims_parses_search_html():
    html = (
        '<li class="iCIMS_JobCardItem"><div class="row">'
        '<div class="col-xs-6 header left">'
        '<span class="sr-only field-label">Job Locations</span>'
        "<span >US-IN-New Castle</span></div>"
        '<div class="col-xs-12 title">'
        '<a href="https://careers-x.icims.com/jobs/5374/dentist/job?in_iframe=1" '
        'class="iCIMS_Anchor" title="5374 - Dentist"><h3 >\nDentist</h3></a></div>'
        '<div class="col-xs-12 description">Great braces &amp; aligner work</div>'
        "</div></li>"
    ).encode()
    parsed = _parse_icims_jobs(html)
    assert parsed == [
        (
            "Dentist",
            "https://careers-x.icims.com/jobs/5374/dentist/job",
            "US-IN-New Castle Great braces & aligner work",
        )
    ]


def test_icims_empty_for_garbage():
    assert _parse_icims_jobs(b"not html at all") == []


def test_usajobs_skips_without_key(monkeypatch):
    import os

    import dotenv

    # prevent reload of .env (which now has the key) and force a clean env
    monkeypatch.setattr(dotenv, "load_dotenv", lambda *a, **k: None)
    env = dict(os.environ)
    env.pop("USAJOBS_API_KEY", None)
    env.pop("USAJOBS_API_EMAIL", None)
    monkeypatch.setattr(os, "environ", env)
    res = fetch_usajobs(BoardConfig(name="USAJOBS", source="usajobs"))
    assert res == []


def test_fetch_board_unknown_source():
    assert fetch_board(BoardConfig(name="X", source="unknown")) == []