cyberyan_linkedin_search/backend/tests/test_search_api.py

"""
End-to-end tests of the search endpoint against the PostgreSQL backend.

They pin down the behaviour the task description asks for: keyword search plus
at least two filters, and a correct contract between frontend and backend.
"""

from __future__ import annotations

import pytest
from django.urls import reverse

pytestmark = pytest.mark.django_db


def search(client, **params):  # noqa: ANN001, ANN201
    return client.get(reverse("v1:search:search"), params)


def names(response) -> list[str]:  # noqa: ANN001
    return [hit["full_name"] for hit in response.data["results"]]


def test_empty_query_returns_every_profile(
    auth_client, analyst, sample_profiles
) -> None:  # noqa: ANN001
    response = search(auth_client(analyst))
    assert response.status_code == 200
    assert response.data["count"] == len(sample_profiles)
    assert response.data["backend"] == "postgres"


def test_keyword_matches_skills_and_summary(
    auth_client, analyst, sample_profiles
) -> None:  # noqa: ANN001
    response = search(auth_client(analyst), q="django")
    assert names(response) == ["sara python"]


def test_keyword_matches_a_person_name(
    auth_client, analyst, sample_profiles
) -> None:  # noqa: ANN001
    assert names(search(auth_client(analyst), q="maria")) == ["maria frontend"]


def test_keyword_tolerates_a_typo_in_the_name(
    auth_client, analyst, sample_profiles
) -> None:  # noqa: ANN001
    """Trigram similarity is what makes ``jhon`` still find ``john``."""
    assert "john recruiter" in names(search(auth_client(analyst), q="jhon recruiter"))


def test_skill_filter(auth_client, analyst, sample_profiles) -> None:  # noqa: ANN001
    assert names(search(auth_client(analyst), skills="react")) == ["maria frontend"]


def test_multiple_values_inside_one_filter_are_or_combined(
    auth_client, analyst, sample_profiles
) -> None:  # noqa: ANN001
    response = search(auth_client(analyst), skills="react,django")
    assert sorted(names(response)) == ["maria frontend", "sara python"]


def test_different_filters_are_and_combined(
    auth_client, analyst, sample_profiles
) -> None:  # noqa: ANN001
    response = search(auth_client(analyst), skills="react,django", countries="spain")
    assert names(response) == ["maria frontend"]


def test_role_and_experience_filters(auth_client, analyst, sample_profiles) -> None:  # noqa: ANN001
    response = search(auth_client(analyst), roles="engineering", min_experience=5)
    assert names(response) == ["sara python"]


def test_repeated_query_parameters_are_supported(
    auth_client, analyst, sample_profiles
) -> None:  # noqa: ANN001
    client = auth_client(analyst)
    response = client.get(f"{reverse('v1:search:search')}?skills=react&skills=django")
    assert sorted(names(response)) == ["maria frontend", "sara python"]


def test_facets_reflect_the_filtered_result_set(
    auth_client, analyst, sample_profiles
) -> None:  # noqa: ANN001
    response = search(auth_client(analyst), roles="engineering")
    countries = {bucket["key"] for bucket in response.data["facets"]["country"]}
    assert countries == {"germany", "spain"}
    assert response.data["facets"]["skills"]


def test_selecting_a_facet_value_keeps_its_own_sibling_values_visible(
    auth_client, analyst, sample_profiles
) -> None:  # noqa: ANN001
    """
    Multi-select facets: ticking "Engineering" under Job family must not make
    "Human resources" disappear from that same checkbox list - a facet is
    scoped by every *other* active filter, never by its own.  Every other
    facet (country, in ``test_facets_reflect_the_filtered_result_set`` above)
    keeps narrowing to the selection as before.
    """
    response = search(auth_client(analyst), roles="engineering")
    assert sorted(names(response)) == ["maria frontend", "sara python"]

    roles = {bucket["key"]: bucket["count"] for bucket in response.data["facets"]["job_title_role"]}
    assert roles == {"engineering": 2, "human_resources": 1}


def test_self_exclusion_does_not_ignore_other_active_filters(
    auth_client, analyst, sample_profiles
) -> None:  # noqa: ANN001
    """
    Excluding a facet's own filter must not turn into excluding *all*
    filters: with ``roles=engineering`` AND ``countries=germany`` both active,
    the job-family facet drops only its own ``roles`` filter and keeps
    ``countries=germany``.  "human_resources" (john, based in the United
    States) correctly stays absent here - not because the facet is narrowed by
    its own selection, but because no German profile has that role at all.
    Contrast with ``test_selecting_a_facet_value_keeps_its_own_sibling_values_visible``,
    where the same "human_resources" sibling *does* reappear once
    ``countries`` is not also constraining the result set.
    """
    response = search(auth_client(analyst), roles="engineering", countries="germany")
    assert names(response) == ["sara python"]

    roles = {bucket["key"] for bucket in response.data["facets"]["job_title_role"]}
    assert roles == {"engineering"}

    countries = {bucket["key"] for bucket in response.data["facets"]["country"]}
    # Scoped by roles=engineering (the other active filter) but not by its
    # own countries=germany selection, so spain (maria, also engineering)
    # still shows up.
    assert countries == {"germany", "spain"}


def test_sorting_by_experience(auth_client, analyst, sample_profiles) -> None:  # noqa: ANN001
    assert names(search(auth_client(analyst), sort="-experience"))[0] == "john recruiter"
    assert names(search(auth_client(analyst), sort="experience"))[0] == "maria frontend"
    assert names(search(auth_client(analyst), sort="name"))[0] == "john recruiter"


def test_pagination_metadata(auth_client, analyst, sample_profiles) -> None:  # noqa: ANN001
    response = search(auth_client(analyst), page_size=2, page=2, sort="name")
    assert response.data["page"] == 2
    assert response.data["page_size"] == 2
    assert response.data["total_pages"] == 2
    assert len(response.data["results"]) == 1


def test_highlight_is_returned_for_keyword_searches(
    auth_client, analyst, sample_profiles
) -> None:  # noqa: ANN001
    hit = search(auth_client(analyst), q="django").data["results"][0]
    assert "<mark>" in " ".join(hit["highlight"].get("skills", []))


def test_invalid_experience_range_is_rejected(auth_client, analyst) -> None:  # noqa: ANN001
    response = search(auth_client(analyst), min_experience=10, max_experience=2)
    assert response.status_code == 400
    assert response.data["error"]["code"] == "validation_error"


def test_page_size_is_capped(auth_client, analyst) -> None:  # noqa: ANN001
    assert search(auth_client(analyst), page_size=5000).status_code == 400


def test_deep_pagination_is_refused(auth_client, analyst) -> None:  # noqa: ANN001
    assert search(auth_client(analyst), page=99999, page_size=100).status_code == 400


def test_unknown_filter_values_return_an_empty_page_not_an_error(
    auth_client, analyst, sample_profiles
) -> None:  # noqa: ANN001
    response = search(auth_client(analyst), skills="cobol")
    assert response.status_code == 200
    assert response.data["count"] == 0


def test_suggest_endpoint(auth_client, analyst, sample_profiles) -> None:  # noqa: ANN001
    response = auth_client(analyst).get(reverse("v1:search:suggest"), {"q": "mar"})
    assert response.status_code == 200
    assert "maria frontend" in response.data["suggestions"]


def test_skill_dictionary_endpoint(auth_client, analyst, sample_profiles) -> None:  # noqa: ANN001
    response = auth_client(analyst).get(reverse("v1:profiles:skills"), {"q": "py"})
    assert response.status_code == 200
    assert any(item["name"] == "python" for item in response.data)