CIAgent/apps/api/tests/enrichment/test_ninjapear_provider.py

"""NinjaPearProvider against the real API shape documented at
nubela.co/llms-full.txt: website-only company identification, and the
vendor's actual response field names (executives/employee_count,
total_funds_raised/funding_rounds, competitors keyed by website, three
separately-categorized customer/investor/partner arrays, first_name+domain
for work-email, x_profile_url for the person-profile endpoint)."""

from __future__ import annotations

import httpx
import pytest
import respx

from app.core.config import Settings
from app.enrichment.ninjapear import NinjaPearProvider

_SETTINGS = Settings(ninjapear_api_key="test-key")
_WEBSITE = "https://acme.example.com"


@pytest.mark.asyncio
async def test_get_company_details_parses_executives_and_employee_count():
    payload = {
        "description": "A payments company.",
        "industry": "Fintech",
        "founded_year": 2010,
        "specialties": ["Payments", "APIs"],
        "employee_count": "1001-5000",
        "executives": [
            {"name": "Jane Doe", "title": "CEO"},
            {"name": "", "title": "Ignored - no name"},
        ],
    }
    with respx.mock:
        respx.get("https://nubela.co/api/v1/company/details").mock(
            return_value=httpx.Response(200, json=payload)
        )
        details = await NinjaPearProvider(_SETTINGS).get_company_details("Acme Corp", _WEBSITE)

    assert details.industry == "Fintech"
    assert details.employee_count_range == "1001-5000"
    assert len(details.leadership_team) == 1
    assert details.leadership_team[0].name == "Jane Doe"


@pytest.mark.asyncio
async def test_company_details_raises_without_a_website():
    with pytest.raises(ValueError, match="website"):
        await NinjaPearProvider(_SETTINGS).get_company_details("Acme Corp", None)


@pytest.mark.asyncio
async def test_get_funding_maps_total_funds_raised_and_funding_rounds():
    # total_funds_raised/amount_usd are raw numbers, and each investor is
    # an object (not a plain string) - matches the real API's shape.
    payload = {
        "total_funds_raised_usd": 500_000_000,
        "funding_rounds": [
            {
                "round_type": "Series C",
                "amount_usd": 200_000_000,
                "date": "2024-01-01",
                "investors": [{"name": "VC Co", "type": "company", "website": None}],
            }
        ],
    }
    with respx.mock:
        respx.get("https://nubela.co/api/v1/company/funding").mock(
            return_value=httpx.Response(200, json=payload)
        )
        funding = await NinjaPearProvider(_SETTINGS).get_funding("Acme Corp", _WEBSITE)

    assert funding.total_raised == "500000000"
    assert funding.rounds[0].round_name == "Series C"
    assert funding.rounds[0].amount == "200000000"
    assert funding.rounds[0].investors == ["VC Co"]


@pytest.mark.asyncio
async def test_get_competitors_falls_back_to_website_when_no_name_given():
    payload = {
        "competitors": [{"website": "rival.example.com", "competition_reason": "Same market"}]
    }
    with respx.mock:
        respx.get("https://nubela.co/api/v1/competitor/listing").mock(
            return_value=httpx.Response(200, json=payload)
        )
        competitors = await NinjaPearProvider(_SETTINGS).get_competitors("Acme Corp", _WEBSITE)

    assert competitors[0].name == "rival.example.com"
    assert competitors[0].reason == "Same market"


@pytest.mark.asyncio
async def test_get_customers_merges_the_three_categorized_arrays():
    payload = {
        "customers": [{"name": "BigCo"}],
        "investors": [{"name": "VC Fund"}],
        "partner_platforms": [{"name": "PlatformCo"}],
    }
    with respx.mock:
        respx.get("https://nubela.co/api/v1/customer/listing").mock(
            return_value=httpx.Response(200, json=payload)
        )
        customers = await NinjaPearProvider(_SETTINGS).get_customers("Acme Corp", _WEBSITE)

    by_relationship = {c.relationship: c.name for c in customers}
    assert by_relationship == {"customer": "BigCo", "investor": "VC Fund", "partner": "PlatformCo"}


@pytest.mark.asyncio
async def test_get_products_joins_categories():
    payload = {
        "products": [
            {"name": "Widget", "description": "A widget", "categories": ["Hardware", "IoT"]}
        ]
    }
    with respx.mock:
        respx.get("https://nubela.co/api/v1/product/listing").mock(
            return_value=httpx.Response(200, json=payload)
        )
        products = await NinjaPearProvider(_SETTINGS).get_products("Acme Corp", _WEBSITE)

    assert products[0].name == "Widget"
    assert products[0].category == "Hardware, IoT"


@pytest.mark.asyncio
async def test_get_updates_maps_title_and_source():
    payload = {
        "updates": [
            {
                "title": "We shipped a feature",
                "source": "blog",
                "url": "https://x",
                "timestamp": "2026-01-01",
            }
        ]
    }
    with respx.mock:
        respx.get("https://nubela.co/api/v1/company/updates").mock(
            return_value=httpx.Response(200, json=payload)
        )
        updates = await NinjaPearProvider(_SETTINGS).get_updates("Acme Corp", _WEBSITE)

    assert updates[0].text == "We shipped a feature"
    assert updates[0].type == "blog"


@pytest.mark.asyncio
async def test_get_work_email_sends_first_last_name_and_domain():
    with respx.mock:
        route = respx.get("https://nubela.co/api/v1/employee/work-email").mock(
            return_value=httpx.Response(200, json={"work_email": "[email protected]"})
        )
        email = await NinjaPearProvider(_SETTINGS).get_work_email("Jane Doe", _WEBSITE)

    assert email == "[email protected]"
    sent_params = dict(route.calls[0].request.url.params)
    assert sent_params == {"first_name": "Jane", "last_name": "Doe", "domain": "acme.example.com"}


@pytest.mark.asyncio
async def test_get_person_profile_hits_the_v2_endpoint_and_maps_x_profile_url():
    with respx.mock:
        respx.get("https://nubela.co/api/v2/employee/profile").mock(
            return_value=httpx.Response(
                200, json={"x_profile_url": "https://x.com/janedoe", "bio": "A bio."}
            )
        )
        profile_url, bio = await NinjaPearProvider(_SETTINGS).get_person_profile(
            "Jane Doe", _WEBSITE
        )

    assert profile_url == "https://x.com/janedoe"
    assert bio == "A bio."


@pytest.mark.asyncio
async def test_a_provider_error_response_raises_rather_than_being_swallowed():
    with respx.mock:
        respx.get("https://nubela.co/api/v1/company/details").mock(
            return_value=httpx.Response(429, text="rate limited")
        )
        with pytest.raises(httpx.HTTPStatusError):
            await NinjaPearProvider(_SETTINGS).get_company_details("Acme Corp", _WEBSITE)