tests/test_analytics.py

"""Analytics — in-your-control metrics from implementation_plan/06 §5."""
from tests._helpers import (
    auth_headers,
    create_creator,
    create_persona,
    enable_widget,
    seed_approved_clip,
    signup,
)


def _start_and_send(client, persona, text: str) -> str:
    r = client.post("/widget/sessions", json={"public_key": persona["public_key"]})
    sid = r.json()["id"]
    client.post(f"/widget/sessions/{sid}/message", json={"text": text})
    return sid


def test_analytics_basic_counts(client, fake_storage, no_redis):
    token = signup(client, "an1@example.com")
    creator = create_creator(client, token)
    persona = create_persona(client, token, creator["id"])
    enable_widget(client, token, persona["id"])
    seed_approved_clip(client, token, persona["id"], intent_tag="pricing",
                       transcript="המחיר X")

    # 2 sessions: one served (pricing), one fallback (oos)
    _start_and_send(client, persona, "כמה זה עולה?")
    _start_and_send(client, persona, "מי גילה את אמריקה")

    r = client.get(
        f"/personas/{persona['id']}/analytics", headers=auth_headers(token)
    )
    assert r.status_code == 200
    body = r.json()
    assert body["persona_id"] == persona["id"]
    assert body["sessions_started"] == 2
    assert body["sessions_completed"] == 2
    assert body["avg_questions_per_session"] == 1.0
    # 1 served + 1 fallback → fallback_rate = 0.5
    assert body["fallback_rate"] == 0.5
    # Reasons present (out_of_scope from the second message)
    assert "out_of_scope" in body["fallback_rate_by_reason"]
    # missing_topics — only pricing is covered, the rest of core_intents are missing
    assert "who_you_are" in body["missing_topics"]
    assert "pricing" not in body["missing_topics"]


def test_analytics_lead_conversion(client, fake_storage, no_redis):
    token = signup(client, "an2@example.com")
    creator = create_creator(client, token)
    persona = create_persona(client, token, creator["id"])
    enable_widget(client, token, persona["id"])

    # One session, one lead captured at the end
    sid = _start_and_send(client, persona, "מי גילה את אמריקה")
    client.post(
        f"/widget/sessions/{sid}/lead",
        json={"name": "X", "contact": "x@x.io", "question_context": "ctx"},
    )

    r = client.get(
        f"/personas/{persona['id']}/analytics", headers=auth_headers(token)
    )
    body = r.json()
    assert body["leads_count"] == 1
    assert body["sessions_completed"] == 1
    assert body["lead_conversion_rate"] == 1.0


def test_analytics_cross_tenant_403(client, fake_storage, no_redis):
    token_a = signup(client, "alice-a@example.com", org="AnA")
    token_b = signup(client, "bob-a@example.com", org="AnB")
    creator = create_creator(client, token_a)
    persona = create_persona(client, token_a, creator["id"])

    r = client.get(
        f"/personas/{persona['id']}/analytics", headers=auth_headers(token_b)
    )
    assert r.status_code == 403


def test_analytics_top_intents_orders_by_count(client, fake_storage, no_redis):
    token = signup(client, "an3@example.com")
    creator = create_creator(client, token)
    persona = create_persona(client, token, creator["id"])
    enable_widget(client, token, persona["id"])
    seed_approved_clip(client, token, persona["id"], intent_tag="pricing",
                       transcript="A")
    seed_approved_clip(client, token, persona["id"], intent_tag="how_it_works",
                       transcript="B", filename="b.mp4")

    # 3 pricing questions, 1 how_it_works
    for _ in range(3):
        _start_and_send(client, persona, "כמה זה עולה?")
    _start_and_send(client, persona, "איך התהליך עובד?")

    r = client.get(
        f"/personas/{persona['id']}/analytics", headers=auth_headers(token)
    )
    top = r.json()["top_intents"]
    assert top[0]["intent"] == "pricing"
    assert top[0]["count"] == 3
    intents = {row["intent"]: row["count"] for row in top}
    assert intents.get("how_it_works") == 1