tests/test_deploy_surface.py

"""The deploy surface exists and matches the published contract.

`deploy/orchestrator_registration.yaml` is what the orchestrator merges into its `agents.yaml`.
If a tool is renamed here and not there (or vice versa), the orchestrator dispatches into
nothing — so the descriptor and the live `gr.api` names are checked against each other.
"""

from __future__ import annotations

from pathlib import Path

import gradio_ui


def _descriptor() -> str:
    return Path("deploy/orchestrator_registration.yaml").read_text()


def test_mcp_server_registers_the_contract_tool_names():
    """The MCP tool names ARE the published contract — not the Python wrapper names."""
    import asyncio

    from fastmcp import Client

    from src.server import mcp

    async def _names():
        async with Client(mcp) as client:
            return sorted(t.name for t in await client.list_tools())

    assert mcp.name == "pdac-genomics-agent"
    assert asyncio.run(_names()) == ["query_variant_status", "variant_by_subtype"]


def test_machine_endpoints_exist_with_request_injection():
    """Each `gr.api` target must accept an injected `request` — the token gate needs it."""
    import inspect

    for name in ("query_variant_status", "variant_by_subtype", "panel"):
        fn = getattr(gradio_ui, name)
        params = inspect.signature(fn).parameters
        assert "request" in params, f"{name} cannot gate without an injected request"
        assert params["request"].default is None


def test_api_names_match_the_orchestrator_descriptor():
    """The frozen names in the descriptor must be the ones actually registered."""
    descriptor = _descriptor()
    for tool_name in ("query_variant_status", "variant_by_subtype"):
        assert tool_name in descriptor


def test_descriptor_arg_names_match_the_live_signatures():
    """The published `args` must be the parameters the endpoints actually take.

    String equality on tool names caught renames but not the drift that actually happened:
    the descriptor advertised two args for `variant_by_subtype` (which takes four) and
    `list[str]` for `genes` (which is a comma-separated string). Compare structurally.
    """
    import inspect

    import yaml  # a gradio dependency, so always present where the Space runs

    tools = {t["name"]: t for t in yaml.safe_load(_descriptor())["tools"]}
    for name, entry in tools.items():
        params = inspect.signature(getattr(gradio_ui, name)).parameters
        declared = set(entry["args"] or {})
        live = {p for p in params if p != "request"}
        assert declared == live, f"{name}: descriptor says {declared}, live takes {live}"
        for arg in declared:
            # PEP 563 is on in `gradio_ui`, so annotations arrive as strings.
            assert params[arg].annotation == "str", f"{name}.{arg} is not a plain string"
            assert entry["args"][arg] == "str", (
                f"{name}.{arg} is declared as {entry['args'][arg]!r}, but every machine "
                "argument is a plain string"
            )


def test_descriptor_does_not_restate_the_cohort_set_wrongly():
    """Whatever cohorts the descriptor names must be curated artifacts that exist."""
    import re

    import yaml

    doc = yaml.safe_load(_descriptor())
    curated = {p.stem for p in Path("src/resources/curated").glob("*.json")}
    # Study ids appear in the illustrative tail of `cohorts:` as a comma-separated run.
    named = set(re.findall(r"(?:^|[ ,])([a-z][a-z0-9]+(?:_[a-z0-9]+)+)(?=[,.\s]|$)", doc["cohorts"]))
    named -= {"src", "resources", "curated", "json"}
    assert named == curated, (
        f"descriptor cohort list has drifted — extra: {sorted(named - curated)}, "
        f"missing: {sorted(curated - named)}"
    )


def test_blocks_app_is_importable():
    """The Space's `app_file` must expose a `demo` that built without error."""
    assert gradio_ui.demo is not None
    assert gradio_ui.demo.title == "PDAC Genomics Agent"


def test_panel_endpoint_returns_the_v1_panel(monkeypatch):
    """The orchestrator can ask for the panel rather than hardcoding the gene list.

    Asserted against `load_panel()` rather than a literal count, precisely because the panel is
    config that widens (19 → 29 on 2026-08-05). A hardcoded number here would fail on every
    widening for no reason; what actually matters is that the endpoint and the resource agree.
    """
    import json

    from src.workflows.variant_status import load_panel

    monkeypatch.delenv("ACCESS_CONTROL", raising=False)
    body = json.loads(gradio_ui.panel(request=None))
    assert body["n_genes"] == len(load_panel())
    assert body["panel"] == load_panel()
    assert "KRAS" in body["panel"]