app/backend/tests/api/routes/test_sync.py

"""`fluksio sync` against a real engine: what it stores, and what it refuses."""

import sys
from collections.abc import Generator
from pathlib import Path

import pytest
from fastapi.testclient import TestClient

from fluksio.core.config import settings
from fluksio.sdk import FLOWS, MARKER, SyncError
from fluksio.sdk.client import Client, sync

EXAMPLES = Path(__file__).parents[4] / "examples"
PREFIX = settings.API_V1_STR
ORIGIN = {
    "kind": "python",
    "repo": str(EXAMPLES),
    "commit": "a1b2c3d4",
    "dirty": False,
}


@pytest.fixture(scope="module")
def flows() -> Generator[dict, None, None]:
    """The example research package, imported the way `sync` imports it."""
    sys.path.insert(0, str(EXAMPLES))
    FLOWS.clear()
    import myresearch.pipeline  # noqa: F401

    yield dict(FLOWS)
    FLOWS.clear()
    sys.path.remove(str(EXAMPLES))


@pytest.fixture
def api(client: TestClient, superuser_token_headers: dict[str, str]) -> Client:
    return Client(http=client, token=superuser_token_headers["Authorization"][7:])


def commits() -> int:
    """How many commits the flow store has made."""
    from fluksio.flow.store import FlowStore

    store = FlowStore(settings.FLOWS_DIR)
    result = store._git("rev-list", "--count", "HEAD")
    return int(result.stdout.strip()) if result.returncode == 0 else 0


def test_sync_stores_a_runnable_flow_with_its_origin(api, flows):
    reports = sync([flows["train"]], api, origin=ORIGIN)

    assert [r.flow for r in reports] == ["train"]
    assert reports[0].created and reports[0].published

    stored = api.get_flow("train")
    definition = stored["definition"]
    assert definition["origin"]["commit"] == "a1b2c3d4"
    assert [n["id"] for n in definition["nodes"]] == ["prepare", "fit", "evaluate"]
    assert not stored["has_draft"]
    # The store holds a complete definition, and the bodies say where the real
    # code is rather than being a copy of it.
    source = api.get_source("train", "fit")
    assert source.startswith(MARKER)
    assert "from myresearch.train import fit" in source


def test_a_node_carries_the_modules_its_code_reaches(api, flows):
    """The shim names one function; the cache has to key on what it calls."""
    sync([flows["train"]], api, origin=ORIGIN)

    nodes = {n["id"]: n for n in api.get_flow("train")["definition"]["nodes"]}
    files = nodes["fit"]["code_files"]

    # The module the function is in, and the package whose `__init__` runs on
    # the way to it. Not the engine's own code, and not the standard library.
    assert "myresearch.train" in files
    assert all(path.endswith(".py") for path in files.values())
    assert not any(name.startswith("fluksio") for name in files)
    assert len(nodes["fit"]["code_digest"]) == 64


def test_a_node_whose_helper_moved_is_named_by_sync():
    """Editing a helper changes no shim, so nothing used to say it happened."""
    from fluksio.sdk.client import _moved_code

    stored = {
        "definition": {
            "nodes": [
                {"id": "fit", "code_digest": "aaa"},
                {"id": "prepare", "code_digest": "bbb"},
            ]
        }
    }
    document = {
        "nodes": [
            {"id": "fit", "code_digest": "ccc"},
            {"id": "prepare", "code_digest": "bbb"},
        ]
    }

    assert _moved_code(stored, document) == ["fit"]
    # Nothing to compare against is not a change.
    assert _moved_code(None, document) == []


def test_an_engine_that_records_no_digest_is_not_a_change_every_time():
    """It drops the field on parse, so comparing against it reports for ever.

    Which is what an engine older than this client does, and what a node
    drawn on the canvas looks like — neither is a helper somebody edited.
    """
    from fluksio.sdk.client import _moved_code

    older = {"definition": {"nodes": [{"id": "fit"}, {"id": "prepare"}]}}
    document = {
        "nodes": [
            {"id": "fit", "code_digest": "ccc"},
            {"id": "prepare", "code_digest": ""},
        ]
    }

    assert _moved_code(older, document) == []


def test_a_second_sync_changes_nothing_and_commits_nothing(api, flows):
    sync([flows["train"]], api, origin=ORIGIN)
    before = commits()

    reports = sync([flows["train"]], api, origin=ORIGIN)

    assert reports[0].unchanged
    assert commits() == before


def test_sync_refuses_to_overwrite_a_canvas_edit(api, flows):
    sync([flows["train"]], api, origin=ORIGIN)
    api.put_source("train", "evaluate", "def process(weights):\n    return {}\n")

    with pytest.raises(SyncError, match="edited on the canvas"):
        sync([flows["train"]], api, origin=ORIGIN)

    reports = sync([flows["train"]], api, origin=ORIGIN, force=True)
    assert "evaluate" in reports[0].changed
    assert api.get_source("train", "evaluate").startswith(MARKER)


def test_an_interrupted_sync_finishes_on_the_next_one(api, flows):
    """A sync that died between the document and the bodies must be resumable.

    That state — nodes named by a stored flow, with no source written for them
    yet — used to read as a canvas edit, because the engine answers a new-node
    template for a node with no body and the template carries no marker. The
    advice was `--force`, for a canvas nobody had touched.
    """
    target = flows["train"]
    target.name = "resumed"
    try:
        # Exactly what the document PUT leaves behind before any body is written.
        api.put_flow(target.document(ORIGIN) | {"version": 1})
        assert api.get_source_entry("resumed", "fit")["missing"]

        reports = sync([target], api, origin=ORIGIN)

        assert "fit" in reports[0].changed
        assert api.get_source("resumed", "fit").startswith(MARKER)
        assert not api.get_source_entry("resumed", "fit")["missing"]
        assert not api.get_flow("resumed")["has_draft"]
    finally:
        target.name = "train"


def test_sync_refuses_a_flow_it_did_not_create(api, flows):
    api.put_flow({"name": "drawn", "version": 1, "nodes": [], "mode": "batch"})
    api.publish("drawn", 1)
    theirs = flows["train"]
    theirs.name = "drawn"

    try:
        with pytest.raises(SyncError, match="not created by sync"):
            sync([theirs], api, origin=ORIGIN)
    finally:
        theirs.name = "train"


def test_use_stores_the_rewiring_it_was_given(api, flows):
    sync([flows["finetune"]], api, origin=ORIGIN)

    nodes = {n["id"]: n for n in api.get_flow("finetune")["definition"]["nodes"]}
    assert [p["name"] for p in nodes["fit"]["requires"]] == ["augmented", "lr"]
    assert nodes["fit"]["params"] == {"epochs": 3}


def test_refresh_retires_the_workers(client, superuser_token_headers):
    response = client.post(f"{PREFIX}/modules/refresh", headers=superuser_token_headers)

    assert response.status_code == 200
    assert "afresh" in response.json()["message"]