tests/test_submit.py

"""Unit tests for the submit-tab pending-row builder.

C4 contract: ``_build_pending_row`` defaults the three Bundle 1+2
schema fields (``validation_status="unvalidated"``,
``validation_method=None``, ``hf_username=None``) and keeps the
existing metadata + sha256 fields intact.

Hub I/O (``_resolve_data_revision`` reads from
``HfApi.dataset_info``) is monkeypatched out, so the suite has zero
network traffic.
"""
from __future__ import annotations

import submit


def _stub_meta() -> dict:
    """Minimum meta.json shape that survives ``_load_and_validate_meta``."""
    return {
        "submitter_name": "team-test",
        "submission_name": "Stub Agent v1",
        "agent_url": "https://github.com/example/stub-agent",
        "notes": "test row, not a real submission",
        "agree_to_publish": True,
    }


def test_pending_row_defaults_new_fields(monkeypatch):
    """Three Bundle 1+2 fields land with their schema defaults."""
    monkeypatch.setattr(submit, "_resolve_data_revision", lambda: "test-rev")
    row = submit._build_pending_row(
        submission_id="sub-test-x",
        meta=_stub_meta(),
        blob_url="https://huggingface.co/datasets/example/sub-test-x.zip",
        submission_sha256="a" * 64,
    )
    assert row["validation_status"] == "unvalidated"
    assert row["validation_method"] is None
    assert row["hf_username"] is None


def test_pending_row_preserves_sha256(monkeypatch):
    """Existing dedup path's row-level half: sha256 still gets stamped on the row."""
    monkeypatch.setattr(submit, "_resolve_data_revision", lambda: "test-rev")
    expected_hash = "f" * 64
    row = submit._build_pending_row(
        submission_id="sub-test-x",
        meta=_stub_meta(),
        blob_url="https://huggingface.co/datasets/example/sub-test-x.zip",
        submission_sha256=expected_hash,
    )
    assert row["submission_sha256"] == expected_hash


def test_pending_row_populates_hf_username_when_provided(monkeypatch):
    """C10 OAuth path: profile.username flows into the row's hf_username.

    The submit handler reads ``gr.OAuthProfile`` (injected by Gradio)
    and passes ``profile.username`` through as a kwarg. This test
    exercises just the row builder's side of that handoff so a
    refactor that drops the kwarg gets caught.
    """
    monkeypatch.setattr(submit, "_resolve_data_revision", lambda: "test-rev")
    row = submit._build_pending_row(
        submission_id="sub-test-x",
        meta=_stub_meta(),
        blob_url="https://huggingface.co/datasets/example/sub-test-x.zip",
        submission_sha256="a" * 64,
        hf_username="alice",
    )
    assert row["hf_username"] == "alice"


def test_pending_row_hf_username_defaults_to_none(monkeypatch):
    """Omitting the kwarg keeps `hf_username` null.

    Covers the pre-OAuth callers (test fixtures, scripts) that don't
    have a profile in scope. Pre-C10 row writers and any future
    non-OAuth caller default cleanly.
    """
    monkeypatch.setattr(submit, "_resolve_data_revision", lambda: "test-rev")
    row = submit._build_pending_row(
        submission_id="sub-test-x",
        meta=_stub_meta(),
        blob_url="https://huggingface.co/datasets/example/sub-test-x.zip",
        submission_sha256="a" * 64,
    )
    assert row["hf_username"] is None


def test_pending_row_preserves_existing_metadata(monkeypatch):
    """Pre-Bundle-1+2 fields keep their values from meta + args.

    Regression guard: a future refactor of ``_build_pending_row`` that
    accidentally drops one of these keys would silently change the
    schema of every row the Space writes.
    """
    monkeypatch.setattr(submit, "_resolve_data_revision", lambda: "test-rev")
    meta = _stub_meta()
    row = submit._build_pending_row(
        submission_id="sub-test-x",
        meta=meta,
        blob_url="https://example.test/sub-test-x.zip",
        submission_sha256="0" * 64,
    )
    assert row["submission_id"] == "sub-test-x"
    assert row["status"] == "pending"
    assert row["failure_reason"] is None
    assert row["submitter_name"] == meta["submitter_name"]
    assert row["submission_name"] == meta["submission_name"]
    assert row["agent_url"] == meta["agent_url"]
    assert row["notes"] == meta["notes"]
    assert row["submission_blob_url"] == "https://example.test/sub-test-x.zip"
    assert row["cadgenbench_data_revision"] == "test-rev"
    # Score-shaped fields are null on a fresh pending row.
    for k in (
        "aggregate_score",
        "validity_rate",
        "score_by_task_type",
        "per_task_scores",
        "per_fixture_scores",
        "per_fixture_breakdown",
    ):
        assert row[k] is None