tests/rag/test_pipeline_version.py

"""Tests for `app.rag.pipeline_version` (Task 8, Step 1).

Two things must be provably true before any dedup mechanism gets built on
top of this registry:

1. `rag-naive-v1` is frozen -- exactly `dedup_enabled=False`,
   `overfetch_factor=1` -- forever. If someone edits the registry entry by
   accident while working on `rag-dedup-v2`, this test catches it.
2. The unspecified/default path (`get_pipeline_version(None)`) resolves to
   that exact frozen v1 behavior, so every ordinary product call site that
   never passes a pipeline version is provably unaffected by anything
   registered here.

An unknown, explicitly-requested version name is a hard error, never a
silent fallback to v1 -- see `UnknownPipelineVersionError`.
"""

from __future__ import annotations

import pytest

from app.rag.pipeline_version import (
    DEFAULT_PIPELINE_VERSION,
    PipelineVersion,
    UnknownPipelineVersionError,
    get_pipeline_version,
    registered_versions,
)


def test_naive_v1_is_registered_and_frozen():
    pv = get_pipeline_version("rag-naive-v1")
    assert pv == PipelineVersion(
        name="rag-naive-v1", dedup_enabled=False, overfetch_factor=1
    )


def test_dedup_v2_is_registered_with_overfetch_and_dedup_enabled():
    pv = get_pipeline_version("rag-dedup-v2")
    assert pv.dedup_enabled is True
    assert pv.overfetch_factor > 1


def test_default_none_resolves_to_naive_v1_exactly():
    default_pv = get_pipeline_version(None)
    explicit_v1 = get_pipeline_version("rag-naive-v1")
    assert default_pv == explicit_v1
    assert default_pv.name == DEFAULT_PIPELINE_VERSION
    assert default_pv.dedup_enabled is False
    assert default_pv.overfetch_factor == 1


def test_unknown_explicit_version_raises():
    with pytest.raises(UnknownPipelineVersionError):
        get_pipeline_version("rag-reranked-v3-does-not-exist")


def test_unknown_version_error_message_lists_registered_versions():
    with pytest.raises(UnknownPipelineVersionError) as exc_info:
        get_pipeline_version("not-a-real-version")
    message = str(exc_info.value)
    assert "rag-naive-v1" in message
    assert "rag-dedup-v2" in message


def test_registered_versions_contains_both_known_versions():
    versions = registered_versions()
    assert "rag-naive-v1" in versions
    assert "rag-dedup-v2" in versions