import json
import tempfile
import pytest
from pathlib import Path

import numpy as np

from evaluation.config import GeneratorConfig, PipelineConfig, RetrieverConfig
from evaluation.pipeline import RAGPipeline


class _DummyGenerator:
    """Always returns a fixed answer, ignoring HF pipeline."""

    def generate(self, question: str, contexts: list[str], **kwargs) -> str:
        return "DUMMY_ANSWER"

    def __repr__(self):
        return "DummyGenerator"


@pytest.fixture
def tmp_doc_store(tmp_path_factory):
    docs = [
        {"id": 0, "text": "Retrieval Augmented Generation combines retrieval and generation."},
        {"id": 1, "text": "BM25 is a strong baseline."},
        {"id": 2, "text": "FAISS enables efficient similarity search."},
    ]
    doc_path = tmp_path_factory.mktemp("docs") / "docs.jsonl"
    with doc_path.open("w") as f:
        for row in docs:
            f.write(json.dumps(row) + "\n")
    return doc_path


def test_pipeline_with_dense(tmp_doc_store, monkeypatch, tmp_path):
    import evaluation.generators.hf_generator as hf_module

    monkeypatch.setattr(hf_module, "HFGenerator", _DummyGenerator)

    cfg = PipelineConfig(
        retriever=RetrieverConfig(
            name="dense",
            top_k=2,
            faiss_index=tmp_path / "dense.idx",
            doc_store=tmp_doc_store,
            device="cpu",
            model_name="dummy/ignored", 
        ),
        generator=GeneratorConfig(model_name="dummy"),
    )
    pipeline = RAGPipeline(cfg)

    results = pipeline.run_queries([{"question": "Q?", "id": 0}])
    assert isinstance(results, list)
    assert all("answer" in r for r in results)