from __future__ import annotations
import json
from dataclasses import dataclass
from pathlib import Path
from typing import Any
import pytest
from src.mcp_server.tools.get_document_summary import (
_extract_summary_fields,
_find_metadata_in_jsonl,
_load_one_metadata_by_doc_id,
)
@pytest.mark.unit
def test_find_metadata_in_jsonl_returns_none_when_missing(tmp_path: Path) -> None:
assert _find_metadata_in_jsonl(tmp_path / "missing.jsonl", doc_id="x") is None
@pytest.mark.unit
def test_find_metadata_in_jsonl_finds_first_match(tmp_path: Path) -> None:
doc_id = "doc-123"
path = tmp_path / "store.jsonl"
path.write_text(
"\n".join(
[
json.dumps(
{"id": "c1", "metadata": {"doc_id": "other", "title": "t0"}}
),
json.dumps(
{
"id": "c2",
"metadata": {
"doc_id": doc_id,
"title": "t1",
"summary": "s1",
"tags": ["a", "b"],
},
}
),
json.dumps({"id": "c3", "metadata": {"doc_id": doc_id, "title": "t2"}}),
]
)
+ "\n",
encoding="utf-8",
)
out = _find_metadata_in_jsonl(path, doc_id=doc_id)
assert out is not None
assert out["doc_id"] == doc_id
assert out["title"] == "t1"
@pytest.mark.unit
def test_extract_summary_fields_applies_fallbacks() -> None:
title, summary, tags = _extract_summary_fields({"filename": "a.pdf"})
assert title == "a.pdf"
assert summary == "a.pdf"
assert tags == []
@dataclass
class _VectorStoreCfg:
backend: str
persist_path: str
collection_name: str
@dataclass
class _Settings:
vector_store: Any
@pytest.mark.unit
def test_load_one_metadata_by_doc_id_jsonl(tmp_path: Path) -> None:
collection_name = "knowledge_hub"
doc_id = "doc-1"
jsonl_path = tmp_path / f"{collection_name}.jsonl"
jsonl_path.write_text(
json.dumps(
{
"id": "c1",
"metadata": {
"doc_id": doc_id,
"title": "T",
"summary": "S",
"tags": ["x"],
},
}
)
+ "\n",
encoding="utf-8",
)
settings = _Settings(
vector_store=_VectorStoreCfg(
backend="jsonl", persist_path=str(tmp_path), collection_name=collection_name
)
)
meta = _load_one_metadata_by_doc_id(settings, doc_id=doc_id)
assert meta["doc_id"] == doc_id
assert meta["title"] == "T"
@pytest.mark.unit
def test_load_one_metadata_by_doc_id_raises_for_missing_doc_id(tmp_path: Path) -> None:
collection_name = "knowledge_hub"
jsonl_path = tmp_path / f"{collection_name}.jsonl"
jsonl_path.write_text(
json.dumps({"id": "c1", "metadata": {"doc_id": "x"}}) + "\n", encoding="utf-8"
)
settings = _Settings(
vector_store=_VectorStoreCfg(
backend="jsonl", persist_path=str(tmp_path), collection_name=collection_name
)
)
with pytest.raises(ValueError):
_load_one_metadata_by_doc_id(settings, doc_id="nope")
