Files
daily-paper/tests/test_embedder.py
T

176 lines
7.1 KiB
Python

"""Embedder / Chroma 服务测试 — 初始化、索引、embedding API。"""
from __future__ import annotations
from unittest.mock import MagicMock, patch
import pytest
from app.config import settings
# ═══════════════════════════════════════════════════════════════════════
# 初始化
# ═══════════════════════════════════════════════════════════════════════
class TestEmbedderInit:
"""embedder.py 初始化测试。"""
def test_chroma_disabled_skip_init(self, monkeypatch):
"""CHROMA_ENABLED=false 时不初始化。"""
monkeypatch.setattr(settings, "CHROMA_ENABLED", False)
import app.services.embedder as emb
emb._chroma.reset()
emb.init_chroma()
assert emb._chroma._client is None
def test_chroma_init_success(self, monkeypatch, tmp_path):
"""CHROMA_ENABLED=true 时初始化成功。"""
monkeypatch.setattr(settings, "CHROMA_ENABLED", True)
monkeypatch.setattr(settings, "CHROMA_DIR", str(tmp_path / "chroma"))
import app.services.embedder as emb
emb._chroma.reset()
emb.init_chroma()
assert emb._chroma._client is not None
assert emb._chroma._collection is not None
# 清理
emb._chroma.reset()
def test_get_collection_returns_none_when_disabled(self, monkeypatch):
"""CHROMA_ENABLED=false 时 get_collection 返回 None。"""
monkeypatch.setattr(settings, "CHROMA_ENABLED", False)
import app.services.embedder as emb
emb._chroma.reset()
assert emb.get_collection() is None
# ═══════════════════════════════════════════════════════════════════════
# 索引
# ═══════════════════════════════════════════════════════════════════════
class TestEmbedderIndexing:
"""embedder.py 索引测试。"""
def test_index_paper_disabled(self, monkeypatch):
"""CHROMA_ENABLED=false 时 index_paper 返回 False。"""
monkeypatch.setattr(settings, "CHROMA_ENABLED", False)
import app.services.embedder as emb
emb._chroma.reset()
assert emb.index_paper("test-id") is False
def test_index_paper_no_api_config(self, monkeypatch, tmp_path):
"""没有 EMBED_API_BASE 时返回 False。"""
monkeypatch.setattr(settings, "CHROMA_ENABLED", True)
monkeypatch.setattr(settings, "CHROMA_DIR", str(tmp_path / "chroma"))
monkeypatch.setattr(settings, "EMBED_API_BASE", "")
monkeypatch.setattr(settings, "EMBED_MODEL", "")
import app.services.embedder as emb
emb._chroma.reset()
emb.init_chroma()
result = emb.index_paper("test-id", {"title_zh": "测试", "title_en": "Test"})
assert result is False
emb._chroma.reset()
def test_index_batch_disabled(self, monkeypatch):
"""CHROMA_ENABLED=false 时 index_batch 返回全失败。"""
monkeypatch.setattr(settings, "CHROMA_ENABLED", False)
import app.services.embedder as emb
emb._chroma.reset()
result = emb.index_batch(["a", "b"])
assert result["success"] == 0
assert result["failed"] == 2
def test_index_batch_empty(self, monkeypatch):
"""空列表时返回 0。"""
monkeypatch.setattr(settings, "CHROMA_ENABLED", False)
import app.services.embedder as emb
result = emb.index_batch([])
assert result["total"] == 0
def test_delete_paper_disabled(self, monkeypatch):
"""CHROMA_ENABLED=false 时 delete_paper 返回 False。"""
monkeypatch.setattr(settings, "CHROMA_ENABLED", False)
import app.services.embedder as emb
emb._chroma.reset()
assert emb.delete_paper("test-id") is False
def test_search_similar_disabled(self, monkeypatch):
"""CHROMA_ENABLED=false 时 search_similar 返回空列表。"""
monkeypatch.setattr(settings, "CHROMA_ENABLED", False)
import app.services.embedder as emb
emb._chroma.reset()
assert emb.search_similar("test query") == []
# ═══════════════════════════════════════════════════════════════════════
# Embedding API
# ═══════════════════════════════════════════════════════════════════════
class TestEmbeddingApi:
"""_get_embedding 测试。"""
def test_no_api_base_returns_none(self, monkeypatch):
"""EMBED_API_BASE 为空时返回 None。"""
monkeypatch.setattr(settings, "EMBED_API_BASE", "")
monkeypatch.setattr(settings, "EMBED_MODEL", "")
import app.services.embedder as emb
assert emb._get_embedding("test") is None
def test_dimension_mismatch_returns_none(self, monkeypatch):
"""维度不匹配时返回 None。"""
monkeypatch.setattr(settings, "EMBED_API_BASE", "http://fake")
monkeypatch.setattr(settings, "EMBED_MODEL", "test-model")
monkeypatch.setattr(settings, "EMBED_API_KEY", "")
monkeypatch.setattr(settings, "EMBED_DIMENSIONS", 128)
monkeypatch.setattr(settings, "HTTP_TIMEOUT_SECONDS", 5)
import app.services.embedder as emb
mock_resp = MagicMock()
mock_resp.json.return_value = {"data": [{"embedding": [0.1] * 64}]}
mock_resp.raise_for_status = MagicMock()
with patch("httpx.Client") as mock_client:
mock_client.return_value.__enter__ = MagicMock(return_value=mock_resp)
mock_client.return_value.__exit__ = MagicMock(return_value=False)
result = emb._get_embedding("test")
assert result is None
def test_api_failure_returns_none(self, monkeypatch):
"""API 调用失败时返回 None。"""
monkeypatch.setattr(settings, "EMBED_API_BASE", "http://fake")
monkeypatch.setattr(settings, "EMBED_MODEL", "test-model")
monkeypatch.setattr(settings, "EMBED_API_KEY", "")
monkeypatch.setattr(settings, "EMBED_DIMENSIONS", 0)
monkeypatch.setattr(settings, "HTTP_TIMEOUT_SECONDS", 5)
import app.services.embedder as emb
with patch("httpx.Client") as mock_client:
mock_client.return_value.__enter__ = MagicMock()
mock_client.return_value.__exit__ = MagicMock(return_value=False)
mock_client.return_value.__enter__.return_value.post.side_effect = (
Exception("timeout")
)
result = emb._get_embedding("test")
assert result is None