Spaces:
Running
Running
Merge pull request #174 from Jiya3177/test/backend-rag-pipeline-121
Browse files- .github/workflows/ci.yml +9 -0
- backend/requirements.txt +3 -0
- backend/tests/conftest.py +183 -0
- backend/tests/test_auth.py +81 -0
- backend/tests/test_chat.py +50 -0
- backend/tests/test_chunker.py +38 -0
- backend/tests/test_documents.py +34 -0
.github/workflows/ci.yml
CHANGED
|
@@ -54,6 +54,15 @@ jobs:
|
|
| 54 |
run: |
|
| 55 |
python -c "import sys; sys.path.insert(0, 'backend'); from app.config import settings; print('✅ Config imports OK')" || true
|
| 56 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 57 |
# ── 2. Frontend Build Check ─────────────────────────────
|
| 58 |
frontend-check:
|
| 59 |
name: ⚛️ Frontend — TypeScript & Build
|
|
|
|
| 54 |
run: |
|
| 55 |
python -c "import sys; sys.path.insert(0, 'backend'); from app.config import settings; print('✅ Config imports OK')" || true
|
| 56 |
|
| 57 |
+
- name: Run backend pytest suite
|
| 58 |
+
env:
|
| 59 |
+
SECRET_KEY: ci-dummy-secret
|
| 60 |
+
DATABASE_URL: sqlite:///./ci_test.db
|
| 61 |
+
HF_TOKEN: ci-dummy-token
|
| 62 |
+
UPLOAD_DIR: /tmp/uploads
|
| 63 |
+
CHROMA_PERSIST_DIR: /tmp/chroma
|
| 64 |
+
run: pytest backend/tests -v
|
| 65 |
+
|
| 66 |
# ── 2. Frontend Build Check ─────────────────────────────
|
| 67 |
frontend-check:
|
| 68 |
name: ⚛️ Frontend — TypeScript & Build
|
backend/requirements.txt
CHANGED
|
@@ -18,6 +18,9 @@ google-auth
|
|
| 18 |
# Config
|
| 19 |
pydantic-settings
|
| 20 |
pydantic[email]
|
|
|
|
|
|
|
|
|
|
| 21 |
|
| 22 |
# Document Processing
|
| 23 |
PyMuPDF
|
|
|
|
| 18 |
# Config
|
| 19 |
pydantic-settings
|
| 20 |
pydantic[email]
|
| 21 |
+
pytest
|
| 22 |
+
pytest-cov
|
| 23 |
+
httpx
|
| 24 |
|
| 25 |
# Document Processing
|
| 26 |
PyMuPDF
|
backend/tests/conftest.py
ADDED
|
@@ -0,0 +1,183 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
import os
|
| 2 |
+
import sys
|
| 3 |
+
import types
|
| 4 |
+
from contextlib import asynccontextmanager
|
| 5 |
+
from pathlib import Path
|
| 6 |
+
|
| 7 |
+
import pytest
|
| 8 |
+
from fastapi.testclient import TestClient
|
| 9 |
+
from sqlalchemy import create_engine
|
| 10 |
+
from sqlalchemy.orm import sessionmaker
|
| 11 |
+
|
| 12 |
+
|
| 13 |
+
ROOT = Path(__file__).resolve().parents[2]
|
| 14 |
+
BACKEND_DIR = ROOT / "backend"
|
| 15 |
+
|
| 16 |
+
if str(BACKEND_DIR) not in sys.path:
|
| 17 |
+
sys.path.insert(0, str(BACKEND_DIR))
|
| 18 |
+
|
| 19 |
+
os.environ.setdefault("SECRET_KEY", "test-secret-key")
|
| 20 |
+
os.environ.setdefault("DATABASE_URL", "sqlite:///./test_bootstrap.db")
|
| 21 |
+
os.environ.setdefault("HF_TOKEN", "test-hf-token")
|
| 22 |
+
os.environ.setdefault("UPLOAD_DIR", str(ROOT / "backend" / "test_uploads"))
|
| 23 |
+
os.environ.setdefault("CHROMA_PERSIST_DIR", str(ROOT / "backend" / "test_chroma"))
|
| 24 |
+
|
| 25 |
+
|
| 26 |
+
fake_embeddings = types.ModuleType("app.rag.embeddings")
|
| 27 |
+
fake_embeddings.get_embedding_model = lambda: object()
|
| 28 |
+
fake_embeddings.embed_query = lambda query: [0.0]
|
| 29 |
+
fake_embeddings.embed_texts = lambda texts: [[0.0] for _ in texts]
|
| 30 |
+
sys.modules.setdefault("app.rag.embeddings", fake_embeddings)
|
| 31 |
+
|
| 32 |
+
|
| 33 |
+
class _FakeChromaClient:
|
| 34 |
+
def heartbeat(self):
|
| 35 |
+
return "ok"
|
| 36 |
+
|
| 37 |
+
|
| 38 |
+
fake_vectorstore = types.ModuleType("app.rag.vectorstore")
|
| 39 |
+
fake_vectorstore.get_chroma_client = lambda: _FakeChromaClient()
|
| 40 |
+
fake_vectorstore.store_chunks = lambda chunks, document_id, filename, user_id: len(chunks)
|
| 41 |
+
fake_vectorstore.delete_document_chunks = lambda document_id, user_id: None
|
| 42 |
+
fake_vectorstore.query_chunks = lambda query_embedding, user_id, document_id=None, top_k=10: []
|
| 43 |
+
sys.modules.setdefault("app.rag.vectorstore", fake_vectorstore)
|
| 44 |
+
|
| 45 |
+
slowapi_module = types.ModuleType("slowapi")
|
| 46 |
+
slowapi_errors = types.ModuleType("slowapi.errors")
|
| 47 |
+
slowapi_middleware = types.ModuleType("slowapi.middleware")
|
| 48 |
+
slowapi_util = types.ModuleType("slowapi.util")
|
| 49 |
+
|
| 50 |
+
|
| 51 |
+
class RateLimitExceeded(Exception):
|
| 52 |
+
pass
|
| 53 |
+
|
| 54 |
+
|
| 55 |
+
class SlowAPIMiddleware:
|
| 56 |
+
def __init__(self, app, *args, **kwargs):
|
| 57 |
+
self.app = app
|
| 58 |
+
|
| 59 |
+
async def __call__(self, scope, receive, send):
|
| 60 |
+
await self.app(scope, receive, send)
|
| 61 |
+
|
| 62 |
+
|
| 63 |
+
class Limiter:
|
| 64 |
+
def __init__(self, key_func=None, *args, **kwargs):
|
| 65 |
+
self.key_func = key_func
|
| 66 |
+
|
| 67 |
+
def limit(self, _value):
|
| 68 |
+
def decorator(fn):
|
| 69 |
+
return fn
|
| 70 |
+
return decorator
|
| 71 |
+
|
| 72 |
+
|
| 73 |
+
slowapi_errors.RateLimitExceeded = RateLimitExceeded
|
| 74 |
+
slowapi_middleware.SlowAPIMiddleware = SlowAPIMiddleware
|
| 75 |
+
slowapi_util.get_remote_address = lambda request: "127.0.0.1"
|
| 76 |
+
slowapi_module.Limiter = Limiter
|
| 77 |
+
|
| 78 |
+
sys.modules.setdefault("slowapi", slowapi_module)
|
| 79 |
+
sys.modules.setdefault("slowapi.errors", slowapi_errors)
|
| 80 |
+
sys.modules.setdefault("slowapi.middleware", slowapi_middleware)
|
| 81 |
+
sys.modules.setdefault("slowapi.util", slowapi_util)
|
| 82 |
+
|
| 83 |
+
from app.auth import create_access_token, create_refresh_token, hash_password
|
| 84 |
+
from app.database import Base, get_db
|
| 85 |
+
from app.main import app
|
| 86 |
+
from app.models import Document, User
|
| 87 |
+
|
| 88 |
+
|
| 89 |
+
@pytest.fixture()
|
| 90 |
+
def db_session(tmp_path):
|
| 91 |
+
db_file = tmp_path / "test.db"
|
| 92 |
+
engine = create_engine(
|
| 93 |
+
f"sqlite:///{db_file}",
|
| 94 |
+
connect_args={"check_same_thread": False},
|
| 95 |
+
)
|
| 96 |
+
TestingSessionLocal = sessionmaker(autocommit=False, autoflush=False, bind=engine)
|
| 97 |
+
Base.metadata.create_all(bind=engine)
|
| 98 |
+
|
| 99 |
+
session = TestingSessionLocal()
|
| 100 |
+
try:
|
| 101 |
+
yield session
|
| 102 |
+
finally:
|
| 103 |
+
session.close()
|
| 104 |
+
Base.metadata.drop_all(bind=engine)
|
| 105 |
+
engine.dispose()
|
| 106 |
+
|
| 107 |
+
|
| 108 |
+
@pytest.fixture()
|
| 109 |
+
def client(db_session, monkeypatch):
|
| 110 |
+
def override_get_db():
|
| 111 |
+
try:
|
| 112 |
+
yield db_session
|
| 113 |
+
finally:
|
| 114 |
+
pass
|
| 115 |
+
|
| 116 |
+
@asynccontextmanager
|
| 117 |
+
async def no_lifespan(_app):
|
| 118 |
+
yield
|
| 119 |
+
|
| 120 |
+
monkeypatch.setattr("app.database.SessionLocal", lambda: db_session)
|
| 121 |
+
app.dependency_overrides[get_db] = override_get_db
|
| 122 |
+
app.router.lifespan_context = no_lifespan
|
| 123 |
+
|
| 124 |
+
with TestClient(app) as test_client:
|
| 125 |
+
yield test_client
|
| 126 |
+
|
| 127 |
+
app.dependency_overrides.clear()
|
| 128 |
+
|
| 129 |
+
|
| 130 |
+
@pytest.fixture()
|
| 131 |
+
def user(db_session):
|
| 132 |
+
instance = User(
|
| 133 |
+
username="tester",
|
| 134 |
+
email="tester@example.com",
|
| 135 |
+
hashed_password=hash_password("password123"),
|
| 136 |
+
)
|
| 137 |
+
db_session.add(instance)
|
| 138 |
+
db_session.commit()
|
| 139 |
+
db_session.refresh(instance)
|
| 140 |
+
return instance
|
| 141 |
+
|
| 142 |
+
|
| 143 |
+
@pytest.fixture()
|
| 144 |
+
def auth_headers(user):
|
| 145 |
+
token = create_access_token(user.id)
|
| 146 |
+
return {"Authorization": f"Bearer {token}"}
|
| 147 |
+
|
| 148 |
+
|
| 149 |
+
@pytest.fixture()
|
| 150 |
+
def refresh_token(user):
|
| 151 |
+
return create_refresh_token(user.id)
|
| 152 |
+
|
| 153 |
+
|
| 154 |
+
@pytest.fixture()
|
| 155 |
+
def ready_document(db_session, user):
|
| 156 |
+
instance = Document(
|
| 157 |
+
user_id=user.id,
|
| 158 |
+
filename="ready.txt",
|
| 159 |
+
original_name="ready.txt",
|
| 160 |
+
file_size=128,
|
| 161 |
+
page_count=1,
|
| 162 |
+
chunk_count=2,
|
| 163 |
+
status="ready",
|
| 164 |
+
)
|
| 165 |
+
db_session.add(instance)
|
| 166 |
+
db_session.commit()
|
| 167 |
+
db_session.refresh(instance)
|
| 168 |
+
return instance
|
| 169 |
+
|
| 170 |
+
|
| 171 |
+
@pytest.fixture()
|
| 172 |
+
def pending_document(db_session, user):
|
| 173 |
+
instance = Document(
|
| 174 |
+
user_id=user.id,
|
| 175 |
+
filename="pending.txt",
|
| 176 |
+
original_name="pending.txt",
|
| 177 |
+
file_size=64,
|
| 178 |
+
status="pending",
|
| 179 |
+
)
|
| 180 |
+
db_session.add(instance)
|
| 181 |
+
db_session.commit()
|
| 182 |
+
db_session.refresh(instance)
|
| 183 |
+
return instance
|
backend/tests/test_auth.py
ADDED
|
@@ -0,0 +1,81 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
def test_register_success(client):
|
| 2 |
+
response = client.post(
|
| 3 |
+
"/api/v1/auth/register",
|
| 4 |
+
json={
|
| 5 |
+
"username": "newuser",
|
| 6 |
+
"email": "newuser@example.com",
|
| 7 |
+
"password": "password123",
|
| 8 |
+
},
|
| 9 |
+
)
|
| 10 |
+
|
| 11 |
+
assert response.status_code == 201
|
| 12 |
+
payload = response.json()
|
| 13 |
+
assert payload["access_token"]
|
| 14 |
+
assert payload["refresh_token"]
|
| 15 |
+
assert payload["user"]["email"] == "newuser@example.com"
|
| 16 |
+
|
| 17 |
+
|
| 18 |
+
def test_register_duplicate_email_or_username_conflict(client):
|
| 19 |
+
payload = {
|
| 20 |
+
"username": "dupuser",
|
| 21 |
+
"email": "dup@example.com",
|
| 22 |
+
"password": "password123",
|
| 23 |
+
}
|
| 24 |
+
first = client.post("/api/v1/auth/register", json=payload)
|
| 25 |
+
assert first.status_code == 201
|
| 26 |
+
|
| 27 |
+
duplicate_email = client.post(
|
| 28 |
+
"/api/v1/auth/register",
|
| 29 |
+
json={**payload, "username": "anotheruser"},
|
| 30 |
+
)
|
| 31 |
+
assert duplicate_email.status_code == 409
|
| 32 |
+
assert duplicate_email.json()["detail"] == "Email already registered"
|
| 33 |
+
|
| 34 |
+
duplicate_username = client.post(
|
| 35 |
+
"/api/v1/auth/register",
|
| 36 |
+
json={**payload, "email": "another@example.com"},
|
| 37 |
+
)
|
| 38 |
+
assert duplicate_username.status_code == 409
|
| 39 |
+
assert duplicate_username.json()["detail"] == "Username already taken"
|
| 40 |
+
|
| 41 |
+
|
| 42 |
+
def test_login_success(client, user):
|
| 43 |
+
response = client.post(
|
| 44 |
+
"/api/v1/auth/login",
|
| 45 |
+
json={"email": user.email, "password": "password123"},
|
| 46 |
+
)
|
| 47 |
+
|
| 48 |
+
assert response.status_code == 200
|
| 49 |
+
payload = response.json()
|
| 50 |
+
assert payload["access_token"]
|
| 51 |
+
assert payload["refresh_token"]
|
| 52 |
+
assert payload["user"]["username"] == user.username
|
| 53 |
+
|
| 54 |
+
|
| 55 |
+
def test_login_invalid_password(client, user):
|
| 56 |
+
response = client.post(
|
| 57 |
+
"/api/v1/auth/login",
|
| 58 |
+
json={"email": user.email, "password": "wrong-password"},
|
| 59 |
+
)
|
| 60 |
+
|
| 61 |
+
assert response.status_code == 401
|
| 62 |
+
assert response.json()["detail"] == "Invalid email or password"
|
| 63 |
+
|
| 64 |
+
|
| 65 |
+
def test_auth_me_requires_auth(client):
|
| 66 |
+
response = client.get("/api/v1/auth/me")
|
| 67 |
+
|
| 68 |
+
assert response.status_code in (401, 403)
|
| 69 |
+
|
| 70 |
+
|
| 71 |
+
def test_refresh_token_success(client, refresh_token):
|
| 72 |
+
response = client.post(
|
| 73 |
+
"/api/v1/auth/refresh",
|
| 74 |
+
json={"refresh_token": refresh_token},
|
| 75 |
+
)
|
| 76 |
+
|
| 77 |
+
assert response.status_code == 200
|
| 78 |
+
payload = response.json()
|
| 79 |
+
assert payload["access_token"]
|
| 80 |
+
assert payload["refresh_token"]
|
| 81 |
+
assert payload["token_type"] == "bearer"
|
backend/tests/test_chat.py
ADDED
|
@@ -0,0 +1,50 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
def test_chat_ask_success(client, auth_headers, ready_document, monkeypatch):
|
| 2 |
+
monkeypatch.setattr(
|
| 3 |
+
"app.routes.chat.generate_answer",
|
| 4 |
+
lambda question, user_id, document_id=None: {
|
| 5 |
+
"answer": "Mocked answer",
|
| 6 |
+
"sources": [
|
| 7 |
+
{
|
| 8 |
+
"text": "Mock source",
|
| 9 |
+
"filename": "ready.txt",
|
| 10 |
+
"page": 1,
|
| 11 |
+
"score": 0.99,
|
| 12 |
+
"confidence": 99.0,
|
| 13 |
+
}
|
| 14 |
+
],
|
| 15 |
+
},
|
| 16 |
+
)
|
| 17 |
+
|
| 18 |
+
response = client.post(
|
| 19 |
+
"/api/v1/chat/ask",
|
| 20 |
+
headers=auth_headers,
|
| 21 |
+
json={"question": "What is in the doc?", "document_id": ready_document.id},
|
| 22 |
+
)
|
| 23 |
+
|
| 24 |
+
assert response.status_code == 200
|
| 25 |
+
payload = response.json()
|
| 26 |
+
assert payload["answer"] == "Mocked answer"
|
| 27 |
+
assert payload["document_id"] == ready_document.id
|
| 28 |
+
assert payload["sources"][0]["filename"] == "ready.txt"
|
| 29 |
+
|
| 30 |
+
|
| 31 |
+
def test_chat_ask_document_not_found(client, auth_headers):
|
| 32 |
+
response = client.post(
|
| 33 |
+
"/api/v1/chat/ask",
|
| 34 |
+
headers=auth_headers,
|
| 35 |
+
json={"question": "Missing doc?", "document_id": "missing-doc-id"},
|
| 36 |
+
)
|
| 37 |
+
|
| 38 |
+
assert response.status_code == 404
|
| 39 |
+
assert response.json()["detail"] == "Document not found"
|
| 40 |
+
|
| 41 |
+
|
| 42 |
+
def test_chat_ask_document_not_ready(client, auth_headers, pending_document):
|
| 43 |
+
response = client.post(
|
| 44 |
+
"/api/v1/chat/ask",
|
| 45 |
+
headers=auth_headers,
|
| 46 |
+
json={"question": "Pending doc?", "document_id": pending_document.id},
|
| 47 |
+
)
|
| 48 |
+
|
| 49 |
+
assert response.status_code == 400
|
| 50 |
+
assert "Document is still pending" in response.json()["detail"]
|
backend/tests/test_chunker.py
ADDED
|
@@ -0,0 +1,38 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
from pathlib import Path
|
| 2 |
+
|
| 3 |
+
import pytest
|
| 4 |
+
|
| 5 |
+
from app.rag.chunker import chunk_document, get_page_count
|
| 6 |
+
|
| 7 |
+
|
| 8 |
+
def test_txt_extraction_and_chunking(tmp_path):
|
| 9 |
+
file_path = tmp_path / "notes.txt"
|
| 10 |
+
file_path.write_text("This is a sample text file for chunking.", encoding="utf-8")
|
| 11 |
+
|
| 12 |
+
chunks = chunk_document(str(file_path))
|
| 13 |
+
|
| 14 |
+
assert len(chunks) >= 1
|
| 15 |
+
assert chunks[0]["page"] == 1
|
| 16 |
+
assert "sample text file" in chunks[0]["text"]
|
| 17 |
+
|
| 18 |
+
|
| 19 |
+
def test_empty_txt_returns_no_chunks(tmp_path):
|
| 20 |
+
file_path = tmp_path / "empty.txt"
|
| 21 |
+
file_path.write_text(" \n", encoding="utf-8")
|
| 22 |
+
|
| 23 |
+
assert chunk_document(str(file_path)) == []
|
| 24 |
+
|
| 25 |
+
|
| 26 |
+
def test_unsupported_extension_raises_value_error(tmp_path):
|
| 27 |
+
file_path = tmp_path / "data.csv"
|
| 28 |
+
file_path.write_text("a,b,c", encoding="utf-8")
|
| 29 |
+
|
| 30 |
+
with pytest.raises(ValueError, match="Unsupported file type"):
|
| 31 |
+
chunk_document(str(file_path))
|
| 32 |
+
|
| 33 |
+
|
| 34 |
+
def test_get_page_count_for_txt_returns_one(tmp_path):
|
| 35 |
+
file_path = tmp_path / "single.txt"
|
| 36 |
+
file_path.write_text("hello", encoding="utf-8")
|
| 37 |
+
|
| 38 |
+
assert get_page_count(str(file_path)) == 1
|
backend/tests/test_documents.py
ADDED
|
@@ -0,0 +1,34 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
def test_api_health(client):
|
| 2 |
+
response = client.get("/api/health")
|
| 3 |
+
|
| 4 |
+
assert response.status_code == 200
|
| 5 |
+
payload = response.json()
|
| 6 |
+
assert payload["status"] == "healthy"
|
| 7 |
+
assert payload["version"] == "2.0.0"
|
| 8 |
+
|
| 9 |
+
|
| 10 |
+
def test_protected_documents_list_requires_auth(client):
|
| 11 |
+
response = client.get("/api/v1/documents/")
|
| 12 |
+
|
| 13 |
+
assert response.status_code in (401, 403)
|
| 14 |
+
|
| 15 |
+
|
| 16 |
+
def test_documents_list_authenticated(client, auth_headers, ready_document):
|
| 17 |
+
response = client.get("/api/v1/documents/", headers=auth_headers)
|
| 18 |
+
|
| 19 |
+
assert response.status_code == 200
|
| 20 |
+
payload = response.json()
|
| 21 |
+
assert payload["total"] == 1
|
| 22 |
+
assert payload["items"][0]["id"] == ready_document.id
|
| 23 |
+
assert payload["items"][0]["original_name"] == "ready.txt"
|
| 24 |
+
|
| 25 |
+
|
| 26 |
+
def test_upload_rejects_unsupported_extension_before_deep_validation(client, auth_headers):
|
| 27 |
+
response = client.post(
|
| 28 |
+
"/api/v1/documents/upload",
|
| 29 |
+
headers=auth_headers,
|
| 30 |
+
files={"file": ("payload.exe", b"binary-data", "application/octet-stream")},
|
| 31 |
+
)
|
| 32 |
+
|
| 33 |
+
assert response.status_code == 400
|
| 34 |
+
assert "not supported" in response.json()["detail"]
|