apunkt/SurfSense
1
0
Fork 0
mirror of https://github.com/MODSetter/SurfSense.git synced 2026-07-06 22:12:12 +02:00
Code Issues Projects Releases Packages Wiki Activity

test: rename SearchSpace -> Workspace across tests + fixtures (Phase 2 Wave F)

Browse source 
Apply the same rename to surfsense_backend/tests: workspace_id fields/vars,
Workspace* classes/schemas, table name searchspaces -> workspaces in raw SQL,
and the API URL spellings -> /workspaces. Preserves the carve-out wire literals
tests assert (Celery task names, OTel key "search_space.id").
This commit is contained in:
CREDO23 2026-06-26 18:36:46 +02:00
parent 56826a63bc
commit ca9bd28934
124 changed files with 1269 additions and 1269 deletions
Download patch file Download diff file Expand all files Collapse all files

44
surfsense_backend/tests/integration/google_unification/conftest.py
View file

@ -18,7 +18,7 @@ from app.db import (
DocumentType,
SearchSourceConnector,
SearchSourceConnectorType,
SearchSpace,
Workspace,
User,
)
@ -31,7 +31,7 @@ def make_document(
title: str,
document_type: DocumentType,
content: str,
search_space_id: int,
workspace_id: int,
created_by_id: str,
) -> Document:
"""Build a Document instance with unique hashes and a dummy embedding."""
@ -43,7 +43,7 @@ def make_document(
content_hash=f"content-{uid}",
unique_identifier_hash=f"uid-{uid}",
source_markdown=content,
search_space_id=search_space_id,
workspace_id=workspace_id,
created_by_id=created_by_id,
embedding=DUMMY_EMBEDDING,
updated_at=datetime.now(UTC),
@ -66,35 +66,35 @@ def make_chunk(*, content: str, document_id: int) -> Chunk:
@pytest_asyncio.fixture
async def seed_google_docs(
db_session: AsyncSession, db_user: User, db_search_space: SearchSpace
db_session: AsyncSession, db_user: User, db_workspace: Workspace
):
"""Insert a native Drive doc, a legacy Composio Drive doc, and a FILE doc.
Returns a dict with keys ``native_doc``, ``legacy_doc``, ``file_doc``,
plus ``search_space`` and ``user``.
plus ``workspace`` and ``user``.
"""
user_id = str(db_user.id)
space_id = db_search_space.id
space_id = db_workspace.id
native_doc = make_document(
title="Native Drive Document",
document_type=DocumentType.GOOGLE_DRIVE_FILE,
content="quarterly report from native google drive connector",
search_space_id=space_id,
workspace_id=space_id,
created_by_id=user_id,
)
legacy_doc = make_document(
title="Legacy Composio Drive Document",
document_type=DocumentType.COMPOSIO_GOOGLE_DRIVE_CONNECTOR,
content="quarterly report from composio google drive connector",
search_space_id=space_id,
workspace_id=space_id,
created_by_id=user_id,
)
file_doc = make_document(
title="Uploaded PDF",
document_type=DocumentType.FILE,
content="unrelated uploaded file about quarterly reports",
search_space_id=space_id,
workspace_id=space_id,
created_by_id=user_id,
)
@ -121,7 +121,7 @@ async def seed_google_docs(
"native_doc": native_doc,
"legacy_doc": legacy_doc,
"file_doc": file_doc,
"search_space": db_search_space,
"workspace": db_workspace,
"user": db_user,
}
@ -136,8 +136,8 @@ async def seed_google_docs(
async def committed_google_data(async_engine):
"""Insert native, legacy, and FILE docs via a committed transaction.
Yields ``{"search_space_id": int, "user_id": str}``.
Cleans up by deleting the search space (cascades to documents / chunks).
Yields ``{"workspace_id": int, "user_id": str}``.
Cleans up by deleting the workspace (cascades to documents / chunks).
"""
space_id = None
@ -155,7 +155,7 @@ async def committed_google_data(async_engine):
session.add(user)
await session.flush()
space = SearchSpace(name=f"Google Test {uuid.uuid4().hex[:6]}", user_id=user.id)
space = Workspace(name=f"Google Test {uuid.uuid4().hex[:6]}", user_id=user.id)
session.add(space)
await session.flush()
space_id = space.id
@ -165,21 +165,21 @@ async def committed_google_data(async_engine):
title="Native Drive Doc",
document_type=DocumentType.GOOGLE_DRIVE_FILE,
content="quarterly budget from native google drive",
search_space_id=space_id,
workspace_id=space_id,
created_by_id=user_id,
)
legacy_doc = make_document(
title="Legacy Composio Drive Doc",
document_type=DocumentType.COMPOSIO_GOOGLE_DRIVE_CONNECTOR,
content="quarterly budget from composio google drive",
search_space_id=space_id,
workspace_id=space_id,
created_by_id=user_id,
)
file_doc = make_document(
title="Plain File",
document_type=DocumentType.FILE,
content="quarterly budget uploaded as file",
search_space_id=space_id,
workspace_id=space_id,
created_by_id=user_id,
)
session.add_all([native_doc, legacy_doc, file_doc])
@ -195,7 +195,7 @@ async def committed_google_data(async_engine):
)
await session.flush()
yield {"search_space_id": space_id, "user_id": user_id}
yield {"workspace_id": space_id, "user_id": user_id}
async with async_engine.begin() as conn:
await conn.execute(
@ -257,7 +257,7 @@ async def seed_connector(
):
"""Seed a connector with committed data. Returns dict and cleanup function.
Yields ``{"connector_id", "search_space_id", "user_id"}``.
Yields ``{"connector_id", "workspace_id", "user_id"}``.
"""
space_id = None
@ -275,7 +275,7 @@ async def seed_connector(
session.add(user)
await session.flush()
space = SearchSpace(
space = Workspace(
name=f"{name_prefix} {uuid.uuid4().hex[:6]}", user_id=user.id
)
session.add(space)
@ -287,7 +287,7 @@ async def seed_connector(
connector_type=connector_type,
is_indexable=True,
config=config,
search_space_id=space_id,
workspace_id=space_id,
user_id=user.id,
)
session.add(connector)
@ -297,13 +297,13 @@ async def seed_connector(
return {
"connector_id": connector_id,
"search_space_id": space_id,
"workspace_id": space_id,
"user_id": user_id,
}
async def cleanup_space(async_engine, space_id: int):
"""Delete a search space (cascades to connectors/documents)."""
"""Delete a workspace (cascades to connectors/documents)."""
async with async_engine.begin() as conn:
await conn.execute(
text("DELETE FROM workspaces WHERE id = :sid"), {"sid": space_id}

12
surfsense_backend/tests/integration/google_unification/test_calendar_indexer_credentials.py
View file

@ -37,7 +37,7 @@ async def composio_calendar(async_engine):
name_prefix="cal-composio",
)
yield data
await cleanup_space(async_engine, data["search_space_id"])
await cleanup_space(async_engine, data["workspace_id"])
@pytest_asyncio.fixture
@ -49,7 +49,7 @@ async def composio_calendar_no_id(async_engine):
name_prefix="cal-noid",
)
yield data
await cleanup_space(async_engine, data["search_space_id"])
await cleanup_space(async_engine, data["workspace_id"])
@pytest_asyncio.fixture
@ -67,7 +67,7 @@ async def native_calendar(async_engine):
name_prefix="cal-native",
)
yield data
await cleanup_space(async_engine, data["search_space_id"])
await cleanup_space(async_engine, data["workspace_id"])
@patch(_GET_ACCESS_TOKEN)
@ -98,7 +98,7 @@ async def test_composio_calendar_uses_composio_service(
await index_google_calendar_events(
session=session,
connector_id=data["connector_id"],
search_space_id=data["search_space_id"],
workspace_id=data["workspace_id"],
user_id=data["user_id"],
)
@ -137,7 +137,7 @@ async def test_composio_calendar_without_account_id_returns_error(
count, _skipped, error = await index_google_calendar_events(
session=session,
connector_id=data["connector_id"],
search_space_id=data["search_space_id"],
workspace_id=data["workspace_id"],
user_id=data["user_id"],
)
@ -179,7 +179,7 @@ async def test_native_calendar_uses_google_calendar_connector(
await index_google_calendar_events(
session=session,
connector_id=data["connector_id"],
search_space_id=data["search_space_id"],
workspace_id=data["workspace_id"],
user_id=data["user_id"],
)

12
surfsense_backend/tests/integration/google_unification/test_drive_indexer_credentials.py
View file

@ -62,7 +62,7 @@ async def committed_drive_connector(async_engine):
name_prefix="drive-composio",
)
yield data
await cleanup_space(async_engine, data["search_space_id"])
await cleanup_space(async_engine, data["workspace_id"])
@pytest_asyncio.fixture
@ -80,7 +80,7 @@ async def committed_native_drive_connector(async_engine):
name_prefix="drive-native",
)
yield data
await cleanup_space(async_engine, data["search_space_id"])
await cleanup_space(async_engine, data["workspace_id"])
@pytest_asyncio.fixture
@ -92,7 +92,7 @@ async def committed_composio_no_account_id(async_engine):
name_prefix="drive-noid",
)
yield data
await cleanup_space(async_engine, data["search_space_id"])
await cleanup_space(async_engine, data["workspace_id"])
@patch(_GET_ACCESS_TOKEN)
@ -127,7 +127,7 @@ async def test_composio_drive_indexer_uses_composio_drive_client(
await index_google_drive_files(
session=session,
connector_id=data["connector_id"],
search_space_id=data["search_space_id"],
workspace_id=data["workspace_id"],
user_id=data["user_id"],
folder_id="test-folder-id",
)
@ -167,7 +167,7 @@ async def test_composio_connector_without_account_id_returns_error(
count, _skipped, error, _unsupported = await index_google_drive_files(
session=session,
connector_id=data["connector_id"],
search_space_id=data["search_space_id"],
workspace_id=data["workspace_id"],
user_id=data["user_id"],
folder_id="test-folder-id",
)
@ -207,7 +207,7 @@ async def test_native_connector_uses_google_drive_client(
await index_google_drive_files(
session=session,
connector_id=data["connector_id"],
search_space_id=data["search_space_id"],
workspace_id=data["workspace_id"],
user_id=data["user_id"],
folder_id="test-folder-id",
)

12
surfsense_backend/tests/integration/google_unification/test_gmail_indexer_credentials.py
View file

@ -37,7 +37,7 @@ async def composio_gmail(async_engine):
name_prefix="gmail-composio",
)
yield data
await cleanup_space(async_engine, data["search_space_id"])
await cleanup_space(async_engine, data["workspace_id"])
@pytest_asyncio.fixture
@ -49,7 +49,7 @@ async def composio_gmail_no_id(async_engine):
name_prefix="gmail-noid",
)
yield data
await cleanup_space(async_engine, data["search_space_id"])
await cleanup_space(async_engine, data["workspace_id"])
@pytest_asyncio.fixture
@ -67,7 +67,7 @@ async def native_gmail(async_engine):
name_prefix="gmail-native",
)
yield data
await cleanup_space(async_engine, data["search_space_id"])
await cleanup_space(async_engine, data["workspace_id"])
@patch(_GET_ACCESS_TOKEN)
@ -101,7 +101,7 @@ async def test_composio_gmail_uses_composio_service(
await index_google_gmail_messages(
session=session,
connector_id=data["connector_id"],
search_space_id=data["search_space_id"],
workspace_id=data["workspace_id"],
user_id=data["user_id"],
)
@ -140,7 +140,7 @@ async def test_composio_gmail_without_account_id_returns_error(
count, _skipped, error = await index_google_gmail_messages(
session=session,
connector_id=data["connector_id"],
search_space_id=data["search_space_id"],
workspace_id=data["workspace_id"],
user_id=data["user_id"],
)
@ -180,7 +180,7 @@ async def test_native_gmail_uses_google_gmail_connector(
await index_google_gmail_messages(
session=session,
connector_id=data["connector_id"],
search_space_id=data["search_space_id"],
workspace_id=data["workspace_id"],
user_id=data["user_id"],
)

12
surfsense_backend/tests/integration/google_unification/test_hybrid_search_type_filtering.py
View file

@ -19,13 +19,13 @@ async def test_list_of_types_returns_both_matching_doc_types(
db_session, seed_google_docs
):
"""Searching with a list of document types returns documents of ALL listed types."""
space_id = seed_google_docs["search_space"].id
space_id = seed_google_docs["workspace"].id
retriever = ChucksHybridSearchRetriever(db_session)
results = await retriever.hybrid_search(
query_text="quarterly report",
top_k=10,
search_space_id=space_id,
workspace_id=space_id,
document_type=["GOOGLE_DRIVE_FILE", "COMPOSIO_GOOGLE_DRIVE_CONNECTOR"],
query_embedding=DUMMY_EMBEDDING,
)
@ -40,13 +40,13 @@ async def test_list_of_types_returns_both_matching_doc_types(
async def test_single_string_type_returns_only_that_type(db_session, seed_google_docs):
"""Searching with a single string type returns only documents of that exact type."""
space_id = seed_google_docs["search_space"].id
space_id = seed_google_docs["workspace"].id
retriever = ChucksHybridSearchRetriever(db_session)
results = await retriever.hybrid_search(
query_text="quarterly report",
top_k=10,
search_space_id=space_id,
workspace_id=space_id,
document_type="GOOGLE_DRIVE_FILE",
query_embedding=DUMMY_EMBEDDING,
)
@ -59,13 +59,13 @@ async def test_single_string_type_returns_only_that_type(db_session, seed_google
async def test_all_invalid_types_returns_empty(db_session, seed_google_docs):
"""Searching with a list of nonexistent types returns an empty list, no exceptions."""
space_id = seed_google_docs["search_space"].id
space_id = seed_google_docs["workspace"].id
retriever = ChucksHybridSearchRetriever(db_session)
results = await retriever.hybrid_search(
query_text="quarterly report",
top_k=10,
search_space_id=space_id,
workspace_id=space_id,
document_type=["NONEXISTENT_TYPE"],
query_embedding=DUMMY_EMBEDDING,
)

12
surfsense_backend/tests/integration/google_unification/test_search_includes_legacy_docs.py
View file

@ -19,13 +19,13 @@ async def test_search_google_drive_includes_legacy_composio_docs(
async_engine, committed_google_data, patched_session_factory, patched_embed
):
"""search_google_drive returns both GOOGLE_DRIVE_FILE and COMPOSIO_GOOGLE_DRIVE_CONNECTOR docs."""
space_id = committed_google_data["search_space_id"]
space_id = committed_google_data["workspace_id"]
async with patched_session_factory() as session:
service = ConnectorService(session, search_space_id=space_id)
service = ConnectorService(session, workspace_id=space_id)
_, raw_docs = await service.search_google_drive(
user_query="quarterly budget",
search_space_id=space_id,
workspace_id=space_id,
top_k=10,
)
@ -51,13 +51,13 @@ async def test_search_files_does_not_include_google_types(
async_engine, committed_google_data, patched_session_factory, patched_embed
):
"""search_files returns only FILE docs, not Google Drive docs."""
space_id = committed_google_data["search_space_id"]
space_id = committed_google_data["workspace_id"]
async with patched_session_factory() as session:
service = ConnectorService(session, search_space_id=space_id)
service = ConnectorService(session, workspace_id=space_id)
_, raw_docs = await service.search_files(
user_query="quarterly budget",
search_space_id=space_id,
workspace_id=space_id,
top_k=10,
)