zlei6/ragflow
1
0
Fork 0
mirror of https://github.com/infiniflow/ragflow.git synced 2026-07-19 22:21:04 +08:00
Code Issues Packages Projects Releases Wiki Activity

Refactor: reformat all code for lefthook using ruff and gofmt (#16585)

Browse Source
This commit is contained in:
Wang Qi
2026-07-03 12:53:39 +08:00
committed by GitHub
parent 19fcb4a981
commit 6a4b9be426
588 changed files with 11123 additions and 15412 deletions
Download Patch File Download Diff File Expand all files Collapse all files

35
test/unit_test/api/utils/test_doc_validation.py
View File

@@ -53,7 +53,7 @@ def test_validate_immutable_fields_no_changes():
doc.chunk_num = 10
doc.token_num = 100
doc.progress = 0.5
error_msg, error_code = validate_immutable_fields(update_doc_req, doc)
assert error_msg is None
assert error_code is None
@@ -66,7 +66,7 @@ def test_validate_immutable_fields_chunk_count_matches():
doc.chunk_num = 10
doc.token_num = 100
doc.progress = 0.5
error_msg, error_code = validate_immutable_fields(update_doc_req, doc)
assert error_msg is None
assert error_code is None
@@ -79,7 +79,7 @@ def test_validate_immutable_fields_token_count_matches():
doc.chunk_num = 10
doc.token_num = 100
doc.progress = 0.5
error_msg, error_code = validate_immutable_fields(update_doc_req, doc)
assert error_msg is None
assert error_code is None
@@ -92,7 +92,7 @@ def test_validate_immutable_fields_progress_matches():
doc.chunk_num = 10
doc.token_num = 100
doc.progress = 0.5
error_msg, error_code = validate_immutable_fields(update_doc_req, doc)
assert error_msg is None
assert error_code is None
@@ -105,7 +105,7 @@ def test_validate_immutable_fields_chunk_count_mismatch():
doc.chunk_num = 10
doc.token_num = 100
doc.progress = 0.5
error_msg, error_code = validate_immutable_fields(update_doc_req, doc)
assert error_msg == "Can't change `chunk_count`."
assert error_code == RetCode.DATA_ERROR
@@ -118,7 +118,7 @@ def test_validate_immutable_fields_token_count_mismatch():
doc.chunk_num = 10
doc.token_num = 100
doc.progress = 0.5
error_msg, error_code = validate_immutable_fields(update_doc_req, doc)
assert error_msg == "Can't change `token_count`."
assert error_code == RetCode.DATA_ERROR
@@ -131,7 +131,7 @@ def test_validate_immutable_fields_progress_mismatch():
doc.chunk_num = 10
doc.token_num = 100
doc.progress = 0.5
error_msg, error_code = validate_immutable_fields(update_doc_req, doc)
assert error_msg == "Can't change `progress`."
assert error_code == RetCode.DATA_ERROR
@@ -145,18 +145,18 @@ def test_validate_immutable_fields_progress_boundary_values():
doc.chunk_num = 10
doc.token_num = 100
doc.progress = 0.0
error_msg, error_code = validate_immutable_fields(update_doc_req, doc)
assert error_msg is None
assert error_code is None
# Test with 1.0
update_doc_req = UpdateDocumentReq(progress=1.0)
doc = Mock()
doc.chunk_num = 10
doc.token_num = 100
doc.progress = 1.0
error_msg, error_code = validate_immutable_fields(update_doc_req, doc)
assert error_msg is None
assert error_code is None
@@ -169,7 +169,7 @@ def test_validate_immutable_fields_none_values():
doc.chunk_num = 10
doc.token_num = 100
doc.progress = 0.5
error_msg, error_code = validate_immutable_fields(update_doc_req, doc)
assert error_msg is None
assert error_code is None
@@ -240,6 +240,7 @@ def test_validate_document_name_valid():
assert error_msg is None
assert error_code is None
def test_validate_document_name_attr_error():
"""Test valid document name update."""
req_doc_name = 0
@@ -312,7 +313,7 @@ def test_validate_chunk_method_valid():
doc = Mock()
doc.type = FileType.PDF
doc.name = "document.pdf"
error_msg, error_code = validate_chunk_method(doc)
assert error_msg is None
assert error_code is None
@@ -323,7 +324,7 @@ def test_validate_chunk_method_visual_not_supported():
doc = Mock()
doc.type = FileType.VISUAL
doc.name = "image.jpg"
error_msg, error_code = validate_chunk_method(doc)
assert "Not supported yet!" in error_msg
assert error_code == RetCode.DATA_ERROR
@@ -334,7 +335,7 @@ def test_validate_chunk_method_ppt_not_supported():
doc = Mock()
doc.type = FileType.PDF
doc.name = "presentation.ppt"
error_msg, error_code = validate_chunk_method(doc)
assert "Not supported yet!" in error_msg
assert error_code == RetCode.DATA_ERROR
@@ -345,7 +346,7 @@ def test_validate_chunk_method_pptx_not_supported():
doc = Mock()
doc.type = FileType.PDF
doc.name = "presentation.pptx"
error_msg, error_code = validate_chunk_method(doc)
assert "Not supported yet!" in error_msg
assert error_code == RetCode.DATA_ERROR
@@ -356,7 +357,7 @@ def test_validate_chunk_method_pages_not_supported():
doc = Mock()
doc.type = FileType.PDF
doc.name = "document.pages"
error_msg, error_code = validate_chunk_method(doc)
assert "Not supported yet!" in error_msg
assert error_code == RetCode.DATA_ERROR
@@ -367,7 +368,7 @@ def test_validate_chunk_method_other_extensions_still_valid():
doc = Mock()
doc.type = FileType.PDF
doc.name = "document.docx"
error_msg, error_code = validate_chunk_method(doc)
assert error_msg is None
assert error_code is None

13
test/unit_test/api/utils/test_health_utils_minio.py
View File

@@ -17,6 +17,7 @@
Unit tests for MinIO health check (check_minio_alive) and scheme/verify helpers.
Covers SSL/HTTPS and certificate verification (issues #13158, #13159).
"""
from unittest.mock import patch, Mock
@@ -27,6 +28,7 @@ class TestMinioSchemeAndVerify:
def test_scheme_http_when_secure_false(self, mock_settings):
mock_settings.MINIO = {"host": "minio:9000", "secure": False}
from api.utils.health_utils import _minio_scheme_and_verify
scheme, verify = _minio_scheme_and_verify()
assert scheme == "http"
assert verify is True
@@ -35,6 +37,7 @@ class TestMinioSchemeAndVerify:
def test_scheme_https_when_secure_true(self, mock_settings):
mock_settings.MINIO = {"host": "minio:9000", "secure": True}
from api.utils.health_utils import _minio_scheme_and_verify
scheme, verify = _minio_scheme_and_verify()
assert scheme == "https"
assert verify is True
@@ -43,6 +46,7 @@ class TestMinioSchemeAndVerify:
def test_scheme_https_when_secure_string_true(self, mock_settings):
mock_settings.MINIO = {"host": "minio:9000", "secure": "true"}
from api.utils.health_utils import _minio_scheme_and_verify
scheme, verify = _minio_scheme_and_verify()
assert scheme == "https"
@@ -50,6 +54,7 @@ class TestMinioSchemeAndVerify:
def test_verify_false_for_self_signed(self, mock_settings):
mock_settings.MINIO = {"host": "minio:9000", "secure": True, "verify": False}
from api.utils.health_utils import _minio_scheme_and_verify
scheme, verify = _minio_scheme_and_verify()
assert scheme == "https"
assert verify is False
@@ -58,6 +63,7 @@ class TestMinioSchemeAndVerify:
def test_verify_string_false(self, mock_settings):
mock_settings.MINIO = {"host": "minio:9000", "verify": "false"}
from api.utils.health_utils import _minio_scheme_and_verify
_, verify = _minio_scheme_and_verify()
assert verify is False
@@ -65,6 +71,7 @@ class TestMinioSchemeAndVerify:
def test_default_verify_true_when_key_missing(self, mock_settings):
mock_settings.MINIO = {"host": "minio:9000"}
from api.utils.health_utils import _minio_scheme_and_verify
_, verify = _minio_scheme_and_verify()
assert verify is True
@@ -80,6 +87,7 @@ class TestCheckMinioAlive:
mock_response.status_code = 200
mock_get.return_value = mock_response
from api.utils.health_utils import check_minio_alive
result = check_minio_alive()
assert result["status"] == "alive"
assert "elapsed" in result["message"]
@@ -96,6 +104,7 @@ class TestCheckMinioAlive:
mock_response.status_code = 200
mock_get.return_value = mock_response
from api.utils.health_utils import check_minio_alive
check_minio_alive()
call_args = mock_get.call_args
assert call_args[0][0] == "https://minio:9000/minio/health/live"
@@ -108,6 +117,7 @@ class TestCheckMinioAlive:
mock_response.status_code = 200
mock_get.return_value = mock_response
from api.utils.health_utils import check_minio_alive
check_minio_alive()
call_args = mock_get.call_args
assert call_args[1]["verify"] is False
@@ -120,6 +130,7 @@ class TestCheckMinioAlive:
mock_response.status_code = 503
mock_get.return_value = mock_response
from api.utils.health_utils import check_minio_alive
result = check_minio_alive()
assert result["status"] == "timeout"
@@ -129,6 +140,7 @@ class TestCheckMinioAlive:
mock_settings.MINIO = {"host": "minio:9000"}
mock_get.side_effect = ConnectionError("Connection refused")
from api.utils.health_utils import check_minio_alive
result = check_minio_alive()
assert result["status"] == "timeout"
assert "error" in result["message"]
@@ -141,6 +153,7 @@ class TestCheckMinioAlive:
mock_response.status_code = 200
mock_get.return_value = mock_response
from api.utils.health_utils import check_minio_alive
check_minio_alive()
call_args = mock_get.call_args
assert call_args[1]["timeout"] == 10

212
test/unit_test/api/utils/test_oceanbase_health.py
View File

@@ -16,6 +16,7 @@
"""
Unit tests for OceanBase health check and performance monitoring functionality.
"""
import inspect
import os
import types
@@ -27,20 +28,15 @@ from api.utils.health_utils import get_oceanbase_status, check_oceanbase_health
class TestOceanBaseHealthCheck:
"""Test cases for OceanBase health check functionality."""
@patch('api.utils.health_utils.OBConnection')
@patch.dict(os.environ, {'DOC_ENGINE': 'oceanbase'})
@patch("api.utils.health_utils.OBConnection")
@patch.dict(os.environ, {"DOC_ENGINE": "oceanbase"})
def test_get_oceanbase_status_success(self, mock_ob_class):
"""Test successful OceanBase status retrieval."""
# Setup mock
mock_ob_connection = Mock()
mock_ob_connection.uri = "localhost:2881"
mock_ob_connection.health.return_value = {
"uri": "localhost:2881",
"version_comment": "OceanBase 4.3.5.1",
"status": "healthy",
"connection": "connected"
}
mock_ob_connection.health.return_value = {"uri": "localhost:2881", "version_comment": "OceanBase 4.3.5.1", "status": "healthy", "connection": "connected"}
mock_ob_connection.get_performance_metrics.return_value = {
"connection": "connected",
"latency_ms": 5.2,
@@ -49,13 +45,13 @@ class TestOceanBaseHealthCheck:
"query_per_second": 150,
"slow_queries": 2,
"active_connections": 10,
"max_connections": 300
"max_connections": 300,
}
mock_ob_class.return_value = mock_ob_connection
# Execute
result = get_oceanbase_status()
# Assert
assert result["status"] == "alive"
assert "message" in result
@@ -63,36 +59,31 @@ class TestOceanBaseHealthCheck:
assert "performance" in result["message"]
assert result["message"]["health"]["status"] == "healthy"
assert result["message"]["performance"]["latency_ms"] == 5.2
@patch.dict(os.environ, {'DOC_ENGINE': 'elasticsearch'})
@patch.dict(os.environ, {"DOC_ENGINE": "elasticsearch"})
def test_get_oceanbase_status_not_configured(self):
"""Test OceanBase status when not configured."""
with pytest.raises(Exception) as exc_info:
get_oceanbase_status()
assert "OceanBase is not in use" in str(exc_info.value)
@patch('api.utils.health_utils.OBConnection')
@patch.dict(os.environ, {'DOC_ENGINE': 'oceanbase'})
@patch("api.utils.health_utils.OBConnection")
@patch.dict(os.environ, {"DOC_ENGINE": "oceanbase"})
def test_get_oceanbase_status_connection_error(self, mock_ob_class):
"""Test OceanBase status when connection fails."""
mock_ob_class.side_effect = Exception("Connection failed")
result = get_oceanbase_status()
assert result["status"] == "timeout"
assert "error" in result["message"]
@patch('api.utils.health_utils.OBConnection')
@patch.dict(os.environ, {'DOC_ENGINE': 'oceanbase'})
@patch("api.utils.health_utils.OBConnection")
@patch.dict(os.environ, {"DOC_ENGINE": "oceanbase"})
def test_check_oceanbase_health_healthy(self, mock_ob_class):
"""Test OceanBase health check returns healthy status."""
mock_ob_connection = Mock()
mock_ob_connection.health.return_value = {
"uri": "localhost:2881",
"version_comment": "OceanBase 4.3.5.1",
"status": "healthy",
"connection": "connected"
}
mock_ob_connection.health.return_value = {"uri": "localhost:2881", "version_comment": "OceanBase 4.3.5.1", "status": "healthy", "connection": "connected"}
mock_ob_connection.get_performance_metrics.return_value = {
"connection": "connected",
"latency_ms": 5.2,
@@ -101,28 +92,23 @@ class TestOceanBaseHealthCheck:
"query_per_second": 150,
"slow_queries": 0,
"active_connections": 10,
"max_connections": 300
"max_connections": 300,
}
mock_ob_class.return_value = mock_ob_connection
result = check_oceanbase_health()
assert result["status"] == "healthy"
assert result["details"]["connection"] == "connected"
assert result["details"]["latency_ms"] == 5.2
assert result["details"]["query_per_second"] == 150
@patch('api.utils.health_utils.OBConnection')
@patch.dict(os.environ, {'DOC_ENGINE': 'oceanbase'})
@patch("api.utils.health_utils.OBConnection")
@patch.dict(os.environ, {"DOC_ENGINE": "oceanbase"})
def test_check_oceanbase_health_degraded(self, mock_ob_class):
"""Test OceanBase health check returns degraded status for high latency."""
mock_ob_connection = Mock()
mock_ob_connection.health.return_value = {
"uri": "localhost:2881",
"version_comment": "OceanBase 4.3.5.1",
"status": "healthy",
"connection": "connected"
}
mock_ob_connection.health.return_value = {"uri": "localhost:2881", "version_comment": "OceanBase 4.3.5.1", "status": "healthy", "connection": "connected"}
mock_ob_connection.get_performance_metrics.return_value = {
"connection": "connected",
"latency_ms": 1500.0, # High latency > 1000ms
@@ -131,43 +117,35 @@ class TestOceanBaseHealthCheck:
"query_per_second": 50,
"slow_queries": 5,
"active_connections": 10,
"max_connections": 300
"max_connections": 300,
}
mock_ob_class.return_value = mock_ob_connection
result = check_oceanbase_health()
assert result["status"] == "degraded"
assert result["details"]["latency_ms"] == 1500.0
@patch('api.utils.health_utils.OBConnection')
@patch.dict(os.environ, {'DOC_ENGINE': 'oceanbase'})
@patch("api.utils.health_utils.OBConnection")
@patch.dict(os.environ, {"DOC_ENGINE": "oceanbase"})
def test_check_oceanbase_health_unhealthy(self, mock_ob_class):
"""Test OceanBase health check returns unhealthy status."""
mock_ob_connection = Mock()
mock_ob_connection.health.return_value = {
"uri": "localhost:2881",
"status": "unhealthy",
"connection": "disconnected",
"error": "Connection timeout"
}
mock_ob_connection.get_performance_metrics.return_value = {
"connection": "disconnected",
"error": "Connection timeout"
}
mock_ob_connection.health.return_value = {"uri": "localhost:2881", "status": "unhealthy", "connection": "disconnected", "error": "Connection timeout"}
mock_ob_connection.get_performance_metrics.return_value = {"connection": "disconnected", "error": "Connection timeout"}
mock_ob_class.return_value = mock_ob_connection
result = check_oceanbase_health()
assert result["status"] == "unhealthy"
assert result["details"]["connection"] == "disconnected"
assert "error" in result["details"]
@patch.dict(os.environ, {'DOC_ENGINE': 'elasticsearch'})
@patch.dict(os.environ, {"DOC_ENGINE": "elasticsearch"})
def test_check_oceanbase_health_not_configured(self):
"""Test OceanBase health check when not configured."""
result = check_oceanbase_health()
assert result["status"] == "not_configured"
assert result["details"]["connection"] == "not_configured"
assert "not configured" in result["details"]["message"].lower()
@@ -175,29 +153,32 @@ class TestOceanBaseHealthCheck:
class TestOBConnectionPerformanceMetrics:
"""Test cases for OBConnection performance metrics methods."""
def _create_mock_connection(self):
"""Create a mock OBConnection with actual methods."""
# Create a simple object and bind the real methods to it
class MockConn:
pass
conn = MockConn()
# Get the actual class from the singleton wrapper's closure
from rag.utils import ob_conn
# OBConnection is wrapped by @singleton decorator, so it's a function
# The original class is stored in the closure of the singleton function
# Find the class by checking all closure cells
ob_connection_class = None
if hasattr(ob_conn.OBConnection, '__closure__') and ob_conn.OBConnection.__closure__:
if hasattr(ob_conn.OBConnection, "__closure__") and ob_conn.OBConnection.__closure__:
for cell in ob_conn.OBConnection.__closure__:
cell_value = cell.cell_contents
if inspect.isclass(cell_value):
ob_connection_class = cell_value
break
if ob_connection_class is None:
raise ValueError("Could not find OBConnection class in closure")
# Bind the actual methods to our mock object
conn.get_performance_metrics = types.MethodType(ob_connection_class.get_performance_metrics, conn)
conn._get_storage_info = types.MethodType(ob_connection_class._get_storage_info, conn)
@@ -205,7 +186,7 @@ class TestOBConnectionPerformanceMetrics:
conn._get_slow_query_count = types.MethodType(ob_connection_class._get_slow_query_count, conn)
conn._estimate_qps = types.MethodType(ob_connection_class._estimate_qps, conn)
return conn
def test_get_performance_metrics_success(self):
"""Test successful retrieval of performance metrics."""
# Create mock connection with actual methods
@@ -214,29 +195,27 @@ class TestOBConnectionPerformanceMetrics:
conn.client = mock_client
conn.uri = "localhost:2881"
conn.db_name = "test"
# Mock client methods - create separate mock results for each call
mock_result1 = Mock()
mock_result1.fetchone.return_value = (1,)
mock_result2 = Mock()
mock_result2.fetchone.return_value = (100.5,)
mock_result3 = Mock()
mock_result3.fetchone.return_value = (100.0,)
mock_result4 = Mock()
mock_result4.fetchall.return_value = [
(1, 'user', 'host', 'db', 'Query', 0, 'executing', 'SELECT 1')
]
mock_result4.fetchone.return_value = ('max_connections', '300')
mock_result4.fetchall.return_value = [(1, "user", "host", "db", "Query", 0, "executing", "SELECT 1")]
mock_result4.fetchone.return_value = ("max_connections", "300")
mock_result5 = Mock()
mock_result5.fetchone.return_value = (0,)
mock_result6 = Mock()
mock_result6.fetchone.return_value = (5,)
# Setup side_effect to return different mocks for different queries
def sql_side_effect(query):
if "SELECT 1" in query:
@@ -254,21 +233,22 @@ class TestOBConnectionPerformanceMetrics:
elif "information_schema.processlist" in query and "COUNT" in query:
return mock_result6
return Mock()
mock_client.perform_raw_text_sql.side_effect = sql_side_effect
mock_client.pool_size = 300
# Mock logger
import logging
conn.logger = logging.getLogger('test')
conn.logger = logging.getLogger("test")
result = conn.get_performance_metrics()
assert result["connection"] == "connected"
assert result["latency_ms"] >= 0
assert "storage_used" in result
assert "storage_total" in result
def test_get_performance_metrics_connection_error(self):
"""Test performance metrics when connection fails."""
# Create mock connection with actual methods
@@ -277,14 +257,14 @@ class TestOBConnectionPerformanceMetrics:
conn.client = mock_client
conn.uri = "localhost:2881"
conn.logger = Mock()
mock_client.perform_raw_text_sql.side_effect = Exception("Connection failed")
result = conn.get_performance_metrics()
assert result["connection"] == "disconnected"
assert "error" in result
def test_get_storage_info_success(self):
"""Test successful retrieval of storage information."""
# Create mock connection with actual methods
@@ -293,27 +273,27 @@ class TestOBConnectionPerformanceMetrics:
conn.client = mock_client
conn.db_name = "test"
conn.logger = Mock()
mock_result1 = Mock()
mock_result1.fetchone.return_value = (100.5,)
mock_result2 = Mock()
mock_result2.fetchone.return_value = (100.0,)
def sql_side_effect(query):
if "information_schema.tables" in query:
return mock_result1
elif "__all_disk_stat" in query:
return mock_result2
return Mock()
mock_client.perform_raw_text_sql.side_effect = sql_side_effect
result = conn._get_storage_info()
assert "storage_used" in result
assert "storage_total" in result
assert "MB" in result["storage_used"]
def test_get_storage_info_fallback(self):
"""Test storage info with fallback when total space unavailable."""
# Create mock connection with actual methods
@@ -322,7 +302,7 @@ class TestOBConnectionPerformanceMetrics:
conn.client = mock_client
conn.db_name = "test"
conn.logger = Mock()
# First query succeeds, second fails
def side_effect(query):
if "information_schema.tables" in query:
@@ -331,14 +311,14 @@ class TestOBConnectionPerformanceMetrics:
return mock_result
else:
raise Exception("Table not found")
mock_client.perform_raw_text_sql.side_effect = side_effect
result = conn._get_storage_info()
assert "storage_used" in result
assert "storage_total" in result
def test_get_connection_pool_stats(self):
"""Test retrieval of connection pool statistics."""
# Create mock connection with actual methods
@@ -347,31 +327,28 @@ class TestOBConnectionPerformanceMetrics:
conn.client = mock_client
conn.logger = Mock()
mock_client.pool_size = 300
mock_result1 = Mock()
mock_result1.fetchall.return_value = [
(1, 'user', 'host', 'db', 'Query', 0, 'executing', 'SELECT 1'),
(2, 'user', 'host', 'db', 'Sleep', 10, None, None)
]
mock_result1.fetchall.return_value = [(1, "user", "host", "db", "Query", 0, "executing", "SELECT 1"), (2, "user", "host", "db", "Sleep", 10, None, None)]
mock_result2 = Mock()
mock_result2.fetchone.return_value = ('max_connections', '300')
mock_result2.fetchone.return_value = ("max_connections", "300")
def sql_side_effect(query):
if "SHOW PROCESSLIST" in query:
return mock_result1
elif "SHOW VARIABLES LIKE 'max_connections'" in query:
return mock_result2
return Mock()
mock_client.perform_raw_text_sql.side_effect = sql_side_effect
result = conn._get_connection_pool_stats()
assert "active_connections" in result
assert "max_connections" in result
assert result["active_connections"] >= 0
def test_get_slow_query_count(self):
"""Test retrieval of slow query count."""
# Create mock connection with actual methods
@@ -379,16 +356,16 @@ class TestOBConnectionPerformanceMetrics:
mock_client = Mock()
conn.client = mock_client
conn.logger = Mock()
mock_result = Mock()
mock_result.fetchone.return_value = (5,)
mock_client.perform_raw_text_sql.return_value = mock_result
result = conn._get_slow_query_count(threshold_seconds=1)
assert isinstance(result, int)
assert result >= 0
def test_estimate_qps(self):
"""Test QPS estimation."""
# Create mock connection with actual methods
@@ -396,17 +373,16 @@ class TestOBConnectionPerformanceMetrics:
mock_client = Mock()
conn.client = mock_client
conn.logger = Mock()
mock_result = Mock()
mock_result.fetchone.return_value = (10,)
mock_client.perform_raw_text_sql.return_value = mock_result
result = conn._estimate_qps()
assert isinstance(result, int)
assert result >= 0
if __name__ == "__main__":
pytest.main([__file__, "-v"])