Skip to content

armgit / dify

Public Imported from GitHub
0

Build Agentic workflows, RAG pipelines, with rich AI model and tool support on one collaborative workspace. Deploy on cloud, VPC, or self-hosted, so teams move from prototype to production without rebuilding the stack.

test(api): strictly type plugin model and chunk merger tests (#45230)

Eddy ZHANGed49462a42f0
diff --git a/api/tests/unit_tests/core/plugin/backwards_invocation/test_model.py b/api/tests/unit_tests/core/plugin/backwards_invocation/test_model.py
index 0bd803a2f32..5e871f8c136 100644
--- a/api/tests/unit_tests/core/plugin/backwards_invocation/test_model.py
+++ b/api/tests/unit_tests/core/plugin/backwards_invocation/test_model.py
@@ -1,6 +1,7 @@
import json
from collections.abc import Callable, Iterator
from dataclasses import dataclass, field
+from typing import Literal
from unittest.mock import patch
import httpx
@@ -93,7 +94,7 @@ def _tts_schema(audio_type: str) -> AIModelEntity:
)
-def test_system_model_helpers_forward_user_id():
+def test_system_model_helpers_forward_user_id() -> None:
with (
patch(
"core.plugin.backwards_invocation.model.ModelInvocationUtils.get_max_llm_context_tokens",
@@ -122,7 +123,7 @@ def test_system_model_helpers_forward_user_id():
)
-def test_invoke_summary_uses_same_user_scope_for_token_helpers():
+def test_invoke_summary_uses_same_user_scope_for_token_helpers() -> None:
tenant = Tenant(name="Test Workspace")
tenant.id = "tenant-1"
payload = RequestInvokeSummary(text="short", instruction="keep it concise")
@@ -239,7 +240,7 @@ class _SchemaModelInstance:
return self.schema
-def test_invoke_text_embedding_forwards_input_type_to_bound_model_instance():
+def test_invoke_text_embedding_forwards_input_type_to_bound_model_instance() -> None:
tenant = _backwards_invocation_model_tenant()
payload = RequestInvokeTextEmbedding(
provider="langgenius/openai/openai",
@@ -270,7 +271,7 @@ def test_invoke_text_embedding_forwards_input_type_to_bound_model_instance():
]
-def test_invoke_multimodal_embedding_uses_bound_model_instance():
+def test_invoke_multimodal_embedding_uses_bound_model_instance() -> None:
tenant = _backwards_invocation_model_tenant()
documents = [{"content": "AQID", "content_type": "image", "file_id": "asset.png"}]
payload = RequestInvokeMultimodalEmbedding(
@@ -305,7 +306,7 @@ def test_invoke_multimodal_embedding_uses_bound_model_instance():
]
-def test_invoke_rerank_accepts_omitted_threshold_and_top_n():
+def test_invoke_rerank_accepts_omitted_threshold_and_top_n() -> None:
tenant = _backwards_invocation_model_tenant()
payload = RequestInvokeRerank(
provider="langgenius/cohere/cohere",
@@ -335,7 +336,9 @@ def test_invoke_rerank_accepts_omitted_threshold_and_top_n():
("model_type", "selected"),
[(ModelType.TEXT_EMBEDDING, False), (ModelType.LLM, False), (ModelType.LLM, True)],
)
-def test_list_models_returns_only_active_tenant_models_with_installed_identity(model_type, selected):
+def test_list_models_returns_only_active_tenant_models_with_installed_identity(
+ model_type: Literal[ModelType.LLM, ModelType.TEXT_EMBEDDING], selected: bool
+) -> None:
model_name = "selected-model"
payload = RequestListModels(model_type=model_type, model=model_name if selected else None, limit=10)
provider = _ModelProvider(provider="langgenius/openai/openai")
@@ -395,7 +398,7 @@ def test_list_models_returns_only_active_tenant_models_with_installed_identity(m
@pytest.mark.parametrize("parameter", ["max_tokens", "max_completion_tokens", "maxtoken", "custom_limit"])
-def test_model_token_limits_use_schema_maximum_not_default(parameter):
+def test_model_token_limits_use_schema_maximum_not_default(parameter: str) -> None:
schema = _ModelSchema(
model_properties={"context_size": 131072},
parameter_rules=[_ParameterRule(name=parameter, use_template="max_tokens", max=32768.0, default=1024)],
@@ -405,6 +408,7 @@ def test_model_token_limits_use_schema_maximum_not_default(parameter):
limits = PluginModelBackwardsInvocation._model_token_limits(
tenant_id="tenant-1", user_id="user-1", provider="plugin/provider", model="selected-model"
)
+ assert limits is not None
assert limits.context_tokens == 131072
assert limits.max_output_tokens == 32768
assert limits.output_parameter == parameter
@@ -418,7 +422,7 @@ def test_model_token_limits_use_schema_maximum_not_default(parameter):
@pytest.mark.parametrize("maximum", [None, 0, -1, True, float("inf"), float("nan"), 10**400, 12.5, "32768"])
-def test_model_token_limits_do_not_invent_unknown_output_capacity(maximum):
+def test_model_token_limits_do_not_invent_unknown_output_capacity(maximum: object) -> None:
schema = _ModelSchema(
model_properties={"context_size": 8192},
parameter_rules=[_ParameterRule(name="max_tokens", use_template=None, max=maximum, default=4096)],
@@ -431,11 +435,12 @@ def test_model_token_limits_do_not_invent_unknown_output_capacity(maximum):
limits = PluginModelBackwardsInvocation._model_token_limits(
tenant_id="tenant-1", user_id="user-1", provider="plugin/provider", model="selected-model"
)
+ assert limits is not None
assert limits.context_tokens == 8192
assert limits.max_output_tokens is None
-def test_model_token_limits_respect_the_tighter_declared_alias():
+def test_model_token_limits_respect_the_tighter_declared_alias() -> None:
schema = _ModelSchema(
model_properties={},
parameter_rules=[
@@ -451,11 +456,12 @@ def test_model_token_limits_respect_the_tighter_declared_alias():
limits = PluginModelBackwardsInvocation._model_token_limits(
tenant_id="tenant-1", user_id="user-1", provider="plugin/provider", model="selected-model"
)
+ assert limits is not None
assert limits.max_output_tokens == 8192
assert limits.output_parameter == "max_output_tokens"
-def test_invoke_tts_emits_the_verified_mime_type_for_backwards_invocation():
+def test_invoke_tts_emits_the_verified_mime_type_for_backwards_invocation() -> None:
tenant = Tenant(name="Test Workspace")
tenant.id = "tenant-1"
model_instance = _tts_model_instance()
@@ -485,7 +491,7 @@ def test_invoke_tts_emits_the_verified_mime_type_for_backwards_invocation():
]
-def test_invoke_tts_defers_provider_errors_until_the_response_generator_is_consumed():
+def test_invoke_tts_defers_provider_errors_until_the_response_generator_is_consumed() -> None:
tenant = Tenant(name="Test Workspace")
tenant.id = "tenant-1"
model_instance = _tts_model_instance()
@@ -509,7 +515,7 @@ def test_invoke_tts_defers_provider_errors_until_the_response_generator_is_consu
next(response)
-def test_backwards_event_stream_serializes_a_deferred_mime_error():
+def test_backwards_event_stream_serializes_a_deferred_mime_error() -> None:
tenant = Tenant(name="Test Workspace")
tenant.id = "tenant-1"
model_instance = _tts_model_instance()
diff --git a/api/tests/unit_tests/core/plugin/utils/test_chunk_merger.py b/api/tests/unit_tests/core/plugin/utils/test_chunk_merger.py
index 314f6bff026..8ca1a7118e3 100644
--- a/api/tests/unit_tests/core/plugin/utils/test_chunk_merger.py
+++ b/api/tests/unit_tests/core/plugin/utils/test_chunk_merger.py
@@ -10,14 +10,14 @@ from graphon.file import File, FileTransferMethod, FileType
class TestChunkMerger:
- def test_file_chunk_initialization(self):
+ def test_file_chunk_initialization(self) -> None:
"""Test FileChunk initialization."""
chunk = FileChunk(1024)
assert chunk.bytes_written == 0
assert chunk.total_length == 1024
assert len(chunk.data) == 1024
- def test_merge_blob_chunks_with_single_complete_chunk(self):
+ def test_merge_blob_chunks_with_single_complete_chunk(self) -> None:
"""Test merging a single complete blob chunk."""
def mock_generator() -> Generator[ToolInvokeMessage]:
@@ -43,7 +43,7 @@ class TestChunkMerger:
# The buffer should contain the complete data
assert result[0].message.blob[:10] == b"HelloWorld"
- def test_merge_blob_chunks_with_multiple_files(self):
+ def test_merge_blob_chunks_with_multiple_files(self) -> None:
"""Test merging chunks from multiple files."""
def mock_generator() -> Generator[ToolInvokeMessage]:
@@ -81,7 +81,7 @@ class TestChunkMerger:
# Check that both files are properly merged
assert all(r.type == ToolInvokeMessage.MessageType.BLOB for r in result)
- def test_merge_blob_chunks_passes_through_non_blob_messages(self):
+ def test_merge_blob_chunks_passes_through_non_blob_messages(self) -> None:
"""Test that non-blob messages pass through unchanged."""
def mock_generator() -> Generator[ToolInvokeMessage]:
@@ -113,7 +113,7 @@ class TestChunkMerger:
assert isinstance(result[2].message, ToolInvokeMessage.TextMessage)
assert result[2].message.text == "World"
- def test_merge_blob_chunks_file_too_large(self):
+ def test_merge_blob_chunks_file_too_large(self) -> None:
"""Test that error is raised when file exceeds max size."""
def mock_generator() -> Generator[ToolInvokeMessage]:
@@ -131,7 +131,7 @@ class TestChunkMerger:
list(merge_blob_chunks(mock_generator(), max_file_size=1000))
assert "File is too large" in str(exc_info.value)
- def test_merge_blob_chunks_chunk_too_large(self):
+ def test_merge_blob_chunks_chunk_too_large(self) -> None:
"""Test that error is raised when chunk exceeds max chunk size."""
def mock_generator() -> Generator[ToolInvokeMessage]:
@@ -147,7 +147,7 @@ class TestChunkMerger:
list(merge_blob_chunks(mock_generator(), max_chunk_size=8192))
assert "File chunk is too large" in str(exc_info.value)
- def test_merge_blob_chunks_with_agent_invoke_message(self):
+ def test_merge_blob_chunks_with_agent_invoke_message(self) -> None:
"""Test that merge_blob_chunks works with AgentInvokeMessage."""
def mock_generator() -> Generator[AgentInvokeMessage]:
@@ -171,7 +171,7 @@ class TestChunkMerger:
assert isinstance(result[0], AgentInvokeMessage)
assert result[0].type == AgentInvokeMessage.MessageType.BLOB
- def test_merge_blob_chunks_preserves_meta(self):
+ def test_merge_blob_chunks_preserves_meta(self) -> None:
"""Test that meta information is preserved in merged messages."""
def mock_generator() -> Generator[ToolInvokeMessage]:
@@ -187,7 +187,7 @@ class TestChunkMerger:
assert len(result) == 1
assert result[0].meta == {"key": "value"}
- def test_merge_blob_chunks_custom_limits(self):
+ def test_merge_blob_chunks_custom_limits(self) -> None:
"""Test merge_blob_chunks with custom size limits."""
def mock_generator() -> Generator[ToolInvokeMessage]:
@@ -221,7 +221,7 @@ class TestChunkMerger:
with pytest.raises(ValueError, match="File is too large which reached the limit of 0\\.000286102294921875MB"):
list(merge_blob_chunks(mock_generator2(), max_file_size=300))
- def test_merge_blob_chunks_data_integrity(self):
+ def test_merge_blob_chunks_data_integrity(self) -> None:
"""Test that merged chunks exactly match the original data."""
# Create original data
original_data = b"This is a test message that will be split into chunks for testing purposes."
@@ -255,7 +255,7 @@ class TestChunkMerger:
# Verify the merged data exactly matches the original
assert result[0].message.blob == original_data
- def test_merge_blob_chunks_empty_chunk(self):
+ def test_merge_blob_chunks_empty_chunk(self) -> None:
"""Test handling of empty chunks."""
def mock_generator() -> Generator[ToolInvokeMessage]:
@@ -288,7 +288,7 @@ class TestChunkMerger:
# The final blob should contain "Hello" followed by "World"
assert result[0].message.blob[:10] == b"HelloWorld"
- def test_merge_blob_chunks_single_chunk_file(self):
+ def test_merge_blob_chunks_single_chunk_file(self) -> None:
"""Test file that arrives as a single complete chunk."""
def mock_generator() -> Generator[ToolInvokeMessage]:
@@ -310,7 +310,7 @@ class TestChunkMerger:
assert isinstance(result[0].message, ToolInvokeMessage.BlobMessage)
assert result[0].message.blob == b"Single Data"
- def test_merge_blob_chunks_concurrent_files(self):
+ def test_merge_blob_chunks_concurrent_files(self) -> None:
"""Test that chunks from different files are properly separated."""
def mock_generator() -> Generator[ToolInvokeMessage]:
@@ -358,7 +358,7 @@ class TestChunkMerger:
expected = {b"First file content", b"Second file data", b"Third file"}
assert blobs == expected
- def test_merge_blob_chunks_exact_buffer_size(self):
+ def test_merge_blob_chunks_exact_buffer_size(self) -> None:
"""Test that data fitting exactly in buffer works correctly."""
def mock_generator() -> Generator[ToolInvokeMessage]:
@@ -392,7 +392,7 @@ class TestChunkMerger:
assert len(result[0].message.blob) == 100
assert result[0].message.blob == b"X" * 100
- def test_merge_blob_chunks_large_file_simulation(self):
+ def test_merge_blob_chunks_large_file_simulation(self) -> None:
"""Test handling of a large file split into many chunks."""
def mock_generator() -> Generator[ToolInvokeMessage]:
@@ -433,7 +433,7 @@ class TestChunkMerger:
chunk = merged_data[chunk_start:chunk_end]
assert all(b == expected_byte for b in chunk), f"Chunk {i} has incorrect data"
- def test_merge_blob_chunks_sequential_order_required(self):
+ def test_merge_blob_chunks_sequential_order_required(self) -> None:
"""
Test note: The current implementation assumes chunks arrive in sequential order.
Out-of-order chunks would need additional logic to handle properly.
@@ -465,7 +465,7 @@ class TestChunkMerger:
class TestConverter:
- def test_convert_parameters_to_plugin_format_with_single_file_and_selector(self):
+ def test_convert_parameters_to_plugin_format_with_single_file_and_selector(self) -> None:
file_param = File(
tenant_id="tenant-1",
file_type=FileType.IMAGE,
@@ -498,7 +498,7 @@ class TestConverter:
assert converted["selector"]["provider_id"] == "org/plugin/provider"
assert converted["plain"] == 123
- def test_convert_parameters_to_plugin_format_with_lists_and_passthrough_values(self):
+ def test_convert_parameters_to_plugin_format_with_lists_and_passthrough_values(self) -> None:
file_one = File(
tenant_id="tenant-1",
file_type=FileType.DOCUMENT,
@@ -530,10 +530,11 @@ class TestConverter:
tool_parameters={},
)
+ empty_list: list[object] = []
params = {
"files": [file_one, file_two],
"selectors": [selector_one, selector_two],
- "empty_list": [],
+ "empty_list": empty_list,
"mixed_list": [file_one, "raw"],
"none_value": None,
}
@@ -549,7 +550,7 @@ class TestConverter:
assert converted["mixed_list"] == [file_one, "raw"]
assert converted["none_value"] is None
- def test_merge_blob_chunks_rejects_declared_size_above_limit(self):
+ def test_merge_blob_chunks_rejects_declared_size_above_limit(self) -> None:
"""A declared total_length above the limit is rejected before allocation."""
def mock_generator() -> Generator[ToolInvokeMessage]:
@@ -563,7 +564,7 @@ class TestConverter:
with pytest.raises(ValueError, match="File is too large"):
list(merge_blob_chunks(mock_generator(), max_file_size=1024))
- def test_merge_blob_chunks_rejects_absurd_declared_size(self):
+ def test_merge_blob_chunks_rejects_absurd_declared_size(self) -> None:
"""An absurd declared total_length raises ValueError, not MemoryError/OverflowError."""
def mock_generator() -> Generator[ToolInvokeMessage]:
diff --git a/api/tests/unit_tests/pyrefly.toml b/api/tests/unit_tests/pyrefly.toml
index 218bad532bb..df499fa46f8 100644
--- a/api/tests/unit_tests/pyrefly.toml
+++ b/api/tests/unit_tests/pyrefly.toml
@@ -186,11 +186,9 @@ project-excludes = [
"core/plugin/impl/test_oauth.py",
"core/plugin/impl/test_trigger.py",
"core/plugin/backwards_invocation/test_app.py",
- "core/plugin/backwards_invocation/test_model.py",
"core/plugin/impl/test_model_runtime.py",
"core/plugin/entities/test_plugin_daemon.py",
"core/plugin/impl/test_plugin.py",
- "core/plugin/utils/test_chunk_merger.py",
"core/prompt/test_advanced_prompt_transform.py",
"core/prompt/utils/test_extract_thread_messages.py",
"core/rag/cleaner/test_clean_processor.py",