Simplified the preprocessing interface.

This commit is contained in:
ilya-kolchinsky 2025-03-11 13:05:48 +01:00
parent 1eeba2cc8a
commit ad4cf97604
8 changed files with 31 additions and 43 deletions

View file

@ -52,7 +52,7 @@ class InclineBasicPreprocessorImpl(Preprocessing, PreprocessorsProtocolPrivate):
async def unregister_preprocessor(self, preprocessor_id: str) -> None: ...
async def preprocess(
async def do_preprocess(
self,
preprocessor_id: str,
preprocessor_inputs: List[PreprocessingDataElement],
@ -98,12 +98,12 @@ class InclineBasicPreprocessorImpl(Preprocessing, PreprocessorsProtocolPrivate):
success=True, output_data_type=PreprocessingDataType.raw_text_document, results=results
)
async def chain_preprocess(
async def preprocess(
self,
preprocessors: PreprocessorChain,
preprocessor_inputs: List[PreprocessingDataElement],
) -> PreprocessorResponse:
return await self.preprocess(preprocessor_id="", preprocessor_inputs=preprocessor_inputs)
return await self.do_preprocess(preprocessor_id="", preprocessor_inputs=preprocessor_inputs)
@staticmethod
def _resolve_input_type(preprocessor_input: PreprocessingDataElement) -> PreprocessingDataType:

View file

@ -47,7 +47,7 @@ class InclineDoclingPreprocessorImpl(Preprocessing, PreprocessorsProtocolPrivate
async def unregister_preprocessor(self, preprocessor_id: str) -> None: ...
async def preprocess(
async def do_preprocess(
self,
preprocessor_id: str,
preprocessor_inputs: List[PreprocessingDataElement],
@ -106,9 +106,9 @@ class InclineDoclingPreprocessorImpl(Preprocessing, PreprocessorsProtocolPrivate
)
return PreprocessorResponse(success=True, output_data_type=output_data_type, results=results)
async def chain_preprocess(
async def preprocess(
self,
preprocessors: PreprocessorChain,
preprocessor_inputs: List[PreprocessingDataElement],
) -> PreprocessorResponse:
return await self.preprocess(preprocessor_id="", preprocessor_inputs=preprocessor_inputs)
return await self.do_preprocess(preprocessor_id="", preprocessor_inputs=preprocessor_inputs)

View file

@ -47,7 +47,7 @@ class InclineSimpleChunkingImpl(Preprocessing, PreprocessorsProtocolPrivate):
async def unregister_preprocessor(self, preprocessor_id: str) -> None: ...
async def preprocess(
async def do_preprocess(
self,
preprocessor_id: str,
preprocessor_inputs: List[PreprocessingDataElement],
@ -72,12 +72,12 @@ class InclineSimpleChunkingImpl(Preprocessing, PreprocessorsProtocolPrivate):
return PreprocessorResponse(success=True, output_data_type=PreprocessingDataType.chunks, results=chunks)
async def chain_preprocess(
async def preprocess(
self,
preprocessors: PreprocessorChain,
preprocessor_inputs: List[PreprocessingDataElement],
) -> PreprocessorResponse:
return await self.preprocess(preprocessor_id="", preprocessor_inputs=preprocessor_inputs)
return await self.do_preprocess(preprocessor_id="", preprocessor_inputs=preprocessor_inputs)
def _resolve_chunk_size_params(self, options: PreprocessorOptions) -> Tuple[int, int]:
window_len = (options or {}).get(

View file

@ -81,7 +81,7 @@ class MemoryToolRuntimeImpl(ToolsProtocolPrivate, ToolRuntime, RAGToolRuntime):
preprocessor_chain: Optional[PreprocessorChain] = None,
) -> None:
preprocessor_inputs = [self._rag_document_to_preprocessor_input(d) for d in documents]
preprocessor_response = await self.preprocessing_api.chain_preprocess(
preprocessor_response = await self.preprocessing_api.preprocess(
preprocessors=preprocessor_chain or self.DEFAULT_PREPROCESSING_CHAIN,
preprocessor_inputs=preprocessor_inputs,
)