Pipeline

This section contains auto-generated documentation for pipeline components.

Pipeline Module

class routir.pipeline.pipeline.SearchPipeline(pipeline, collection, runtime_kwargs=None, verify=True)[source]

Bases: object

Executes search pipelines defined using a custom DSL.

Supports sequential and parallel execution of search, rerank, query expansion, and fusion operations.

pipeline: Parsed pipeline component tree

collection: Document collection to search

runtime_kwargs: Runtime parameters passed to pipeline components

doc_content_cache: Cache for retrieved document content

__init__(pipeline, collection, runtime_kwargs=None, verify=True)[source]

Initialize search pipeline.

Parameters:

pipeline (SystemCall | CallSequence | ParallelCallSequences) – Parsed pipeline component
collection (str) – Collection name
runtime_kwargs (Dict[str, Dict[str, Any]]) – Runtime parameters for pipeline components
verify (bool) – Whether to verify all services exist

verify()[source]: Verify that all required services exist in the registry.

classmethod from_string(pipeline_string, collection, runtime_kwargs=None, verify=True)[source]

Create pipeline from string specification.

Parameters:

pipeline_string (str) – Pipeline DSL string
collection (str) – Collection name
runtime_kwargs (Dict[str, Dict[str, Any]]) – Runtime parameters
verify (bool) – Whether to verify services

Returns:

SearchPipeline instance

Return type:

SearchPipeline

async get_doc_content(doc_id)[source]

Retrieve and cache document content.

Parameters:: doc_id (str) – Document identifier
Returns:: Document text content

async run(query, last_output=None, current_node=None, scratch=None)[source]

Recursively execute the pipeline for a query.

Parameters:

query (str) – Search query
last_output (Dict[str, Any]) – Output from previous stage
current_node (SystemCall | CallSequence | ParallelCallSequences) – Current pipeline component to execute
scratch (Dict[str, Dict[str, Any]]) – Scratch space for intermediate results

Returns:

Final search results

Return type:

List[Dict[str, float]]

Parser

class routir.pipeline.parser.SystemCall(name: str, alias: str | None = None, limit: int | None = None, role: str | None = 'search')[source]