Adapters API Reference¶

Searcher Interface¶

The searcher interface provides a unified API for different search backends.

BaseSearcher¶

Abstract base class for all searcher implementations.

Methods: - search(query: str, k: int = 10) -> List[SearchHit] - Search for documents - batch_search(queries: List[str], k: int = 10) -> Dict[str, List[SearchHit]] - Batch search

SearchHit¶

Represents a single search result.

Attributes: - docid: str - Document ID - score: float - Relevance score - text: Optional[str] - Document text (if available)

Pyserini Adapter¶

PyseriniSearcher¶

Adapter for Pyserini's LuceneSearcher and LuceneImpactSearcher.

from querygym.adapters import PyseriniSearcher

searcher = PyseriniSearcher(index_path="path/to/index")
results = searcher.search("query text", k=10)

PyTerrier Adapter¶

PyTerrierSearcher¶

Adapter for PyTerrier retrievers.

from querygym.adapters import PyTerrierSearcher
import pyterrier as pt

pt.init()
retriever = pt.BatchRetrieve(index, wmodel="BM25")
searcher = PyTerrierSearcher(retriever)
results = searcher.search("query text", k=10)

Wrapper Functions¶

wrap_pyserini_searcher¶

Wrap a Pyserini searcher instance.

from querygym import wrap_pyserini_searcher
from pyserini.search.lucene import LuceneSearcher

lucene_searcher = LuceneSearcher("path/to/index")
searcher = wrap_pyserini_searcher(lucene_searcher)

wrap_pyterrier_retriever¶

Wrap a PyTerrier retriever instance.

from querygym import wrap_pyterrier_retriever
import pyterrier as pt

pt.init()
retriever = pt.BatchRetrieve(index, wmodel="BM25")
searcher = wrap_pyterrier_retriever(retriever)

wrap_custom_searcher¶

Wrap a custom search function.

from querygym import wrap_custom_searcher

def my_search(query: str, k: int) -> List[dict]:
    # Your custom search logic
    return [{"docid": "doc1", "score": 0.9, "text": "..."}]

searcher = wrap_custom_searcher(my_search)