SummarySearcher API

The SummarySearcher class provides a simple way to query an index built by DocstringIndexer. It takes a search query, embeds it using the same embedding function used for indexing, and retrieves the most semantically similar summaries from the vector database.

Constructor

Class: SummarySearcher (defined in kit/docstring_indexer.py)

The SummarySearcher is typically initialized with an instance of DocstringIndexer. It uses the DocstringIndexer’s configured backend and embedding function to perform searches.

from kit.docstring_indexer import DocstringIndexer, SummarySearcher

# Assuming 'indexer' is an already initialized DocstringIndexer instance
# indexer = DocstringIndexer(repo=my_repo, summarizer=my_summarizer)
# indexer.build() # Ensure the index is built

searcher = SummarySearcher(indexer=indexer)

Parameters:

• indexer (DocstringIndexer, required): An instance of DocstringIndexer that has been configured and preferably has had its build() method called. The SummarySearcher will use this indexer’s backend and embed_fn. See the DocstringIndexer API docs for more details on the indexer.

Methods

search

Method: SummarySearcher.search (defined in kit/docstring_indexer.py)

Embeds the given query string and searches the vector database (via the indexer’s backend) for the top_k most similar document summaries.

query_text = "How is user authentication handled?"
results = searcher.search(query=query_text, top_k=3)

for result in results:
    print(f"Found in: {result.get('file_path')} ({result.get('symbol_name')})")
    print(f"Score: {result.get('score')}")
    print(f"Summary: {result.get('summary')}")
    print("----")}

Parameters:

• query (str, required): The natural language query string to search for.
• top_k (int, default: 5): The maximum number of search results to return.

Returns: List[Dict[str, Any]]

A list of dictionaries, where each dictionary represents a search hit. Each hit typically includes metadata, a score, an ID, and the summary text.