langdash.search package

Submodules

langdash.search.embedding_search module

Classes:

EmbeddingSearch(embd_session[, embd_formatter])

A search engine that uses vector embeddings from a language model.

class langdash.search.embedding_search.EmbeddingSearch(embd_session: LLMEmbeddingSession, embd_formatter: str = '{text}')

Bases: Engine

A search engine that uses vector embeddings from a language model.

Parameters:

• embd_session (LLMEmbeddingSession) – An LLMSession object for models that supports embeddings.

• embd_formatter (str) –

  Formatting string which will be used to preprocess raw text before passing into the model. This string must contain the text variable as formatting argument.

  Defaults to “{text}”.

Methods:

add(text_raw)	Add one or more documents to the search engine.
search(text[, max_documents])	Search for documents relating to text in the engine.

add(text_raw: str | List[str])

Add one or more documents to the search engine.

Arg:

texts (Union[str, List[str]]):

a document or a list of documents to add to the search engine.

search(text: str, max_documents: int = 1) → Generator[Tuple[int, str, float], None, None]

Search for documents relating to text in the engine.

Parameters:

• text (str) – The text to search for in the documents.

• max_documents (int) – The maximum number of documents to search through. Default is 1.

Returns:

A generator of tuples containing the index of the document, its content and its relevance score to the search text.

langdash.search.engine module

Classes:

Engine()

Base class representing a search engine that can search for texts in a collection of documents.

class langdash.search.engine.Engine

Bases: object

Base class representing a search engine that can search for texts in a collection of documents.

Methods:

add(texts)	Add one or more documents to the search engine.
search(text[, max_documents])	Search for documents relating to text in the engine.

add(texts: str | List[str])

Add one or more documents to the search engine.

Arg:

texts (Union[str, List[str]]):

a document or a list of documents to add to the search engine.

search(text: str, max_documents: int = 1) → Generator[Tuple[int, str, float], None, None]

Search for documents relating to text in the engine.

Parameters:

• text (str) – The text to search for in the documents.

• max_documents (int) – The maximum number of documents to search through. Default is 1.

Returns:

A generator of tuples containing the index of the document, its content and its relevance score to the search text.

langdash.search.multichoice_search module

Classes:

MultichoiceSearch(prompt_chain, ...)

A search engine that uses multiple choice questions to ask the most relevant documents from a generative language model.

class langdash.search.multichoice_search.MultichoiceSearch(prompt_chain: ~langdash.chains.chains.LDChainCached, document_prompt: ~typing.Callable[[~langdash.search.multichoice_search.MultichoiceSearch, int, str], ~typing.Tuple[str, str]] = <function number_based_prompt>)

Bases: Engine

A search engine that uses multiple choice questions to ask the most relevant documents from a generative language model.

Methods:

add(texts)	Add one or more documents to the search engine.
search(text[, max_documents])	Search for documents relating to text in the engine.

add(texts: str | List[str])

Add one or more documents to the search engine.

Arg:

texts (Union[str, List[str]]):

a document or a list of documents to add to the search engine.

search(text: str, max_documents: int = 1) → Generator[Tuple[int, str, float], None, None]

Search for documents relating to text in the engine.

Parameters:

• text (str) – The text to search for in the documents.

• max_documents (int) – The maximum number of documents to search through. Default is 1.

Returns:

A generator of tuples containing the index of the document, its content and its relevance score to the search text.

Module contents