RAGchain.schema package

Submodules

RAGchain.schema.db_origin module

class RAGchain.schema.db_origin.DBOrigin(*, db_type: str, db_path: dict)

Bases: Serializable

Class for storing a db_type and db_path: dict

db_path: dict

db_type: str

to_dict() → dict

RAGchain.schema.evaluate_result module

class RAGchain.schema.evaluate_result.EvaluateResult(results: dict, use_metrics: List[str], each_results: pandas.core.frame.DataFrame)

Bases: object

each_results: DataFrame: DataFrame of each results question by question

results: dict: Dictionary of metrics and their results

use_metrics: List[str]: List of metrics to use

RAGchain.schema.passage module

class RAGchain.schema.passage.Passage(*, id: UUID | str = None, content: str, filepath: str, content_datetime: datetime = None, importance: int = 0, previous_passage_id: UUID | str | None = None, next_passage_id: UUID | str | None = None, metadata_etc: dict = None)

Bases: Serializable

Class for storing a passage and metadatas

class Config

Bases: object

extra = 'forbid'

content: str: String text.

content_datetime: datetime: Datetime when the passage content is created or edited. Everytime passge content changes, this value should be updated.

copy(*args, **kwargs)

Duplicate a model, optionally choose which fields to include, exclude and change.

Parameters:

include – fields to include in new model
exclude – fields to exclude from new model, as with values this takes precedence over include
update – values to change/add in the new model. Note: the data is not validated before creating the new model: you should trust this data
deep – set to True to make a deep copy of the model

Returns:

new model instance

filepath: str: Filepath of the passage.

classmethod from_documents(documents: List[Document]) → List[Passage]: Convert a list of documents to a list of passages. metadata with ‘source’ key is required. It will convert to filepath filed. metadat with ‘content_datetime’ key is optional. It will convert to content_datetime field. It can be datetime.datetime object, or string with ‘%Y-%m-%d %H:%M:%S’ format. metadata with ‘importance’ key is optional. It will convert to importance field. It must be int. :param documents: A list of documents.

classmethod from_search(search_results: List[Dict[str, str]]) → List[Passage]: Convert a list of search results to a list of passages. :param search_results: A list of search results, it requires ‘title’, ‘link’, ‘snippet’ keys.

id: UUID | str: Unique identifier for the passage. You can use string or UUID. By default, create new UUID for new passage.

importance: int: Importance of the passage. The higher the value, the more important the passage is. It can be minus value. The default is 0.

is_exactly_same(other)

static make_prompts(passages: List[Passage]) → str

metadata_etc: dict: Arbitrary metadata about the passage.

next_passage_id: UUID | str | None: Next passage’s id. If this is the last passage, this value should be None.

previous_passage_id: UUID | str | None: Previous passage’s id. If this is the first passage, this value should be None.

reset_id()

to_dict()

to_document() → Document

RAGchain.schema.prompt module

class RAGchain.schema.prompt.RAGchainChatPromptTemplate(*, name: str | None = None, input_variables: List[str], input_types: Dict[str, Any] = None, output_parser: BaseOutputParser | None = None, partial_variables: Mapping[str, str | Callable[[], str]] = None, messages: List[BaseMessagePromptTemplate | BaseMessage | BaseChatPromptTemplate], validate_template: bool = False)

Bases: ChatPromptTemplate

input_variables: List[str]: List of input variables in template messages. Used for validation.

messages: List[MessageLike]: List of messages consisting of either message prompt templates or messages.

validate_template: bool: Whether or not to try validating the template.

class RAGchain.schema.prompt.RAGchainPromptTemplate(*, name: str | None = None, input_variables: List[str], input_types: Dict[str, Any] = None, output_parser: BaseOutputParser | None = None, partial_variables: Mapping[str, str | Callable[[], str]] = None, template: str, template_format: Literal['f-string', 'jinja2'] = 'f-string', validate_template: bool = False)

Bases: PromptTemplate

input_variables: List[str]: A list of the names of the variables the prompt template expects.

template: str: The prompt template.

template_format: Literal['f-string', 'jinja2']: The format of the prompt template. Options are: ‘f-string’, ‘jinja2’.

validate_template: bool: Whether or not to try validating the template.

RAGchain.schema.retrieval_result module

class RAGchain.schema.retrieval_result.RetrievalResult(*, query: str, passages: List[Passage], scores: List[float], metadata: dict = None)

Bases: BaseModel

class for storing retrieval result

metadata: dict: metadata that you can store anything you want

passages: List[Passage]: list of passage retrieved

query: str: query string used for retrieval

scores: List[float]: list of scores for each passage

slice(start: int = 0, end: int | None = None): Slice passages and scores. :param start: int, start index of slice. Default is 0. :param end: int, end index of slice. Default is the length of passages.

to_dict()

to_prompt_input(passage_convert_func: ~typing.Callable[[~typing.List[~RAGchain.schema.passage.Passage]], str] = <function Passage.make_prompts>) → dict