TextSplitter#

class langchain_text_splitters.base.TextSplitter(chunk_size: int = 4000, chunk_overlap: int = 200, length_function: ~typing.Callable[[str], int] = <built-in function len>, keep_separator: bool | ~typing.Literal['start', 'end'] = False, add_start_index: bool = False, strip_whitespace: bool = True)[source]#

Interface for splitting text into chunks.

Create a new TextSplitter.

Parameters:

  • chunk_size (int) – Maximum size of chunks to return

  • chunk_overlap (int) – Overlap in characters between chunks

  • length_function (Callable[[str], int]) – Function that measures the length of given chunks

  • keep_separator (Union[bool, Literal['start', 'end']]) – Whether to keep the separator and where to place it in each corresponding chunk (True=’start’)

  • add_start_index (bool) – If True, includes chunk’s start index in metadata

  • strip_whitespace (bool) – If True, strips whitespace from the start and end of every document

Methods

__init__([chunk_size, chunk_overlap, ...])

Create a new TextSplitter.

atransform_documents(documents, **kwargs)

Asynchronously transform a list of documents.

create_documents(texts[, metadatas])

Create documents from a list of texts.

from_huggingface_tokenizer(tokenizer, **kwargs)

Text splitter that uses HuggingFace tokenizer to count length.

from_tiktoken_encoder([encoding_name, ...])

Text splitter that uses tiktoken encoder to count length.

split_documents(documents)

Split documents.

split_text(text)

Split text into multiple components.

transform_documents(documents, **kwargs)

Transform sequence of documents by splitting them.
__init__(chunk_size: int = 4000, chunk_overlap: int = 200, length_function: ~typing.Callable[[str], int] = <built-in function len>, keep_separator: bool | ~typing.Literal['start', 'end'] = False, add_start_index: bool = False, strip_whitespace: bool = True) → None[source]#

Create a new TextSplitter.

Parameters:

  • chunk_size (int) – Maximum size of chunks to return

  • chunk_overlap (int) – Overlap in characters between chunks

  • length_function (Callable[[str], int]) – Function that measures the length of given chunks

  • keep_separator (bool | Literal['start', 'end']) – Whether to keep the separator and where to place it in each corresponding chunk (True=’start’)

  • add_start_index (bool) – If True, includes chunk’s start index in metadata

  • strip_whitespace (bool) – If True, strips whitespace from the start and end of every document

Return type:

None

async atransform_documents(
documents: Sequence[Document],
**kwargs: Any,
) → Sequence[Document]#

Asynchronously transform a list of documents.

Parameters:

  • documents (Sequence[Document]) – A sequence of Documents to be transformed.

  • kwargs (Any)

Returns:

A sequence of transformed Documents.

Return type:

Sequence[Document]

create_documents(
texts: list[str],
metadatas: list[dict[Any, Any]] | None = None,
) → List[Document][source]#

Create documents from a list of texts.

Parameters:

  • texts (list[str])

  • metadatas (list[dict[Any, Any]] | None)

Return type:

List[Document]

classmethod from_huggingface_tokenizer(
tokenizer: Any,
**kwargs: Any,
) → TextSplitter[source]#

Text splitter that uses HuggingFace tokenizer to count length.

Parameters:

  • tokenizer (Any)

  • kwargs (Any)

Return type:

TextSplitter

classmethod from_tiktoken_encoder(
encoding_name: str = 'gpt2',
model_name: str | None = None,
allowed_special: Literal['all'] | AbstractSet[str] = {},
disallowed_special: Literal['all'] | Collection[str] = 'all',
**kwargs: Any,
) → TS[source]#

Text splitter that uses tiktoken encoder to count length.

Parameters:

  • encoding_name (str)

  • model_name (str | None)

  • allowed_special (Literal['all'] | ~typing.AbstractSet[str])

  • disallowed_special (Literal['all'] | ~typing.Collection[str])

  • kwargs (Any)

Return type:

TS

split_documents(
documents: Iterable[Document],
) → List[Document][source]#

Split documents.

Parameters:

documents (Iterable[Document])

Return type:

List[Document]

abstractmethod split_text(text: str) → List[str][source]#

Split text into multiple components.

Parameters:

text (str)

Return type:

List[str]

transform_documents(
documents: Sequence[Document],
**kwargs: Any,
) → Sequence[Document][source]#

Transform sequence of documents by splitting them.

Parameters:

  • documents (Sequence[Document])

  • kwargs (Any)

Return type:

Sequence[Document]