langchain-upstage

langchain_upstage

ChatUpstage

Bases: BaseChatOpenAI

ChatUpstage chat model.

To use, you should have the environment variable UPSTAGE_API_KEY set with your API key or pass it as a named parameter to the constructor.

Example

from langchain_upstage import ChatUpstage

model = ChatUpstage()

METHOD	DESCRIPTION
validate_environment	Validate that api key and python package exists in environment.
get_token_ids	Get the tokens present in the text.
get_num_tokens_from_messages	Calculate num tokens for solar model.

model_name class-attribute instance-attribute

model_name: str = Field(default='solar-mini', alias='model')

Model name to use.

upstage_api_key class-attribute instance-attribute

upstage_api_key: SecretStr = Field(
    default_factory=secret_from_env(
        "UPSTAGE_API_KEY",
        error_message="You must specify an api key. You can pass it an argument as `api_key=...` or set the environment variable `UPSTAGE_API_KEY`.",
    ),
    alias="api_key",
)

Automatically inferred from env are UPSTAGE_API_KEY if not provided.

upstage_api_base class-attribute instance-attribute

upstage_api_base: str | None = Field(
    default_factory=from_env(
        "UPSTAGE_API_BASE", default="https://api.upstage.ai/v1/solar"
    ),
    alias="base_url",
)

Base URL path for API requests.

Leave blank if not using a proxy or service emulator.

openai_api_key class-attribute instance-attribute

openai_api_key: SecretStr | None = Field(default=None)

OpenAI api key is not supported for upstage. use upstage_api_key instead.

openai_api_base class-attribute instance-attribute

openai_api_base: str | None = Field(default=None)

OpenAI api base is not supported for upstage. use upstage_api_base instead.

openai_organization class-attribute instance-attribute

openai_organization: str | None = Field(default=None)

OpenAI organization is not supported for upstage.

tiktoken_model_name class-attribute instance-attribute

tiktoken_model_name: str | None = None

Tiktoken is not supported for upstage.

tokenizer_name class-attribute instance-attribute

tokenizer_name: str | None = 'upstage/solar-pro2-tokenizer'

Huggingface tokenizer name.

Solar tokenizer is opened in huggingface https://huggingface.co/upstage/solar-pro-tokenizer

default_headers class-attribute instance-attribute

default_headers: Mapping[str, str] | None = DEFAULT_HEADERS

Add trace headers.

validate_environment

validate_environment() -> Self

Validate that api key and python package exists in environment.

get_token_ids

get_token_ids(text: str) -> list[int]

Get the tokens present in the text.

get_num_tokens_from_messages

get_num_tokens_from_messages(
    messages: Sequence[BaseMessage], tools: Sequence[Any] | None = None
) -> int

Calculate num tokens for solar model.

UpstageDocumentParseLoader

Bases: BaseLoader

Upstage Document Parse Loader.

To use, you should have the environment variable UPSTAGE_API_KEY set with your API key or pass it as a named parameter to the constructor.

Example

from langchain_upstage import UpstageDocumentParseLoader

file_path = "/PATH/TO/YOUR/FILE.pdf"
loader = UpstageDocumentParseLoader(
    file_path, split="page", output_format="text"
)

METHOD	DESCRIPTION
aload	Load data into Document objects.
load_and_split	Load Document and split into chunks. Chunks are returned as Document.
alazy_load	A lazy loader for Document.
__init__	Initializes an instance of the Upstage document parse loader.
load	Loads and parses the document using the UpstageDocumentParseParser.
lazy_load	Lazily loads and parses the document using the UpstageDocumentParseParser.
merge_and_split	Merges the page content and metadata of multiple documents into a single

aload async

aload() -> list[Document]

Load data into Document objects.

RETURNS	DESCRIPTION
list[Document]	The documents.

load_and_split

load_and_split(text_splitter: TextSplitter | None = None) -> list[Document]

Load Document and split into chunks. Chunks are returned as Document.

Danger

Do not override this method. It should be considered to be deprecated!

PARAMETER	DESCRIPTION
text_splitter	TextSplitter instance to use for splitting documents. Defaults to RecursiveCharacterTextSplitter. TYPE: TextSplitter | None DEFAULT: None

RAISES	DESCRIPTION
ImportError	If langchain-text-splitters is not installed and no text_splitter is provided.

RETURNS	DESCRIPTION
list[Document]	List of Document.

alazy_load async

alazy_load() -> AsyncIterator[Document]

A lazy loader for Document.

YIELDS	DESCRIPTION
AsyncIterator[Document]	The Document objects.

__init__

__init__(
    file_path: str | Path | list[str] | list[Path],
    split: SplitType = "none",
    api_key: str | None = None,
    base_url: str = DOCUMENT_PARSE_BASE_URL,
    model: str = DOCUMENT_PARSE_DEFAULT_MODEL,
    chart_recognition: bool = True,
    ocr: OCR = "auto",
    output_format: OutputFormat = "html",
    coordinates: bool = True,
    base64_encoding: list[Category] | None = None,
)

Initializes an instance of the Upstage document parse loader.

PARAMETER	DESCRIPTION
file_path	The path to the document to be loaded. TYPE: str | Path | list[str] | list[Path]
split	The type of splitting to be applied. TYPE: SplitType DEFAULT: 'none'
api_key	The API key for accessing the Upstage API. Defaults to None, in which case it will be fetched from the environment variable UPSTAGE_API_KEY. TYPE: str DEFAULT: None
base_url	The base URL for accessing the Upstage API. TYPE: str DEFAULT: DOCUMENT_PARSE_BASE_URL
model	The model to be used for the document parse. Defaults to 'document-parse'. TYPE: str DEFAULT: DOCUMENT_PARSE_DEFAULT_MODEL
ocr	Extract text from images in the document using OCR. If the value is 'force', OCR is used to extract text from an image. If the value is 'auto', text is extracted from a PDF. (An error will occur if the value is 'auto' and the input is NOT in PDF format) TYPE: OCRMode DEFAULT: 'auto'
output_format	Format of the inference results. TYPE: OutputFormat DEFAULT: 'html'
coordinates	Whether to include the coordinates of the OCR in the output. TYPE: bool DEFAULT: True
base64_encoding	The category of the elements to be encoded in base64. TYPE: list[Category] DEFAULT: None

load

load() -> list[Document]

Loads and parses the document using the UpstageDocumentParseParser.

RETURNS	DESCRIPTION
list[Document]	Document objects representing the parsed layout analysis.

lazy_load

lazy_load() -> Iterator[Document]

Lazily loads and parses the document using the UpstageDocumentParseParser.

RETURNS	DESCRIPTION
Iterator[Document]	An iterator of Document objects representing the parsed layout analysis.

merge_and_split

merge_and_split(
    documents: list[Document], splitter: object | None = None
) -> list[Document]

Merges the page content and metadata of multiple documents into a single document, or splits the documents using a custom splitter.

PARAMETER	DESCRIPTION
documents	A list of Document objects to be merged and split. TYPE: list
splitter	An optional splitter object that implements the split_documents method. If provided, the documents will be split using this splitter. Defaults to None, in which case the documents are merged. TYPE: object DEFAULT: None

RETURNS	DESCRIPTION
list	A list of Document objects. If no splitter is provided, a single TYPE: list[Document]
list[Document]	Document object is returned with the merged content and combined metadata.
list[Document]	If a splitter is provided, the documents are split and a list of Document
list[Document]	objects is returned.

RAISES	DESCRIPTION
AssertionError	If a splitter is provided but it does not implement the split_documents method.

UpstageDocumentParseParser

Bases: BaseBlobParser

Upstage Document Parse Parser.

To use, you should have the environment variable UPSTAGE_API_KEY set with your API key or pass it as a named parameter to the constructor.

Example

from langchain_upstage import UpstageDocumentParseParser

loader = UpstageDocumentParseParser(split="page", output_format="text")

METHOD	DESCRIPTION
parse	Eagerly parse the blob into a Document or list of Document objects.
__init__	Initializes an instance of the Upstage class.
lazy_parse	Lazily parses a document and yields Document objects based on the specified

parse

parse(blob: Blob) -> list[Document]

Eagerly parse the blob into a Document or list of Document objects.

This is a convenience method for interactive development environment.

Production applications should favor the lazy_parse method instead.

Subclasses should generally not over-ride this parse method.

PARAMETER	DESCRIPTION
blob	Blob instance TYPE: Blob

RETURNS	DESCRIPTION
list[Document]	List of Document objects

__init__

__init__(
    api_key: str | None = None,
    base_url: str = DOCUMENT_PARSE_BASE_URL,
    model: str = DOCUMENT_PARSE_DEFAULT_MODEL,
    split: SplitType = "none",
    chart_recognition: bool = True,
    ocr: OCR = "auto",
    output_format: OutputFormat = "html",
    coordinates: bool = True,
    base64_encoding: list[Category] | None = None,
)

Initializes an instance of the Upstage class.

PARAMETER	DESCRIPTION
api_key	The API key for accessing the Upstage API. Defaults to None, in which case it will be fetched from the environment variable UPSTAGE_API_KEY. TYPE: str DEFAULT: None
base_url	The base URL for accessing the Upstage API. TYPE: str DEFAULT: DOCUMENT_PARSE_BASE_URL
model	The model to be used for the document parse. Defaults to "document-parse". TYPE: str DEFAULT: DOCUMENT_PARSE_DEFAULT_MODEL
split	The type of splitting to be applied. Defaults to "none" (no splitting). TYPE: SplitType DEFAULT: 'none'
ocr	Extract text from images in the document using OCR. If the value is "force", OCR is used to extract text from an image. If the value is "auto", text is extracted from a PDF. (An error will occur if the value is "auto" and the input is NOT in PDF format) TYPE: OCRMode DEFAULT: 'auto'
output_format	Format of the inference results. TYPE: OutputFormat DEFAULT: 'html'
coordinates	Whether to include the coordinates of the OCR in the output. TYPE: bool DEFAULT: True
base64_encoding	The category of the elements to be encoded in base64. TYPE: list[Category] DEFAULT: None

lazy_parse

lazy_parse(blob: Blob, is_batch: bool = False) -> Iterator[Document]

Lazily parses a document and yields Document objects based on the specified split type.

PARAMETER	DESCRIPTION
blob	The input document blob to parse. TYPE: Blob
is_batch	Whether to parse the document in batches. Defaults to False (single page parsing) TYPE: bool DEFAULT: False

YIELDS	DESCRIPTION
Document	The parsed document object. TYPE:: Document

RAISES	DESCRIPTION
ValueError	If an invalid split type is provided.

UpstageEmbeddings

Bases: BaseModel, Embeddings

UpstageEmbeddings embedding model.

To use, set the environment variable UPSTAGE_API_KEY with your API key or pass it as a named parameter to the constructor.

Example

from langchain_upstage import UpstageEmbeddings

model = UpstageEmbeddings(model='solar-embedding-1-large')

METHOD	DESCRIPTION
build_extra	Build extra kwargs from additional params that were passed in.
validate_environment	Validate that api key and python package exists in environment.
embed_documents	Embed a list of document texts using passage model.
embed_query	Embed query text using query model.
aembed_documents	Embed a list of document texts using passage model asynchronously.
aembed_query	Asynchronous Embed query text using query model.

model class-attribute instance-attribute

model: str = Field(...)

Embeddings model name to use. Do not add suffixes like -query and -passage.

Instead, use 'solar-embedding-1-large' for example.

dimensions class-attribute instance-attribute

dimensions: int | None = None

The number of dimensions the resulting output embeddings should have.

Not yet supported.

upstage_api_key class-attribute instance-attribute

upstage_api_key: SecretStr = Field(
    default_factory=secret_from_env(
        "UPSTAGE_API_KEY",
        error_message="You must specify an api key. You can pass it an argument as `api_key=...` or set the environment variable `UPSTAGE_API_KEY`.",
    ),
    alias="api_key",
)

Automatically inferred from env are UPSTAGE_API_KEY if not provided.

upstage_api_base class-attribute instance-attribute

upstage_api_base: str | None = Field(
    default_factory=from_env(
        "UPSTAGE_API_BASE", default="https://api.upstage.ai/v1/solar"
    ),
    alias="base_url",
)

Endpoint URL to use.

embedding_ctx_length class-attribute instance-attribute

embedding_ctx_length: int = 4096

The maximum number of tokens to embed at once.

Not yet supported.

allowed_special class-attribute instance-attribute

allowed_special: Literal['all'] | set[str] = set()

Not yet supported.

disallowed_special class-attribute instance-attribute

disallowed_special: Literal['all'] | set[str] | Sequence[str] = 'all'

Not yet supported.

chunk_size class-attribute instance-attribute

chunk_size: int = 1000

Maximum number of texts to embed in each batch.

Not yet supported.

max_retries class-attribute instance-attribute

max_retries: int = 2

Maximum number of retries to make when generating.

request_timeout class-attribute instance-attribute

request_timeout: float | tuple[float, float] | Any | None = Field(
    default=None, alias="timeout"
)

Timeout for requests to Upstage embedding API. Can be float, httpx.Timeout or None.

show_progress_bar class-attribute instance-attribute

show_progress_bar: bool = False

Whether to show a progress bar when embedding.

Not yet supported.

model_kwargs class-attribute instance-attribute

model_kwargs: dict[str, Any] = Field(default_factory=dict)

Holds any model parameters valid for create call not explicitly specified.

skip_empty class-attribute instance-attribute

skip_empty: bool = False

Whether to skip empty strings when embedding or raise an error. Defaults to not skipping.

Not yet supported.

default_headers class-attribute instance-attribute

default_headers: Mapping[str, str] | None = DEFAULT_HEADERS

add trace header.

http_client class-attribute instance-attribute

http_client: Any | None = None

Optional httpx.Client.

Only used for sync invocations.

Must specify http_async_client as well if you'd like a custom client for async invocations.

http_async_client class-attribute instance-attribute

http_async_client: Any | None = None

Optional httpx.AsyncClient.

Only used for async invocations.

Must specify http_client as well if you'd like a custom client for sync invocations.

build_extra classmethod

build_extra(values: dict[str, Any]) -> Any

Build extra kwargs from additional params that were passed in.

validate_environment

validate_environment() -> Self

Validate that api key and python package exists in environment.

embed_documents

embed_documents(texts: list[str]) -> list[list[float]]

Embed a list of document texts using passage model.

PARAMETER	DESCRIPTION
texts	The list of texts to embed. TYPE: list[str]

RETURNS	DESCRIPTION
list[list[float]]	List of embeddings, one for each text.

embed_query

embed_query(text: str) -> list[float]

Embed query text using query model.

PARAMETER	DESCRIPTION
text	The text to embed. TYPE: str

RETURNS	DESCRIPTION
list[float]	Embedding for the text.

aembed_documents async

aembed_documents(texts: list[str]) -> list[list[float]]

Embed a list of document texts using passage model asynchronously.

PARAMETER	DESCRIPTION
texts	The list of texts to embed. TYPE: list[str]

RETURNS	DESCRIPTION
list[list[float]]	List of embeddings, one for each text.

aembed_query async

aembed_query(text: str) -> list[float]

Asynchronous Embed query text using query model.

PARAMETER	DESCRIPTION
text	The text to embed. TYPE: str

RETURNS	DESCRIPTION
list[float]	Embedding for the text.

UpstagePrebuiltInformationExtraction

UpstagePrebuiltInformationExtraction Information extraction model.

To use, set the environment variable UPSTAGE_API_KEY with your API key or pass it as a named parameter to the constructor.

Example

from langchain_upstage import UpstagePrebuiltInformationExtraction

model = UpstagePrebuiltInformationExtraction(model='receipt-extraction')

UpstageUniversalInformationExtraction

UpstageUniversalInformationExtraction Information extraction model.

To use, set the environment variable UPSTAGE_API_KEY with your API key or pass it as a named parameter to the constructor.

Example

from langchain_upstage import UpstageUniversalInformationExtraction

model = UpstageUniversalInformationExtraction(model='information-extract')