Skip to main content
For the complete documentation index for agents and LLMs, see llms.txt.

Module files

Sync client for files workflow.

upload​

def upload(paths: List[Path],
           api_key: Optional[str] = None,
           api_url: Optional[str] = None,
           workspace_name: str = DEFAULT_WORKSPACE_NAME,
           write_mode: WriteMode = WriteMode.KEEP,
           blocking: bool = True,
           timeout_s: Optional[int] = None,
           show_progress: bool = True,
           recursive: bool = False,
           desired_file_types: Optional[List[str]] = None,
           enable_parallel_processing: bool = False,
           safe_mode: bool = False) -> S3UploadSummary

Upload a folder to deepset AI Platform.

Arguments:

  • paths: Path to the folder to upload. If the folder contains unsupported file types, they're skipped. deepset AI Platform supports csv, docx, html, json, md, txt, pdf, pptx, xlsx, xml.
  • api_key: deepset API key to use for authentication.
  • api_url: API URL to use for authentication.
  • workspace_name: Name of the workspace to upload the files to. It uses the workspace from the .ENV file by default.
  • write_mode: Specifies what to do when a file with the same name already exists in the workspace. Possible options are: KEEP - uploads the file with the same name and keeps both files in the workspace. OVERWRITE - overwrites the file that is in the workspace. FAIL - fails to upload the file with the same name.
  • blocking: Whether to wait for the files to be uploaded and displayed in deepset AI Platform.
  • timeout_s: Timeout in seconds for the blocking parameter.
  • show_progress: Shows the upload progress.
  • recursive: Uploads files from subfolders as well.
  • desired_file_types: A list of allowed file types to upload. If not provided, all files are uploaded.
  • enable_parallel_processing: If True, deepset AI Platform ingests files in parallel. Use this to speed up the upload process. Make sure you are not running concurrent uploads for the same files.
  • safe_mode: If True, disables ingesting files in parallel.

download​

def download(workspace_name: str = DEFAULT_WORKSPACE_NAME,
             file_dir: Optional[Union[Path, str]] = None,
             name: Optional[str] = None,
             odata_filter: Optional[str] = None,
             include_meta: bool = True,
             batch_size: int = 50,
             api_key: Optional[str] = None,
             api_url: Optional[str] = None,
             show_progress: bool = True,
             timeout_s: Optional[int] = None,
             safe_mode: bool = False) -> None

Download a folder to deepset AI Platform.

Downloads all files from a workspace to a local folder.

Arguments:

  • workspace_name: Name of the workspace to upload the files to. It uses the workspace from the .ENV file by default.
  • file_dir: Path to the folder to download.
  • name: Name of the file to odata_filter by.
  • odata_filter: odata_filter by file meta data.
  • include_meta: Whether to include the file meta in the folder.
  • batch_size: Batch size for the listing.
  • api_key: API key to use for authentication.
  • api_url: API URL to use for authentication.
  • show_progress: Shows the upload progress.
  • timeout_s: Timeout in seconds for the API requests.
  • safe_mode: If True, disables ingesting files in parallel.

upload_texts​

def upload_texts(files: List[DeepsetCloudFile],
                 api_key: Optional[str] = None,
                 api_url: Optional[str] = None,
                 workspace_name: str = DEFAULT_WORKSPACE_NAME,
                 write_mode: WriteMode = WriteMode.KEEP,
                 blocking: bool = True,
                 timeout_s: Optional[int] = None,
                 show_progress: bool = True,
                 enable_parallel_processing: bool = False) -> S3UploadSummary

Upload texts to deepset AI Platform.

Arguments:

  • files: List of DeepsetCloudFiles to upload.
  • api_key: deepset API key to use for authentication.
  • api_url: API URL to use for authentication.
  • workspace_name: Name of the workspace to upload the files to. It uses the workspace from the .ENV file by default.
  • write_mode: Specifies what to do when a file with the same name already exists in the workspace. Possible options are: KEEP - uploads the file with the same name and keeps both files in the workspace. OVERWRITE - overwrites the file that is in the workspace. FAIL - fails to upload the file with the same name.
  • blocking: Whether to wait for the files to be uploaded and listed in deepset AI Platform.
  • timeout_s: Timeout in seconds for the blocking parameter.
  • show_progress: Shows the upload progress.
  • enable_parallel_processing: If True, deepset AI Platform ingests files in parallel. Use this to speed up the upload process. Make sure you are not running concurrent uploads for the same files.

Example:

from deepset_cloud_sdk.workflows.sync_client.files import upload_texts, DeepsetCloudFile

upload_texts(
    api_key="<deepsetCloud_API_key>",
    workspace_name="<default_workspace>", # optional, by default the environment variable "DEFAULT_WORKSPACE_NAME" is used
    files=[
        DeepsetCloudFile(
            name="example.txt",
            text="this is text",
            meta={"key": "value"},  # optional
        )
    ],
    blocking=True,  # optional, by default True
    timeout_s=300,  # optional, by default 300
)

upload_bytes​

def upload_bytes(files: List[DeepsetCloudFileBytes],
                 api_key: Optional[str] = None,
                 api_url: Optional[str] = None,
                 workspace_name: str = DEFAULT_WORKSPACE_NAME,
                 write_mode: WriteMode = WriteMode.KEEP,
                 blocking: bool = True,
                 timeout_s: Optional[int] = None,
                 show_progress: bool = True,
                 enable_parallel_processing: bool = False) -> S3UploadSummary

Upload any supported file types to deepset AI Platform. These include .csv, .docx, .html, .json, .md, .txt, .pdf, .pptx, .xlsx and .xml.

Arguments:

  • files: List of DeepsetCloudFilesBytes to upload.
  • api_key: deepset API key to use for authentication.
  • api_url: API URL to use for authentication.
  • workspace_name: Name of the workspace to upload the files to. It uses the workspace from the .ENV file by default.
  • write_mode: Specifies what to do when a file with the same name already exists in the workspace. Possible options are: KEEP - uploads the file with the same name and keeps both files in the workspace. OVERWRITE - overwrites the file that is in the workspace. FAIL - fails to upload the file with the same name.
  • blocking: Whether to wait for the files to be uploaded and listed in deepset AI Platform.
  • timeout_s: Timeout in seconds for the blocking parameter.
  • show_progress: Shows the upload progress.
  • enable_parallel_processing: If True, deepset AI Platform ingests files in parallel. Use this to speed up the upload process. Make sure you are not running concurrent uploads for the same files.

get_upload_session​

def get_upload_session(
        session_id: UUID,
        api_key: Optional[str] = None,
        api_url: Optional[str] = None,
        workspace_name: str = DEFAULT_WORKSPACE_NAME) -> UploadSessionStatus

Get the status of an upload session.

Arguments:

  • session_id: ID of the upload session to get the status for.
  • api_key: deepset API key to use for authentication.
  • api_url: API URL to use for authentication.
  • workspace_name: Name of the workspace to upload the files to.

list_files​

def list_files(
        api_key: Optional[str] = None,
        api_url: Optional[str] = None,
        workspace_name: str = DEFAULT_WORKSPACE_NAME,
        name: Optional[str] = None,
        odata_filter: Optional[str] = None,
        batch_size: int = 100,
        timeout_s: Optional[int] = None) -> Generator[List[File], None, None]

List files in a deepset workspace.

Arguments:

  • api_key: deepset API key to use for authentication.
  • api_url: API URL to use for authentication.
  • workspace_name: Name of the workspace to list the files from. It uses the workspace from the .ENV file by default.
  • name: Name of the file to odata_filter for.
  • odata_filter: odata_filter to apply to the file list. For example, odata_filter="category eq 'news'" lists files with metadata:
{"meta": {"category": "news"}}
  • batch_size: Batch size to use for the file list.
  • timeout_s: Timeout in seconds for the API requests.

list_upload_sessions​

def list_upload_sessions(
    api_key: Optional[str] = None,
    api_url: Optional[str] = None,
    workspace_name: str = DEFAULT_WORKSPACE_NAME,
    is_expired: Optional[bool] = False,
    batch_size: int = 100,
    timeout_s: Optional[int] = None
) -> Generator[List[UploadSessionDetail], None, None]

List the details of all upload sessions, including the closed ones.

Arguments:

  • api_key: deepset API key to use for authentication.
  • api_url: API URL to use for authentication.
  • workspace_name: Name of the workspace whose sessions you want to list. It uses the workspace from the .ENV file by default.
  • is_expired: Lists expired sessions.
  • batch_size: Batch size to use for the session list.
  • timeout_s: Timeout in seconds for the API request.

Module pipeline_client

Sync pipeline client for importing pipelines and indexes to deepset AI Platform.

PipelineClient​

class PipelineClient()

Sync client for importing Haystack pipelines and indexes to deepset AI platform.

Example for importing a Haystack pipeline or index to deepset AI platform:

from deepset_cloud_sdk import (
    PipelineClient,
    PipelineConfig,
    PipelineInputs,
    PipelineOutputs,
    IndexConfig,
    IndexInputs,
)
from haystack import Pipeline

# Initialize the client with configuration from environment variables (after running `deepset-cloud login`)
client = PipelineClient()

# or initialize the client with explicit configuration
client = PipelineClient(
    api_key="your-api-key",
    workspace_name="your-workspace",
    api_url="https://api.cloud.deepset.ai/api/v1"
)

# Configure your pipeline
pipeline = Pipeline()

# Configure import
# if importing a pipeline, use PipelineConfig
config = PipelineConfig(
    name="my-pipeline",
    inputs=PipelineInputs(
        query=["prompt_builder.query"],
        filters=["bm25_retriever.filters", "embedding_retriever.filters"],
    ),
    outputs=PipelineOutputs(
        answers="answers_builder.answers",
        documents="ranker.documents",
    ),
    strict_validation=False,  # Fail on validation errors (default: False, warnings only)
    overwrite=False,  # Overwrite existing pipelines with the same name. If True, creates if it doesn't exist (default: False)
)

# if importing an index, use IndexConfig
config = IndexConfig(
    name="my-index",
    inputs=IndexInputs(files=["file_type_router.sources"]),
    strict_validation=False,  # Fail on validation errors (default: False, warnings only)
    overwrite=False,  # Overwrite existing indexes with the same name. If True, creates if it doesn't exist (default: False)
)

# sync execution
client.import_into_deepset(pipeline, config)

PipelineClient.__init__​

def __init__(api_key: str | None = None,
             workspace_name: str | None = None,
             api_url: str | None = None) -> None

Initialize the Pipeline Client.

The client can be configured in two ways:

  1. Using environment variables (recommended):

    • Run deepset-cloud login to set up the following environment variables:
      • API_KEY: Your deepset AI platform API key
      • API_URL: The URL of the deepset AI platform API
      • DEFAULT_WORKSPACE_NAME: The workspace name to use.

  2. Using explicit parameters:

    • Provide the values directly to this constructor
    • Any missing parameters will fall back to environment variables

Arguments:

  • api_key: Your deepset AI platform API key. Falls back to API_KEY environment variable.
  • workspace_name: The workspace to use. Falls back to DEFAULT_WORKSPACE_NAME environment variable.
  • api_url: The URL of the deepset AI platform API. Falls back to API_URL environment variable.

Raises:

  • ValueError: If no api key or workspace name is provided and API_KEY or DEFAULT_WORKSPACE_NAME is not set in the environment.

PipelineClient.import_into_deepset​

def import_into_deepset(pipeline: PipelineProtocol,
                        config: IndexConfig | PipelineConfig) -> None

Import a Haystack Pipeline or AsyncPipeline into deepset AI Platform synchronously.

The pipeline must be imported as either an index or a pipeline:

  • An index: Processes files and stores them in a document store, making them available for pipelines to search.
  • A pipeline: For other use cases, for example, searching through documents stored by index pipelines.

Arguments:

  • pipeline: The Haystack Pipeline or AsyncPipeline to import.
  • config: Configuration for importing into deepset, use either IndexConfig or PipelineConfig. If importing an index, the config argument is expected to be of type IndexConfig, if importing a pipeline, the config argument is expected to be of type PipelineConfig.