nv-ingest-client 25.8.0rc7

Python client for the nv-ingest service

NV-Ingest-Client

NV-Ingest-Client is a tool designed for efficient ingestion and processing of large datasets. It provides both a Python API and a command-line interface to cater to various ingestion needs.

[!Note] NV-Ingest is also known as NVIDIA Ingest and NeMo Retriever extraction.

Table of Contents

1. Installation
2. Usage
   • CLI Tool
   • API Libraries
3. Command Line Interface (CLI)
   • Command Overview
   • Options
4. Examples

Installation

To install NV-Ingest-Client, run the following command in your terminal:

pip install [REPO_ROOT]/client

This command installs both the API libraries and the nv-ingest-cli tool which can subsequently be called from the command line.

API Libraries

nv_ingest_client.primitives.jobs

JobSpec

Specification for creating a job for submission to the nv-ingest microservice.

• Parameters:

  • payload (Dict): The payload data for the job.
  • tasks (Optional[List], optional): A list of tasks to be added to the job. Defaults to None.
  • source_id (Optional[str], optional): An identifier for the source of the job. Defaults to None.
  • source_name (Optional[str], optional): A name for the source of the job. Defaults to None.
  • document_type (Optional[str], optional): Type of the document. Defaults to 'txt'.
  • job_id (Optional[Union[UUID, str]], optional): A unique identifier for the job. Defaults to a new UUID.
  • extended_options (Optional[Dict], optional): Additional options for job processing. Defaults to None.

• Attributes:

  • _payload (Dict): Storage for the payload data.
  • _tasks (List): Storage for the list of tasks.
  • _source_id (str): Storage for the source identifier.
  • _job_id (UUID): Storage for the job's unique identifier.
  • _extended_options (Dict): Storage for the additional options.

• Methods:

  • to_dict() -> Dict:
    • Description: Converts the job specification to a dictionary for JSON serialization.
    • Returns: Dict: Dictionary representation of the job specification.
  • add_task(task):
    • Description: Adds a task to the job specification.
    • Parameters:
      • task: The task to be added. Assumes the task has a to_dict() method.
    • Raises:
      • ValueError: If the task does not have a to_dict() method or is not an instance of Task.

• Properties:

  • payload: Getter/Setter for the payload data.
  • job_id: Getter/Setter for the job's unique identifier.
  • source_id: Getter/Setter for the source identifier.
  • source_name: Getter/Setter for the source name.

• Example Usage:

  job_spec = JobSpec(
      payload={"data": "Example data"},
      tasks=[extract_task, split_task],
      source_id="12345",
      job_id="abcd-efgh-ijkl-mnop",
      extended_options={"tracing_options": {"trace": True}}
  )
  print(job_spec.to_dict())
  

nv_ingest_client.primitives.tasks

Task Factory

• Function: task_factory(task_type, **kwargs)

  • Description: Factory method for creating task objects based on the provided task type. It dynamically selects the appropriate task class from a mapping and initializes it with any additional keyword arguments.
  • Parameters:
    • task_type (TaskType or str): The type of the task to create. Can be an enum member of TaskType or a string representing a valid task type.
    • **kwargs (dict): Additional keyword arguments to pass to the task's constructor.
  • Returns:
    • Task: An instance of the task corresponding to the given task type.
  • Raises:
    • ValueError: If an invalid task type is provided, or if any unexpected keyword arguments are passed that do not match the task constructor's parameters.

• Example:

  # Assuming TaskType has 'Extract' and 'Split' as valid members and corresponding classes are defined.
  extract_task = task_factory('extract', document_type='PDF', extract_text=True)
  split_task = task_factory('split', split_by='sentence', split_length=100)
  

ExtractTask

Object for document extraction tasks, extending the Task class.

• Method: __init__(document_type, extract_method='pdfium', extract_text=False, extract_images=False, extract_tables=False)

  • Parameters:
    • document_type: Type of document.
    • extract_method: Method used for extraction. Default is 'pdfium'.
    • extract_text: Boolean indicating if text should be extracted. Default is False.
    • extract_images: Boolean indicating if images should be extracted. Default is False.
    • extract_tables: Boolean indicating if tables should be extracted. Default is False.
    • extract_page_as_image: Boolean indicating if each page should be rendered as an image for embedding. Default is False.
  • Description: Sets up configuration for the extraction task.

• Method: to_dict()

  • Description: Converts task details to a dictionary for submission to message client. Includes handling for specific methods and document types.
  • Returns: Dict: Dictionary containing task type and properties.

• Example:

  extract_task = ExtractTask(
    document_type=file_type,
    extract_text=True,
    extract_images=True,
    extract_tables=True
  )
  

SplitTask

Object for document splitting tasks, extending the Task class.

• Method: __init__(split_by=None, split_length=None, split_overlap=None, max_character_length=None, sentence_window_size=None)

  • Parameters:
    • split_by: Criterion for splitting, e.g., 'word', 'sentence', 'passage'.
    • split_length: The length of each split segment.
    • split_overlap: Overlap length between segments.
    • max_character_length: Maximum character length for a split.
    • sentence_window_size: Window size for sentence-based splits.
  • Description: Sets up configuration for the splitting task.

• Method: to_dict()

  • Description: Converts task details to a dictionary for submission to message client.
  • Returns: Dict: Dictionary containing task type and properties.

• Example:

  split_task = SplitTask(
      split_by="word",
      split_length=300,
      split_overlap=10,
      max_character_length=5000,
      sentence_window_size=0,
  )
  

nv_ingest_client.client.client

The NvIngestClient class provides a comprehensive suite of methods to handle job submission and retrieval processes efficiently. Below are the public methods available:

Initialization

• __init__: Initializes the NvIngestClient with customizable client allocator and Redis configuration.

  • Parameters:
    • message_client_allocator: A callable that returns an instance of the client used for communication.
    • message_client_hostname: Hostname of the message client server. Defaults to "localhost".
    • message_client_port: Port number of the message client server. Defaults to 7670.
    • message_client_kwargs: Additional keyword arguments for the message client.
    • msg_counter_id: Redis key for tracking message counts. Defaults to "nv-ingest-message-id".
    • worker_pool_size: Number of worker processes in the pool. Defaults to 1.

• Example:

  client = NvIngestClient(
    message_client_hostname="localhost", # Host where nv-ingest-ms-runtime is running
    message_client_port=7670 # REST port, defaults to 7670
  )
  

Submission Methods

submit_job

Submits a job to a specified job queue. This method can optionally wait for a response if blocking is set to True.

• Parameters:

  • job_id: The unique identifier of the job to be submitted.
  • job_queue_id: The ID of the job queue where the job will be submitted.

• Returns:

  • Optional[Dict]: The job result if blocking is True and a result is available before the timeout, otherwise None.

• Raises:

  • Exception: If submitting the job fails.

• Example:

  client.submit_job(job_id, "ingest_task_queue")
  

submit_jobs

Submits multiple jobs to a specified job queue. This method does not wait for any of the jobs to complete.

• Parameters:

  • job_ids: A list of job IDs to be submitted.
  • job_queue_id: The ID of the job queue where the jobs will be submitted.

• Returns:

  • List[Union[Dict, None]]: A list of job results if blocking is True and results are available before the timeout, otherwise None.

• Example:

  client.submit_jobs([job_id0, job_id1], "ingest_task_queue")
  

submit_job_async

Asynchronously submits one or more jobs to a specified job queue using a thread pool. This method handles both single job ID or a list of job IDs.

• Parameters:

  • job_ids: A single job ID or a list of job IDs to be submitted.
  • job_queue_id: The ID of the job queue where the jobs will be submitted.

• Returns:

  • Dict[Future, str]: A dictionary mapping futures to their respective job IDs for later retrieval of outcomes.

• Notes:

  • This method queues the jobs for asynchronous submission and returns a mapping of futures to job IDs.
  • It does not wait for any of the jobs to complete.
  • Ensure that each job is in the proper state before submission.

• Example:

  client.submit_job_async(job_id, "ingest_task_queue")
  

Job Retrieval

fetch_job_result

• Description: Fetches the job result from a message client, handling potential errors and state changes.

• Method: fetch_job_result(job_id, timeout=10, data_only=True)

• Parameters:

  • job_id (str): The identifier of the job.
  • timeout (float, optional): Timeout for the fetch operation in seconds. Defaults to 10.
  • data_only (bool, optional): If true, only returns the data part of the job result.

• Returns:

  • Tuple[Dict, str]: The job result and the job ID.

• Raises:

  • ValueError: If there is an error in decoding the job result.
  • TimeoutError: If the fetch operation times out.
  • Exception: For all other unexpected issues.

• Example:

  job_id = client.add_job(job_spec)
  client.submit_job(job_id, TASK_QUEUE)
  generated_metadata = client.fetch_job_result(
      job_id, timeout=DEFAULT_JOB_TIMEOUT
  )
  

fetch_job_result_async

• Description: Fetches job results for a list or a single job ID asynchronously and returns a mapping of futures to job IDs.

• Method: fetch_job_result_async(job_ids, timeout=10, data_only=True)

• Parameters:

  • job_ids (Union[str, List[str]]): A single job ID or a list of job IDs.
  • timeout (float, optional): Timeout for fetching each job result, in seconds. Defaults to 10.
  • data_only (bool, optional): Whether to return only the data part of the job result.

• Returns:

  • Dict[Future, str]: A dictionary mapping each future to its corresponding job ID.

• Raises:

  • No explicit exceptions raised but leverages the exceptions from fetch_job_result.

• Example:

  job_id = client.add_job(job_spec)
  client.submit_job(job_id, TASK_QUEUE)
  generated_metadata = client.fetch_job_result_async(
      job_id, timeout=DEFAULT_JOB_TIMEOUT
  )
  

Job and Task Management

job_count

• Description: Returns the number of jobs currently tracked by the client.

• Method: job_count()

• Returns: Integer representing the total number of jobs.

• Example:

  client.job_count()
  

add_job

• Description: Adds a job specification to the job tracking system.

• Method: add_job(job_spec)

• Parameters:

  • job_spec (JobSpec, optional): The job specification to add. If not provided, a new job ID will be generated.

• Returns: String representing the job ID of the added job.

• Raises:

  • ValueError: If a job with the specified job ID already exists.

• Example:

  extract_task = ExtractTask(
    document_type=file_type,
    extract_text=True,
    extract_images=True,
    extract_tables=True,
    text_depth="document",
    extract_tables_method="yolox",
  )
  job_spec.add_task(extract_task)
  job_id = client.add_job(job_spec)
  

create_job

• Description: Creates a new job with specified parameters and adds it to the job tracking dictionary.
• Method: create_job(payload, source_id, source_name, document_type, tasks, job_id, extended_options)
• Parameters:
  • payload (str): The payload associated with the job.
  • source_id (str): The source identifier for the job.
  • source_name (str): The unique name of the job's source data.
  • document_type (str, optional): The type of document to be processed.
  • tasks (list, optional): A list of tasks to be associated with the job.
  • job_id (uuid.UUID | str, optional): The unique identifier for the job.
  • extended_options (dict, optional): Additional options for job creation.
• Returns: String representing the job ID.
• Raises:
  • ValueError: If a job with the specified job ID already exists.

add_task

• Description: Adds a task to an existing job.

• Method: add_task(job_id, task)

• Parameters:

  • job_id (str): The job ID to which the task will be added.
  • task (Task): The task to add.

• Raises:

  • ValueError: If the job does not exist or is not in the correct state.

• Example:

  job_spec = JobSpec(
      document_type=file_type,
      payload=file_content,
      source_id=SAMPLE_PDF,
      source_name=SAMPLE_PDF,
      extended_options={
          "tracing_options": {
              "trace": True,
              "ts_send": time.time_ns(),
          }
      },
  )
  extract_task = ExtractTask(
      document_type=file_type,
      extract_text=True,
      extract_images=True,
      extract_tables=True,
      text_depth="document",
      extract_tables_method="yolox",
  )
  job_spec.add_task(extract_task)
  

create_task

• Description: Creates a task with specified parameters and adds it to an existing job.

• Method: create_task(job_id, task_type, task_params)

• Parameters:

  • job_id (uuid.UUID | str): The unique identifier of the job.
  • task_type (TaskType): The type of the task.
  • task_params (dict, optional): Parameters for the task.

• Raises:

  • ValueError: If the job does not exist or if an attempt is made to modify a job after its submission.

• Example:

  job_id = client.add_job(job_spec)
  client.create_task(job_id, DedupTask, {content_type: "image", filter: True})
  

CLI Tool

After installation, you can use the nv-ingest-cli tool from the command line to manage and process datasets.

CLI Options

Here are the options provided by the CLI, explained:

• --batch_size: Specifies the number of documents to process in a single batch. Default is 10. Must be 1 or more.
• --doc: Adds a new document to be processed. Supports multiple entries. Files must exist.
• --dataset: Specifies the path to a dataset definition file.
• --client: Sets the client type with choices including REST, Redis, Kafka. Default is Redis.
• --client_host: Specifies the DNS name or URL for the endpoint.
• --client_port: Sets the port number for the client endpoint.
• --client_kwargs: Provides additional arguments to pass to the client. Default is {}.
• --concurrency_n: Defines the number of inflight jobs to maintain at one time. Default is 1.
• --dry_run: Enables a dry run without executing actions.
• --output_directory: Specifies the output directory for results.
• --log_level: Sets the log level. Choices are DEBUG, INFO, WARNING, ERROR, CRITICAL. Default is INFO.
• --shuffle_dataset: Shuffles the dataset before processing if enabled. Default is true.
• --task: Allows for specification of tasks in JSON format. Supports multiple tasks.
• --collect_profiling_traces: Collect the tracing profile for the run after processing.
• --zipkin_host: Host used to connect to Zipkin to gather tracing profiles.
• --zipkin_port: Port used to connect to Zipkin to gether tracing profiles.

Examples

You can find a notebook with examples that use the CLI client in the client examples folder.