​
Datalab SDK Documentation

Our SDK is designed to help you quickly get started converting your documents. It supports:
  • Marker (document conversion) and OCR endpoints
  • Sync and async mode
  • Usage from Python or the CLI
  • Single file or directory conversion
  • Automatic saving of output
This guide will go into the SDK in more detail.

​
Installation

pip install datalab-python-sdk

​
Authentication

Set your API key as an environment variable: 
export DATALAB_API_KEY="your_api_key_here"
Or pass it directly to the client: 
from datalab_sdk import DatalabClient
client = DatalabClient(api_key="your_api_key_here")

​
Python Usage

​
Convert

The convert functionality uses the marker endpoint to convert documents to markdown, HTML, or JSON format.

​
Basic Usage

from datalab_sdk import DatalabClient
from datalab_sdk.models import ConvertOptions

# Synchronous client
client = DatalabClient()

# Basic conversion to markdown
result = client.convert("document.pdf")
print(result.markdown)

# With options
options = ConvertOptions(
    output_format="html",
    max_pages=10,
    force_ocr=True,
    use_llm=True
)
result = client.convert("document.pdf", options=options)
print(result.html)

​
Async Usage

import asyncio
from datalab_sdk import AsyncDatalabClient
from datalab_sdk.models import ConvertOptions

async def convert_document():
    async with AsyncDatalabClient() as client:
        options = ConvertOptions(
            output_format="json",
            paginate=True,
            max_pages=5
        )
        result = await client.convert("document.pdf", options=options)
        return result.json

# Run async function
result = asyncio.run(convert_document())

​
Convert Options

The ConvertOptions class supports all marker endpoint parameters. See more information on the parameters here.

​
Conversion Result

The ConversionResult object contains the converted content and metadata. See more information on the return fields here.

​
Output Saving

You can save conversion results: 
# Save during conversion
result = client.convert("document.pdf", save_output="output_folder/document")

# Or save afterward
result.save_output("output_folder/document", save_images=True)

​
OCR

The OCR functionality extracts text with detailed positional information from documents.

​
Basic Usage

from datalab_sdk import DatalabClient
from datalab_sdk.models import OCROptions

# Synchronous client
client = DatalabClient()

# Basic OCR
result = client.ocr("document.pdf")
text = result.get_text()
print(text)

# With options
options = OCROptions(max_pages=5)
result = client.ocr("document.pdf", options=options)

​
Async Usage

import asyncio
from datalab_sdk import AsyncDatalabClient
from datalab_sdk.models import OCROptions

async def ocr_document():
    async with AsyncDatalabClient() as client:
        options = OCROptions(max_pages=3)
        result = await client.ocr("document.pdf", options=options)
        return result.get_text()

# Run async function
text = asyncio.run(ocr_document())

​
OCR Options

The OCROptions class supports OCR-specific parameters. See more information on the parameters here.

​
OCR Result

The OCRResult object contains detailed text and positional information. See more information on the return fields here.

​
Output Saving

# Save during OCR
result = client.ocr("document.pdf", save_output="output_folder/ocr_result")

# Or save afterward
result.save_output("output_folder/ocr_result")
# Creates: ocr_result.txt (plain text) and ocr_result.ocr.json (detailed data)

​
CLI Usage

​
Authentication

Pass the --api_key option or set the DATALAB_API_KEY environment variable.

​
Convert

​
Examples

# Convert to JSON format with LLM enhancement
datalab-sdk convert document.pdf --format json --use_llm

# Process only PDFs and Word docs in a directory
datalab-sdk convert /docs/ --extensions pdf,docx --max_concurrent 10

# Force OCR with custom page range
datalab-sdk convert document.pdf --force_ocr --page_range "0-5"

# Save to specific output directory
datalab-sdk convert document.pdf --output_dir /output/ --paginate

​
OCR

​
Examples

# OCR specific pages
datalab-sdk ocr document.pdf --page_range "0-2,5,7"

# Process images with custom extensions
datalab-sdk ocr /images/ --extensions png,jpg --max_concurrent 5

# OCR with custom output directory
datalab-sdk ocr document.pdf --output_dir /ocr_results/