In this sample we will use the Cloudmersive Document AI API to intelligently split an input document into its component sub-documents and output them to the same folder as the input document, as separate PDFs.

Prior to running the script, install the SDK:

pip install cloudmersive-documentai-api-client requests

Then configure the variables INPUT_PDF_PATH, CLOUDMERSIVE_API_BASE_PATH, and CLOUDMERSIVE_API_KEY appropriately.

"""
Intelligently split a combined PDF into constituent sub-documents using
Cloudmersive Document AI (cloudmersive-documentai-api-client).

Outputs:
  <inputfilename>_part1_output.pdf
  <inputfilename>_part2_output.pdf
  ...

Install:
  pip install cloudmersive-documentai-api-client requests
"""

import base64
import os
from pathlib import Path
from urllib.parse import urlparse

import requests

import cloudmersive_documentai_api_client
from cloudmersive_documentai_api_client.rest import ApiException


# =========================
# User-configurable variables
# =========================
INPUT_PDF_PATH = r"C:\path\to\input.pdf"
CLOUDMERSIVE_API_BASE_PATH = "https://api.cloudmersive.com"  # e.g. "https://api.cloudmersive.com" or "api.cloudmersive.com"
CLOUDMERSIVE_API_KEY = "YOUR_API_KEY_HERE"
# =========================



def _normalize_host(base_path: str) -> str:
    s = (base_path or "").strip()
    if not s:
        return "https://api.cloudmersive.com"

    # ensure it parses as a URL; strip any path/query and keep only scheme + netloc
    if "://" not in s:
        s = "https://" + s

    u = urlparse(s)
    return f"{u.scheme}://{u.netloc}"


def main():
    input_path = Path(INPUT_PDF_PATH)
    if not input_path.exists():
        raise FileNotFoundError(f"Input file not found: {input_path}")

    out_dir = input_path.parent
    base_name = input_path.stem  # inputfilename (no extension)

    # Configure SDK
    configuration = cloudmersive_documentai_api_client.Configuration()
    configuration.api_key["Apikey"] = CLOUDMERSIVE_API_KEY
    configuration.host = _normalize_host(CLOUDMERSIVE_API_BASE_PATH)

    api_client = cloudmersive_documentai_api_client.ApiClient(configuration)
    extract_api = cloudmersive_documentai_api_client.ExtractApi(api_client)

    # Call Intelligent Split (SDK expects a FILE PATH, not an open file object)
    try:
        resp = extract_api.extract_split(input_file=str(input_path))
    except ApiException as e:
        raise RuntimeError(f"Cloudmersive API error: {e}") from e

    # The response model typically has a list under resp.sub_documents
    sub_docs = getattr(resp, "sub_documents", None) or getattr(resp, "subDocuments", None)
    if not sub_docs:
        # fallback: sometimes the list is directly called "documents"
        sub_docs = getattr(resp, "documents", None)

    if not sub_docs:
        raise RuntimeError("No sub-documents were returned by extract_split().")

    written = 0
    for idx, sub in enumerate(sub_docs, start=1):
        out_path = out_dir / f"{base_name}_part{idx}_output.pdf"

        # In your case the model exposes: start_page, end_page, document_description, file_bytes
        file_b64 = getattr(sub, "file_bytes", None) or getattr(sub, "_file_bytes", None)

        if not file_b64:
            raise RuntimeError(
                f"Sub-document #{idx} did not include file_bytes. "
                f"Available fields: {dir(sub)}"
            )

        # file_bytes is base64 for the PDF
        try:
            pdf_bytes = base64.b64decode(file_b64, validate=False)
        except Exception as ex:
            raise RuntimeError(f"Failed to base64-decode file_bytes for sub-document #{idx}") from ex

        out_path.parent.mkdir(parents=True, exist_ok=True)
        with open(out_path, "wb") as f:
            f.write(pdf_bytes)

        start_page = getattr(sub, "start_page", None)
        end_page = getattr(sub, "end_page", None)
        desc = getattr(sub, "document_description", None)

        print(f"Wrote: {out_path}  (pages {start_page}-{end_page})" + (f"  [{desc}]" if desc else ""))
        written += 1

    print(f"\nDone. Wrote {written} file(s) to: {out_dir}")


if __name__ == "__main__":
    main()