PDF OCR: How to Extract Text from Scanned PDFs with an API

Scanned PDFs are everywhere — contracts, invoices, academic papers, government forms — yet the text inside them is locked in pixel data. You cannot search it, copy it, or pipe it into a database. PDF OCR solves this by running Optical Character Recognition directly on PDF pages, giving you structured, selectable text without converting to images first. In this tutorial you will learn how to extract text from scanned PDFs using a single API endpoint, with working code in cURL, Python, and JavaScript.

Why PDF OCR Is Different from Image OCR

You might be tempted to export each PDF page as a JPEG and send it to a regular image OCR endpoint. That works, but it adds unnecessary steps and loses information:

• Multiple pages in a single request — A dedicated PDF OCR endpoint accepts the entire file and returns text for every page at once. No need to split, convert, and re-assemble.
• Page-range control — Process only the pages you need (e.g. pages 3–7 of a 200-page document) to save time and API credits.
• Preserved page structure — The response is organized by page number, so you always know which text belongs to which page.
• No image conversion overhead — Skip the PDF-to-image pipeline entirely. Upload the PDF and get text back.

The /ocr-pdf Endpoint

The OCR Wizard API exposes a dedicated /ocr-pdf endpoint for scanned PDF files. Here is what it accepts and returns:

Request Parameters

• pdf_file (required) — The PDF file to process, sent as multipart form data.
• first_page (optional) — First page to process (1-indexed). Defaults to 1.
• last_page (optional) — Last page to process. Defaults to the last page of the document.

Response Schema

The response JSON contains a body.pages array. Each element represents one page:

{
  "body": {
    "pages": [
      {
        "pageNumber": 1,
        "fullText": "Extracted text from page 1...",
        "detectedLanguage": "en"
      },
      {
        "pageNumber": 2,
        "fullText": "Extracted text from page 2...",
        "detectedLanguage": "en"
      }
    ]
  }
}

Quick Start: Extract Text from a PDF

cURL

Upload a PDF and extract text from pages 1 through 3:

curl -X POST "https://ocr-wizard.p.rapidapi.com/ocr-pdf" \
  -H "x-rapidapi-key: YOUR_API_KEY" \
  -H "x-rapidapi-host: ocr-wizard.p.rapidapi.com" \
  -F "pdf_file=@document.pdf" \
  -F "first_page=1" \
  -F "last_page=3"

Python

A minimal Python script that uploads a scanned PDF and prints the text from each page:

import requests

HEADERS = {
    "x-rapidapi-host": "ocr-wizard.p.rapidapi.com",
    "x-rapidapi-key": "YOUR_API_KEY",
}

with open("scanned.pdf", "rb") as f:
    resp = requests.post(
        "https://ocr-wizard.p.rapidapi.com/ocr-pdf",
        headers=HEADERS,
        files={"pdf_file": f},
        data={"first_page": 1, "last_page": 5},
    )

for i, page in enumerate(resp.json()["body"]["pages"]):
    print(f"--- Page {i + 1} ({page['detectedLanguage']}) ---")
    print(page["fullText"])

JavaScript (fetch)

Upload a PDF from a browser file input or a Node.js buffer:

const formData = new FormData();
formData.append("pdf_file", fileInput.files[0]);
formData.append("first_page", "1");
formData.append("last_page", "5");

const response = await fetch(
  "https://ocr-wizard.p.rapidapi.com/ocr-pdf",
  {
    method: "POST",
    headers: {
      "x-rapidapi-host": "ocr-wizard.p.rapidapi.com",
      "x-rapidapi-key": "YOUR_API_KEY",
    },
    body: formData,
  }
);

const data = await response.json();
data.body.pages.forEach((page) => {
  console.log(`--- Page ${page.pageNumber} (${page.detectedLanguage}) ---`);
  console.log(page.fullText);
});

Process a Multi-Page PDF

For large documents, you may want to process all pages and save the extracted text to a file. Here is a complete Python script that does exactly that:

import json
import requests

HEADERS = {
    "x-rapidapi-host": "ocr-wizard.p.rapidapi.com",
    "x-rapidapi-key": "YOUR_API_KEY",
}
API_URL = "https://ocr-wizard.p.rapidapi.com/ocr-pdf"

def extract_pdf_text(pdf_path, first_page=None, last_page=None):
    """Extract text from a scanned PDF and return a list of page dicts."""
    with open(pdf_path, "rb") as f:
        payload = {}
        if first_page is not None:
            payload["first_page"] = first_page
        if last_page is not None:
            payload["last_page"] = last_page

        resp = requests.post(
            API_URL,
            headers=HEADERS,
            files={"pdf_file": f},
            data=payload,
        )
        resp.raise_for_status()

    return resp.json()["body"]["pages"]


# Process the full document
pages = extract_pdf_text("annual-report.pdf")

# Print a summary
for page in pages:
    preview = page["fullText"][:80].replace("\n", " ")
    print(f"Page {page['pageNumber']} [{page['detectedLanguage']}]: {preview}...")

# Save full output to a JSON file
with open("extracted_text.json", "w") as out:
    json.dump(pages, out, indent=2, ensure_ascii=False)

print(f"\nDone — {len(pages)} pages saved to extracted_text.json")

Combine PDF OCR with Image OCR

In many real-world workflows your input is a mix of scanned PDFs and standalone images (JPEG photos of receipts, PNG screenshots, etc.). You can use both endpoints from the same API to handle everything:

import os
import requests

HEADERS = {
    "x-rapidapi-host": "ocr-wizard.p.rapidapi.com",
    "x-rapidapi-key": "YOUR_API_KEY",
}

def ocr_file(path):
    """Route to the right endpoint based on file extension."""
    ext = os.path.splitext(path)[1].lower()

    if ext == ".pdf":
        with open(path, "rb") as f:
            resp = requests.post(
                "https://ocr-wizard.p.rapidapi.com/ocr-pdf",
                headers=HEADERS,
                files={"pdf_file": f},
            )
        pages = resp.json()["body"]["pages"]
        return "\n\n".join(p["fullText"] for p in pages)
    else:
        with open(path, "rb") as f:
            resp = requests.post(
                "https://ocr-wizard.p.rapidapi.com/ocr",
                headers=HEADERS,
                files={"image": f},
            )
        return resp.json()["body"]["fullText"]


# Process a mixed batch
for filename in ["invoice.pdf", "receipt.jpg", "contract.pdf", "note.png"]:
    text = ocr_file(filename)
    print(f"=== {filename} ===")
    print(text[:200], "\n")

Real-World Use Cases

PDF OCR unlocks automation for any workflow that still depends on paper or scanned documents:

• Invoice and receipt processing — Extract vendor names, dates, line items, and totals from scanned invoices. Feed the data into your accounting software or ERP system automatically.
• Contract digitization — Convert signed contracts and legal agreements into searchable text. Enable full-text search across thousands of documents without reading each one manually.
• Archive scanning — Libraries, government agencies, and hospitals sit on decades of paper records. PDF OCR turns those scans into a searchable, indexed digital archive.
• Academic paper extraction — Researchers often deal with older papers that are only available as scanned PDFs. Extract the text to feed into citation tools, summarization pipelines, or literature review databases.
• Compliance and audit — Financial and regulatory documents frequently arrive as scanned PDFs. Automate data extraction to speed up audits and reduce manual entry errors.

Tips for Best Results

1. Use page ranges for large PDFs. If you only need data from specific pages, set first_page and last_page to avoid processing hundreds of irrelevant pages. This saves time and API credits.
2. Ensure good scan quality. OCR accuracy depends on the quality of the input. 300 DPI scans produce significantly better results than 72 DPI thumbnails. If you control the scanning pipeline, aim for 300 DPI or higher.
3. Check the detected language. Each page in the response includes a detectedLanguage field. Use it to route multilingual documents to the right downstream processing (translation, NLP, etc.).
4. Handle errors gracefully. If a page is blank or too degraded to read, the API will still return a page entry but with empty or minimal text. Check for empty fullText values and flag those pages for manual review.
5. Batch with concurrency. If you have many PDFs to process, send requests in parallel (5–10 at a time) instead of sequentially. This dramatically reduces total processing time while staying within rate limits.

Pricing

The OCR Wizard API is available on RapidAPI with several tiers:

• Free — 100 requests/month. Great for testing and prototyping.
• Pro — Higher limits for production workloads.
• Ultra / Mega — Enterprise-grade volume with priority support.

Each /ocr-pdf request counts as one API call regardless of the number of pages processed. Check the API pricing page for current rates.

Next Steps

PDF OCR turns static scanned documents into structured, searchable data with a single API call. Whether you are automating invoice processing, digitizing archives, or building a document search engine, the /ocr-pdf endpoint handles the heavy lifting.

Ready to go further? Check out the image OCR guide for extracting text from photos and screenshots, or read the OCR API comparison to see how managed APIs stack up against open-source engines. Visit the OCR Wizard API page to get your API key and start extracting text from your PDFs today.