Skip to content
PDF Oxide

Async PDF Processing

PDF Oxide ships first-class async APIs in Python, C#, and Node.js so extraction never blocks your event loop or HTTP request handler. The same PdfDocument methods are exposed as async wrappers that dispatch to a background thread / thread pool.

Python — AsyncPdfDocument

Every method on PdfDocument has an awaitable counterpart on AsyncPdfDocument. Each call wraps the sync method with asyncio.to_thread().

Python

import asyncio
from pdf_oxide import AsyncPdfDocument

async def extract(path):
    doc = await AsyncPdfDocument.open(path)
    text = await doc.extract_text(0)
    return text

asyncio.run(extract("report.pdf"))

Fan-out across pages

Python

import asyncio
from pdf_oxide import AsyncPdfDocument

async def extract_all(path):
    doc = await AsyncPdfDocument.open(path)
    page_count = await doc.page_count()
    pages = await asyncio.gather(*[doc.extract_text(i) for i in range(page_count)])
    return pages

Creation

AsyncPdf mirrors the Pdf class for creation flows:

Python

from pdf_oxide import AsyncPdf

pdf = await AsyncPdf.from_markdown("# Hello")
await pdf.save_async("out.pdf")

Office conversion

AsyncOfficeConverter handles async DOCX / XLSX / PPTX → PDF.

Python

from pdf_oxide import AsyncOfficeConverter

converter = AsyncOfficeConverter()
pdf_bytes = await converter.docx_to_pdf_bytes("input.docx")

Free-threaded Python (cp314t)

The pdf_oxide extension module declares gil_used = false, making it safe to use under cp314t (Python 3.14 free-threaded builds). Multiple threads can call PdfDocument methods in parallel without GIL serialisation.

Python

from concurrent.futures import ThreadPoolExecutor
from pdf_oxide import PdfDocument

doc = PdfDocument("large.pdf")

with ThreadPoolExecutor(max_workers=8) as pool:
    pages = list(pool.map(doc.extract_text, range(doc.page_count())))

C# — async Task<T> with CancellationToken

Every extraction and save method has an *Async variant returning Task<T> and accepting an optional CancellationToken.

C#

using PdfOxide.Core;

using var doc = PdfDocument.Open("report.pdf");

string text = await doc.ExtractTextAsync(0);

using var cts = new CancellationTokenSource(TimeSpan.FromSeconds(30));
var tasks = Enumerable.Range(0, doc.PageCount)
    .Select(i => doc.ExtractTextAsync(i, cts.Token));
string[] pages = await Task.WhenAll(tasks);

Editor

C#

using var editor = DocumentEditor.Open("form.pdf");
editor.SetFormFieldValue("name", "Jane Doe");
await editor.SaveAsync("filled.pdf");

Node.js — *Async methods

Every sync method has an async sibling — extractText → extractTextAsync, save → saveAsync, etc. Async calls run on the libuv thread pool.

Node.js

const { PdfDocument } = require("pdf-oxide");

async function extractAll(path) {
  const doc = new PdfDocument(path);
  try {
    const pageCount = doc.getPageCount();
    const pages = await Promise.all(
      Array.from({ length: pageCount }, (_, i) => doc.extractTextAsync(i))
    );
    return pages;
  } finally {
    doc.close();
  }
}

HTTP handler example

Node.js

import express from "express";
import { PdfDocument } from "pdf-oxide";

const app = express();

app.post("/extract", express.raw({ type: "application/pdf", limit: "50mb" }), async (req, res) => {
  const doc = PdfDocument.openFromBytes(req.body);
  try {
    const text = await doc.extractTextAsync(0);
    res.json({ text });
  } finally {
    doc.close();
  }
});

Concurrency model

Language Mechanism
Python asyncio.to_thread dispatches sync call to the default executor
Python (cp314t) True thread parallelism — GIL is optional
C# Task-based Asynchronous Pattern, dispatched to ThreadPool
Node.js libuv worker threads (Napi::AsyncWorker)
Go Not needed — goroutines call sync methods directly
Rust Not provided — use tokio::task::spawn_blocking or your executor of choice

See the concurrency guide for sharing a single PdfDocument across threads.