Create an interactive reading experience for PDF's in your react application! PaperCraft is used and maintained by the Semantic Scholar team to create Semantic Reader. It is built on top of React-PDF, with additional features for visual augmentations, interactivities, and performance improvements.



The PaperCraft library (a.k.a., pdf-Component-Libraray) provides the means of displaying PDF’s in your React application with several built in reading features. It is built on top of the React-PDF library, with some added components to help with creating an interactive reading experience. This library is built with research papers in mind, and aims at providing researchers with helpful tools to help their reading experience. It allowed us to create the Semantic Reader which you can demo here


  • Citation Popovers
  • PDF Outlines
  • Thumbnails
  • Zoom Buttons
  • Page number control
  • Text highlighting
  • integration for note taking
  • Scroll-to logic
  • Browser performance optimization

How does it work?

We render each pdf page with several layers in order to display and render interactable components. First an image is generated of each PDF page, which is used for a background image on the bottom most layer. Then we place a transparent “text layer” in front of the background image. The text layer allows for text selection, copy/pasting, and highlighting. Above the text layer is the overlay layer. The overlay is where most of the interactive features are located, such as page highlights and citation popovers.

When to use PDF-Component-Library

You should use this library if you want a React based application with an emphasis on PDF reading. I recommend first glossing over the React-PDF library, then if you decide you like that library and want several added features to help with creating an interactive PDF experience, then try our library.

Getting Started

Install by executing npm install @allenai/pdf-components. Refer to directory ui/demo for examples of how to import the components.

Running the Demo Locally via Docker

In this demo, you will see several added features to the PDF reading experience. This includes Citation Cards, Thumbnails, Outlines, Zoom, and Notetaking.

  1. Download and install Docker
  2. Download or clone the Pdf-Components-Library repo
  3. In your terminal, navigate to ui/demo then run docker-compose up --build. This process launches 2 services, the ui and a proxy responsible for forwarding traffic to the appropriate services. You'll see output from each.
  4. Once docker is done running, you can access the demo on http://localhost:8080/

Modifying the library

Made changes to the library and want to apply them to your own project? The below steps will help you build and apply your changes

Building/Modifying the library

Want to make changes the library? Most code changes will go into ui/library directory. After making your changes, navigate to the ui/library directory and run yarn build. This will copy artifacts over to ui/library/dist folder. Run npm pack in this folder to create a package, for example pdf-component-library/ui/library/dist/allenai-pdf-components-0.2.3.tgz. You can then use this package in your projects by listing it in your package.json like so:

"dependencies": {
    "@allenai/pdf-components": "file:~/pdf-component-library/ui/library/dist/allenai-pdf-components-0.2.3.tgz"

Make sure to re-install your package after this update via yarn install or npm install

If this is the first time building the library, you might have to follow the steps listed in this Docker file to install the dependencies.

Components / Contexts

DocumentWrapperLoads the pdf file, and does several necessary initialization code such as setting scroll observers and setting the render type.
PageWrapperRenders a page, should be placed in the <DocumentWrapper />
ContextProviderContains several necessary contexts which I will go into below. These contexts help us control and get data from the pdf which allows us to create several of the features.
DocumentContextHas helpful info about the pdf including the number of pages, the pdf outline (table of contents), and the page dimensions.
TransformContextAllows setting the scale of pages (zoom), and the rotation
UIContextHelpful ui info such as if the pdf is currently loading and if thumbnails/outlines/highlights are currently showing.
ScrollContextBy far our most complex and worked on context. It provides scroll logic and scroll data, including if at the top of the pdf, how many pages are currently visible, and functionality to scroll to certain parts of the pdf. This is all possible via Intersection Observers, which tell us if a certain spot is currently visible in the user’s viewport. We place several of these on every page so we know exactly how visible reader pages are.
PageRenderContextContains logic for rendering the images on every page. Once the page images are done rendering, you can grab the image’s blobURL via getObjectURLForPage and render anywhere like this <img src={getObjectURLForPage()} />. This is how we made thumbnails.


This project is licensed under the Apache License 2.0.