Implementing truly usable PDF text editing in the browser turned out to be much harder than we expected.
Over the past months, we’ve been working on a project called ElasticPDF. The original motivation was simple: most existing PDF editors either rely heavily on backend services or make text editing feel like manually adjusting coordinates rather than actually editing text.
We wanted something that:
- runs entirely in the browser
- never uploads PDF files
- still feels close to editing text in Word
We chose PDF.js as the foundation, but quickly realized that rendering PDFs is the easy part — editing existing text objects is where things get messy.
Some of the hardest problems we ran into:
- Editing existing PDF text instead of overlaying fake text layers
- Keeping text layout stable without manually tweaking coordinates
- Handling multilingual input, especially Chinese, Japanese, and Korean
- Editing rotated text (90°, 180°, or arbitrary angles)
- Making the experience visual instead of coordinate-based
To address these, we ended up building a pure frontend PDF text editing SDK that:
- Lets you add, delete, and modify existing PDF text
- Supports multilingual text (including CJK)
-
Automatically reflows text to avoid overflow and misalignment
-
Supports rotated text editing
- Works with no backend at all
Over time, we also added annotation tools (highlight, underline, pen, signatures) and basic page management, since most real-world PDFs need more than just text edits.
The result is a cross-platform, framework-agnostic SDK that can be embedded into web apps with a single frontend codebase, while keeping all PDF data local and private.
This project is still evolving, but it’s been a deep dive into how PDFs actually work under the hood — and how far frontend-only solutions can go.
If anyone’s curious, we put together a demo and more details here: www.elasticpdf.com
Happy to answer questions or discuss PDF internals with anyone interested.
