Static marketing and documentation website for DocuExtract, the PDF Invoice to Excel Extractor.
- Website: https://manayab.github.io/DocuExtract.tech/
- Live application: https://manayab.github.io/DocuExtract/
- Application source: https://github.com/MANayab/DocuExtract
- Issues: https://github.com/MANayab/DocuExtract/issues
- Developer GitHub: https://github.com/MANayab
/— product overview/developer/— developer profile and project context/how-it-works/— processing pipeline, file support and limitations/faq/— frequently asked questions/roadmap/— released capabilities and planned work/privacy/— website and application privacy information/404.html— not-found page
The website is static HTML and CSS. It does not require a frontend build step or runtime JavaScript. Pages use relative internal links so they can work under the configured GitHub Pages project path and a correctly configured custom domain.
The site uses a shared stylesheet at assets/styles.css. The favicon and social preview are referenced at assets/logo.svg and assets/social-preview.png. Supply those original assets in the repository; they are not generated by the website.
The product's invoice-processing implementation is a separate application. Keep product claims aligned with its source and disclose that OCR is optional and build-dependent. The deployed app's OCR assets must not be assumed present without a live check.
The software source declares version 1.3.1. DocuExtract processes batches of up to 50 PDFs in the browser, with a documented per-file limit of 25 MB and 100 pages.
- XLSX: combined workbook with invoice rows and supporting sheets.
- CSV: invoice-level or line-item export.
- JSON: includes validation status.
- Multiple invoices in one PDF: split attempt when invoice numbers change; boundary edge cases are not fully tested.
- GST-aware fields include GSTIN, HSN, taxable value, tax rate, tax amount, CGST, SGST and IGST.
- Locale parsing supports en-IN, en-US, en-GB and de-DE; ambiguous dates are flagged and reprocessing protects manual edits.
- OCR is optional and build-dependent. When bundled, it is English-only, limited to 20 pages, capped at 75% confidence and always marked for review. Deployed OCR availability is NOT VERIFIED.
- Extraction is heuristic. Review all values before accounting, tax, payment or legal use.
- The software LICENSE and TERMS.md state MIT licensing without warranties; this website does not promise future commercial licensing.
See the processing guide and FAQ for details.
Use Node.js 22 or a compatible later release.
node --version
node scripts/validate.mjs