Skip to content

Open source

bloom-parser

A service that turns images, PDFs and spreadsheets into one structured document.

I built it and published it as open source.

Code

Role
Sole author
Context
Personal project, open source
When
2026

What I did

  • Routed 11 input formats through 4 adapters into one document model.
  • Made OCR optional at build time: the default binary needs no native libraries, and a build tag adds Tesseract.

Impact

Public MIT-licensed code; a new format is an adapter plus a detection rule, and the pipeline, OCR and exporters don't change.

bloom-parser pipeline A request is validated and its format detected, then routed to one adapter per format: image (with an optional OCR engine), PDF, spreadsheet or text. Every adapter produces the same Document, which can be exported to CSV or XLSX, or published to Power BI. requestvalidate · detectimagepdfxlsxtextDocumentCSV / XLSXPower BIOCR enginebehind an interface

When things go wrong

  1. When the OCR engine isn't installed,

    OCR requests get a clear per-page error and every other path keeps working.

Built with

Go, gRPC, Python, Tesseract, Vue