Self-hosted personal finance for India. Upload your bank and credit-card statements, get every transaction categorised automatically, and budget against categories that nest three levels deep.
Built for Indian statement formats — the ones generic finance apps don't read.
Status: early development. Statement parsing works and two banks are supported; categorisation and budgeting are next. Star the repo to follow along, or add support for your bank — it's three files and no changes to shared code.
Firefly III and Actual Budget are excellent self-hosted finance apps. Neither can read an HDFC savings statement or an ICICI credit-card PDF.
If you bank in India, you currently export to CSV by hand, wrestle with column layouts that differ per bank, and re-categorise the same merchants every month. This project does that part for you:
- Statement parsing built for Indian banks — bank and credit-card formats, including password-protected PDFs.
- Categorisation that learns — deterministic rules over merchant names and UPI VPAs. Recategorise once and it remembers. Works fully offline.
- Three-level categories —
Food & Dining → Eating Out → Weekend, budgeted and reported at any level. - Your data stays yours — runs on your own machine or server. No account, no cloud, no third party unless you explicitly opt in.
Your bank isn't supported yet? That's the whole point of the architecture — adding one is three files and no changes to shared code. See docs/adding-a-bank-parser.md.
git clone https://github.com/varunjuneja/expense-analyser.git
cd expense-analyser
cp .env.example .env # set DJANGO_SECRET_KEY before exposing to a network
docker compose -f deploy/docker-compose.yml upOpen http://localhost:8000 and register. The first account is yours; everything lives in Postgres inside the stack.
No Docker needed. You need Python 3.11+, Node 20+, Poetry and a local Postgres.
git clone https://github.com/varunjuneja/expense-analyser.git
cd expense-analyser
make setup # installs deps, creates the databases, runs migrations
make dev # backend on :8000, frontend on :5173Open http://localhost:5173. Both servers hot-reload. Full guide: CONTRIBUTING.md.
| Bank | Savings/current | Credit card | Formats |
|---|---|---|---|
| HDFC Bank | ✅ | — | xls (delimited export), csv, txt |
| ICICI Bank | ✅ | ✅ | pdf (incl. password-protected) |
Adding yours is the most useful contribution you can make, and the codebase is arranged specifically to make it cheap. Start with docs/adding-a-bank-parser.md, or open a bank support request if you'd rather someone else write it.
Statement (PDF/XLS/CSV)
│
▼
Parser registry ──► bank-specific parser ──► canonical transactions
│ │
│ ▼
│ dedupe (re-uploading an
│ overlapping period is a no-op)
│ │
▼ ▼
Rules engine ──► category (3-level tree) ──► budgets and charts
The rules engine runs in this order, first match wins: your own rules → rules learned from your past recategorisations → bundled community merchant patterns. Everything is deterministic and offline.
An optional AI categoriser can label whatever the rules miss. It is off by default, needs your own API key, and sends transaction descriptions to a third party — see docs/faq.md before enabling it.
More detail: docs/architecture.md.
Django 5 + Django REST Framework · PostgreSQL · React + TypeScript + Vite
Contributions are very welcome — especially bank parsers, but also bug reports, docs and UI work.
- CONTRIBUTING.md — setup and workflow
- docs/adding-a-bank-parser.md — the most wanted contribution
- Good first issues
AGPL-3.0. Use it, modify it, self-host it freely. If you run a modified version as a network service, you must publish your changes.