PBZ PDF to CSV parser

This parser converts PBZ statements to CSV.

Requirements

• Python 3.9+ (stdlib only)
• pdftotext from Poppler

Install Poppler:

• macOS: brew install poppler
• Ubuntu/Debian: sudo apt-get install poppler-utils

Python setup

python3 -m venv .venv
source .venv/bin/activate

Usage

python parser.py data/pdfs/*.pdf \
  --out out/pbz_statements.csv \
  --validation-out out/pbz_validation.csv

Notes:

• Output includes raw_text_block for audit and tx_fingerprint for idempotent imports.
• Dedupe by tx_fingerprint is on by default. Use --no-dedupe to keep all rows.
• raw_text_block contains embedded newlines and will be CSV-quoted.

JavaScript parser (Bun)

Requires Bun 1.0+ and pdftotext (Poppler).

bun parser.js data/pdfs/*.pdf \
  --out out/pbz_statements_js.csv \
  --validation-out out/pbz_validation_js.csv

Run tests (compares JS output to the sample CSVs in out/):

bun test

Static app (Bun build)

Install dependencies:

bun install

Build the static app:

bun run build

The output is written to dist/. Host that folder on any static provider.

Notes:

• The browser app uses PDF.js to rebuild a fixed-width layout before parsing.
• If columns drift, tune the line grouping and x-position tolerance in src/app.js.
• UI strings live in src/strings.json (Croatian by default).