Prima cifră / The first digit: files to repeat the calculation
https://mariuscomper.uk/prima-cifra/  (RO)   https://mariuscomper.uk/prima-cifra/en/  (EN)

1. fetch.py        downloads the eight CSV exports from data.gov.ro (CC BY 4.0, Autoritatea pentru Digitalizarea Romaniei) and writes manifest.json with each file's SHA-256.
2. aggregate.py    streams them row by row and writes derived/aggregate.json (digit counts, a log-value histogram, last digits). Rules: PREREGISTRATION.md.
3. controls.py     first-digit counts for the control sets in raw/ (World Bank indicators, the 2021 census residents per administrative unit, Fibonacci terms).
4. analyze.py      writes derived/site-data.json, the numbers the page draws and quotes (MAD, exact expected MAD for each n, chi-square p, the funnel, the detectability bound).
5. crosscheck.py   re-derives the first-digit counts of two quarters with a different method (regex over the raw lines); the counts agree to within the rows the CSV parser rejected as malformed.

Needs Python 3 only (standard library). Raw CSVs are about 900 MB and are not stored here; fetch.py downloads them to $PRIMA_RAW.