Weekly automated web scraper for Euribor 3M rate (EUR365) → CSV output for Microsoft Fabric

We need a small, reliable scraper that runs once a week (or on-demand) and extracts a single interest rate value from a public Italian financial data page, writing the result to a CSV file ready to be ingested into Microsoft Fabric. Source website (Italian): https://mutuionline.24oreborsaonline.ilsole24ore.com/guide-mutui/euribor.asp?refresh_ce=1 On this page there is a section titled "STORICO DEGLI ULTIMI DIECI ANNI" (historical data, last 10 years) with three dropdown menus: Indice (Index) → select "Euribor 3 Mesi" Anno (Year) → select the current year Mese (Month) → select the month prior to the current month Selecting these dropdowns loads a table. In that table: Column 1 is "Data" (Date) — identify the last row, i.e. the last calendar day of the selected month. Column 3 is "EUR365" — this is the value we need, the Euribor 3M rate on a EUR/365 day-count basis. Worked example: for the "July" run, the scraper should select Anno=2026, Mese=Giugno (June), find the row for 30/06/2026, and extract the EUR365 value, which was 2.36%. What we need built: A script (Python preferred; Playwright/Selenium likely required since dropdown selections probably trigger a JS/AJAX call — please inspect network requests to see if there's a direct API/endpoint you can call instead of full browser automation, as that would be more robust and cheaper to run). Logic to always compute "current year" and "month prior to current month" dynamically (no hardcoding), so the script works unattended every week/month. Error handling: if the site structure changes, dropdown values aren't found, or the expected last-day row is missing, the script should fail loudly (exit code / log / notification) rather than silently returning wrong data. Scheduling: either a built-in scheduler (weekly cron-style trigger) or clear instructions/config so we can run it in our own orchestrator (Azure Function, Fabric pipeline, GitHub Actions, etc. — your recommendation welcome). Output: a CSV file with one row appended per run, in this exact format: csv reference_date,index_name,value_eur365,source_url,extraction_timestamp 2026-06-30,Euribor 3M,2.36,https://mutuionline.24oreborsaonline.ilsole24ore.com/guide-mutui/euribor.asp,2026-07-27T09:00:00Z reference_date: ISO 8601 date (last day of the reference month) — used as the natural key for dedup/idempotency. index_name: fixed string "Euribor 3M". value_eur365: numeric, dot as decimal separator (not comma — avoid the Italian "2,36%" formatting), no % sign. source_url / extraction_timestamp: for auditability/lineage. Re-running for the same month should not create duplicate rows (idempotent append, or overwrite existing row for that reference_date). CSV should be written to a location we can plug into Microsoft Fabric (e.g., Azure Blob Storage / OneLake) — happy to discuss your suggested approach. Documentation: brief README on how to run locally, how scheduling works, and how to update selectors if the site's HTML changes. Nice to have: Basic retry/backoff logic for transient network failures. Deliverables: Source code (Git repo or zip) Short documentation One demo run showing successful extraction and the resulting CSV output