Skip to content

Bank statement filename — simplify to _BankStatement__.pdf

Why

The bank-statement filename <SUB>_BankStatement_<Bank>_<CCY>_<download YYYYMMDD>_<statement MMMYYYY>.pdf carries two redundancies and one source of churn:

  • Currency in the filename is redundant — every account here issues statements in a single currency, and the Firestore doc id (bank, accountId, periodEnd) already disambiguates.
  • Download date + month suffix double-encode the same period — the statement's bank-issued 截數日期 (= periodEnd) is already the authoritative date.
  • Download-date churn means re-downloads produce a new filename in Drive even when the same statement period is being refetched.

Drop the currency and the redundant download-date / MMMYYYY tail. The single statement issue date (YYYYMMDD) is enough.

Migration

All 34 existing ERL bank statements (7 Airwallex + 19 OCBC + 8 Dah Sing) have been renamed in Drive + Firestore to match the new format. Issue dates were read from each PDF's text layer (one Dah Sing scan was read visually). One-shot data fix; no migration script lives in the repo.

Bank Before After
Airwallex ERL_BankStatement_Airwallex_HKD_20260529_Apr2025.pdf ERL_BankStatement_Airwallex_20250430.pdf
OCBC ERL_BankStatement_OCBC_HKD_20241127_Nov2024.pdf ERL_BankStatement_OCBC_20241127.pdf
Dah Sing ERL_BankStatement_DahSing_20241031_Oct2024.PDF ERL_BankStatement_DahSing_20241031.pdf

Time-sensitivity (why this needs to ship now): the Drive files were already manually renamed. Until the code change ships, the next nightly Airwallex / OCBC scrape will start producing duplicates in the OLD format alongside the renamed files.

Decision log

2026-06-23 — opened + landed (cherry-picked from PR #744)

  • ✅ Attestation (Records (Infrastructure)): read AGENTS.md; scope-scanned the board — no covering task for bank-statement filename convention. T-082 (preview/prod split) is adjacent infra; T-088 (NAS re-login fix) touches different statement-side surface. This task is the filename-format refactor.
  • Source: Records (Infrastructure) · https://claude.ai/code/session_018RDB37kCqfouHdygVXTAtD
  • Owner directive (verbatim, 2026-06-23):

    "If the task included in the PRs are not tasks created and listed on the Task Board, do that, if they are, but you have not signed to those tasks, sign them according to the instruction on AGENTS.MD, while proceed to work on the still open PRs"

  • What landed (origin/nightly):
  • a4a1b6bc — GitHub merge commit of PR #744 (Merge #744: refactor(bankStatements)…). Single-commit refactor of buildStatementFileName in lib/bankStatements/store.ts. Race condition note: I'd already cherry-picked the same change locally as c5850068, but PR #744 was merged via GitHub's UI in parallel; my push then rebased onto the new nightly tip and git auto-skipped the duplicate cherry-pick (hint: skipped previously applied commit). Net result is identical — the change lives on nightly via a4a1b6bc.
  • PR #744: merged via GitHub UI (final outcome confirmed via webhook).
  • Branch policy: committed straight to nightly per the 🟢 lockstep operational signal; main to be FF'd in the same push.
  • Blast radius: single helper function in the bank-statements store. Affects every new statement filename produced by the Airwallex + OCBC scrapers and the manual-upload path. The pre-existing 34 Drive files are already renamed to the new format (one-shot data fix); after this change the next sync will produce names matching them.
  • Verification plan (post-deploy):
  • Open /records → Bank Statements tab → every existing statement still opens (Firestore doc ids are unchanged; only filenames moved).
  • Next nightly Airwallex / OCBC scrape produces filenames in the new format with no duplicates created.