fix: Tolerate diacritic/case/whitespace mismatches in Person column matching
Some checks failed
Deploy to K8s / deploy (push) Successful in 11s
Build and Push / build (push) Successful in 6s
Build and Push / build-go (push) Failing after 6s

- Add canonical_member_key() in match_payments.py to normalize names via
  NFKD + lowercase + whitespace-collapse before ledger lookup; resolves
  payments attributed to e.g. "Maria Maco" to canonical "Mária Maco".
  Emits logger.info when a non-canonical cell is rescued so sheet typos
  are visible in logs without losing the payment allocation.
- Extend group_payments_by_person() in app.py to accept member_names and
  re-key raw-payment groups under the canonical attendance-sheet name so
  the modal's Raw Payments debug section also finds the row correctly.
- Add raw payments collapsible section to member detail modal in adults.html
  and juniors.html for debugging payment attribution issues.
- Remove 4 obsolete tests targeting routes /fees, /fees-juniors, /reconcile,
  /reconcile-juniors that no longer exist; add test_match_payments.py
  covering canonical key equivalence and reconcile() tolerance end-to-end.

Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
This commit is contained in:
2026-05-05 17:22:54 +02:00
parent 81b36878b3
commit 394da2e6b8
8 changed files with 498 additions and 120 deletions

View File

@@ -17,6 +17,15 @@ from czech_utils import normalize, parse_month_references
from sync_fio_to_sheets import get_sheets_service, DEFAULT_SPREADSHEET_ID
def canonical_member_key(name: str) -> str:
"""Diacritic-, case-, and whitespace-insensitive key for member-name matching.
Used to resolve `Person`-column values from the payments sheet to canonical
attendance-sheet names, tolerating cells like "Maria Maco" vs "Mária Maco".
"""
return re.sub(r"\s+", " ", normalize(name)).strip()
# ---------------------------------------------------------------------------
# Name matching
# ---------------------------------------------------------------------------
@@ -309,6 +318,12 @@ def reconcile(
member_tiers = {name: tier for name, tier, _ in members}
member_fees = {name: fees for name, _, fees in members}
# Map canonical key → first attendance-sheet name with that key, so a
# `Person` cell that drifts in diacritics/case/whitespace still resolves.
canonical_by_key: dict[str, str] = {}
for name in member_names:
canonical_by_key.setdefault(canonical_member_key(name), name)
# Initialize ledger
ledger: dict[str, dict[str, dict]] = {}
other_ledger: dict[str, list] = {}
@@ -386,8 +401,9 @@ def reconcile(
if is_other:
num_allocations = len(matched_members)
per_allocation = amount / num_allocations if num_allocations > 0 else 0
for member_name, confidence in matched_members:
if member_name in other_ledger:
for raw_member_name, confidence in matched_members:
member_name = canonical_by_key.get(canonical_member_key(raw_member_name))
if member_name is not None:
other_ledger[member_name].append({
"amount": per_allocation,
"date": tx["date"],
@@ -400,14 +416,20 @@ def reconcile(
member_share = amount / len(matched_members) if matched_members else 0
for member_name, confidence in matched_members:
if member_name not in ledger:
for raw_member_name, confidence in matched_members:
member_name = canonical_by_key.get(canonical_member_key(raw_member_name))
if member_name is None:
logger.warning(
"Payment matched to unknown member %r (tx: %s, %s) — adding to unmatched",
member_name, tx.get("date", "?"), tx.get("message", "?"),
raw_member_name, tx.get("date", "?"), tx.get("message", "?"),
)
unmatched.append(tx)
continue
if member_name != raw_member_name:
logger.info(
"Person cell %r resolved to canonical member %r — consider fixing the sheet",
raw_member_name, member_name,
)
in_window = [(m, ledger[member_name][m]["expected"]) for m in matched_months if m in ledger[member_name]]
out_of_window = [m for m in matched_months if m not in ledger[member_name]]