Merge pull request 'feat(go): M5.4 — parity diff binary + make parity' (#19) from feat/go-m5-4-parity-binary into main

Reviewed-on: #19

CHANGELOG.md

@@ -1,5 +1,30 @@
 # Changelog
 
+## 2026-05-07 23:17 CEST — fix(go): pass raw value to FormatDate so numeric serial-day dates format
+
+- `go/internal/services/membership/sources.go`: transaction-row parser now passes `row[idxDate]` directly to `matching.FormatDate` (via a new `getRaw` helper) instead of stringifying first via `getVal`. The Sheets API returns numeric serial-day values as `float64` for date-formatted cells; pre-stringifying them defeated `FormatDate`'s `case float64:` dispatch, causing all numeric dates to leak through as `"46147"` style strings instead of `"2026-05-05"`.
+- Surfaced by `make parity` (M5.4): every `transactions[].date` field on `/api/adults` and `/api/juniors` differed between Python and Go.
+- `sources_test.go::TestLoadTransactions` extended with a numeric-serial-day row covering the regression.
+
+## 2026-05-07 23:05 CEST — fix(go): default CacheDir to `tmp/go` to avoid Python collision
+
+- `go/internal/config/config.go`: `CacheDir` default changed from `tmp` to `tmp/go`. Override via `CACHE_DIR` env var still works.
+- Why: both backends used `tmp/<key>_cache.json` with the same keys (`attendance_regular`, `attendance_juniors`, `payments_transactions`, `exceptions_dict`) but different shapes — Python caches post-processed view-model tuples, Go caches raw rows. Whichever wrote last poisoned the cache; running both in parallel produced `ValueError: too many values to unpack (expected 2, got 68)` on Python's `/adults` after the Go server populated `attendance_regular_cache.json` with raw CSV rows.
+- After upgrading: stop the Go server, hit `/flush-cache` on the Python side once (rewrites `tmp/*.json` with correct shapes), then restart `make web-go` — it will use `tmp/go/` going forward. Required for the M5.4 `make parity` workflow which assumes both backends run side-by-side.
+
+## 2026-05-07 22:55 CEST — feat(go): M5.4 — parity diff binary + `make parity`
+
+- `go/cmd/parity/main.go`: new standalone binary that GETs `/api/adults`, `/api/juniors`, `/api/payments` from both Python (:5001) and Go (:8080) backends, scrubs an allowlist (`render_time.total`), and prints `cmp.Diff` for any remaining differences. Exits 0 on full match, 1 on diffs, 2 on fetch/parse errors — CI-friendly for M7.2. `/api/version` is excluded by design (returns binary identity — tag/commit/build_date — which differs between independently built backends); still accessible via `make parity ARGS="-route /api/version"`.
+- `go/cmd/parity/scrub_test.go`: 4 unit tests covering top-level delete, nested delete, missing path, and non-map parent.
+- `go/go.mod`: `github.com/google/go-cmp` promoted to direct dependency.
+- `Makefile`: `parity` target added (`.PHONY`, help, `cd go && go run ./cmd/parity`).
+- `docs/plans/2026-05-07-2254-m5-4-parity-binary.md`: plan archived.
+
+## 2026-05-07 22:37 CEST — feat(py): M5.3 — Python /api/* shadow endpoints
+
+- `app.py`: four new JSON routes (`/api/version`, `/api/adults`, `/api/juniors`, `/api/payments`) mirroring the Go `/api/*` handlers; `_unwrap_view_model_for_api()` helper expands pre-serialised JSON strings and renames `month_labels_json` → `month_labels`, `raw_payments_json` → `raw_payments` to match Go wire contract.
+- `tests/test_app.py`: four new smoke tests asserting top-level key sets and that unwrapped fields are objects (not strings).
+
 ## 2026-05-07 20:13 CEST — feat(go): M5.2 — HTTP handlers for /api/adults, /api/juniors, /api/payments, /api/version
 
 - `web/api/handler.go`: `Handler` struct + `ServeAdults`, `ServeJuniors`, `ServePayments`, `ServeVersion` using `membership.Sources`.

Makefile

@@ -1,4 +1,4 @@
-.PHONY: help fees match web web-py web-debug web-go go-build go-test go-test-all go-parity go-run go-sync-debug go-lint capture-fixtures image run sync sync-2026 test test-v docs
+.PHONY: help fees match web web-py web-debug web-go go-build go-test go-test-all go-parity go-run go-sync-debug go-lint capture-fixtures parity image run sync sync-2026 test test-v docs
 
 export PYTHONPATH := scripts:$(PYTHONPATH)
 VENV := .venv
@@ -29,6 +29,7 @@ help:
 	@echo "  make go-lint     - Run golangci-lint on Go code"
 	@echo "  make go-sync-debug [DAYS=N] - Dry-run Go sync with Fio debug logs and txn table (default DAYS=30)"
 	@echo "  make capture-fixtures - Regenerate parity fixture corpus from live Python"
+	@echo "  make parity      - Diff /api/* between web-py (:5001) and web-go (:8080); both must be running"
 	@echo "  make image       - Build Python OCI container image"
 	@echo "  make run         - Run the built Python Docker image locally"
 	@echo "  make sync        - Sync Fio transactions to Google Sheets"
@@ -102,6 +103,9 @@ go-lint:
 web-go: go-build
 	./$(GO_BIN) server
 
+parity:
+	cd $(GO_SRC) && go run ./cmd/parity $(ARGS)
+
 image:
 	docker build -t fuj-management:latest \
 		--build-arg GIT_TAG=$$(git describe --tags --always 2>/dev/null || echo "untagged") \

app.py

@@ -7,7 +7,7 @@ import os
 import io
 import qrcode
 import logging
-from flask import Flask, render_template, g, send_file, request
+from flask import Flask, render_template, g, send_file, request, jsonify
 
 # Configure logging, allowing override via LOG_LEVEL environment variable
 log_level = os.environ.get("LOG_LEVEL", "INFO").upper()
@@ -68,6 +68,16 @@ BUILD_META = _json.loads(_meta_path.read_text()) if _meta_path.exists() else {
     "tag": "dev", "commit": "local", "build_date": ""
 }
 
+
+def _unwrap_view_model_for_api(vm: dict) -> dict:
+    """Expand pre-stringified JSON fields and rename to match Go API contract."""
+    out = dict(vm)
+    out["member_data"] = _json.loads(out.pop("member_data"))
+    out["month_labels"] = _json.loads(out.pop("month_labels_json"))
+    out["raw_payments"] = _json.loads(out.pop("raw_payments_json"))
+    return out
+
+
 warmup_cache()
 
 @app.before_request
@@ -144,6 +154,75 @@ def sync_bank():
 def version():
     return BUILD_META
 
+@app.route("/api/version")
+def api_version():
+    return jsonify(BUILD_META)
+
+@app.route("/api/adults")
+def api_adults():
+    attendance_url = f"https://docs.google.com/spreadsheets/d/{ATTENDANCE_SHEET_ID}/edit"
+    payments_url = f"https://docs.google.com/spreadsheets/d/{PAYMENTS_SHEET_ID}/edit"
+    members_data = get_cached_data("attendance_regular", ATTENDANCE_SHEET_ID, get_members_with_fees)
+    if not members_data:
+        return jsonify({"error": "no data"}), 503
+    members, sorted_months = members_data
+    transactions = get_cached_data("payments_transactions", PAYMENTS_SHEET_ID, fetch_sheet_data, PAYMENTS_SHEET_ID, CREDENTIALS_PATH)
+    exceptions = get_cached_data(
+        "exceptions_dict", PAYMENTS_SHEET_ID, fetch_exceptions,
+        PAYMENTS_SHEET_ID, CREDENTIALS_PATH,
+        serialize=lambda d: [[list(k), v] for k, v in d.items()],
+        deserialize=lambda c: {tuple(k): v for k, v in c},
+    )
+    result = reconcile(members, sorted_months, transactions, exceptions)
+    vm = build_adults_view_model(
+        members, sorted_months, result, transactions,
+        datetime.now().strftime("%Y-%m"),
+        attendance_url=attendance_url, payments_url=payments_url, bank_account=BANK_ACCOUNT,
+    )
+    return jsonify(_unwrap_view_model_for_api(vm))
+
+@app.route("/api/juniors")
+def api_juniors():
+    attendance_url = f"https://docs.google.com/spreadsheets/d/{ATTENDANCE_SHEET_ID}/edit#gid={JUNIOR_SHEET_GID}"
+    payments_url = f"https://docs.google.com/spreadsheets/d/{PAYMENTS_SHEET_ID}/edit"
+    junior_members_data = get_cached_data("attendance_juniors", ATTENDANCE_SHEET_ID, get_junior_members_with_fees)
+    if not junior_members_data:
+        return jsonify({"error": "no data"}), 503
+    junior_members, sorted_months = junior_members_data
+    transactions = get_cached_data("payments_transactions", PAYMENTS_SHEET_ID, fetch_sheet_data, PAYMENTS_SHEET_ID, CREDENTIALS_PATH)
+    exceptions = get_cached_data(
+        "exceptions_dict", PAYMENTS_SHEET_ID, fetch_exceptions,
+        PAYMENTS_SHEET_ID, CREDENTIALS_PATH,
+        serialize=lambda d: [[list(k), v] for k, v in d.items()],
+        deserialize=lambda c: {tuple(k): v for k, v in c},
+    )
+    adapted_members = adapt_junior_members(junior_members)
+    result = reconcile(adapted_members, sorted_months, transactions, exceptions)
+    vm = build_juniors_view_model(
+        junior_members, adapted_members, sorted_months, result, transactions,
+        datetime.now().strftime("%Y-%m"),
+        attendance_url=attendance_url, payments_url=payments_url, bank_account=BANK_ACCOUNT,
+    )
+    return jsonify(_unwrap_view_model_for_api(vm))
+
+@app.route("/api/payments")
+def api_payments():
+    attendance_url = f"https://docs.google.com/spreadsheets/d/{ATTENDANCE_SHEET_ID}/edit"
+    payments_url = f"https://docs.google.com/spreadsheets/d/{PAYMENTS_SHEET_ID}/edit"
+    transactions = get_cached_data("payments_transactions", PAYMENTS_SHEET_ID, fetch_sheet_data, PAYMENTS_SHEET_ID, CREDENTIALS_PATH)
+    adults_data = get_cached_data("attendance_regular", ATTENDANCE_SHEET_ID, get_members_with_fees)
+    juniors_data = get_cached_data("attendance_juniors", ATTENDANCE_SHEET_ID, get_junior_members_with_fees)
+    member_names = []
+    if adults_data:
+        member_names.extend(name for name, _, _ in adults_data[0])
+    if juniors_data:
+        member_names.extend(name for name, _, _ in juniors_data[0])
+    vm = build_payments_view_model(
+        transactions, member_names,
+        attendance_url=attendance_url, payments_url=payments_url,
+    )
+    return jsonify(vm)
+
 @app.route("/adults")
 def adults_view():
     attendance_url = f"https://docs.google.com/spreadsheets/d/{ATTENDANCE_SHEET_ID}/edit"

docs/plans/2026-05-03-2349-go-backend-rewrite-progress.md

@@ -2,9 +2,9 @@
 
 Companion to [2026-05-03-2349-go-backend-rewrite.md](2026-05-03-2349-go-backend-rewrite.md).
 
-**Current milestone:** M4 — IO layer behind interfaces ✅
+**Current milestone:** M5 — JSON-only `/api/...` routes ✅
 **Started:** 2026-05-04
-**Last updated:** 2026-05-07
+**Last updated:** 2026-05-07 (M5.4)
 
 ## How to use
 
@@ -99,8 +99,8 @@ Goal: byte-equal JSON between Python and Go for every route. This is the parity
 
 - [x] **M5.1** Hand-author Go structs for `/api/adults`, `/api/juniors`, `/api/payments`, `/api/version` with explicit `json:` tags matching Python keys; emit JSON Schemas via `github.com/invopop/jsonschema` to `tests/fixtures/api-schema/` — `f253e3f`
 - [x] **M5.2** Implement Go handlers for `/api/*` routes composing `services/*` results into the JSON structs — `7d48e8f`
-- [ ] **M5.3** Add Python `/api/X` shadow endpoints in [app.py](app.py): `jsonify(view_model_dict)` — no transformation
-- [ ] **M5.4** Build `cmd/parity/main.go`: hits both backends' `/api/X`, normalizes allowlist (`render_time.total`, `build_meta`), prints `cmp.Diff`. Add `make parity` target
+- [x] **M5.3** Add Python `/api/X` shadow endpoints in [app.py](app.py): `jsonify(view_model_dict)` — no transformation — `40e4a9e`
+- [x] **M5.4** Build `cmd/parity/main.go`: hits both backends' `/api/X`, normalizes allowlist (`render_time.total`, `build_meta`), prints `cmp.Diff`. Add `make parity` target
 
 **Gate:** For each route, `make parity` reports zero non-allowlisted diffs across the M3 fixture corpus.
 
@@ -154,6 +154,7 @@ Goal: Go is the one true backend.
 
 (Add entries as you go. Format: `YYYY-MM-DD — short note`.)
 
+- 2026-05-07 — `/api/version` excluded from parity diff by design. Each binary's tag/commit/build_date is identity, not a data contract — diffing it would always flag a diff between independently built backends. Route remains reachable via `make parity ARGS="-route /api/version"` for manual inspection.
 - 2026-05-04 — Plan approved. Versioning policy: latest stable for Go and all libs at the time M1 starts. Frontends explicitly allowed to diverge between Python and Go; only the JSON API contract is parity-locked. No reverse proxy — both backends run on different ports via `make web-py` / `make web-go`.
 - 2026-05-07 — M4 complete. Chose fakes-only unit tests (no live integration tests) and CSV-via-public-URL for attendance (no Sheets API auth required for read-only). golangci-lint gofumpt extra-rules differ slightly from standalone gofumpt; used `golangci-lint run --fix --enable-only gofumpt` to auto-resolve formatting.
 - 2026-05-04 — M1 complete. Dockerfile base changed from `distroless/static:nonroot` → `alpine:3` for debuggability (can tighten later). CLI dispatcher uses stdlib `flag`; module path `fuj-management/go`. golangci-lint v1 embedded gofumpt merges all imports into one group (no stdlib/local split) — accepted as the project style.

docs/plans/2026-05-07-2114-go-rewrite-m5-3-python-api-shadow.md

@@ -0,0 +1,113 @@
+# M5.3 — Python `/api/X` shadow endpoints
+
+Companion to:
+- [2026-05-03-2349-go-backend-rewrite.md](2026-05-03-2349-go-backend-rewrite.md) (master design)
+- [2026-05-03-2349-go-backend-rewrite-progress.md](2026-05-03-2349-go-backend-rewrite-progress.md) (M5.3 row)
+- [2026-05-07-1431-m5-json-api-parity.md](2026-05-07-1431-m5-json-api-parity.md) (Python view-model extraction prep)
+- [2026-05-07-1650-go-rewrite-m5-1-api-structs-schemas.md](2026-05-07-1650-go-rewrite-m5-1-api-structs-schemas.md) (Go wire types)
+
+## Context
+
+M5.1 (Go wire types + JSON Schemas) and M5.2 (Go HTTP handlers for `/api/adults` `/api/juniors` `/api/payments` `/api/version`) have merged. M5.3 mirrors the same four endpoints on the Python Flask side so M5.4's `cmd/parity` tool can hit both backends and diff the JSON. After M5.3, every byte the Go side emits has a Python counterpart to compare against.
+
+The Python view-model builders ([scripts/views.py](scripts/views.py)) already produce dicts very close to the wire shape — except three template-only fields (`member_data`, `month_labels_json`, `raw_payments_json`) are pre-`json.dumps`'d for inline `<script>` blocks. M5.1's plan called this out explicitly: M5.3's `/api/X` is `jsonify(unwrap_json_strings(view_model_dict))` — a 4-line shim, not real transformation logic.
+
+## Approach
+
+Add four shadow routes to [app.py](app.py) and one private unwrap helper. Builders and templates are untouched.
+
+### Decisions
+
+1. **Unwrap shim lives in `app.py`**, not `scripts/views.py`. It's 4 lines, only the API routes use it, and it's parity-only scaffolding that M8 will delete. Keeping `views.py` free of HTTP-layer concerns means cleaner deletion later.
+2. **No data-loading helper extraction.** Each shadow route duplicates ~8 lines of cache loads from its sibling HTML route. A helper would have to thread `attendance_url` / `payments_url` / `bank_account` and the adults-vs-juniors-vs-payments branching back out — net negative for code that M8 will erase wholesale.
+3. **Drop `record_step` calls in API routes.** `record_step` only feeds `inject_render_time` (a Jinja `context_processor`); JSON responses don't go through templates, so timing breakdown has no consumer.
+4. **`/api/version` is a one-liner.** `BUILD_META` already has the keys (`tag`, `commit`, `build_date`) Go emits. Just `jsonify(BUILD_META)`. Existing `/version` route stays as-is — the new endpoint sits alongside.
+5. **Tests assert key sets and unwrap, not values.** Hard-code `EXPECTED_ADULTS_KEYS` etc. as module constants in [tests/test_app.py](tests/test_app.py). Catches drift in unit tests rather than waiting for M5.4 parity diffs. Type-check the unwrapped fields (`isinstance(member_data, dict)` etc.) to prove the shim ran.
+
+### The shim
+
+```python
+def _unwrap_view_model_for_api(vm: dict) -> dict:
+    out = dict(vm)
+    out["member_data"] = _json.loads(out.pop("member_data"))
+    out["month_labels"] = _json.loads(out.pop("month_labels_json"))
+    out["raw_payments"] = _json.loads(out.pop("raw_payments_json"))
+    return out
+```
+
+Note the **rename**: `month_labels_json` → `month_labels`, `raw_payments_json` → `raw_payments` (matches Go contract per [adults.go](go/internal/web/api/adults.go) JSON tags).
+
+### Route skeleton (`/api/adults`)
+
+```python
+@app.route("/api/adults")
+def api_adults():
+    attendance_url = f"https://docs.google.com/spreadsheets/d/{ATTENDANCE_SHEET_ID}/edit"
+    payments_url = f"https://docs.google.com/spreadsheets/d/{PAYMENTS_SHEET_ID}/edit"
+    members_data = get_cached_data("attendance_regular", ATTENDANCE_SHEET_ID, get_members_with_fees)
+    if not members_data:
+        return jsonify({"error": "no data"}), 503
+    members, sorted_months = members_data
+    transactions = get_cached_data("payments_transactions", PAYMENTS_SHEET_ID, fetch_sheet_data, PAYMENTS_SHEET_ID, CREDENTIALS_PATH)
+    exceptions = get_cached_data(
+        "exceptions_dict", PAYMENTS_SHEET_ID, fetch_exceptions,
+        PAYMENTS_SHEET_ID, CREDENTIALS_PATH,
+        serialize=lambda d: [[list(k), v] for k, v in d.items()],
+        deserialize=lambda c: {tuple(k): v for k, v in c},
+    )
+    result = reconcile(members, sorted_months, transactions, exceptions)
+    vm = build_adults_view_model(
+        members, sorted_months, result, transactions,
+        datetime.now().strftime("%Y-%m"),
+        attendance_url=attendance_url, payments_url=payments_url, bank_account=BANK_ACCOUNT,
+    )
+    return jsonify(_unwrap_view_model_for_api(vm))
+```
+
+`/api/juniors` mirrors with `adapt_junior_members` + `JUNIOR_SHEET_GID` + `build_juniors_view_model`. `/api/payments` skips the unwrap (its builder has no JSON-string fields): `return jsonify(vm)`. `/api/version`: `return jsonify(BUILD_META)`.
+
+## Files to modify
+
+- [app.py](app.py)
+  - L10: add `jsonify` to `from flask import ...`.
+  - Add `_unwrap_view_model_for_api` near [BUILD_META](app.py#L67) (it already imports `json as _json`).
+  - Add four routes: `/api/version`, `/api/adults`, `/api/juniors`, `/api/payments`. Place them after [`/version`](app.py#L143) for grouping.
+- [tests/test_app.py](tests/test_app.py)
+  - Add `EXPECTED_ADULTS_KEYS`, `EXPECTED_JUNIORS_KEYS`, `EXPECTED_PAYMENTS_KEYS`, `EXPECTED_VERSION_KEYS` module constants (sourced from [adults.go](go/internal/web/api/adults.go) / [juniors.go](go/internal/web/api/juniors.go) / [payments.go](go/internal/web/api/payments.go) / [version.go](go/internal/web/api/version.go) JSON tags).
+  - Four new test functions: `test_api_adults`, `test_api_juniors`, `test_api_payments`, `test_api_version`. Reuse the existing `_bypass_cache` patcher and the `fetch_sheet_data` / `fetch_exceptions` / `get_members_with_fees` / `get_junior_members_with_fees` mocks already in the file.
+  - Each adults/juniors test asserts: `200`, `response.is_json`, `set(json.keys()) == EXPECTED_*_KEYS`, `isinstance(json["member_data"], dict)`, `isinstance(json["month_labels"], dict)`, `isinstance(json["raw_payments"], dict)`.
+- [CHANGELOG.md](CHANGELOG.md): post-merge entry per CLAUDE.md format.
+- [docs/plans/2026-05-03-2349-go-backend-rewrite-progress.md:102](docs/plans/2026-05-03-2349-go-backend-rewrite-progress.md#L102): tick M5.3 with merge SHA.
+
+## Reusable existing code
+
+- [`build_adults_view_model`](scripts/views.py#L64), [`build_juniors_view_model`](scripts/views.py#L240), [`build_payments_view_model`](scripts/views.py#L432) — call as-is, no changes.
+- [`reconcile`](scripts/match_payments.py) and [`adapt_junior_members`](scripts/views.py#L48) — same.
+- [`get_cached_data`](app.py#L36), [`fetch_sheet_data`](scripts/match_payments.py), [`fetch_exceptions`](scripts/match_payments.py), [`get_members_with_fees`](scripts/attendance.py), [`get_junior_members_with_fees`](scripts/attendance.py) — same call sites as the HTML routes.
+- [`BUILD_META`](app.py#L67) — already shaped to match Go's `VersionResponse`.
+
+## Verification
+
+1. `make test` — all existing tests still pass; four new `test_api_*` tests pass.
+2. `python -c "import app"` — catches the missing `jsonify` import.
+3. `make web-py`, then:
+   - `curl -s localhost:5001/api/version | jq .` → `{tag, commit, build_date}`.
+   - `curl -s localhost:5001/api/adults | jq 'keys'` → 14 keys, no `_json` suffix anywhere.
+   - `curl -s localhost:5001/api/adults | jq '.member_data | type, .month_labels | type, .raw_payments | type'` → all `"object"` (proves unwrap).
+   - Same checks on `/api/juniors` and `/api/payments`.
+4. Visit `/adults`, `/juniors`, `/payments` in browser — HTML still renders identically (regression check on builders).
+5. **Optional pre-M5.4 peek:** `make web-go` on :8080 + `make web-py` on :5001, then `diff <(curl -s :5001/api/adults | jq -S .) <(curl -s :8080/api/adults | jq -S .)`. Expect non-zero diff (raw transaction key shape, see below) — that is fine; M5.4 surfaces and resolves these.
+
+## Out of scope (M5.4 will surface and resolve)
+
+These are known parity friction points; **don't fix in M5.3** — the whole point of M5.4 is to enumerate them:
+
+- **`raw_payments[name][i]` row shape**: Python emits raw Google Sheets row dicts with column-header keys (e.g. `"VS"`, `"Sync ID"`); Go's `RawTransaction` uses snake_case (`vs`, `sync_id`). Keys and types will diverge.
+- **`unmatched[]`**: same divergence (same raw row dicts).
+- **`null` vs missing keys**: Python omits keys never set; Go zero-value structs may emit `null`/`""` depending on `omitempty`.
+- **`Decimal` / float precision** in `grouped_payments` amounts.
+- **Field insertion order**: `jsonify` preserves insertion order (Python 3.7+); Go marshals struct fields in declaration order. Likely fine, parity tool will tell.
+
+## Branch + MR
+
+Per CLAUDE.md: branch `feat/go-m5-3-python-api-shadow`, push with `-u`, open MR via `tea pr create --title ... --description ... --base main --head feat/go-m5-3-python-api-shadow`. Do not merge from CLI.

docs/plans/2026-05-07-2254-m5-4-parity-binary.md

@@ -0,0 +1,152 @@
+# M5.4 — Parity diff binary (`cmd/parity`) + `make parity`
+
+## Context
+
+Per [docs/plans/2026-05-03-2349-go-backend-rewrite-progress.md:103](docs/plans/2026-05-03-2349-go-backend-rewrite-progress.md#L103), M5.4 is the next milestone in the Go rewrite. M5.1–M5.3 already landed:
+
+- M5.1: hand-authored Go response structs at [go/internal/web/api/](go/internal/web/api/) + JSON Schemas in [go/tests/fixtures/api-schema/](go/tests/fixtures/api-schema/).
+- M5.2: Go `/api/version|adults|juniors|payments` handlers in [go/internal/web/api/handler.go](go/internal/web/api/handler.go).
+- M5.3: Python shadow endpoints in [app.py:157-224](app.py#L157-L224) using `_unwrap_view_model_for_api` for the same JSON shape.
+
+What's missing: a tool that proves the two backends actually agree on the wire. The M5 gate says "byte-equal JSON between Python and Go for every route." Without a diffing tool, drift between the two implementations slips in silently and we can't gate further milestones (M6 frontend, M7 watch period) on parity.
+
+This task delivers the **parity contract enforcer**: a Go binary that fetches `/api/*` from both backends, scrubs an allowlist of expected diffs, and prints `cmp.Diff` for everything else. Both backends read the same live Google Sheets — that shared state is what makes parity meaningful, no scenario fixtures needed.
+
+## Scope
+
+In scope:
+- New binary `go/cmd/parity/main.go` plus a small support package for the allowlist scrubber.
+- A unit test for the scrubber.
+- New `make parity` target.
+- `go-cmp` promoted to a direct dependency.
+
+Out of scope (explicitly):
+- CI integration / nightly job — that's M7.2.
+- Fixture-driven offline parity — pure-fn parity already runs via `make go-parity` (M3).
+- Hooking `make parity` into `make go-test-all` — leave it manual since it requires two live servers.
+
+## Approach
+
+### 1. New binary: `go/cmd/parity/main.go`
+
+Standalone binary (mirrors task spec — not a subcommand of `fuj`). Stdlib `flag`, no third-party CLI lib.
+
+**Flags:**
+- `-py` — Python base URL, default `http://localhost:5001`
+- `-go` — Go base URL, default `http://localhost:8080`
+- `-route` — optional single route to diff (e.g. `/api/adults`); empty means iterate all four
+- `-timeout` — per-request timeout, default `30s` (sheet fetches are slow on cold cache)
+
+**Routes**, hard-coded:
+```go
+var routes = []string{"/api/version", "/api/adults", "/api/juniors", "/api/payments"}
+```
+All `GET`, no query params (verified in [app.py:157-224](app.py#L157-L224) and [go/internal/web/api/handler.go:27-68](go/internal/web/api/handler.go#L27-L68)).
+
+**Per route, the binary:**
+1. `GET py+route` and `GET go+route` (sequential — keep it simple; total wall time ~1–2s on warm cache).
+2. Verify both return HTTP 200; on non-200, print body and mark route as ERROR.
+3. Decode each body into `map[string]any` (NOT into `api.AdultsResponse` — using the typed struct would silently drop unknown Python-side keys, defeating the diff. Map gives `cmp.Diff` clean dotted-path field names for free.).
+4. Run `scrub(m, allowlist)` on both decoded maps.
+5. `diff := cmp.Diff(pyMap, goMap, cmpopts.EquateEmpty())`.
+6. If `diff != ""`, print `=== /api/X ===` header followed by the diff; track for exit code.
+
+**Exit codes:**
+- `0` — all routes match (or the only diffs were under the allowlist).
+- `1` — at least one route had a non-allowlisted diff.
+- `2` — at least one route failed to fetch / parse (HTTP error, timeout, non-JSON body).
+
+This makes the binary CI-friendly when M7.2 lands.
+
+**Output**: human-readable to stdout — header per route, then "OK" or the diff. Final summary line: `parity: 4/4 routes match` or `parity: 2/4 match, 1 diff, 1 error`.
+
+### 2. Allowlist scrubber
+
+Live in same package (`main`). Keep it tiny — no need for a separate sub-package.
+
+```go
+var defaultAllowlist = []string{"render_time.total", "build_meta"}
+
+// scrub walks m and deletes any key whose dotted path matches an allowlist entry.
+// "render_time.total" deletes m["render_time"]["total"] only (preserves render_time.breakdown).
+// "build_meta" (no dot) deletes m["build_meta"] entirely.
+func scrub(m map[string]any, paths []string) { ... }
+```
+
+Today these fields **don't appear in the JSON** — they're Jinja template-only ([app.py:91-110](app.py#L91-L110)) and the Go side only logs render time via [go/internal/web/middleware/timer.go](go/internal/web/middleware/timer.go). So today the scrub is a no-op. The implementation is forward-compatible insurance: if someone later adds either field to a JSON response on one side only, the parity binary already tolerates it.
+
+The implementation note in [go/internal/web/middleware/timer.go](go/internal/web/middleware/timer.go) ("the elapsed value maps to render_time.total in the M5 JSON allowlist") confirms this is the intended design.
+
+### 3. Unit test
+
+Add `go/cmd/parity/scrub_test.go` with a couple of cases:
+- Scrubbing top-level key (`build_meta`) — key removed; siblings untouched.
+- Scrubbing nested key (`render_time.total`) — only `total` removed; `breakdown` preserved.
+- Path that doesn't exist — no-op, no error.
+- Map without the parent key — no-op (don't panic).
+
+Runs as part of normal `make go-test` (no `-tags` needed).
+
+### 4. Makefile target
+
+Add to [Makefile](Makefile):
+
+```make
+parity:
+	cd $(GO_SRC) && go run ./cmd/parity $(ARGS)
+```
+
+Add `parity` to the `.PHONY` line at [Makefile:1](Makefile#L1) and a `help` entry like:
+
+```
+@echo "  make parity      - Diff /api/* between web-py (:5001) and web-go (:8080); both must be running"
+```
+
+Don't depend on `go-build` — `go run` compiles ad-hoc, and parity is interactive enough that the slight rebuild cost doesn't matter.
+
+### 5. Dependency
+
+`github.com/google/go-cmp v0.7.0` is already a transitive dep ([go/go.sum:24-25](go/go.sum#L24-L25)) but not in `go.mod`'s `require` block. After adding the import, run `go mod tidy` inside `go/` to promote it to direct. No new external deps.
+
+## Files to add / modify
+
+**Add:**
+- [go/cmd/parity/main.go](go/cmd/parity/main.go) — flags, route loop, fetch+scrub+diff, exit codes
+- [go/cmd/parity/scrub_test.go](go/cmd/parity/scrub_test.go) — unit tests for the scrubber
+
+**Modify:**
+- [Makefile](Makefile) — `.PHONY`, help block, `parity` target
+- [go/go.mod](go/go.mod) — promote `go-cmp` to direct (via `go mod tidy`)
+- [go/go.sum](go/go.sum) — likely unchanged (already pinned)
+- [docs/plans/2026-05-03-2349-go-backend-rewrite-progress.md](docs/plans/2026-05-03-2349-go-backend-rewrite-progress.md) — tick M5.4, add commit SHA
+- [CHANGELOG.md](CHANGELOG.md) — entry per project convention
+
+**Per [CLAUDE.md](CLAUDE.md) plans convention:** copy this plan to `docs/plans/2026-05-07-HHMM-m5-4-parity-binary.md` (timestamp via `date "+%Y-%m-%d-%H%M"`) before opening the MR, since plan files are committed for posterity.
+
+## Existing utilities to reuse
+
+- `cmp.Diff` + `cmpopts.EquateEmpty()` from `github.com/google/go-cmp/cmp` (already in go.sum).
+- Stdlib `net/http`, `encoding/json`, `flag` — sufficient; no need to pull a CLI framework.
+- No existing scrubber to reuse — the one in [scripts/scrub_fixtures.py](scripts/scrub_fixtures.py) operates on the *capture* path and renames PII; the parity scrubber is a different concern (path-based deletion of expected diffs).
+
+## Verification
+
+Manual smoke test (the binary is meant to be run against live servers):
+
+1. **Sanity / build:** `cd go && go build ./cmd/parity && go test ./cmd/parity/...` — compiles + scrubber unit test passes.
+2. **Both backends up:**
+   - Terminal A: `make web-py`
+   - Terminal B: `make web-go`
+   - Terminal C: `make parity`
+   Expected: `parity: 4/4 routes match`, exit 0.
+3. **Negative test:** add a literal extra field to one Go handler temporarily (e.g. `"diagnostic": "test"` in `VersionResponse`), rebuild, re-run `make parity`. Expected: non-zero exit, diff shown for `/api/version`. Revert.
+4. **Allowlist test:** in a unit test (or by manually constructing a payload with `render_time.total` injected), confirm the scrubber removes it before the diff stage.
+5. **CHANGELOG** entry added; progress tracker ticked with the merge commit SHA.
+
+Per [CLAUDE.md](CLAUDE.md) branching policy: this is a feature, so work happens on `feat/go-m5-4-parity-binary`, push with `-u`, open MR with `tea pr create --base main --head feat/go-m5-4-parity-binary`. User merges in Gitea.
+
+## Notes & risks
+
+- **Cache-warmth dependency:** Both backends cache aggressively. A cold-cache fetch on Python can take 30s+ — hence the configurable timeout. If parity is run immediately after `flush-cache`, expect the first run to be slow.
+- **Order-sensitive lists:** `cmp.Diff` on `map[string]any` treats slices as ordered. If either backend ever returns members/transactions in a different order, the diff will flag it — that's a real parity bug, not a false positive, so this is correct behavior. If we hit ordering instability later, fix the source, don't add `cmpopts.SortSlices`.
+- **Float formatting:** Both sides go through `encoding/json` (Go) and `jsonify` (Python `json.dumps`). Floats in totals/balances may format differently (`123` vs `123.0`). If this surfaces, the Go side already uses typed structs with `int` where appropriate — investigate at the source rather than allowlisting.

go/cmd/parity/main.go

@@ -0,0 +1,133 @@
+package main
+
+import (
+	"encoding/json"
+	"flag"
+	"fmt"
+	"io"
+	"net/http"
+	"os"
+	"strings"
+	"time"
+
+	"github.com/google/go-cmp/cmp"
+	"github.com/google/go-cmp/cmp/cmpopts"
+)
+
+// /api/version is intentionally excluded — it returns each binary's own build
+// identity (tag/commit/build_date), which differs by design between independently
+// built backends. Pass -route /api/version to inspect it manually.
+var allRoutes = []string{"/api/adults", "/api/juniors", "/api/payments"}
+
+// defaultAllowlist holds dotted key paths to strip before diffing.
+// render_time.total is forward-compatible insurance: today it lives in the Jinja
+// template context only (app.py inject_render_time) and is logged via
+// middleware/timer.go on the Go side, so it isn't in any JSON response. If either
+// side ever surfaces it under a render_time envelope, the scrubber handles it.
+var defaultAllowlist = []string{"render_time.total"}
+
+func main() {
+	pyURL := flag.String("py", "http://localhost:5001", "Python backend base URL")
+	goURL := flag.String("go", "http://localhost:8080", "Go backend base URL")
+	route := flag.String("route", "", "single route to diff, e.g. /api/adults (default: all)")
+	timeout := flag.Duration("timeout", 30*time.Second, "per-request HTTP timeout")
+	flag.Parse()
+
+	client := &http.Client{Timeout: *timeout}
+
+	targets := allRoutes
+	if *route != "" {
+		targets = []string{*route}
+	}
+
+	matched, diffs, errs := 0, 0, 0
+
+	for _, r := range targets {
+		pyMap, err1 := fetch(client, *pyURL+r)
+		goMap, err2 := fetch(client, *goURL+r)
+
+		if err1 != nil || err2 != nil {
+			fmt.Printf("=== %s ===\n", r)
+			if err1 != nil {
+				fmt.Printf("ERROR (py): %v\n", err1)
+			}
+			if err2 != nil {
+				fmt.Printf("ERROR (go): %v\n", err2)
+			}
+			fmt.Println()
+			errs++
+			continue
+		}
+
+		scrub(pyMap, defaultAllowlist)
+		scrub(goMap, defaultAllowlist)
+
+		diff := cmp.Diff(pyMap, goMap, cmpopts.EquateEmpty())
+		if diff == "" {
+			fmt.Printf("=== %s ===\nOK\n\n", r)
+			matched++
+		} else {
+			fmt.Printf("=== %s ===\n%s\n", r, diff)
+			diffs++
+		}
+	}
+
+	total := len(targets)
+	fmt.Printf("parity: %d/%d routes match", matched, total)
+	if diffs > 0 {
+		fmt.Printf(", %d diff", diffs)
+		if diffs > 1 {
+			fmt.Print("s")
+		}
+	}
+	if errs > 0 {
+		fmt.Printf(", %d error", errs)
+		if errs > 1 {
+			fmt.Print("s")
+		}
+	}
+	fmt.Println()
+
+	if errs > 0 {
+		os.Exit(2)
+	}
+	if diffs > 0 {
+		os.Exit(1)
+	}
+}
+
+func fetch(client *http.Client, url string) (map[string]any, error) {
+	resp, err := client.Get(url) //nolint:noctx
+	if err != nil {
+		return nil, err
+	}
+	defer resp.Body.Close()
+	body, err := io.ReadAll(resp.Body)
+	if err != nil {
+		return nil, fmt.Errorf("read body: %w", err)
+	}
+	if resp.StatusCode != http.StatusOK {
+		return nil, fmt.Errorf("HTTP %d: %s", resp.StatusCode, strings.TrimSpace(string(body)))
+	}
+	var m map[string]any
+	if err := json.Unmarshal(body, &m); err != nil {
+		return nil, fmt.Errorf("decode JSON: %w", err)
+	}
+	return m, nil
+}
+
+// scrub removes keys from m whose dotted paths appear in paths.
+// A bare segment (no dot) deletes a top-level key.
+// A two-segment path "parent.child" deletes child from m["parent"] if it is a map.
+func scrub(m map[string]any, paths []string) {
+	for _, path := range paths {
+		parts := strings.SplitN(path, ".", 2)
+		if len(parts) == 1 {
+			delete(m, parts[0])
+		} else {
+			if child, ok := m[parts[0]].(map[string]any); ok {
+				delete(child, parts[1])
+			}
+		}
+	}
+}

go/cmd/parity/scrub_test.go

@@ -0,0 +1,57 @@
+package main
+
+import "testing"
+
+func TestScrubTopLevel(t *testing.T) {
+	m := map[string]any{
+		"build_meta": map[string]any{"tag": "v1"},
+		"other":      "keep",
+	}
+	scrub(m, []string{"build_meta"})
+	if _, ok := m["build_meta"]; ok {
+		t.Error("expected build_meta to be removed")
+	}
+	if m["other"] != "keep" {
+		t.Error("expected other to be preserved")
+	}
+}
+
+func TestScrubNested(t *testing.T) {
+	m := map[string]any{
+		"render_time": map[string]any{
+			"total":     "0.123",
+			"breakdown": "fetch:0.1s",
+		},
+		"other": "keep",
+	}
+	scrub(m, []string{"render_time.total"})
+	rt, ok := m["render_time"].(map[string]any)
+	if !ok {
+		t.Fatal("render_time should still be present")
+	}
+	if _, ok := rt["total"]; ok {
+		t.Error("expected render_time.total to be removed")
+	}
+	if rt["breakdown"] != "fetch:0.1s" {
+		t.Error("expected render_time.breakdown to be preserved")
+	}
+	if m["other"] != "keep" {
+		t.Error("expected other to be preserved")
+	}
+}
+
+func TestScrubMissingPath(t *testing.T) {
+	m := map[string]any{"foo": "bar"}
+	scrub(m, []string{"nonexistent", "render_time.total"})
+	if m["foo"] != "bar" {
+		t.Error("expected foo to be preserved")
+	}
+}
+
+func TestScrubNestedParentNotMap(t *testing.T) {
+	m := map[string]any{"render_time": "not-a-map"}
+	scrub(m, []string{"render_time.total"})
+	if m["render_time"] != "not-a-map" {
+		t.Error("expected render_time to be unchanged when it is not a map")
+	}
+}

go/go.mod

@@ -3,6 +3,7 @@ module fuj-management/go
 go 1.26.1
 
 require (
+	github.com/google/go-cmp v0.7.0
 	github.com/invopop/jsonschema v0.14.0
 	golang.org/x/net v0.53.0
 	golang.org/x/text v0.36.0

go/internal/config/config.go

@@ -50,7 +50,7 @@ func Load() Config {
 	return Config{
 		CredentialsPath:  env("CREDENTIALS_PATH", ".secret/fuj-management-bot-credentials.json"),
 		BankAccount:      env("BANK_ACCOUNT", "CZ8520100000002800359168"),
-		CacheDir:         env("CACHE_DIR", "tmp"),
+		CacheDir:         env("CACHE_DIR", "tmp/go"),
 		CacheTTL:         envDuration("CACHE_TTL_SECONDS", 300),
 		CacheAPICheckTTL: envDuration("CACHE_API_CHECK_TTL_SECONDS", 300),
 		DriveTimeout:     envDuration("DRIVE_TIMEOUT_SECONDS", 10),

go/internal/services/membership/sources.go

@@ -394,9 +394,19 @@ func parseTransactionRows(rows [][]any) ([]reconcile.Transaction, error) {
 		return fmt.Sprint(row[i])
 	}
 
+	// getRaw returns row[i] without stringifying — needed for FormatDate to
+	// dispatch on the underlying numeric type (Sheets returns serial-day
+	// numbers as float64). Stringifying first defeats that dispatch.
+	getRaw := func(row []any, i int) any {
+		if i < 0 || i >= len(row) {
+			return nil
+		}
+		return row[i]
+	}
+
 	var txns []reconcile.Transaction
 	for _, row := range rows[1:] {
-		dateStr := matching.FormatDate(getVal(row, idxDate))
+		dateStr := matching.FormatDate(getRaw(row, idxDate))
 		amountRaw := row[idxAmount]
 		if idxAmount < 0 || idxAmount >= len(row) {
 			amountRaw = ""

go/internal/services/membership/sources_test.go

@@ -114,12 +114,15 @@ func TestLoadJuniors(t *testing.T) {
 
 func TestLoadTransactions(t *testing.T) {
 	// Sheets fake keyed by "<spreadsheetID>/<range>" — use the real constant.
+	// Row 1 uses a pre-formatted date string; row 2 uses the numeric Sheets
+	// serial-day form (float64) — the API returns either depending on cell
+	// formatting, and FormatDate must handle both.
 	paymentsKey := config.PaymentsSheetID + "/A1:Z"
 	sh := &sheets.Fake{Values: map[string][][]any{
 		paymentsKey: {
 			{"Date", "Amount", "manual fix", "Person", "Purpose", "Inferred Amount", "Sender", "VS", "Message", "Bank ID", "Sync ID"},
 			{"2026-04-01", 700.0, "", "Alice", "2026-04", "", "Alice Bank", "", "fee", "", "abc"},
-			{"2026-05-01", 500.0, "", "", "", "", "Bob Bank", "", "platba", "", "def"},
+			{46147.0, 500.0, "", "", "", "", "Bob Bank", "", "platba", "", "def"}, // 46147 serial-day = 2026-05-05
 		},
 	}}
 	s := buildSources(t, &attendance.Fake{}, sh)
@@ -137,6 +140,12 @@ func TestLoadTransactions(t *testing.T) {
 	if txns[0].Amount != 700 {
 		t.Errorf("txn[0].Amount: want 700, got %v", txns[0].Amount)
 	}
+	if txns[0].Date != "2026-04-01" {
+		t.Errorf("txn[0].Date: want 2026-04-01, got %q", txns[0].Date)
+	}
+	if txns[1].Date != "2026-05-05" {
+		t.Errorf("txn[1].Date (numeric serial-day): want 2026-05-05, got %q", txns[1].Date)
+	}
 }
 
 func TestLoadExceptions(t *testing.T) {

tests/test_app.py

@@ -1,7 +1,17 @@
 import unittest
+import json
 from unittest.mock import patch
 from app import app
 
+EXPECTED_ADULTS_KEYS = {
+    "months", "raw_months", "results", "totals", "member_data", "month_labels",
+    "raw_payments", "credits", "debts", "unmatched", "attendance_url",
+    "payments_url", "bank_account", "current_month",
+}
+EXPECTED_JUNIORS_KEYS = EXPECTED_ADULTS_KEYS
+EXPECTED_PAYMENTS_KEYS = {"grouped_payments", "sorted_people", "attendance_url", "payments_url"}
+EXPECTED_VERSION_KEYS = {"tag", "commit", "build_date"}
+
 
 def _bypass_cache(cache_key, sheet_id, fetch_func, *args, serialize=None, deserialize=None, **kwargs):
     """Test helper: call fetch_func directly, bypassing the cache layer."""
@@ -97,5 +107,80 @@ class TestWebApp(unittest.TestCase):
         self.assertIn(b'500/500 CZK', response.data)
         self.assertIn(b'?', response.data)
 
+    def test_api_version(self):
+        """Test /api/version returns BUILD_META keys as JSON."""
+        response = self.client.get('/api/version')
+        self.assertEqual(response.status_code, 200)
+        self.assertTrue(response.is_json)
+        data = json.loads(response.data)
+        self.assertEqual(set(data.keys()), EXPECTED_VERSION_KEYS)
+
+    @patch('app.get_cached_data', side_effect=_bypass_cache)
+    @patch('app.fetch_sheet_data')
+    @patch('app.fetch_exceptions', return_value={})
+    @patch('app.get_members_with_fees')
+    def test_api_adults(self, mock_get_members, mock_exceptions, mock_fetch_sheet, mock_cache):
+        """Test /api/adults returns JSON with correct top-level keys and unwrapped fields."""
+        mock_get_members.return_value = (
+            [('Test Member', 'A', {'2026-01': (750, 4)})],
+            ['2026-01']
+        )
+        mock_fetch_sheet.return_value = [{
+            'date': '2026-01-01', 'amount': 750, 'person': 'Test Member',
+            'purpose': '2026-01', 'message': 'test payment',
+            'sender': 'External Bank User', 'inferred_amount': 750,
+        }]
+        response = self.client.get('/api/adults')
+        self.assertEqual(response.status_code, 200)
+        self.assertTrue(response.is_json)
+        data = json.loads(response.data)
+        self.assertEqual(set(data.keys()), EXPECTED_ADULTS_KEYS)
+        self.assertIsInstance(data['member_data'], dict)
+        self.assertIsInstance(data['month_labels'], dict)
+        self.assertIsInstance(data['raw_payments'], dict)
+
+    @patch('app.get_cached_data', side_effect=_bypass_cache)
+    @patch('app.fetch_sheet_data')
+    @patch('app.fetch_exceptions', return_value={})
+    @patch('app.get_junior_members_with_fees')
+    def test_api_juniors(self, mock_get_junior_members, mock_exceptions, mock_fetch_sheet, mock_cache):
+        """Test /api/juniors returns JSON with correct top-level keys and unwrapped fields."""
+        mock_get_junior_members.return_value = (
+            [
+                ('Junior One', 'J', {'2026-01': (500, 3, 0, 3)}),
+                ('Junior Two', 'X', {'2026-01': ('?', 1, 0, 1)}),
+            ],
+            ['2026-01']
+        )
+        mock_fetch_sheet.return_value = [{
+            'date': '2026-01-15', 'amount': 500, 'person': 'Junior One',
+            'purpose': '2026-01', 'message': '', 'sender': 'Parent', 'inferred_amount': 500,
+        }]
+        response = self.client.get('/api/juniors')
+        self.assertEqual(response.status_code, 200)
+        self.assertTrue(response.is_json)
+        data = json.loads(response.data)
+        self.assertEqual(set(data.keys()), EXPECTED_JUNIORS_KEYS)
+        self.assertIsInstance(data['member_data'], dict)
+        self.assertIsInstance(data['month_labels'], dict)
+        self.assertIsInstance(data['raw_payments'], dict)
+
+    @patch('app.get_cached_data', side_effect=_bypass_cache)
+    @patch('app.fetch_sheet_data')
+    def test_api_payments(self, mock_fetch_sheet, mock_cache):
+        """Test /api/payments returns JSON with correct top-level keys."""
+        mock_fetch_sheet.return_value = [{
+            'date': '2026-01-01', 'amount': 750, 'person': 'Test Member',
+            'purpose': '2026-01', 'message': 'test', 'sender': 'Someone',
+        }]
+        response = self.client.get('/api/payments')
+        self.assertEqual(response.status_code, 200)
+        self.assertTrue(response.is_json)
+        data = json.loads(response.data)
+        self.assertEqual(set(data.keys()), EXPECTED_PAYMENTS_KEYS)
+        self.assertIsInstance(data['grouped_payments'], dict)
+        self.assertIsInstance(data['sorted_people'], list)
+
+
 if __name__ == '__main__':
     unittest.main()