[shockalotti]

Closes #216.

Adds internal/i18n/locales/pl.json, so Polish joins every booker-facing surface: booking page, manage page, embed widget, the four emails and the calendar invite.

Nothing else was needed in code, which is the design: init() globs the locale directory, and the switcher takes its display name from x/text CLDR data, so the picker shows "Polski" on its own.

Provenance

The strings are kybeq's proposal from #216. Credit to him: this is the first non-English locale in the repo that did not start as an LLM draft, which matters given the README's standing caveat about machine translations.

Validation

• 169/169 keys, exactly matching en.json, no missing and no extras.
• Same-keys and printf-verb guards pass unchanged.
• TestDateTablesMatchCLDR initially failed on 19 date abbreviations, which I corrected:

key group	proposed	CLDR (Polish)
month_short_* (12)	Sty, Lut, Gru, ...	sty, lut, gru, ... (lowercase)
dow_short_* (7)	Pon, Wt, Śr, ...	pon., wt., śr., ... (lowercase + trailing dot)

date_format (%[1]s %[2]d %[3]s %[4]d) and clock_format (24h) already match Polish convention, so those are used as given.

go test ./... green.

Docs updated for the count: README, ARCHITECTURE §23, and CLAUDE.md.

Note

The other / others keys still only express singular vs plural, which cannot cover Polish paucal forms fully. That is a pre-existing limitation of the key design, not specific to this locale, so I left kybeq's wording alone rather than inventing a scheme here.

🤖 Generated with Claude Code

[pullfrog]

ℹ️ No critical issues — two translation slips inline, plus a docs reconciliation worth a look.

Reviewed changes

• New Polish locale — adds internal/i18n/locales/pl.json (169 keys, parity with en.json) covering the booking page, manage page, embed widget, the four emails, the assistant and the calendar invite.
• Docs count — bumps the shipped-locale count from 9 to 10 in README.md, docs/ARCHITECTURE.md §23 and CLAUDE.md.

ℹ️ Locale docs still contradict the change

The locale file itself is mechanically sound — key parity, printf-verb parity and CLDR date checks all pass — but the docs this PR edits still disagree with both the PR's premise and the tree. §23 still says Polish needs a plural-rule selector before it can ship correctly, still calls every non-English locale an LLM draft, and the counts/lists omit the already-shipping Greek locale. Worth reconciling in this PR so the documented state matches the code.

Technical details

# Locale documentation reconciliation

## Affected sites
- `CONTRIBUTING.md:77` — still says "ships 9 locales (`en es fr fr-CA de it pt nl sv`)" and omits `pl`; README/ARCHITECTURE/CLAUDE were updated but this one was not.
- `docs/ARCHITECTURE.md:1232` — says 10 locales and omits `el`; the tree has 11 `internal/i18n/locales/*.json` files, and `el.json` is a complete Greek locale that `init()` embeds and the matcher serves like any other. Same omission in `README.md:203-206` and `CLAUDE.md:143`.
- `docs/ARCHITECTURE.md:1332` — "Every non-English locale is an LLM draft with no native review" is no longer true for `pl` per the PR provenance.
- `docs/ARCHITECTURE.md:1342-1344` — the known-limitations section explicitly names Polish as needing a real plural-rule selector "before they can ship correctly". This PR ships it; the only affected key in practice is `other`/`others` (see the inline note), so either relax the limitation or defer the locale.

## Required outcome
- Any locale count/list in README, ARCHITECTURE §23, CLAUDE.md and CONTRIBUTING reflects the locales that actually ship, including `el` if it is meant to ship.
- §23's translation-quality and plural-limitation claims agree with shipping Polish.

## Open questions for the human
- Is `el.json` intended to ship? If so it should be listed; if not, this PR's count should not be read as authoritative.

  ^{｜ Fix all ➔ ｜ Fix 👍s ➔ ｜ View workflow run ｜ Using deepseek-v4.1-flash ｜ 𝕏}

[pullfrog]

ponowanie is not a Polish word — both this string and email_rebook_intro (line 138) need ponownie (“again”). As written, the reschedule and rebook emails show “Aby ponowanie zmienić termin…” / “Aby zarezerwować ponowanie…”.

[pullfrog]

others is always rendered after a numeral ({{n-3}} others), and Polish numerals 2+ take the genitive here — 2 innych, 5 innych, not 2 inni. inny on line 39 is fine for the singular case; innych keeps the 2-form limitation from producing ungrammatical text for every count above one.

[kybeq]

Thank you! GOAT!

[shockalotti]

Closing as superseded by #220, which merged as 0817f8f.

This was an earlier attempt at the same file and had the right CLDR dates, but it went in before the review found two things:

• email_rebook_intro / email_manage_link_intro_reschedule still said "ponowanie" (should be "ponownie").
• others was "inni"; it is "innych" — genitive plural after a numeral, confirmed by kybeq on Polish translation #216.

#220 carries those fixes plus the doc updates, so nothing here is lost. The README language-count edit from this branch is picked up in #222 instead, since it was also stale for Greek.