i18n: a missing translation fails no check; cargo xtask i18n-report lists them (#1228)

docs/localization-parity.md

@@ -1,6 +1,6 @@
 # Localization parity
 
-> **Last reviewed:** 2026-10-10 · **Last updated:** 2026-10-17 · **Change:** minor (Russian and Turkish catalogs reviewed by native speakers, Russian one/few/many plurals; Czech and Indonesian drafts; the Japanese draft completed; counts re-measured; Windows UI language from the first preferred language; draft catalogs catch up: every draft translates every string again; Simplified Chinese draft catalog; macOS finds its downloadable CJK system fonts (#406); string counts re-measured: 2,272 entries, the drafts at 95–99%; Swedish command labels and menu wordings; Russian, Swedish, Turkish, Czech and Indonesian listed; a duplicated paragraph removed; draft catalogs catch up: every draft translates every string again; Simplified Chinese draft catalog) · **Target:** Figma Design (desktop app 126.10.7)
+> **Last reviewed:** 2026-10-10 · **Last updated:** 2026-10-23 · **Change:** minor (a missing translation fails no check: `cargo xtask i18n-report` lists them; Russian and Turkish catalogs reviewed by native speakers, Russian one/few/many plurals; Czech and Indonesian drafts; the Japanese draft completed; counts re-measured; Windows UI language from the first preferred language; draft catalogs catch up: every draft translates every string again; Simplified Chinese draft catalog; macOS finds its downloadable CJK system fonts (#406); string counts re-measured: 2,272 entries, the drafts at 95–99%; Swedish command labels and menu wordings; Russian, Swedish, Turkish, Czech and Indonesian listed; a duplicated paragraph removed; draft catalogs catch up: every draft translates every string again; Simplified Chinese draft catalog) · **Target:** Figma Design (desktop app 126.10.7)
 
 FigCraft's interface has a localization layer with an English catalog, a Russian catalog reviewed
 by a native speaker, and Spanish (with a Latin American overlay), French, German, Brazilian
@@ -46,11 +46,11 @@ supported by both apps (CJK, Arabic and Devanagari shaping, IME); Figma has no v
   (`es-419.toml` over `es.toml`); its UI strings, command labels and menu wordings read the
   overlay, then the base language, then English.
 - **Check:** `cargo test -p figcraft-ui-egui i18n` fails on keys the code uses that `en.toml`
-  lacks, keys no code uses, unknown command ids, placeholders that differ from English, overlay
-  entries that read as in their base language and, for complete languages (Spanish), anything
-  untranslated. The other catalogs are not held to it: a string added later shows English there
-  until someone translates it. Every draft translated every string on 2026-10-12; by 2026-10-17
-  the strings added since left them at 95–99% (table below).
+  lacks, keys no code uses, unknown command ids, placeholders that differ from English and overlay
+  entries that read as in their base language. No language is held to full coverage: a string
+  added later shows English until someone translates it, and `cargo xtask i18n-report` (part of
+  `cargo xtask ci`) lists what each language lacks. Every draft translated every string on
+  2026-10-12; by 2026-10-17 the strings added since left them at 95–99% (table below).
 - **Not yet localized:** messages the engine writes (errors, a few toasts, Dev Mode property values
   such as "inside"), number formatting (decimal points stay `.`), and decimal plural rules
   (`trn` takes integers: Czech uses one/few/other and Russian one/few/many; French and Brazilian
@@ -122,6 +122,7 @@ Hindi, Vietnamese and Arabic. Traditional Chinese is not planned (no catalog; it
 
 | Date | Change | Summary |
 |---|---|---|
+| 2026-10-23 | minor | A missing translation fails no check: Spanish is no longer held to full coverage, and the tests of one language's wording, names and plural forms are gone (the plural rules and language-tag tests stay). `cargo xtask i18n-report`, run by `cargo xtask ci`, lists each language's missing UI strings, command labels and menu wordings. Grade unchanged |
 | 2026-10-15 | minor | Community catalogs (#257, #263, #469, #477, #288): Russian (`ru`, Русский, reviewed and approved by VanTan) and Turkish (`tr`, Türkçe, reviewed by a native speaker), Czech (`cs`, Čeština) and Indonesian (`id`, Bahasa Indonesia) drafts, and the Japanese draft caught up; Preferences ▸ Language lists them and System maps their regional tags. Catalogs may give few and many plural forms; Russian picks one, few or many by its integer rule (11–14 take many), the catalog supplying the text choosing the rule. Only Spanish is held to completeness. Counts re-measured: 1,587 UI keys (38 plural), 559 command labels and 31 menu wordings, 2,177 entries; Spanish 100%, Japanese 2,176, Czech and Indonesian 2,166, Russian and Turkish 2,165, the five older drafts 2,088–2,089 (95%). Localization 39% and 29–51 h unchanged at this rounding |
 | 2026-10-17 | minor | Chinese showed as boxes on macOS (#406): the font scan also walks the system's downloaded font assets (`/System/Library/AssetsV2` and the older `/System/Library/Assets`, only their `com_apple_MobileAsset_Font*` folders), where current versions keep PingFang SC, and Hiragino Sans GB and Heiti SC join the Chinese UI faces and the text fallbacks. Tested with a folder laid out the same way; not yet run on a Mac |
 | 2026-10-17 | minor | String counts re-measured from the catalogs: 1,680 UI keys, 561 command labels and 31 menu wordings (2,272 entries); Spanish complete, the other drafts at 95.3–99.2% (strings added since 2026-10-12 show English), es-419 101 entries. Indonesian has a draft (98.8%; it was listed as none); Russian (98.8%, reviewed), Swedish (99.6%, command labels and menu wordings from #599), Turkish (98.7%) and Czech (98.8%) listed with counts; the language picker's list updated; a paragraph repeated by a merge removed. Grade and hours unchanged (Indonesian's draft is offset by the drafts' new gaps) |

docs/translating.md

@@ -91,13 +91,9 @@ picker leaves Georgian out. A new CJK or other non-Latin language also needs its
    three forms. Czech's CLDR `many` applies only to decimals; `trn` takes integer counts.
    Polish uses `one` for 1 only, `few` like Russian (2–4, 22–24… except 12–14) and `many` for
    the other integers, including 0 and 21 (e.g. `{n} warstwa`, `{n} warstwy`, `{n} warstw`).
-4. To require full coverage, add the code to `COMPLETE` in
-   [`crates/ui-egui/src/i18n_tests.rs`](../crates/ui-egui/src/i18n_tests.rs). Every later PR that
-   adds a string then has to translate it into each complete language, so only Spanish is there:
-   French, German, Brazilian Portuguese, Japanese, Korean, Simplified Chinese, Czech, Indonesian and
-   Georgian are registered as drafts, Russian and Turkish are reviewed but not required, and a string added
-   later shows English in them until it is translated. To list what a draft lacks,
-   add its code to `COMPLETE` locally and run the check.
+4. A string added later shows English in every language until someone translates it; a missing
+   translation never fails a check or a build. `cargo xtask i18n-report` lists what each language
+   lacks (UI strings, command labels, menu wordings), and `cargo xtask ci` prints the same list.
 5. Run the check and look at the panels in the app (longer words must fit the right panel):
 
        cargo test -p figcraft-ui-egui i18n
@@ -106,9 +102,9 @@ picker leaves Georgian out. A new CJK or other non-Latin language also needs its
    with "…" and show it in full on hover; elsewhere, look for clipped or wrapped text.
 
 The check fails on keys the code uses that `en.toml` lacks (it prints the lines to paste), keys no
-code uses, unknown command ids, placeholders that don't match English, and, for complete languages,
-anything untranslated. Machine-drafted catalogs stay marked as drafts until a native speaker has
-reviewed them; a reviewed catalog stays a draft until a maintainer approves it (its `status` says which).
+code uses, unknown command ids and placeholders that don't match English. Machine-drafted catalogs
+stay marked as drafts until a native speaker has reviewed them; a reviewed catalog stays a draft
+until a maintainer approves it (its `status` says which).
 
 ## Adding strings in code
 
@@ -116,11 +112,11 @@ Wrap every string a user sees or a screen reader hears: `tr` for plain text, `tr
 placeholders, `trn` for counts, `tr_noop` for literals in `const` tables that are looked up later
 with `tr(label)`. The key must be a string literal as the first argument (the check reads the
 source). Don't wrap ids, command ids, JSON, log messages, file extensions or document content. Then
-add the key to `en.toml` and to each complete language (Spanish and Russian); the drafts may follow later.
+add the key to `en.toml`; translations may follow later.
 
 Labels from the engine and document crates (tool names, blend modes, easing curves…) are looked up
 at display with `tr(x.label())`; they are listed in `dynamic_labels` in `i18n_tests.rs`.
 
 The importers' fixed notes (the string literals passed to `warn` in `crates/format/src/fig` and
 `crates/format/src/rest`) are keys too: the import report shows them in the UI language, so a new
-note goes into `en.toml` and each complete language. Notes built with `format!` show in English.
+note goes into `en.toml`. Notes built with `format!` show in English.