runk · runk · Jul 12, 2026 · Jun 26, 2026 · Jul 11, 2026 · Jul 12, 2026
diff --git a/.gitignore b/.gitignore
@@ -7,3 +7,4 @@ lib
 TODO.md
 package-lock.json
 .vscode
+BASELINE.md
diff --git a/BASELINE.md b/BASELINE.md
@@ -0,0 +1,115 @@
+# Generated model migration baseline
+
+Recorded on 2026-07-12 at commit
+`df1e8dd0088f5d4b71790a54ee2fc1f2afe6cb0a` (`v3`), using Node.js v24.10.0
+and npm 11.6.1.
+
+`TODO.md` is locally present but ignored by the repository. There were no
+tracked or unignored changes when this baseline was recorded. Generated `lib/`
+and coverage output are also ignored.
+
+## Checks
+
+| Check                             | Result                                                             |
+| --------------------------------- | ------------------------------------------------------------------ |
+| `npm test`                        | Pass: 8 files, 48 tests passed, 8 todo                             |
+| TypeScript                        | Pass                                                               |
+| Coverage                          | 93.04% statements, 85.44% branches, 95.86% functions, 93.13% lines |
+| `npm run models:verify`           | Pass; models and evaluation are reproducible                       |
+| `npm run validate -- --split=all` | Completes; 0/138 encoding and language predictions correct         |
+| `npm run benchmark`               | 7,848.13 files/sec; 1.42 MiB/sec                                   |
+| `npm run build`                   | Pass                                                               |
+| Built `lib/`                      | 341,661 bytes across files; 448 KiB allocated on disk              |
+| `npm pack --dry-run`              | 42,359 bytes packed; 348,964 bytes unpacked; 42 entries            |
+
+The benchmark processed 13,800 calls and 2,622,400 bytes in 1,758.38 ms,
+producing 100,200 matches. These figures are a local performance reference, not
+a hard CI threshold; comparisons should use the same machine and runtime.
+
+## Runtime corpus validation
+
+The validation corpus currently contains only newly generated models that are
+not registered with the runtime. The 0% result is therefore the expected
+pre-migration baseline, not a model-evaluator failure. Every input still
+received at least one prediction.
+
+### By split
+
+| Split      |   Files | Encoding correct | Language correct |
+| ---------- | ------: | ---------------: | ---------------: |
+| train      |      92 |        0 (0.00%) |        0 (0.00%) |
+| validation |      23 |        0 (0.00%) |        0 (0.00%) |
+| test       |      23 |        0 (0.00%) |        0 (0.00%) |
+| **Total**  | **138** |    **0 (0.00%)** |    **0 (0.00%)** |
+
+### By expected encoding
+
+| Expected encoding | Files | Encoding correct | Most common current prediction |
+| ----------------- | ----: | ---------------: | ------------------------------ |
+| CP850             |    18 |                0 | windows-1252                   |
+| CP852             |     6 |                0 | windows-1250                   |
+| CP949             |     6 |                0 | EUC-KR                         |
+| IBM855            |    12 |                0 | windows-1252                   |
+| IBM866            |     6 |                0 | windows-1252                   |
+| ISO-8859-10       |     6 |                0 | ISO-8859-1                     |
+| ISO-8859-13       |     6 |                0 | windows-1257                   |
+| ISO-8859-14       |     6 |                0 | ISO-8859-1                     |
+| ISO-8859-15       |    18 |                0 | ISO-8859-1                     |
+| ISO-8859-16       |     6 |                0 | ISO-8859-2                     |
+| ISO-8859-3        |     6 |                0 | ISO-8859-1                     |
+| ISO-8859-4        |     6 |                0 | windows-1257                   |
+| KOI8-U            |     6 |                0 | KOI8-R                         |
+| macintosh         |    18 |                0 | windows-1252                   |
+| x-mac-cyrillic    |    12 |                0 | windows-1251                   |
+
+The most common predictions above are useful collision targets, but do not
+describe every document. Notably, weak IBM855 input can also be classified as
+Shift_JIS or windows-1255.
+
+## Public naming decisions
+
+Existing names remain exactly as declared by `EncodingName`; the migration must
+not silently recase or rename them. The new SBCS canonical result names are:
+
+```text
+KOI8-U
+IBM866
+IBM855
+macintosh
+x-mac-cyrillic
+ISO-8859-3
+ISO-8859-4
+ISO-8859-10
+ISO-8859-13
+ISO-8859-14
+ISO-8859-15
+ISO-8859-16
+CP850
+CP852
+```
+
+CP949 is reserved as the canonical future result name for the separate
+multibyte follow-up.
+
+The current public API does not accept encoding-name filters and does not expose
+alias normalization. Consequently, the initial migration adds only these
+canonical `EncodingName` values. Iconv/compiler input labels such as `CP866`,
+`CP855`, `MACINTOSH`, and `MAC-CYRILLIC` are implementation aliases and must not
+be returned by `detect` or `analyse`.
+
+## Migration scope decision
+
+All existing SBCS recognizers will move to generated models. The legacy byte
+maps and n-gram tables in `src/encoding/sbcs.ts` will not remain as a parallel
+model system. Existing encodings must first be added to the corpus/model inputs,
+generated, and validated before their hard-coded tables are removed.
+
+## Known pre-migration limitations
+
+- Generated SBCS models are not consumed or registered by the detector.
+- New encodings therefore lose to the closest existing recognizer.
+- Confidence values are not yet calibrated between generated SBCS and existing
+  UTF, ISO-2022, or MBCS recognizers.
+- The existing test suite has eight explicitly pending tests.
+- CP949 is represented in the corpus but intentionally excluded from the SBCS
+  model compiler and requires separate MBCS work.
diff --git a/corpus/README.md b/corpus/README.md
@@ -4,8 +4,9 @@ This directory contains a deterministic seed corpus for developing additional
 encoding recognisers. The prose is synthetic and should not be treated as a
 replacement for a large, independently sourced natural-language corpus.
 
-`sources.json` stores UTF-8 originals. `manifest.json` maps languages to iconv
-encodings. Generated byte samples and statistics live under `generated/`.
+`sources.json` stores UTF-8 originals. Held-out material is kept separately in
+`test-sources.json`. `manifest.json` maps languages to iconv encodings.
+Generated byte samples and statistics live under `generated/`.
 
 Build and verify the corpus with:
 
@@ -17,8 +18,9 @@ npm run models:evaluate
 npm run models:verify
 ```
 
-The build script generates the corpus and performs an exact iconv round trip for
-every document. The check script independently rebuilds into a temporary
+The build script generates the corpus and performs an iconv round trip for every
+document, allowing canonically equivalent Unicode composition where an encoding
+such as windows-1258 stores combining marks. The check script independently rebuilds into a temporary
 directory and byte-compares the result with `generated/`. Four documents per
 language are assigned to training, one to validation, and one independent
 document to testing. Statistics are calculated from training documents only.
@@ -37,6 +39,9 @@ learns Laplace-smoothed high-byte distributions from the training split. The
 evaluator uses byte likelihood to rank candidates whose trigram hit rates are
 statistically competitive at a 95% margin. `models:evaluate` reports every
 held-out test document, while `models:verify` checks reproducibility and fails if
-any encoding or language result is incorrect. CP949 is excluded because its
+any encoding or language result is incorrect. A different encoding label is
+accepted only when both encodings decode the actual held-out bytes identically;
+the report distinguishes these byte-equivalent results from exact matches.
+CP949 is excluded because its
 multibyte recogniser requires character frequency data rather than byte
 trigrams.
diff --git a/corpus/generated/ISO-8859-1/da/test/community-garden.bin b/corpus/generated/ISO-8859-1/da/test/community-garden.bin
@@ -0,0 +1,3 @@
+F�lleshaven
+
+Naboerne gjorde en tom g�rd til en gr�n have, b�rnene malede skilte, og familierne vander planterne hver uge.
diff --git a/corpus/generated/ISO-8859-1/da/train/city-morning.bin b/corpus/generated/ISO-8859-1/da/train/city-morning.bin
@@ -0,0 +1,3 @@
+Morgen i byen
+
+Byen v�gner langsomt, caf�erne �bner, og folk venter p� de f�rste busser.
diff --git a/corpus/generated/ISO-8859-1/da/train/library.bin b/corpus/generated/ISO-8859-1/da/train/library.bin
@@ -0,0 +1,3 @@
+Det stille bibliotek
+
+L�sere finder b�ger om historie og videnskab, mens bibliotekaren hj�lper eleverne.
diff --git a/corpus/generated/ISO-8859-1/da/train/market.bin b/corpus/generated/ISO-8859-1/da/train/market.bin
@@ -0,0 +1,3 @@
+L�rdagsmarked
+
+B�nder s�lger frisk br�d, �bler, ost og honning p� den travle plads.
diff --git a/corpus/generated/ISO-8859-1/da/train/workshop.bin b/corpus/generated/ISO-8859-1/da/train/workshop.bin
@@ -0,0 +1,3 @@
+Digitalt v�rksted
+
+Et lille hold bygger en museumsguide, unders�ger koden og skriver tydelige tekster.
diff --git a/corpus/generated/ISO-8859-1/da/validation/river-trip.bin b/corpus/generated/ISO-8859-1/da/validation/river-trip.bin
@@ -0,0 +1,3 @@
+Tur til floden
+
+Vennerne tog et kort og varm te med p� en s�ndag og hvilede bagefter ved floden.
diff --git a/corpus/generated/ISO-8859-1/de/test/community-garden.bin b/corpus/generated/ISO-8859-1/de/test/community-garden.bin
@@ -0,0 +1,3 @@
+Der Gemeinschaftsgarten
+
+Die Nachbarn haben einen leeren Hof begr�nt und dort Kr�uter, Gem�se und junge B�ume gepflanzt. Die Sch�ler malten bunte Schilder, w�hrend �ltere Bewohner regelm��ig gie�en und Ger�te aufr�umen.
diff --git a/corpus/generated/ISO-8859-1/de/train/city-morning.bin b/corpus/generated/ISO-8859-1/de/train/city-morning.bin
@@ -0,0 +1,3 @@
+Morgen in der Stadt
+
+Die Stadt erwacht langsam. B�ckereien �ffnen ihre T�ren, Busse beginnen ihre Fahrt und viele Menschen lesen unterwegs die Nachrichten. �ber den D�chern wird der Himmel allm�hlich heller.
diff --git a/corpus/generated/ISO-8859-1/de/train/library.bin b/corpus/generated/ISO-8859-1/de/train/library.bin
@@ -0,0 +1,3 @@
+Die ruhige Bibliothek
+
+In der kleinen Bibliothek rascheln leise die Seiten. Besucher suchen B�cher �ber Geschichte, Technik und ferne L�nder. Eine Mitarbeiterin bereitet eine Ausstellung f�r Sch�ler vor.
diff --git a/corpus/generated/ISO-8859-1/de/train/market.bin b/corpus/generated/ISO-8859-1/de/train/market.bin
@@ -0,0 +1,3 @@
+Markt am Samstag
+
+Auf dem Platz bieten Bauern �pfel, K�se, Honig und frisches Brot an. Die Kunden vergleichen Gem�se, sprechen �ber die Ernte und tauschen einfache Rezepte aus.
diff --git a/corpus/generated/ISO-8859-1/de/train/workshop.bin b/corpus/generated/ISO-8859-1/de/train/workshop.bin
@@ -0,0 +1,3 @@
+Digitale Werkstatt
+
+Ein kleines Team entwickelt einen digitalen Museumsf�hrer. Designer ordnen das Men�, Entwickler pr�fen den Code und Redakteure schreiben verst�ndliche Texte f�r G�ste.
diff --git a/corpus/generated/ISO-8859-1/de/validation/river-trip.bin b/corpus/generated/ISO-8859-1/de/validation/river-trip.bin
@@ -0,0 +1,3 @@
+Ausflug an den Fluss
+
+Am Sonntag wandern Freunde zu einem ruhigen Fluss. Sie nehmen eine Karte, hei�en Tee und eine Kamera mit. Nach dem langen Weg sitzen sie am Ufer und h�ren den V�geln zu.
diff --git a/corpus/generated/ISO-8859-1/en/test/community-garden.bin b/corpus/generated/ISO-8859-1/en/test/community-garden.bin
@@ -0,0 +1,3 @@
+The community garden
+
+Neighbours turned an empty courtyard into a garden with flowers, vegetables, young trees, and a small caf� table.
diff --git a/corpus/generated/ISO-8859-1/en/train/city-morning.bin b/corpus/generated/ISO-8859-1/en/train/city-morning.bin
@@ -0,0 +1,3 @@
+Morning in the city
+
+The city wakes slowly; caf�s open their doors and commuters read news beside a fa�ade.
diff --git a/corpus/generated/ISO-8859-1/en/train/library.bin b/corpus/generated/ISO-8859-1/en/train/library.bin
@@ -0,0 +1,3 @@
+The quiet library
+
+Readers browse books about history, science, and na�ve art while the librarian prepares a r�sum�.
diff --git a/corpus/generated/ISO-8859-1/en/train/market.bin b/corpus/generated/ISO-8859-1/en/train/market.bin
@@ -0,0 +1,3 @@
+Saturday market
+
+Farmers sell bread, apples, jalape�os, and p�t� while neighbours exchange recipes.
diff --git a/corpus/generated/ISO-8859-1/en/train/workshop.bin b/corpus/generated/ISO-8859-1/en/train/workshop.bin
@@ -0,0 +1,3 @@
+Digital workshop
+
+A small team designs the museum guide and discusses its r�le, code, and clear descriptions.
diff --git a/corpus/generated/ISO-8859-1/en/validation/river-trip.bin b/corpus/generated/ISO-8859-1/en/validation/river-trip.bin
@@ -0,0 +1,3 @@
+Journey to the river
+
+Friends carried a map and a thermos, walked past the caf�, and rested by the river.
diff --git a/corpus/generated/ISO-8859-1/es/test/community-garden.bin b/corpus/generated/ISO-8859-1/es/test/community-garden.bin
@@ -0,0 +1,3 @@
+El jard�n comunitario
+
+Los vecinos convirtieron un patio vac�o en un jard�n lleno de �rboles, hierbas y verduras. Los ni�os pintaron carteles de colores y cada s�bado varias familias riegan las plantas.
diff --git a/corpus/generated/ISO-8859-1/es/train/city-morning.bin b/corpus/generated/ISO-8859-1/es/train/city-morning.bin
@@ -0,0 +1,3 @@
+La ma�ana en la ciudad
+
+La ciudad despierta poco a poco. Las cafeter�as abren sus puertas, los autobuses inician la ruta y los vecinos leen las noticias. El cielo se vuelve m�s claro sobre los tejados.
diff --git a/corpus/generated/ISO-8859-1/es/train/library.bin b/corpus/generated/ISO-8859-1/es/train/library.bin
@@ -0,0 +1,3 @@
+Una biblioteca tranquila
+
+En la biblioteca del barrio se oyen p�ginas y pasos suaves. Los lectores buscan libros de historia, ciencia y viajes. La bibliotecaria prepara una exposici�n para los alumnos.
diff --git a/corpus/generated/ISO-8859-1/es/train/market.bin b/corpus/generated/ISO-8859-1/es/train/market.bin
@@ -0,0 +1,3 @@
+El mercado del s�bado
+
+El mercado llena la plaza de colores. Los agricultores venden manzanas, miel, queso y pan reci�n hecho. Los clientes comparan las verduras y comparten recetas de sus familias.
diff --git a/corpus/generated/ISO-8859-1/es/train/workshop.bin b/corpus/generated/ISO-8859-1/es/train/workshop.bin
@@ -0,0 +1,3 @@
+Taller digital
+
+Un equipo peque�o crea el cat�logo digital del museo. Los dise�adores ordenan el men�, los programadores revisan el c�digo y los editores escriben textos claros para el p�blico.
diff --git a/corpus/generated/ISO-8859-1/es/validation/river-trip.bin b/corpus/generated/ISO-8859-1/es/validation/river-trip.bin
@@ -0,0 +1,3 @@
+Paseo junto al r�o
+
+El domingo unos amigos caminan hacia un r�o tranquilo. Llevan un mapa, t� caliente y una c�mara. Despu�s del paseo descansan en la orilla y escuchan a los p�jaros.
diff --git a/corpus/generated/ISO-8859-1/fr/test/community-garden.bin b/corpus/generated/ISO-8859-1/fr/test/community-garden.bin
@@ -0,0 +1,3 @@
+Le jardin partag�
+
+Les habitants ont transform� une cour oubli�e en jardin fleuri. Les enfants ont pr�par� des �tiquettes color�es, tandis que les voisins se r�unissent chaque semaine pour arroser les l�gumes et les jeunes arbres.
diff --git a/corpus/generated/ISO-8859-1/fr/train/city-morning.bin b/corpus/generated/ISO-8859-1/fr/train/city-morning.bin
@@ -0,0 +1,3 @@
+Le matin en ville
+
+La ville se r�veille lentement. Les caf�s ouvrent leurs portes, les autobus commencent leur trajet et les passants consultent les nouvelles. Chacun pr�pare sa journ�e sous un ciel encore p�le.
diff --git a/corpus/generated/ISO-8859-1/fr/train/library.bin b/corpus/generated/ISO-8859-1/fr/train/library.bin
@@ -0,0 +1,3 @@
+La biblioth�que calme
+
+Dans la biblioth�que du quartier, les pages tournent en silence. Les lecteurs cherchent des livres sur l'histoire, les sciences et les voyages. Une biblioth�caire pr�pare une exposition pour les �l�ves.
diff --git a/corpus/generated/ISO-8859-1/fr/train/market.bin b/corpus/generated/ISO-8859-1/fr/train/market.bin
@@ -0,0 +1,3 @@
+Le march� du samedi
+
+Le march� remplit la place de couleurs et de parfums. Les producteurs proposent des pommes, du miel, du fromage et du pain frais. Les clients discutent des saisons et �changent des recettes familiales.
diff --git a/corpus/generated/ISO-8859-1/fr/train/workshop.bin b/corpus/generated/ISO-8859-1/fr/train/workshop.bin
@@ -0,0 +1,3 @@
+Un atelier num�rique
+
+Une petite �quipe construit le catalogue num�rique du mus�e. Les graphistes organisent le menu, les d�veloppeurs v�rifient le code et les r�dacteurs pr�parent des textes clairs pour le public.
diff --git a/corpus/generated/ISO-8859-1/fr/validation/river-trip.bin b/corpus/generated/ISO-8859-1/fr/validation/river-trip.bin
@@ -0,0 +1,3 @@
+Une promenade pr�s de la rivi�re
+
+Dimanche, des amis partent vers une rivi�re tranquille. Ils emportent une carte, du th� chaud et un appareil photo. Apr�s une longue marche, ils se reposent sur la rive et �coutent les oiseaux.
diff --git a/corpus/generated/ISO-8859-1/it/test/community-garden.bin b/corpus/generated/ISO-8859-1/it/test/community-garden.bin
@@ -0,0 +1,3 @@
+Il giardino comune
+
+I vicini hanno trasformato un cortile vuoto in un giardino della citt�, i bambini hanno dipinto cartelli e le famiglie annaffiano le piante.
diff --git a/corpus/generated/ISO-8859-1/it/train/city-morning.bin b/corpus/generated/ISO-8859-1/it/train/city-morning.bin
@@ -0,0 +1,3 @@
+Mattina in citt�
+
+La citt� si sveglia lentamente, i caff� aprono e la gente aspetta i primi autobus.
diff --git a/corpus/generated/ISO-8859-1/it/train/library.bin b/corpus/generated/ISO-8859-1/it/train/library.bin
@@ -0,0 +1,3 @@
+La biblioteca tranquilla
+
+I lettori cercano pi� libri di storia, geografia e scienza mentre la bibliotecaria aiuta gli studenti pi� giovani.
diff --git a/corpus/generated/ISO-8859-1/it/train/market.bin b/corpus/generated/ISO-8859-1/it/train/market.bin
@@ -0,0 +1,3 @@
+Mercato del sabato
+
+Gli agricoltori vendono pane fresco, caff�, mele, formaggio e miele nella piazza della citt�.
diff --git a/corpus/generated/ISO-8859-1/it/train/workshop.bin b/corpus/generated/ISO-8859-1/it/train/workshop.bin
@@ -0,0 +1,3 @@
+Laboratorio digitale
+
+Una piccola squadra prepara la guida del museo, controlla il codice e scrive descrizioni chiare perch� siano utili a pi� persone.
diff --git a/corpus/generated/ISO-8859-1/it/validation/river-trip.bin b/corpus/generated/ISO-8859-1/it/validation/river-trip.bin
@@ -0,0 +1,3 @@
+Gita al fiume
+
+Gli amici portano una mappa e t� caldo, camminano nel bosco e riposano sulla riva.
diff --git a/corpus/generated/ISO-8859-1/nl/test/community-garden.bin b/corpus/generated/ISO-8859-1/nl/test/community-garden.bin
@@ -0,0 +1,3 @@
+De buurttuin
+
+Buren veranderden een lege binnenplaats naast een caf� in een groene tuin, kinderen schilderden borden en gezinnen geven de planten water.
diff --git a/corpus/generated/ISO-8859-1/nl/train/city-morning.bin b/corpus/generated/ISO-8859-1/nl/train/city-morning.bin
@@ -0,0 +1,3 @@
+Ochtend in de stad
+
+De stad ontwaakt langzaam, caf�s openen hun deuren en mensen wachten op de eerste bussen.
diff --git a/corpus/generated/ISO-8859-1/nl/train/library.bin b/corpus/generated/ISO-8859-1/nl/train/library.bin
@@ -0,0 +1,3 @@
+De stille bibliotheek
+
+Lezers zoeken boeken over geschiedenis en wetenschap terwijl de bibliothecaris ge�nteresseerde leerlingen en ��n leraar helpt.
diff --git a/corpus/generated/ISO-8859-1/nl/train/market.bin b/corpus/generated/ISO-8859-1/nl/train/market.bin
@@ -0,0 +1,3 @@
+Zaterdagmarkt
+
+Boeren verkopen vers brood, appels, kaas en honing naast twee caf�s op het drukke plein.
diff --git a/corpus/generated/ISO-8859-1/nl/train/workshop.bin b/corpus/generated/ISO-8859-1/nl/train/workshop.bin
@@ -0,0 +1,3 @@
+Digitale werkplaats
+
+Een klein team maakt een museumgids, controleert de code en schrijft duidelijke idee�n over po�zie en teksten.
diff --git a/corpus/generated/ISO-8859-1/nl/validation/river-trip.bin b/corpus/generated/ISO-8859-1/nl/validation/river-trip.bin
@@ -0,0 +1,3 @@
+Reis naar de rivier
+
+Vrienden namen een kaart en warme thee mee en rustten na ��n lange wandeling bij de rivier.
diff --git a/corpus/generated/ISO-8859-1/no/test/community-garden.bin b/corpus/generated/ISO-8859-1/no/test/community-garden.bin
@@ -0,0 +1,3 @@
+Felleshagen
+
+Naboene gjorde en tom g�rd til en gr�nn hage, barna malte skilt og familiene vanner plantene hver uke.
diff --git a/corpus/generated/ISO-8859-1/no/train/city-morning.bin b/corpus/generated/ISO-8859-1/no/train/city-morning.bin
@@ -0,0 +1,3 @@
+Morgen i byen
+
+Byen v�kner langsomt, kafeene �pner d�rene og folk venter p� de f�rste bussene n�r bryggen og sj�en.
diff --git a/corpus/generated/ISO-8859-1/no/train/library.bin b/corpus/generated/ISO-8859-1/no/train/library.bin
@@ -0,0 +1,3 @@
+Det stille biblioteket
+
+Lesere finner b�ker om historie, �rlige fortellinger og vitenskap mens bibliotekaren hjelper elevene.
diff --git a/corpus/generated/ISO-8859-1/no/train/market.bin b/corpus/generated/ISO-8859-1/no/train/market.bin
@@ -0,0 +1,3 @@
+L�rdagsmarked
+
+B�nder selger ferskt br�d, epler, ost og honning p� det travle torget.
diff --git a/corpus/generated/ISO-8859-1/no/train/workshop.bin b/corpus/generated/ISO-8859-1/no/train/workshop.bin
@@ -0,0 +1,3 @@
+Digitalt verksted
+
+Et lite lag lager en museumsguide, unders�ker koden og skriver tydelige tekster.
diff --git a/corpus/generated/ISO-8859-1/no/validation/river-trip.bin b/corpus/generated/ISO-8859-1/no/validation/river-trip.bin
@@ -0,0 +1,3 @@
+Tur til elven
+
+Vennene tok med kart og varm te p� s�ndag og hvilte senere ved elven.
diff --git a/corpus/generated/ISO-8859-1/pt/test/community-garden.bin b/corpus/generated/ISO-8859-1/pt/test/community-garden.bin
@@ -0,0 +1,3 @@
+A horta comunit�ria
+
+Os vizinhos transformaram um p�tio vazio em jardim, as crian�as pintaram placas e as fam�lias regam as plantas toda semana.
diff --git a/corpus/generated/ISO-8859-1/pt/train/city-morning.bin b/corpus/generated/ISO-8859-1/pt/train/city-morning.bin
@@ -0,0 +1,3 @@
+Manh� na cidade
+
+A cidade acorda devagar, os caf�s abrem e as pessoas esperam os primeiros �nibus.
diff --git a/corpus/generated/ISO-8859-1/pt/train/library.bin b/corpus/generated/ISO-8859-1/pt/train/library.bin
@@ -0,0 +1,3 @@
+A biblioteca tranquila
+
+Os leitores procuram livros de hist�ria e ci�ncia enquanto a bibliotec�ria ajuda os alunos.
diff --git a/corpus/generated/ISO-8859-1/pt/train/market.bin b/corpus/generated/ISO-8859-1/pt/train/market.bin
@@ -0,0 +1,3 @@
+Mercado de s�bado
+
+Os agricultores vendem p�o fresco, ma��s, queijo e mel na pra�a movimentada.
diff --git a/corpus/generated/ISO-8859-1/pt/train/workshop.bin b/corpus/generated/ISO-8859-1/pt/train/workshop.bin
@@ -0,0 +1,3 @@
+Oficina digital
+
+Uma pequena equipe prepara o guia do museu, verifica o c�digo e escreve descri��es claras.
diff --git a/corpus/generated/ISO-8859-1/pt/validation/river-trip.bin b/corpus/generated/ISO-8859-1/pt/validation/river-trip.bin
@@ -0,0 +1,3 @@
+Viagem ao rio
+
+Os amigos levaram um mapa e ch� quente, caminharam pela mata e descansaram junto ao rio.
diff --git a/corpus/generated/ISO-8859-1/sv/test/community-garden.bin b/corpus/generated/ISO-8859-1/sv/test/community-garden.bin
@@ -0,0 +1,3 @@
+Gemensam tr�dg�rd
+
+Grannarna gjorde en tom g�rd till en gr�n tr�dg�rd, barnen m�lade skyltar och familjerna vattnar v�xterna varje vecka.
diff --git a/corpus/generated/ISO-8859-1/sv/train/city-morning.bin b/corpus/generated/ISO-8859-1/sv/train/city-morning.bin
@@ -0,0 +1,3 @@
+Morgon i staden
+
+Staden vaknar l�ngsamt, kaf�erna �ppnar och m�nniskor v�ntar p� de f�rsta bussarna.
diff --git a/corpus/generated/ISO-8859-1/sv/train/library.bin b/corpus/generated/ISO-8859-1/sv/train/library.bin
@@ -0,0 +1,3 @@
+Det tysta biblioteket
+
+L�sare s�ker b�cker om historia och vetenskap medan bibliotekarien hj�lper eleverna.
diff --git a/corpus/generated/ISO-8859-1/sv/train/market.bin b/corpus/generated/ISO-8859-1/sv/train/market.bin
@@ -0,0 +1,3 @@
+L�rdagsmarknad
+
+B�nder s�ljer f�rskt br�d, �pplen, ost och honung p� det livliga torget.
diff --git a/corpus/generated/ISO-8859-1/sv/train/workshop.bin b/corpus/generated/ISO-8859-1/sv/train/workshop.bin
@@ -0,0 +1,3 @@
+Digital verkstad
+
+Ett litet lag skapar en museiguide, granskar koden och skriver tydliga texter f�r bes�kare.
diff --git a/corpus/generated/ISO-8859-1/sv/validation/river-trip.bin b/corpus/generated/ISO-8859-1/sv/validation/river-trip.bin
@@ -0,0 +1,3 @@
+Utflykt till floden
+
+V�nnerna tog med karta och varmt te och vilade sedan vid floden.
-Original file line number
+Diff line change
@@ Expand Up / @@ -7,3 +7,4 @@ lib @@
     TODO.md
     package-lock.json
     .vscode
+    BASELINE.md
Original file line number	Diff line number	Diff line change
		@@ -0,0 +1,3 @@
		F�lleshaven

		Naboerne gjorde en tom g�rd til en gr�n have, b�rnene malede skilte, og familierne vander planterne hver uge.
Original file line number	Diff line number	Diff line change
		@@ -0,0 +1,3 @@
		Morgen i byen

		Byen v�gner langsomt, caf�erne �bner, og folk venter p� de f�rste busser.
Original file line number	Diff line number	Diff line change
		@@ -0,0 +1,3 @@
		Det stille bibliotek

		L�sere finder b�ger om historie og videnskab, mens bibliotekaren hj�lper eleverne.
Original file line number	Diff line number	Diff line change
		@@ -0,0 +1,3 @@
		L�rdagsmarked

		B�nder s�lger frisk br�d, �bler, ost og honning p� den travle plads.
Original file line number	Diff line number	Diff line change
		@@ -0,0 +1,3 @@
		Digitalt v�rksted

		Et lille hold bygger en museumsguide, unders�ger koden og skriver tydelige tekster.
Original file line number	Diff line number	Diff line change
		@@ -0,0 +1,3 @@
		Tur til floden

		Vennerne tog et kort og varm te med p� en s�ndag og hvilede bagefter ved floden.
Original file line number	Diff line number	Diff line change
		@@ -0,0 +1,3 @@
		Der Gemeinschaftsgarten

		Die Nachbarn haben einen leeren Hof begr�nt und dort Kr�uter, Gem�se und junge B�ume gepflanzt. Die Sch�ler malten bunte Schilder, w�hrend �ltere Bewohner regelm��ig gie�en und Ger�te aufr�umen.
Original file line number	Diff line number	Diff line change
		@@ -0,0 +1,3 @@
		Morgen in der Stadt

		Die Stadt erwacht langsam. B�ckereien �ffnen ihre T�ren, Busse beginnen ihre Fahrt und viele Menschen lesen unterwegs die Nachrichten. �ber den D�chern wird der Himmel allm�hlich heller.
Original file line number	Diff line number	Diff line change
		@@ -0,0 +1,3 @@
		Die ruhige Bibliothek

		In der kleinen Bibliothek rascheln leise die Seiten. Besucher suchen B�cher �ber Geschichte, Technik und ferne L�nder. Eine Mitarbeiterin bereitet eine Ausstellung f�r Sch�ler vor.