README

Versioned English language rules and a legally redistributable development lexicon for Crosseno.

The package requires PHP ^8.5, Intl, mbstring, and the Crosseno core, lexicon, SQLite catalog, and solver-index packages.

use Crosseno\Core\ResourceLimits;
use Crosseno\LanguageEn\EnglishLanguagePack;

$pack = EnglishLanguagePack::load(ResourceLimits::standard());
$gridText = $pack->normalizer()->normalize('ice cream'); // ICECREAM
$cells = $pack->tokenizer()->tokenize($gridText);

// Full runtime composition, without parsing artifacts in the application:
$catalog = $pack->catalog();
$solverIndex = $pack->solverIndex();
$ordinalKeys = $pack->answerKeysByOrdinal();
$identity = $pack->runtimeIdentity();
$coverage = $pack->clueCoverage();

The checked-in artifacts are under resources/: manifest.json, catalog.sqlite, and solver.idx. profile.json is the normative profile publication. It pins NFC and locale-independent Unicode uppercase behavior to Unicode 15.1, removes Unicode separator and punctuation categories from grid text, and assigns one extended grapheme cluster per cell. It does not transliterate. Straight and curly apostrophes, hyphens, spaces, and other punctuation remain in display text but do not occupy cells.

Dialect tags are canonical BCP 47 tags. Selection uses RFC 4647 lookup, so an exact region is preferred and may fall back to en; it never falls sideways from one region to another. Thus theater is en-US, while theatre is available for en-GB, en-CA, and en-AU.

The code is MIT-licensed. The native source data is independently dedicated under CC0-1.0; its source and technical transformation inventory is in resources/source-inventory.json. The batch plan and editorial status are recorded in resources/source/batch-plan.json and resources/review-ledger.json. Automated draft measurements and known clue quality are in resources/draft-quality-report.json; the multi-request generation matrix is in resources/benchmark-report.json. Dataset size and update decisions are in resources/pack-policy.json.

Editorial disclosure

Definitions and clue candidates were initially drafted with generative-AI assistance. Crosseno reviewed, edited, approved, and assumes editorial responsibility for the published content.

Data readiness

Readiness level: expanded development draft. Data version 2026.07.3 contains 1,000 compiler-accepted answers, 1,000 senses, and 2,000 English draft clues. Every sense has exactly two clues, and all answers advertise en clue coverage. Crosseno completed editorial review of the complete batch on 2026-07-26.

This pack is intended for deterministic integration, API examples, and small-grid benchmark work. Its answers span 3 through 10 cells. Every answer has an individually authored sense definition and two meaningfully distinct semantic clue candidates. Crosseno has accepted the content for the development release; final contractual, legal, packaging, and release checks remain.

The next gate is final release review and publication of the 1,000-answer development pack. Preview targets are 25,000 answers with two approved clues for at least 90% of included senses and representative dialect coverage. These are acceptance gates, not claims about the current pack.

See runtime composition and data readiness. The executable builder proof is crosseno/builder/examples/generate-english.php.

Build and verify deterministically with:

composer install
composer source:build
composer artifacts:build
composer check

crosseno / language-en

Maintainers

Package info

Statistics

Security

README

Editorial disclosure

Data readiness