langtell 0.0.1 → 0.3.0

package/dist/profiles.d.ts

@@ -0,0 +1,47 @@
+import { b as LanguageCode, L as LanguageProfile } from './types-BIXrkuAr.js';
+
+/**
+ * `langtell/profiles` — ready-to-use {@link LanguageProfile} data.
+ *
+ * This is the heavy DATA half of the library: alphabets, curated function-word
+ * lists, and corpus-frequent word lists. It is deliberately kept behind its own
+ * subpath, OUT of the zero-dependency core, so `import { compile } from
+ * "langtell"` never drags the word corpora into a bundle that only needs the
+ * script/letter rungs. Pass these into `compile({ candidates: [...] })`.
+ *
+ * Each profile is declarative and auditable:
+ *   - `alphabet`        — the language's lowercased alphabet (raw; distinctiveness
+ *                         is computed at runtime per candidate set).
+ *   - `marks`           — orthographic marks that count as rung-1 evidence but
+ *                         are not alphabet letters (the intra-word apostrophe).
+ *   - `words.function`  — curated grammatical markers, hand-verified.
+ *   - `words.frequent`  — common everyday words from a subtitle-frequency corpus.
+ *   - `iso6393`         — ISO 639-3 code for the optional franc engine.
+ *
+ * Curation rule for `function`: a token may appear in exactly one candidate's
+ * list ONLY if that form is genuinely used by only that language among those we
+ * support. Shared forms must be in every list that uses them (set-difference
+ * then cancels them) or omitted from all. When in doubt, omit: a missing marker
+ * only costs recall.
+ */
+
+declare const uk: LanguageProfile;
+declare const ru: LanguageProfile;
+declare const be: LanguageProfile;
+declare const bg: LanguageProfile;
+declare const en: LanguageProfile;
+
+/** Registry of shipped profiles, keyed by BCP-47 code. */
+declare const PROFILES: Readonly<Record<LanguageCode, LanguageProfile>>;
+/** BCP-47 codes for which langtell ships a ready-made {@link LanguageProfile}.
+ *  Handy for narrowing a caller's roster to codes that can actually classify —
+ *  e.g. `codes.filter(hasProfile)`. Derived from {@link PROFILES}. */
+declare const PROFILED_CODES: readonly LanguageCode[];
+/** Whether langtell ships a ready-made {@link LanguageProfile} for `code`. An
+ *  own-property check, so inherited names (`"toString"`, `"constructor"`) read
+ *  as absent. */
+declare function hasProfile(code: LanguageCode): boolean;
+/** Resolve profiles for the given codes, skipping any without a shipped profile. */
+declare function getProfiles(codes: readonly LanguageCode[]): LanguageProfile[];
+
+export { PROFILED_CODES, PROFILES, be, bg, en, getProfiles, hasProfile, ru, uk };