nl2time 0.1.0 → 0.3.0

package/dist/parse/languages.d.ts

@@ -0,0 +1,17 @@
+/**
+ * Language registry: rule sets and merge connectors per BCP-47 language.
+ * Unknown languages fall back to English. Adding a language = a rules module
+ * (Latin-factory lexicon or bespoke rules for non-Latin scripts) plus its
+ * imported corpus + baseline (docs/evals.md, issue #13).
+ */
+import { type Rule } from './en.js';
+export interface LanguageDef {
+    ruleEntries: readonly {
+        name: string;
+        rule: Rule;
+    }[];
+    /** Words transparent between a date part and a time part when merging. */
+    connectors: readonly string[];
+}
+export declare function languageDef(language: string): LanguageDef;
+export declare const SUPPORTED_LANGUAGES: string[];

package/dist/parse/languages.js

@@ -0,0 +1,25 @@
+/**
+ * Language registry: rule sets and merge connectors per BCP-47 language.
+ * Unknown languages fall back to English. Adding a language = a rules module
+ * (Latin-factory lexicon or bespoke rules for non-Latin scripts) plus its
+ * imported corpus + baseline (docs/evals.md, issue #13).
+ */
+import { EN_RULE_ENTRIES } from './en.js';
+import { ES_CONNECTORS, ES_RULE_ENTRIES } from './es.js';
+import { FR_CONNECTORS, FR_RULE_ENTRIES } from './fr.js';
+import { DE_CONNECTORS, DE_RULE_ENTRIES } from './de.js';
+import { JA_CONNECTORS, JA_RULE_ENTRIES } from './ja.js';
+import { ZH_CONNECTORS, ZH_RULE_ENTRIES } from './zh.js';
+const EN_CONNECTORS = ['at', 'on', 'in', 'of', 'the', 'for', 'from', 'around', 'about', '-'];
+const REGISTRY = {
+    en: { ruleEntries: EN_RULE_ENTRIES, connectors: EN_CONNECTORS },
+    es: { ruleEntries: ES_RULE_ENTRIES, connectors: ES_CONNECTORS },
+    fr: { ruleEntries: FR_RULE_ENTRIES, connectors: FR_CONNECTORS },
+    de: { ruleEntries: DE_RULE_ENTRIES, connectors: DE_CONNECTORS },
+    ja: { ruleEntries: JA_RULE_ENTRIES, connectors: JA_CONNECTORS },
+    zh: { ruleEntries: ZH_RULE_ENTRIES, connectors: ZH_CONNECTORS },
+};
+export function languageDef(language) {
+    return REGISTRY[language] ?? REGISTRY['en'];
+}
+export const SUPPORTED_LANGUAGES = Object.keys(REGISTRY);

package/dist/parse/latin.d.ts

@@ -0,0 +1,97 @@
+/**
+ * Parameterized rule factory for Latin-script languages (Spanish, French,
+ * German, …). A LatinLexicon supplies the vocabulary; the factory supplies
+ * the grammar patterns these languages share:
+ *
+ *   deictic days        hoy / hier / übermorgen
+ *   rel + unit          la semana pasada · nächste Woche · le mois dernier
+ *   rel + weekday       el próximo martes · letzten Montag
+ *   ago / in            hace 3 días · il y a 3 jours · vor 3 Tagen · in 3 Tagen
+ *   last/next N units   los últimos 3 días · les 3 derniers jours
+ *   month-day dates     15 de marzo de 2026 · 15. März 2026 · le 15 mars
+ *   day periods         por la mañana · le matin · am Abend
+ *   clock times         a las 15:30 · à 15h30 (via extras) · um 15 Uhr
+ *   weekend             el fin de semana · le week-end · das Wochenende
+ *
+ * Modifier placement is handled on both sides of the noun (Romance languages
+ * postpose: "semana pasada"). Language-specific oddities go in `extras`,
+ * which run before the factory rules.
+ */
+import type { DayPeriod, TimeExpr, Unit, Weekday } from '../ir/types.js';
+import type { Rule } from './en.js';
+export interface LatinLexicon {
+    /** Articles / filler tokens skippable inside phrases (el, la, le, der, die…). */
+    articles: string[];
+    units: Record<string, Unit>;
+    weekdays: Record<string, Weekday>;
+    months: Record<string, number>;
+    /** Day-period nouns. Words also used deictically (Spanish "mañana") must be
+     * listed here AND in `deictic`; bare uses read as the deictic. */
+    periods: Record<string, DayPeriod>;
+    smallNumbers: Record<string, number>;
+    /** Single-word deictics: hoy 0, ayer −1, mañana +1, anteayer −2, vorgestern −2… */
+    deictic: Record<string, number>;
+    /** Multi-word deictics: "pasado mañana" +2. */
+    deicticPhrases?: {
+        words: string[];
+        delta: number;
+    }[];
+    /** Modifiers before the noun: próximo +1, este 0, letzte −1. */
+    preMods: Record<string, -1 | 0 | 1>;
+    /** Modifiers after the noun: pasada −1, dernier −1, prochain +1. */
+    postMods: Record<string, -1 | 0 | 1>;
+    /** Multi-word post-modifiers: "que viene" +1. */
+    postModPhrases?: {
+        words: string[];
+        delta: -1 | 0 | 1;
+    }[];
+    /** "ago" marker phrases that PRECEDE the amount: hace, il y a, vor. */
+    agoPrefixes: string[][];
+    /** "in/within" marker phrases: en, dans, in, dentro de. */
+    inPrefixes: string[][];
+    /** Plural adjectives for "last N units": últimos, derniers, letzten. */
+    lastNAdjs: string[];
+    /** Plural adjectives for "next N units": próximos, prochains, nächsten. */
+    nextNAdjs: string[];
+    /** Clock-time lead-ins licensing a bare hour: a las, à, um. */
+    atPhrases: string[][];
+    /** Period markers before a period noun: por la, de la, am, le. */
+    periodMarkers: string[][];
+    /** Word(s) linking day-month-year: de ("15 de marzo de 2026"). Empty for German. */
+    dateSep: string[];
+    /** Weekend noun phrase: fin de semana, week-end, wochenende. */
+    weekendPhrases: string[][];
+    /** Special single/multi-word datetimes: anoche → yesterday∩night. */
+    specialPhrases?: {
+        words: string[];
+        expr: TimeExpr;
+        role?: 'date' | 'time' | 'datetime';
+    }[];
+    /** noon / midnight words. */
+    noonWords?: string[];
+    midnightWords?: string[];
+    /** "mil" / "mille" / "tausend" for word-number years. */
+    thousandWords?: string[];
+    /** Range markers: desde/entre/de … hasta/a/y. */
+    rangeFrom?: string[][];
+    rangeTo?: string[][];
+    rangeAnd?: string[];
+    /** Duration trigger words: durante, pendant, für. */
+    durationTriggers?: string[];
+    /** Numeric-date order for numdate tokens (default 'DMY'). */
+    dateOrder?: 'DMY' | 'MDY';
+    /** Ordinal day words usable as day-of-month: primero 1, premier 1, erste 1. */
+    dayOrdinals?: Record<string, number>;
+    /** "end of" marker phrases: al final de, à la fin de, am Ende von. */
+    endOfMarkers?: string[][];
+    /** First/last adjectives for "última semana de julio" style scoped weeks. */
+    firstAdjs?: string[];
+    lastAdjs?: string[];
+}
+export declare function makeLatinRules(lex: LatinLexicon, extras?: {
+    name: string;
+    rule: Rule;
+}[]): {
+    name: string;
+    rule: Rule;
+}[];