@keymanapp/kmc-model/build/src/model-defaults.d.ts

Keyman Developer lexical model compiler

import { LexicalModelTypes } from '@keymanapp/common-types';
import CasingForm = LexicalModelTypes.CasingForm;
import CasingFunction = LexicalModelTypes.CasingFunction;
/**
 * Converts wordforms into an indexable form. It does this by
 * normalizing the letter case of characters INDIVIDUALLY (to disregard
 * context-sensitive case transformations), normalizing to NFKD form,
 * and removing common diacritical marks.
 *
 * This is a very speculative implementation, that might work with
 * your language. We don't guarantee that this will be perfect for your
 * language, but it's a start.
 *
 * This uses String.prototype.normalize() to convert normalize into NFKD.
 * NFKD neutralizes some funky distinctions, e.g., ꬲ, ｅ, e should all be the
 * same character; plus, it's an easy way to separate a Latin character from
 * its diacritics; Even then, orthographies regularly use code points
 * that, under NFKD normalization, do NOT decompose appropriately for your
 * language (e.g., SENĆOŦEN, Plains Cree in syllabics).
 *
 * Use this in early iterations of the model. For a production lexical model,
 * you will probably write/generate your own key function, tailored to your
 * language. There is a chance the default will work properly out of the box.
 */
export declare function defaultSearchTermToKey(wordform: string): string;
/**
 * Converts wordforms into an indexable form. It does this by
 * normalizing the letter case of characters INDIVIDUALLY (to disregard
 * context-sensitive case transformations), normalizing to NFKD form,
 * and removing common diacritical marks.
 *
 * This is a very speculative implementation, that might work with
 * your language. We don't guarantee that this will be perfect for your
 * language, but it's a start.
 *
 * This uses String.prototype.normalize() to convert normalize into NFKD.
 * NFKD neutralizes some funky distinctions, e.g., ꬲ, ｅ, e should all be the
 * same character; plus, it's an easy way to separate a Latin character from
 * its diacritics; Even then, orthographies regularly use code points
 * that, under NFKD normalization, do NOT decompose appropriately for your
 * language (e.g., SENĆOŦEN, Plains Cree in syllabics).
 *
 * Use this in early iterations of the model. For a production lexical model,
 * you will probably write/generate your own key function, tailored to your
 * language. There is a chance the default will work properly out of the box.
 */
export declare function defaultCasedSearchTermToKey(wordform: string, applyCasing: CasingFunction): string;
/**
 * Specifies default casing behavior for lexical models when `languageUsesCasing` is
 * set to true.
 * @param casing One of 'lower' (lowercased), 'upper' (uppercased), or 'initial'.
 *
 * 'initial' is designed to cover cases like sentence-initial & proper noun capitalization in English.
 * This may be overwritten as appropriate in model-specific implementations.
 * @param text The text to be modified.
 */
export declare function defaultApplyCasing(casing: CasingForm, text: string): string;
//# sourceMappingURL=model-defaults.d.ts.map