lao-word-splitter/dist/LaoWordSplitter.d.ts

A utility to split Lao language sentences into individual words

/**
 * Splits Lao text into individual words.
 * A class for splitting Lao text into individual words.
 * This class handles various Lao language specific rules for word segmentation,
 * including consonant clusters, leading vowels, and special character sequences.
 */
export declare class LaoWordSplitter {
    private readonly LAO_CONSONANTS;
    private readonly LEADING_VOWELS;
    private readonly MIDDLE_CHARS;
    private readonly DIGRAPH_FOLLOWERS;
    private readonly MAI_YAMOK;
    /**
     * Checks if a character is part of Lao script
     *
     * Examples:
     * isLaoCharacter('ກ') => true
     * isLaoCharacter('a') => false
     * isLaoCharacter('1') => false
     */
    private isLaoCharacter;
    /**
     * Helper function to add a word to the result array and start a new word
     *
     * Examples:
     * addWordToResult('ປະ', words, 'ເ') => 'ເ' and words = ['ປະ']
     * addWordToResult('ລາວ', words, 'ເ') => 'ເ' and words = ['ປະ', 'ລາວ']
     */
    private addWordToResult;
    /**
     * Helper function to handle consonant + 'ວ' sequences
     *
     * Examples:
     * "ຈົນກວ່າຈະ" when encountering '່' after 'ກວ':
     * handleConsonantVSequence('ຈົນກວ', '່', words) => 'ກວ່' and words = ['ຈົນ']
     *
     * "ຄວາມຮັກ" when encountering 'າ' after 'ຄວ':
     * handleConsonantVSequence('ຄວ', 'າ', words) => 'ຄວາ' and words = []
     *
     * "ຂວາງາມ" when encountering 'າ' after 'ຂວ':
     * handleConsonantVSequence('ຂວ', 'າ', words) => 'ຂວາ' and words = []
     */
    private handleConsonantVSequence;
    /**
     * Helper function to handle consonant + 'ຣ' sequences
     *
     * Examples:
     * "ທຣັມ" when encountering 'ັ' after 'ຣ':
     * handleConsonantRSequence('ທຣ', 'ັ', words) => 'ທຣັ' and words = []
     *
     * "ປຣິນເຕີ" when encountering 'ິ' after 'ຣ':
     * handleConsonantRSequence('ປຣ', 'ິ', words) => 'ປຣິ' and words = []
     *
     * "ກຣາມ" when encountering 'າ' after 'ຣ':
     * handleConsonantRSequence('ກຣ', 'າ', words) => 'ກຣາ' and words = []
     */
    private handleConsonantRSequence;
    /**
     * Helper function to handle digraphs with 'ຫ'
     *
     * Examples:
     * "ຫວານຫລາຍ" when encountering 'າ' after 'ຫວ':
     * handleDigraphSequence('ຫວ', 'າ', words) => 'ຫວາ' and words = []
     */
    private handleDigraphSequence;
    /**
     * Helper function to handle regular middle character processing
     *
     * Examples:
     * "ເທດລາວ" when encountering 'າ' after 'ລ':
     * handleRegularMiddleChar('ເທດລ', 'າ', words) => 'ລາ' and words = ['ເທດ']
     */
    private handleRegularMiddleChar;
    /**
     * Helper function to handle 'ວ' or 'ອ' between consonants
     *
     * Examples:
     * "ສຶ່ງສວຍງາມ" when encountering "ວ" after "ສ" and followed by "ຍ":
     * handleVaOrOSequence('ສຶ່ງສ', 'ວ', words) => 'ສວ' and words = ['ສຶ່ງ']
     */
    private handleVaOrOSequence;
    /**
     * Remove Zero Width Spaces
     *
     * Examples:
     * removeZeroWidthSpaces("ສະ​ບາຍ​ດີ") => "ສະບາຍດີ"
     */
    private removeZeroWidthSpaces;
    /**
     * Splits a Lao language sentence into individual words based on syllable structure rules.
     *
     * @param sentence The Lao sentence to be segmented.
     * @returns An array of segmented words.
     *
     * Examples:
     * splitLao("ປະເທດລາວເປັນສິ່ງສວຍງາມ") => ["ປະ", "ເທດ", "ລາວ", "ເປັນ", "ສິ່ງ", "ສວຍ", "ງາມ"]
     * splitLao("ຈົນກວ່າຈະ") => ["ຈົນ", "ກວ່າ", "ຈະ"]
     * splitLao("ຫວຽດນາມ") => ["ຫວຽດ", "ນາມ"]
     * splitLao("ພາສາລາວ 101") => ["ພາ", "ສາ", "ລາວ", "101"]
     */
    split(sentence: string): string[];
}