chrono-node/src/locales/de/constants.ts

A natural language date parser in Javascript

import { OpUnitType, QUnitType } from "dayjs";
import { matchAnyPattern, repeatedTimeunitPattern } from "../../utils/pattern";
import { findMostLikelyADYear } from "../../calculation/years";
import { TimeUnits } from "../../utils/timeunits";

export const WEEKDAY_DICTIONARY: { [word: string]: number } = {
    "sonntag": 0,
    "so": 0,
    "montag": 1,
    "mo": 1,
    "dienstag": 2,
    "di": 2,
    "mittwoch": 3,
    "mi": 3,
    "donnerstag": 4,
    "do": 4,
    "freitag": 5,
    "fr": 5,
    "samstag": 6,
    "sa": 6,
};

export const MONTH_DICTIONARY: { [word: string]: number } = {
    "januar": 1,
    "jänner": 1,
    "janner": 1,
    "jan": 1,
    "jan.": 1,
    "februar": 2,
    "feber": 2,
    "feb": 2,
    "feb.": 2,
    "märz": 3,
    "maerz": 3,
    "mär": 3,
    "mär.": 3,
    "mrz": 3,
    "mrz.": 3,
    "april": 4,
    "apr": 4,
    "apr.": 4,
    "mai": 5,
    "juni": 6,
    "jun": 6,
    "jun.": 6,
    "juli": 7,
    "jul": 7,
    "jul.": 7,
    "august": 8,
    "aug": 8,
    "aug.": 8,
    "september": 9,
    "sep": 9,
    "sep.": 9,
    "sept": 9,
    "sept.": 9,
    "oktober": 10,
    "okt": 10,
    "okt.": 10,
    "november": 11,
    "nov": 11,
    "nov.": 11,
    "dezember": 12,
    "dez": 12,
    "dez.": 12,
};

export const INTEGER_WORD_DICTIONARY: { [word: string]: number } = {
    "eins": 1,
    "eine": 1,
    "einem": 1,
    "einen": 1,
    "einer": 1,
    "zwei": 2,
    "drei": 3,
    "vier": 4,
    "fünf": 5,
    "fuenf": 5,
    "sechs": 6,
    "sieben": 7,
    "acht": 8,
    "neun": 9,
    "zehn": 10,
    "elf": 11,
    "zwölf": 12,
    "zwoelf": 12,
};

export const TIME_UNIT_DICTIONARY: { [word: string]: OpUnitType | QUnitType } = {
    sek: "second",
    sekunde: "second",
    sekunden: "second",
    min: "minute",
    minute: "minute",
    minuten: "minute",
    h: "hour",
    std: "hour",
    stunde: "hour",
    stunden: "hour",
    tag: "d",
    tage: "d",
    tagen: "d",
    woche: "week",
    wochen: "week",
    monat: "month",
    monate: "month",
    monaten: "month",
    monats: "month",
    quartal: "quarter",
    quartals: "quarter",
    quartale: "quarter",
    quartalen: "quarter",
    a: "year",
    j: "year",
    jr: "year",
    jahr: "year",
    jahre: "year",
    jahren: "year",
    jahres: "year",
};

//-----------------------------

export const NUMBER_PATTERN = `(?:${matchAnyPattern(
    INTEGER_WORD_DICTIONARY
)}|[0-9]+|[0-9]+\\.[0-9]+|halb?|halbe?|einigen?|wenigen?|mehreren?)`;

export function parseNumberPattern(match: string): number {
    const num = match.toLowerCase();
    if (INTEGER_WORD_DICTIONARY[num] !== undefined) {
        return INTEGER_WORD_DICTIONARY[num];
    } else if (num === "ein" || num === "einer" || num === "einem" || num === "einen" || num === "eine") {
        return 1;
    } else if (num.match(/wenigen/)) {
        return 2;
    } else if (num.match(/halb/) || num.match(/halben/)) {
        return 0.5;
    } else if (num.match(/einigen/)) {
        return 3;
    } else if (num.match(/mehreren/)) {
        return 7;
    }

    return parseFloat(num);
}

//-----------------------------

export const YEAR_PATTERN = `(?:[0-9]{1,4}(?:\\s*[vn]\\.?\\s*(?:C(?:hr)?|(?:u\\.?|d\\.?(?:\\s*g\\.?)?)?\\s*Z)\\.?|\\s*(?:u\\.?|d\\.?(?:\\s*g\\.)?)\\s*Z\\.?)?)`;
export function parseYear(match: string): number {
    if (/v/i.test(match)) {
        // v.Chr.
        return -parseInt(match.replace(/[^0-9]+/gi, ""));
    }

    if (/n/i.test(match)) {
        // n.Chr.
        return parseInt(match.replace(/[^0-9]+/gi, ""));
    }

    if (/z/i.test(match)) {
        // n.Chr. as "uZ" or "dgZ"
        return parseInt(match.replace(/[^0-9]+/gi, ""));
    }

    const rawYearNumber = parseInt(match);
    return findMostLikelyADYear(rawYearNumber);
}

//-----------------------------

const SINGLE_TIME_UNIT_PATTERN = `(${NUMBER_PATTERN})\\s{0,5}(${matchAnyPattern(TIME_UNIT_DICTIONARY)})\\s{0,5}`;
const SINGLE_TIME_UNIT_REGEX = new RegExp(SINGLE_TIME_UNIT_PATTERN, "i");

export const TIME_UNITS_PATTERN = repeatedTimeunitPattern("", SINGLE_TIME_UNIT_PATTERN);

export function parseTimeUnits(timeunitText): TimeUnits {
    const fragments = {};
    let remainingText = timeunitText;
    let match = SINGLE_TIME_UNIT_REGEX.exec(remainingText);
    while (match) {
        collectDateTimeFragment(fragments, match);
        remainingText = remainingText.substring(match[0].length);
        match = SINGLE_TIME_UNIT_REGEX.exec(remainingText);
    }
    return fragments;
}

function collectDateTimeFragment(fragments, match) {
    const num = parseNumberPattern(match[1]);
    const unit = TIME_UNIT_DICTIONARY[match[2].toLowerCase()];
    fragments[unit] = num;
}