Merge pull request #398 from jebbench/enhance-dutch-support

Added more support for Dutch.
wanasit · Jun 18, 2021 · 0cfef4e · 0cfef4e
2 parents 405a27b + 7bb00af
commit 0cfef4e
Show file tree

Hide file tree

Showing 11 changed files with 1,012 additions and 8 deletions.
diff --git a/src/locales/nl/constants.ts b/src/locales/nl/constants.ts
@@ -141,11 +141,14 @@ export const TIME_UNIT_DICTIONARY: { [word: string]: OpUnitType } = {
     min: "minute",
     mins: "minute",
     minute: "minute",
+    minuut: "minute",
     minuten: "minute",
+    minuutje: "minute",
     h: "hour",
     hr: "hour",
     hrs: "hour",
     uur: "hour",
+    u: "hour",
     uren: "hour",
     dag: "d",
     dagen: "d",
@@ -160,18 +163,21 @@ export const TIME_UNIT_DICTIONARY: { [word: string]: OpUnitType } = {
 
 //-----------------------------
 
-export const NUMBER_PATTERN = `(?:${matchAnyPattern(INTEGER_WORD_DICTIONARY)}|[0-9]+|[0-9]+\\.[0-9]+|een?|halve?)`;
+export const NUMBER_PATTERN = `(?:${matchAnyPattern(
+    INTEGER_WORD_DICTIONARY
+)}|[0-9]+|[0-9]+[\\.,][0-9]+|halve?|half|paar)`;
 
 export function parseNumberPattern(match: string): number {
     const num = match.toLowerCase();
     if (INTEGER_WORD_DICTIONARY[num] !== undefined) {
         return INTEGER_WORD_DICTIONARY[num];
-    } else if (num === "een") {
-        return 1;
-    } else if (num.match(/halve?/)) {
+    } else if (num === "paar") {
+        return 2;
+    } else if (num === "half" || num.match(/halve?/)) {
         return 0.5;
     }
-    return parseFloat(num);
+    // Replace "," with "." to support some European languages
+    return parseFloat(num.replace(",", "."));
 }
 
 //-----------------------------

diff --git a/src/locales/nl/index.ts b/src/locales/nl/index.ts
@@ -20,6 +20,10 @@ import NLSlashMonthFormatParser from "./parsers/NLSlashMonthFormatParser";
 import NLTimeExpressionParser from "./parsers/NLTimeExpressionParser";
 import NLCasualYearMonthDayParser from "./parsers/NLCasualYearMonthDayParser";
 import NLCasualDateTimeParser from "./parsers/NLCasualDateTimeParser";
+import NLTimeUnitCasualRelativeFormatParser from "./parsers/NLTimeUnitCasualRelativeFormatParser";
+import NLRelativeDateFormatParser from "./parsers/NLRelativeDateFormatParser";
+import NLTimeUnitAgoFormatParser from "./parsers/NLTimeUnitAgoFormatParser";
+import NLTimeUnitLaterFormatParser from "./parsers/NLTimeUnitLaterFormatParser";
 
 // Shortcuts
 export const casual = new Chrono(createCasualConfiguration());
@@ -38,6 +42,9 @@ export function createCasualConfiguration(littleEndian = true): Configuration {
     option.parsers.unshift(new NLCasualDateParser());
     option.parsers.unshift(new NLCasualTimeParser());
     option.parsers.unshift(new NLCasualDateTimeParser());
+    option.parsers.unshift(new NLMonthNameParser());
+    option.parsers.unshift(new NLRelativeDateFormatParser());
+    option.parsers.unshift(new NLTimeUnitCasualRelativeFormatParser());
     return option;
 }
 
@@ -49,13 +56,15 @@ export function createConfiguration(strictMode = true, littleEndian = true): Con
         {
             parsers: [
                 new SlashDateFormatParser(littleEndian),
+                new NLTimeUnitWithinFormatParser(),
                 new NLMonthNameMiddleEndianParser(),
                 new NLMonthNameParser(),
-                new NLTimeExpressionParser(),
-                new NLTimeUnitWithinFormatParser(),
-                new NLSlashMonthFormatParser(),
                 new NLWeekdayParser(),
                 new NLCasualYearMonthDayParser(),
+                new NLSlashMonthFormatParser(),
+                new NLTimeExpressionParser(strictMode),
+                new NLTimeUnitAgoFormatParser(strictMode),
+                new NLTimeUnitLaterFormatParser(strictMode),
             ],
             refiners: [new NLMergeDateTimeRefiner(), new NLMergeDateRangeRefiner()],
         },

diff --git a/src/locales/nl/parsers/NLRelativeDateFormatParser.ts b/src/locales/nl/parsers/NLRelativeDateFormatParser.ts
@@ -0,0 +1,70 @@
+import { TIME_UNIT_DICTIONARY } from "../constants";
+import { ParsingContext } from "../../../chrono";
+import { ParsingComponents } from "../../../results";
+import dayjs from "dayjs";
+import { AbstractParserWithWordBoundaryChecking } from "../../../common/parsers/AbstractParserWithWordBoundary";
+import { matchAnyPattern } from "../../../utils/pattern";
+
+const PATTERN = new RegExp(
+    `(dit|deze|komende|volgend|volgende|afgelopen|vorige)\\s*(${matchAnyPattern(TIME_UNIT_DICTIONARY)})(?=\\s*)` +
+        "(?=\\W|$)",
+    "i"
+);
+
+const MODIFIER_WORD_GROUP = 1;
+const RELATIVE_WORD_GROUP = 2;
+
+export default class NLRelativeDateFormatParser extends AbstractParserWithWordBoundaryChecking {
+    innerPattern(): RegExp {
+        return PATTERN;
+    }
+
+    innerExtract(context: ParsingContext, match: RegExpMatchArray): ParsingComponents {
+        const modifier = match[MODIFIER_WORD_GROUP].toLowerCase();
+        const unitWord = match[RELATIVE_WORD_GROUP].toLowerCase();
+        const timeunit = TIME_UNIT_DICTIONARY[unitWord];
+
+        if (modifier == "volgend" || modifier == "volgende" || modifier == "komende") {
+            const timeUnits = {};
+            timeUnits[timeunit] = 1;
+            return ParsingComponents.createRelativeFromRefDate(context.refDate, timeUnits);
+        }
+
+        if (modifier == "afgelopen" || modifier == "vorige") {
+            const timeUnits = {};
+            timeUnits[timeunit] = -1;
+            return ParsingComponents.createRelativeFromRefDate(context.refDate, timeUnits);
+        }
+
+        const components = context.createParsingComponents();
+        let date = dayjs(context.refDate);
+
+        // This week
+        if (unitWord.match(/week/i)) {
+            date = date.add(-date.get("d"), "d");
+            components.imply("day", date.date());
+            components.imply("month", date.month() + 1);
+            components.imply("year", date.year());
+        }
+
+        // This month
+        else if (unitWord.match(/maand/i)) {
+            date = date.add(-date.date() + 1, "d");
+            components.imply("day", date.date());
+            components.assign("year", date.year());
+            components.assign("month", date.month() + 1);
+        }
+
+        // This year
+        else if (unitWord.match(/jaar/i)) {
+            date = date.add(-date.date() + 1, "d");
+            date = date.add(-date.month(), "month");
+
+            components.imply("day", date.date());
+            components.imply("month", date.month() + 1);
+            components.assign("year", date.year());
+        }
+
+        return components;
+    }
+}
diff --git a/src/locales/nl/parsers/NLTimeExpressionParser.ts b/src/locales/nl/parsers/NLTimeExpressionParser.ts
@@ -11,6 +11,10 @@ export default class NLTimeExpressionParser extends AbstractTimeExpressionParser
         return "\\s*(?:\\-|\\–|\\~|\\〜|om|\\?)\\s*";
     }
 
+    primarySuffix(): string {
+        return "(?:\\s*(?:uur))?(?!/)(?=\\W|$)";
+    }
+
     extractPrimaryTimeComponents(context: ParsingContext, match: RegExpMatchArray): ParsingComponents | null {
         // This looks more like a year e.g. 2020
         if (match[0].match(/^\s*\d{4}\s*$/)) {

diff --git a/src/locales/nl/parsers/NLTimeUnitAgoFormatParser.ts b/src/locales/nl/parsers/NLTimeUnitAgoFormatParser.ts
@@ -0,0 +1,26 @@
+import { ParsingContext } from "../../../chrono";
+import { parseTimeUnits, TIME_UNITS_PATTERN } from "../constants";
+import { ParsingComponents } from "../../../results";
+import { AbstractParserWithWordBoundaryChecking } from "../../../common/parsers/AbstractParserWithWordBoundary";
+import { reverseTimeUnits } from "../../../utils/timeunits";
+
+const PATTERN = new RegExp("" + "(" + TIME_UNITS_PATTERN + ")" + "(?:geleden|voor|eerder)(?=(?:\\W|$))", "i");
+
+const STRICT_PATTERN = new RegExp("" + "(" + TIME_UNITS_PATTERN + ")" + "geleden(?=(?:\\W|$))", "i");
+
+export default class NLTimeUnitAgoFormatParser extends AbstractParserWithWordBoundaryChecking {
+    constructor(private strictMode: boolean) {
+        super();
+    }
+
+    innerPattern(): RegExp {
+        return this.strictMode ? STRICT_PATTERN : PATTERN;
+    }
+
+    innerExtract(context: ParsingContext, match: RegExpMatchArray) {
+        const timeUnits = parseTimeUnits(match[1]);
+        const outputTimeUnits = reverseTimeUnits(timeUnits);
+
+        return ParsingComponents.createRelativeFromRefDate(context.refDate, outputTimeUnits);
+    }
+}
diff --git a/src/locales/nl/parsers/NLTimeUnitCasualRelativeFormatParser.ts b/src/locales/nl/parsers/NLTimeUnitCasualRelativeFormatParser.ts
@@ -0,0 +1,27 @@
+import { TIME_UNITS_PATTERN, parseTimeUnits } from "../constants";
+import { ParsingContext } from "../../../chrono";
+import { ParsingComponents } from "../../../results";
+import { AbstractParserWithWordBoundaryChecking } from "../../../common/parsers/AbstractParserWithWordBoundary";
+import { reverseTimeUnits } from "../../../utils/timeunits";
+
+const PATTERN = new RegExp(`(deze|vorige|afgelopen|komende|over|\\+|-)\\s*(${TIME_UNITS_PATTERN})(?=\\W|$)`, "i");
+
+export default class NLTimeUnitCasualRelativeFormatParser extends AbstractParserWithWordBoundaryChecking {
+    innerPattern(): RegExp {
+        return PATTERN;
+    }
+
+    innerExtract(context: ParsingContext, match: RegExpMatchArray): ParsingComponents {
+        const prefix = match[1].toLowerCase();
+        let timeUnits = parseTimeUnits(match[2]);
+        switch (prefix) {
+            case "vorige":
+            case "afgelopen":
+            case "-":
+                timeUnits = reverseTimeUnits(timeUnits);
+                break;
+        }
+
+        return ParsingComponents.createRelativeFromRefDate(context.refDate, timeUnits);
+    }
+}
diff --git a/src/locales/nl/parsers/NLTimeUnitLaterFormatParser.ts b/src/locales/nl/parsers/NLTimeUnitLaterFormatParser.ts
@@ -0,0 +1,27 @@
+import { ParsingContext } from "../../../chrono";
+import { parseTimeUnits, TIME_UNITS_PATTERN } from "../constants";
+import { ParsingComponents } from "../../../results";
+import { AbstractParserWithWordBoundaryChecking } from "../../../common/parsers/AbstractParserWithWordBoundary";
+
+const PATTERN = new RegExp(
+    "" + "(" + TIME_UNITS_PATTERN + ")" + "(later|na|vanaf nu|voortaan|vooruit|uit)" + "(?=(?:\\W|$))",
+    "i"
+);
+
+const STRICT_PATTERN = new RegExp("" + "(" + TIME_UNITS_PATTERN + ")" + "(later|vanaf nu)" + "(?=(?:\\W|$))", "i");
+const GROUP_NUM_TIMEUNITS = 1;
+
+export default class NLTimeUnitLaterFormatParser extends AbstractParserWithWordBoundaryChecking {
+    constructor(private strictMode: boolean) {
+        super();
+    }
+
+    innerPattern(): RegExp {
+        return this.strictMode ? STRICT_PATTERN : PATTERN;
+    }
+
+    innerExtract(context: ParsingContext, match: RegExpMatchArray) {
+        const fragments = parseTimeUnits(match[GROUP_NUM_TIMEUNITS]);
+        return ParsingComponents.createRelativeFromRefDate(context.refDate, fragments);
+    }
+}