Merge pull request #373 from JeroenVdb/master

Fixed and extend NL locale + add test for all
wanasit · Feb 15, 2021 · fbbd996 · fbbd996
2 parents 911588c + 269fa37
commit fbbd996
Show file tree

Hide file tree

Showing 21 changed files with 1,898 additions and 64 deletions.
diff --git a/src/common/parsers/ISOFormatParser.ts b/src/common/parsers/ISOFormatParser.ts
@@ -11,11 +11,11 @@ import { AbstractParserWithWordBoundaryChecking } from "./AbstractParserWithWord
 // - TZD = (Z or +hh:mm or -hh:mm)
 const PATTERN = new RegExp(
     "([0-9]{4})\\-([0-9]{1,2})\\-([0-9]{1,2})" +
-    "(?:T" + //..
-    "([0-9]{1,2}):([0-9]{1,2})" + // hh:mm
-    "(?::([0-9]{1,2})(?:\\.(\\d{1,4}))?)?" + // :ss.s
-    "(?:Z|([+-]\\d{2}):?(\\d{2})?)?" + // TZD (Z or ±hh:mm or ±hhmm or ±hh)
-    ")?" + //..
+        "(?:T" + //..
+        "([0-9]{1,2}):([0-9]{1,2})" + // hh:mm
+        "(?::([0-9]{1,2})(?:\\.(\\d{1,4}))?)?" + // :ss.s
+        "(?:Z|([+-]\\d{2}):?(\\d{2})?)?" + // TZD (Z or ±hh:mm or ±hhmm or ±hh)
+        ")?" + //..
         "(?=\\W|$)",
     "i"
 );

diff --git a/src/locales/nl/constants.ts b/src/locales/nl/constants.ts
@@ -213,7 +213,7 @@ const SINGLE_TIME_UNIT_REGEX = new RegExp(SINGLE_TIME_UNIT_PATTERN, "i");
 const SINGLE_TIME_UNIT_PATTERN_NO_CAPTURE = SINGLE_TIME_UNIT_PATTERN.replace(/\((?!\?)/g, "(?:");
 
 export const TIME_UNITS_PATTERN =
-    `(?:(?:about|around)\\s*)?` +
+    `(?:(?:binnen|in)\\s*)?` +
     `${SINGLE_TIME_UNIT_PATTERN_NO_CAPTURE}\\s*(?:,?\\s*${SINGLE_TIME_UNIT_PATTERN_NO_CAPTURE})*`;
 
 export function parseTimeUnits(timeunitText): TimeUnits {

diff --git a/src/locales/nl/index.ts b/src/locales/nl/index.ts
@@ -11,6 +11,9 @@ import NLWeekdayParser from "./parsers/NLWeekdayParser";
 import NLMonthNameMiddleEndianParser from "./parsers/NLMonthNameMiddleEndianParser";
 import NLMonthNameParser from "./parsers/NLMonthNameParser";
 import NLSlashMonthFormatParser from "./parsers/NLSlashMonthFormatParser";
+import NLTimeExpressionParser from "./parsers/NLTimeExpressionParser";
+import NLCasualYearMonthDayParser from "./parsers/NLCasualYearMonthDayParser";
+import NLCasualDateTimeParser from "./parsers/NLCasualDateTimeParser";
 
 // Shortcuts
 export const casual = new Chrono(createCasualConfiguration());
@@ -28,6 +31,7 @@ export function createCasualConfiguration(littleEndian = true): Configuration {
     const option = createConfiguration(false, littleEndian);
     option.parsers.unshift(new NLCasualDateParser());
     option.parsers.unshift(new NLCasualTimeParser());
+    option.parsers.unshift(new NLCasualDateTimeParser());
     return option;
 }
 
@@ -36,11 +40,13 @@ export function createConfiguration(strictMode = true, littleEndian = true): Con
         {
             parsers: [
                 new SlashDateFormatParser(littleEndian),
-                new NLTimeUnitWithinFormatParser(),
-                new NLWeekdayParser(),
                 new NLMonthNameMiddleEndianParser(),
                 new NLMonthNameParser(),
+                new NLTimeExpressionParser(),
+                new NLTimeUnitWithinFormatParser(),
                 new NLSlashMonthFormatParser(),
+                new NLWeekdayParser(),
+                new NLCasualYearMonthDayParser(),
             ],
             refiners: [new NLMergeDateTimeRefiner(), new NLMergeDateRangeRefiner()],
         },

diff --git a/src/locales/nl/parsers/NLCasualDateParser.ts b/src/locales/nl/parsers/NLCasualDateParser.ts
@@ -3,9 +3,9 @@ import { ParsingComponents, ParsingResult } from "../../../results";
 import { AbstractParserWithWordBoundaryChecking } from "../../../common/parsers/AbstractParserWithWordBoundary";
 import * as references from "../../../common/casualReferences";
 
-export default class ENCasualDateParser extends AbstractParserWithWordBoundaryChecking {
+export default class NLCasualDateParser extends AbstractParserWithWordBoundaryChecking {
     innerPattern(context: ParsingContext): RegExp {
-        return /(nu|vandaag|vanacht|morgen|morgend|gisteren)(?=\W|$)/i;
+        return /(nu|vandaag|morgen|morgend|gisteren)(?=\W|$)/i;
     }
 
     innerExtract(context: ParsingContext, match: RegExpMatchArray): ParsingComponents | ParsingResult {
@@ -25,9 +25,6 @@ export default class ENCasualDateParser extends AbstractParserWithWordBoundaryCh
 
             case "gisteren":
                 return references.yesterday(context.refDate);
-
-            case "vanacht":
-                return references.tonight(context.refDate);
         }
 
         return component;

diff --git a/src/locales/nl/parsers/NLCasualDateTimeParser.ts b/src/locales/nl/parsers/NLCasualDateTimeParser.ts
@@ -0,0 +1,75 @@
+import { ParsingContext } from "../../../chrono";
+import { ParsingComponents, ParsingResult } from "../../../results";
+import { AbstractParserWithWordBoundaryChecking } from "../../../common/parsers/AbstractParserWithWordBoundary";
+import { Meridiem } from "../../../index";
+import { assignSimilarDate, assignTheNextDay } from "../../../utils/dayjs";
+import dayjs from "dayjs";
+
+/*
+ * Find combined words
+ * - morgenochtend
+ * - morgenmiddag
+ * - morgennamiddag
+ * - morgenavond
+ * - morgennacht
+ * - vanochtend
+ * - vanmiddag
+ * - vannamiddag
+ * - vanavond
+ * - vannacht
+ * - gisterenochtend
+ * - gisterenmiddag
+ * - gisterennamiddag
+ * - gisterenavond
+ * - gisterennacht
+ * */
+
+const DATE_GROUP = 1;
+const TIME_OF_DAY_GROUP = 2;
+
+export default class NLCasualDateTimeParser extends AbstractParserWithWordBoundaryChecking {
+    innerPattern(context: ParsingContext): RegExp {
+        return /(gisteren|morgen|van)(ochtend|middag|namiddag|avond|nacht)(?=\W|$)/i;
+    }
+
+    innerExtract(context: ParsingContext, match: RegExpMatchArray): ParsingComponents | ParsingResult {
+        const dateText = match[DATE_GROUP].toLowerCase();
+        const timeText = match[TIME_OF_DAY_GROUP].toLowerCase();
+        const component = context.createParsingComponents();
+        const targetDate = dayjs(context.refDate);
+
+        switch (dateText) {
+            case "gisteren":
+                assignSimilarDate(component, targetDate.add(-1, "day"));
+                break;
+            case "van":
+                assignSimilarDate(component, targetDate);
+                break;
+            case "morgen":
+                assignTheNextDay(component, targetDate);
+                break;
+        }
+
+        switch (timeText) {
+            case "ochtend":
+                component.imply("meridiem", Meridiem.AM);
+                component.imply("hour", 6);
+                break;
+            case "middag":
+                component.imply("meridiem", Meridiem.AM);
+                component.imply("hour", 12);
+                break;
+            case "namiddag":
+                component.imply("meridiem", Meridiem.PM);
+                component.imply("hour", 15);
+                break;
+
+            case "avond":
+                component.imply("meridiem", Meridiem.PM);
+                component.imply("hour", 20);
+                break;
+        }
+
+        return component;
+    }
+}
diff --git a/src/locales/nl/parsers/NLCasualTimeParser.ts b/src/locales/nl/parsers/NLCasualTimeParser.ts
@@ -4,16 +4,25 @@ import { AbstractParserWithWordBoundaryChecking } from "../../../common/parsers/
 import dayjs from "dayjs";
 import { assignTheNextDay } from "../../../utils/dayjs";
 
-export default class ENCasualTimeParser extends AbstractParserWithWordBoundaryChecking {
+const DAY_GROUP = 1;
+const MOMENT_GROUP = 2;
+
+export default class NLCasualTimeParser extends AbstractParserWithWordBoundaryChecking {
     innerPattern() {
-        return /(?:this)?\s*(namiddag|avond|middernacht|ochtend|middag|'s middags|'s avonds|'s ochtends)(?=\W|$)/i;
+        return /(deze)?\s*(namiddag|avond|middernacht|ochtend|middag|'s middags|'s avonds|'s ochtends)(?=\W|$)/i;
     }
 
     innerExtract(context: ParsingContext, match: RegExpMatchArray) {
         const targetDate = dayjs(context.refDate);
         const component = context.createParsingComponents();
 
-        switch (match[1].toLowerCase()) {
+        if (match[DAY_GROUP] === "deze") {
+            component.assign("day", context.refDate.getDate());
+            component.assign("month", context.refDate.getMonth() + 1);
+            component.assign("year", context.refDate.getFullYear());
+        }
+
+        switch (match[MOMENT_GROUP].toLowerCase()) {
             case "namiddag":
             case "'s namiddags":
                 component.imply("meridiem", Meridiem.PM);

diff --git a/src/locales/nl/parsers/NLCasualYearMonthDayParser.ts b/src/locales/nl/parsers/NLCasualYearMonthDayParser.ts
@@ -0,0 +1,49 @@
+import { ParsingContext } from "../../../chrono";
+import { MONTH_DICTIONARY } from "../constants";
+import { matchAnyPattern } from "../../../utils/pattern";
+import { AbstractParserWithWordBoundaryChecking } from "../../../common/parsers/AbstractParserWithWordBoundary";
+
+/*
+    Date format with slash "/" between numbers like ENSlashDateFormatParser,
+    but this parser expect year before month and date.
+    - YYYY/MM/DD
+    - YYYY-MM-DD
+    - YYYY.MM.DD
+*/
+const PATTERN = new RegExp(
+    `([0-9]{4})[\\.\\/\\s]` +
+        `(?:(${matchAnyPattern(MONTH_DICTIONARY)})|([0-9]{1,2}))[\\.\\/\\s]` +
+        `([0-9]{1,2})` +
+        "(?=\\W|$)",
+    "i"
+);
+
+const YEAR_NUMBER_GROUP = 1;
+const MONTH_NAME_GROUP = 2;
+const MONTH_NUMBER_GROUP = 3;
+const DATE_NUMBER_GROUP = 4;
+
+export default class NLCasualYearMonthDayParser extends AbstractParserWithWordBoundaryChecking {
+    innerPattern(): RegExp {
+        return PATTERN;
+    }
+
+    innerExtract(context: ParsingContext, match: RegExpMatchArray) {
+        const month = match[MONTH_NUMBER_GROUP]
+            ? parseInt(match[MONTH_NUMBER_GROUP])
+            : MONTH_DICTIONARY[match[MONTH_NAME_GROUP].toLowerCase()];
+
+        if (month < 1 || month > 12) {
+            return null;
+        }
+
+        const year = parseInt(match[YEAR_NUMBER_GROUP]);
+        const day = parseInt(match[DATE_NUMBER_GROUP]);
+
+        return {
+            day: day,
+            month: month,
+            year: year,
+        };
+    }
+}
diff --git a/src/locales/nl/parsers/NLMonthNameMiddleEndianParser.ts b/src/locales/nl/parsers/NLMonthNameMiddleEndianParser.ts
@@ -1,43 +1,51 @@
 import { ParsingContext } from "../../../chrono";
 import { findYearClosestToRef } from "../../../calculation/years";
-import { MONTH_DICTIONARY } from "../../en/constants";
-import { ORDINAL_NUMBER_PATTERN, parseOrdinalNumberPattern } from "../../en/constants";
-import { YEAR_PATTERN, parseYear } from "../../en/constants";
+import { MONTH_DICTIONARY } from "../constants";
+import { ORDINAL_NUMBER_PATTERN, parseOrdinalNumberPattern } from "../constants";
+import { YEAR_PATTERN, parseYear } from "../constants";
 import { matchAnyPattern } from "../../../utils/pattern";
 import { AbstractParserWithWordBoundaryChecking } from "../../../common/parsers/AbstractParserWithWordBoundary";
 
 const PATTERN = new RegExp(
-    `(${matchAnyPattern(MONTH_DICTIONARY)})` +
-        "(?:-|/|\\s*,?\\s*)" +
-        `(${ORDINAL_NUMBER_PATTERN})(?!\\s*)\\s*` +
-        "(?:" +
-        "(?:to|\\-)\\s*" +
-        `(${ORDINAL_NUMBER_PATTERN})\\s*` +
+    "(?:on\\s*?)?" +
+        `(${ORDINAL_NUMBER_PATTERN})` +
+        "(?:\\s*" +
+        "(?:tot|\\-|\\–|until|through|till|\\s)\\s*" +
+        `(${ORDINAL_NUMBER_PATTERN})` +
         ")?" +
+        "(?:-|/|\\s*(?:of)?\\s*)" +
+        "(" +
+        matchAnyPattern(MONTH_DICTIONARY) +
+        ")" +
         "(?:" +
-        "(?:-|/|\\s*,?\\s*)" +
-        `(${YEAR_PATTERN})` +
+        "(?:-|/|,?\\s*)" +
+        `(${YEAR_PATTERN}(?![^\\s]\\d))` +
         ")?" +
-        "(?=\\W|$)(?!\\:\\d)",
+        "(?=\\W|$)",
     "i"
 );
 
-const MONTH_NAME_GROUP = 1;
-const DATE_GROUP = 2;
-const DATE_TO_GROUP = 3;
+const MONTH_NAME_GROUP = 3;
+const DATE_GROUP = 1;
+const DATE_TO_GROUP = 2;
 const YEAR_GROUP = 4;
 
 /**
- * The parser for parsing US's date format that begin with month's name.
- *  - January 13
- *  - January 13, 2012
- *  - January 13 - 15, 2012
- * Note: Watch out for:
- *  - January 12:00
- *  - January 12.44
- *  - January 1222344
+ * The parser for parsing BE/NL date format with month's name in full writing
+ *  - 1 januari 2019
+ *  - 01 januari 2019
+ *  - 10 januari 2019
+ *  - 13 januari
+ *  - 10 - 25 maart
+ *  - 10 - 25 maart 2019
+ *  - 1 aug 2019
+ *  - 1 september 200 voor Christus
+ *  - 1 september 2002 na Christus
+ *  - 19 januari 87
+ *  - 12de juli 2013
+ *  - 1ste november 2013
  */
-export default class ENMonthNameMiddleEndianParser extends AbstractParserWithWordBoundaryChecking {
+export default class NLMonthNameMiddleEndianParser extends AbstractParserWithWordBoundaryChecking {
     innerPattern(): RegExp {
         return PATTERN;
     }
@@ -46,6 +54,8 @@ export default class ENMonthNameMiddleEndianParser extends AbstractParserWithWor
         const month = MONTH_DICTIONARY[match[MONTH_NAME_GROUP].toLowerCase()];
         const day = parseOrdinalNumberPattern(match[DATE_GROUP]);
         if (day > 31) {
+            // e.g. "[96 Aug]" => "9[6 Aug]", we need to shift away from the next number
+            match.index = match.index + match[DATE_GROUP].length;
             return null;
         }
 

diff --git a/src/locales/nl/parsers/NLMonthNameParser.ts b/src/locales/nl/parsers/NLMonthNameParser.ts
@@ -7,8 +7,8 @@ import { AbstractParserWithWordBoundaryChecking } from "../../../common/parsers/
 
 const PATTERN = new RegExp(
     `(${matchAnyPattern(MONTH_DICTIONARY)})` +
-        "\\s*" +
-        "(?:" +
+        `\\s*` +
+        `(?:` +
         `[,-]?\\s*(${YEAR_PATTERN})?` +
         ")?" +
         "(?=[^\\s\\w]|\\s+[^0-9]|\\s+$|$)",
@@ -20,20 +20,16 @@ const YEAR_GROUP = 2;
 
 /**
  * The parser for parsing month name and year.
- * - January, 2012
- * - January 2012
- * - January
+ * - januari, 2012
+ * - januari 2012
+ * - januari
  */
-export default class ENMonthNameParser extends AbstractParserWithWordBoundaryChecking {
+export default class NLMonthNameParser extends AbstractParserWithWordBoundaryChecking {
     innerPattern(): RegExp {
         return PATTERN;
     }
 
     innerExtract(context: ParsingContext, match: RegExpMatchArray) {
-        if (match[0].length <= 3) {
-            return null;
-        }
-
         const components = context.createParsingComponents();
         components.imply("day", 1);
 

diff --git a/src/locales/nl/parsers/NLSlashMonthFormatParser.ts b/src/locales/nl/parsers/NLSlashMonthFormatParser.ts
@@ -12,7 +12,7 @@ const YEAR_GROUP = 2;
  * - 11/05
  * - 06/2005
  */
-export default class ENSlashMonthFormatParser extends AbstractParserWithWordBoundaryChecking {
+export default class NLSlashMonthFormatParser extends AbstractParserWithWordBoundaryChecking {
     innerPattern(): RegExp {
         return PATTERN;
     }

diff --git a/src/locales/nl/parsers/NLTimeExpressionParser.ts b/src/locales/nl/parsers/NLTimeExpressionParser.ts
@@ -0,0 +1,22 @@
+import { AbstractTimeExpressionParser } from "../../../common/parsers/AbstractTimeExpressionParser";
+import { ParsingComponents } from "../../../results";
+import { ParsingContext } from "../../../chrono";
+
+export default class NLTimeExpressionParser extends AbstractTimeExpressionParser {
+    primaryPrefix(): string {
+        return "(?:(?:om)\\s*)?";
+    }
+
+    followingPhase(): string {
+        return "\\s*(?:\\-|\\–|\\~|\\〜|om|\\?)\\s*";
+    }
+
+    extractPrimaryTimeComponents(context: ParsingContext, match: RegExpMatchArray): ParsingComponents | null {
+        // This looks more like a year e.g. 2020
+        if (match[0].match(/^\s*\d{4}\s*$/)) {
+            return null;
+        }
+
+        return super.extractPrimaryTimeComponents(context, match);
+    }
+}