Fix: little-endian month and year case

wanasit · Dec 30, 2023 · c201115 · c201115
1 parent 3f2e9ad
commit c201115
Show file tree

Hide file tree

Showing 5 changed files with 44 additions and 12 deletions.
diff --git a/src/locales/en/configuration.ts b/src/locales/en/configuration.ts
@@ -31,11 +31,11 @@ export default class ENDefaultConfiguration {
      */
     createCasualConfiguration(littleEndian = false): Configuration {
         const option = this.createConfiguration(false, littleEndian);
-        option.parsers.unshift(new ENCasualDateParser());
-        option.parsers.unshift(new ENCasualTimeParser());
-        option.parsers.unshift(new ENMonthNameParser());
-        option.parsers.unshift(new ENRelativeDateFormatParser());
-        option.parsers.unshift(new ENTimeUnitCasualRelativeFormatParser());
+        option.parsers.push(new ENCasualDateParser());
+        option.parsers.push(new ENCasualTimeParser());
+        option.parsers.push(new ENMonthNameParser());
+        option.parsers.push(new ENRelativeDateFormatParser());
+        option.parsers.push(new ENTimeUnitCasualRelativeFormatParser());
         return option;
     }
 
@@ -52,7 +52,7 @@ export default class ENDefaultConfiguration {
                     new SlashDateFormatParser(littleEndian),
                     new ENTimeUnitWithinFormatParser(strictMode),
                     new ENMonthNameLittleEndianParser(),
-                    new ENMonthNameMiddleEndianParser(),
+                    new ENMonthNameMiddleEndianParser(/*shouldSkipYearLikeDate=*/ littleEndian),
                     new ENWeekdayParser(),
                     new ENCasualYearMonthDayParser(),
                     new ENSlashMonthFormatParser(),

diff --git a/src/locales/en/constants.ts b/src/locales/en/constants.ts
@@ -231,7 +231,7 @@ export function parseOrdinalNumberPattern(match: string): number {
 
 //-----------------------------
 
-export const YEAR_PATTERN = `(?:[1-9][0-9]{0,3}\\s{0,2}(?:BE|AD|BC|BCE|CE)|[1-2][0-9]{3}|[5-9][0-9])`;
+export const YEAR_PATTERN = `(?:[1-9][0-9]{0,3}\\s{0,2}(?:BE|AD|BC|BCE|CE)|[1-2][0-9]{3}|[5-9][0-9]|2[0-5])`;
 export function parseYear(match: string): number {
     if (/BE/i.test(match)) {
         // Buddhist Era

diff --git a/src/locales/en/parsers/ENMonthNameMiddleEndianParser.ts b/src/locales/en/parsers/ENMonthNameMiddleEndianParser.ts
@@ -36,8 +36,16 @@ const YEAR_GROUP = 4;
  *  - January 12:00
  *  - January 12.44
  *  - January 1222344
+ *  - January 21 (when shouldSkipYearLikeDate=true)
  */
 export default class ENMonthNameMiddleEndianParser extends AbstractParserWithWordBoundaryChecking {
+    shouldSkipYearLikeDate: boolean;
+
+    constructor(shouldSkipYearLikeDate: boolean) {
+        super();
+        this.shouldSkipYearLikeDate = shouldSkipYearLikeDate;
+    }
+
     innerPattern(): RegExp {
         return PATTERN;
     }
@@ -49,10 +57,18 @@ export default class ENMonthNameMiddleEndianParser extends AbstractParserWithWor
             return null;
         }
 
-        const components = context.createParsingComponents({
-            day: day,
-            month: month,
-        });
+        // Skip the case where the day looks like a year (ex: January 21)
+        if (this.shouldSkipYearLikeDate) {
+            if (!match[DATE_TO_GROUP] && !match[YEAR_GROUP] && match[DATE_GROUP].match(/^2[0-5]$/)) {
+                return null;
+            }
+        }
+        const components = context
+            .createParsingComponents({
+                day: day,
+                month: month,
+            })
+            .addTag("parser/ENMonthNameMiddleEndianParser");
 
         if (match[YEAR_GROUP]) {
             const year = parseYear(match[YEAR_GROUP]);
@@ -61,7 +77,6 @@ export default class ENMonthNameMiddleEndianParser extends AbstractParserWithWor
             const year = findYearClosestToRef(context.refDate, day, month);
             components.imply("year", year);
         }
-
         if (!match[DATE_TO_GROUP]) {
             return components;
         }

diff --git a/src/locales/en/parsers/ENMonthNameParser.ts b/src/locales/en/parsers/ENMonthNameParser.ts
@@ -45,6 +45,7 @@ export default class ENMonthNameParser extends AbstractParserWithWordBoundaryChe
             match.index + match[0].length
         );
         result.start.imply("day", 1);
+        result.start.addTag("parser/ENMonthNameParser");
 
         const month = MONTH_DICTIONARY[monthName];
         result.start.assign("month", month);

diff --git a/test/en/en_month_name_middle_endian.test.ts b/test/en/en_month_name_middle_endian.test.ts
@@ -1,5 +1,6 @@
 import * as chrono from "../../src";
 import { testSingleCase, testWithExpectedDate, testUnexpectedResult } from "../test_util";
+import { configuration } from "../../src/locales/en";
 
 test("Test - Single Expression", () => {
     testSingleCase(chrono, "She is getting married soon (July 2017).", (result) => {
@@ -357,6 +358,21 @@ test("Test - year 90's parsing", () => {
     });
 });
 
+test("Test - Skip year-like on little-endian configuration", () => {
+    const chronoMiddleEndian = new chrono.Chrono(chrono.en.configuration.createCasualConfiguration(false));
+    testSingleCase(chronoMiddleEndian, "Dec. 21", new Date(2023, 12, 10), (result) => {
+        expect(result.start.get("year")).toBe(2023);
+        expect(result.start.get("month")).toBe(12);
+        expect(result.start.get("day")).toBe(21);
+    });
+
+    const chronoLittleEndian = new chrono.Chrono(chrono.en.configuration.createCasualConfiguration(true));
+    testSingleCase(chronoLittleEndian, "Dec. 21", new Date(2023, 12, 10), (result) => {
+        expect(result.start.get("year")).toBe(2021);
+        expect(result.start.get("month")).toBe(12);
+    });
+});
+
 test("Test - Impossible Dates (Strict Mode)", () => {
     testUnexpectedResult(chrono.strict, "August 32, 2014");