New: (en) Add unlikely format filter specifically for some English words

Author: Wanasit Tanakitrungruang

src/locales/en/configuration.ts

@@ -24,6 +24,7 @@ import ENMergeRelativeAfterDateRefiner from "./refiners/ENMergeRelativeAfterDate
 import ENMergeRelativeFollowByDateRefiner from "./refiners/ENMergeRelativeFollowByDateRefiner";
 import OverlapRemovalRefiner from "../../common/refiners/OverlapRemovalRefiner";
 import ENExtractYearSuffixRefiner from "./refiners/ENExtractYearSuffixRefiner";
+import ENUnlikelyFormatFilter from "./refiners/ENUnlikelyFormatFilter";
 
 export default class ENDefaultConfiguration {
     /**
@@ -37,6 +38,7 @@ export default class ENDefaultConfiguration {
         option.parsers.push(new ENMonthNameParser());
         option.parsers.push(new ENRelativeDateFormatParser());
         option.parsers.push(new ENTimeUnitCasualRelativeFormatParser());
+        option.refiners.push(new ENUnlikelyFormatFilter());
         return option;
     }
 

src/locales/en/refiners/ENUnlikelyFormatFilter.ts

@@ -0,0 +1,33 @@
+import { Filter } from "../../../common/abstractRefiners";
+import { ParsingResult } from "../../../results";
+
+export default class ENUnlikelyFormatFilter extends Filter {
+    constructor() {
+        super();
+    }
+
+    isValid(context, result: ParsingResult): boolean {
+        const text = result.text.trim();
+
+        // If the result is consists of the whole text (e.g. "2024", "May", etc),
+        // then it is unlikely to be a date.
+        if (text === context.text.trim()) {
+            return true;
+        }
+
+        // In English, the word "may" is a month name, but it is also a modal verb.
+        // Check if the text before "may" follows some allowed patterns.
+        if (text.toLowerCase() === "may") {
+            const textBefore = context.text.substring(0, result.index).trim();
+            if (!textBefore.match(/\b(in)$/i)) {
+                context.debug(() => {
+                    console.log(`Removing unlikely result: ${result}`);
+                });
+
+                return false;
+            }
+        }
+
+        return true;
+    }
+}

test/en/en_casual.test.ts

@@ -412,6 +412,8 @@ test("Test - Random negative text", () => {
 
     testUnexpectedResult(chrono, "do I have the money");
 
+    testUnexpectedResult(chrono, "I may by here. May the force be with you. Theresa may become PM soon.");
+
     testUnexpectedResult(chrono.casual, "do I have the money");
 
     testUnexpectedResult(chrono.en.GB, "do I have the money");

test/en/en_month.test.ts

@@ -1,5 +1,5 @@
 import * as chrono from "../../src";
-import { testSingleCase } from "../test_util";
+import { testSingleCase, testUnexpectedResult } from "../test_util";
 
 test("Test - Month-Year expression", function () {
     testSingleCase(chrono, "September 2012", (result) => {
@@ -272,3 +272,16 @@ test("Test - Month should not have timezone", () => {
         }
     );
 });
+
+test("Test - Month only in different context", () => {
+    testSingleCase(chrono.casual, "May", (result) => {
+        expect(result.text).toContain("May");
+    });
+
+    testSingleCase(chrono.casual, "in May", (result) => {
+        expect(result.text).toContain("May");
+    });
+
+    testUnexpectedResult(chrono.casual, "The mountain may not move");
+    testUnexpectedResult(chrono.casual, "May not be correct");
+});