Fix: Time expression parsing after date with number-ending

wanasit · Jun 1, 2024 · 3ae3144 · 3ae3144
1 parent fdcfde0
commit 3ae3144
Show file tree

Hide file tree

Showing 5 changed files with 63 additions and 4 deletions.
diff --git a/src/common/parsers/AbstractTimeExpressionParser.ts b/src/common/parsers/AbstractTimeExpressionParser.ts
@@ -80,6 +80,13 @@ export abstract class AbstractTimeExpressionParser implements Parser {
     extract(context: ParsingContext, match: RegExpMatchArray): ParsingResult {
         const startComponents = this.extractPrimaryTimeComponents(context, match);
         if (!startComponents) {
+            // If the match seem like a year e.g. "2013.12:...",
+            // then skips the year part and try matching again.
+            if (match[0].match(/^\d{4}/)) {
+                match.index += 4; // Skip over potential overlapping pattern
+                return null;
+            }
+
             match.index += match[0].length; // Skip over potential overlapping pattern
             return null;
         }
@@ -94,8 +101,15 @@ export abstract class AbstractTimeExpressionParser implements Parser {
         const followingMatch = followingPattern.exec(remainingText);
 
         // Pattern "456-12", "2022-12" should not be time without proper context
-        if (text.match(/^\d{3,4}/) && followingMatch && followingMatch[0].match(/^\s*([+-])\s*\d{2,4}$/)) {
-            return null;
+        if (text.match(/^\d{3,4}/) && followingMatch) {
+            // e.g. "2022-12"
+            if (followingMatch[0].match(/^\s*([+-])\s*\d{2,4}$/)) {
+                return null;
+            }
+            // e.g. "2022-12:01..."
+            if (followingMatch[0].match(/^\s*([+-])\s*\d{2}\W\d{2}/)) {
+                return null;
+            }
         }
 
         if (

diff --git a/src/locales/en/parsers/ENMonthNameLittleEndianParser.ts b/src/locales/en/parsers/ENMonthNameLittleEndianParser.ts
@@ -19,7 +19,7 @@ const PATTERN = new RegExp(
         `(${matchAnyPattern(MONTH_DICTIONARY)})` +
         "(?:" +
             `(?:-|/|,?\\s{0,3})` +
-            `(${YEAR_PATTERN}(?![^\\s]\\d))` +
+            `(${YEAR_PATTERN}(?!\\w))` +
         ")?" +
         "(?=\\W|$)",
     "i"

diff --git a/src/locales/en/refiners/ENMergeDateTimeRefiner.ts b/src/locales/en/refiners/ENMergeDateTimeRefiner.ts
@@ -8,6 +8,6 @@ import AbstractMergeDateTimeRefiner from "../../../common/refiners/AbstractMerge
  */
 export default class ENMergeDateTimeRefiner extends AbstractMergeDateTimeRefiner {
     patternBetween(): RegExp {
-        return new RegExp("^\\s*(T|at|after|before|on|of|,|-)?\\s*$");
+        return new RegExp("^\\s*(T|at|after|before|on|of|,|-|\\.|:)?\\s*$");
     }
 }
diff --git a/test/en/en_slash.test.ts b/test/en/en_slash.test.ts
@@ -136,6 +136,21 @@ test("Test - Single Expression Little-Endian with Month name", function () {
 
         expect(result.start).toBeDate(new Date(2012, 10 - 1, 8, 12));
     });
+
+    testSingleCase(chrono.strict, "06/Nov/2023", (result) => {
+        expect(result.text).toBe("06/Nov/2023");
+        expect(result).toBeDate(new Date(2023, 11 - 1, 6, 12));
+    });
+
+    testSingleCase(chrono.strict, "06/Nov/2023:06:36:02", (result) => {
+        expect(result.text).toBe("06/Nov/2023:06:36:02");
+        expect(result).toBeDate(new Date(2023, 11 - 1, 6, 6, 36, 2));
+    });
+
+    testSingleCase(chrono.strict, "06/Nov/2023:06:36:02 +0200", (result) => {
+        expect(result.text).toBe("06/Nov/2023:06:36:02 +0200");
+        expect(result).toBeDate(new Date("Mon Nov 06 2023 06:36:02 GMT+0200"));
+    });
 });
 
 test("Test - Range Expression", function () {

diff --git a/test/en/en_time_exp.test.ts b/test/en/en_time_exp.test.ts
@@ -24,6 +24,36 @@ test("Test - Time expression", function () {
     });
 });
 
+test("Test - Time expression after date", function () {
+    testSingleCase(chrono, "05/31/2024 14:15", new Date(2016, 10 - 1, 1, 8), (result, text) => {
+        expect(result.text).toBe(text);
+        expect(result.start.get("hour")).toBe(14);
+        expect(result.start.get("minute")).toBe(15);
+        expect(result.start.get("meridiem")).toBe(Meridiem.PM);
+    });
+
+    testSingleCase(chrono, "05/31/2024.14:15", new Date(2016, 10 - 1, 1, 8), (result, text) => {
+        expect(result.text).toBe(text);
+        expect(result.start.get("hour")).toBe(14);
+        expect(result.start.get("minute")).toBe(15);
+        expect(result.start.get("meridiem")).toBe(Meridiem.PM);
+    });
+
+    testSingleCase(chrono, "05/31/2024:14:15", new Date(2016, 10 - 1, 1, 8), (result, text) => {
+        expect(result.text).toBe(text);
+        expect(result.start.get("hour")).toBe(14);
+        expect(result.start.get("minute")).toBe(15);
+        expect(result.start.get("meridiem")).toBe(Meridiem.PM);
+    });
+
+    testSingleCase(chrono, "05/31/2024-14:15", new Date(2016, 10 - 1, 1, 8), (result, text) => {
+        expect(result.text).toBe(text);
+        expect(result.start.get("hour")).toBe(14);
+        expect(result.start.get("minute")).toBe(15);
+        expect(result.start.get("meridiem")).toBe(Meridiem.PM);
+    });
+});
+
 test("Test - Time range expression", function () {
     testSingleCase(chrono, "10:00:00 - 21:45:00", new Date(2016, 10 - 1, 1, 8), (result, text) => {
         expect(result.text).toBe(text);