Merge pull request #6 from gbassisp/feature/git-format

gbassisp · web-flow · commit 9ff66dc19bd7 · 2024-08-31T18:05:15.000+09:30
Feature/git format
diff --git a/CHANGELOG.md b/CHANGELOG.md
@@ -1,4 +1,8 @@
 
+## 1.0.4
+
+- Added support for git non-sense format, e.g., `Thu May 16 10:18:07 2024 +0930` and `Thu May 16 10:18:07pm 2024 +0930`
+
 ## 1.0.3
 
 - Merged in `0.1.14`:  
diff --git a/lib/src/any_date_rules.dart b/lib/src/any_date_rules.dart
@@ -13,36 +13,45 @@ const _yearPattern = r'(?<year>\d+)';
 const _ambiguousYearPattern = r'(?<year>\d{2})';
 const _dayPattern = r'(?<day>[0-3]?\d)';
 const _textMonthPattern = r'(?<month>\w+)';
-const _monthPattern = r'(?<month>[0-1]?\d)';
+const _monthPattern = r'(?<month>(0?\d)|(1[0-2]))';
 // const _anyMonthPattern = r'(?<month>\d{1,2}|\w+)';
 const _hourPattern = r'(?<hour>[0-2]?\d)';
 const _minutePattern = r'(?<minute>[0-5]?\d)';
 const _secondPattern = r'(?<second>[0-5]?\d)';
-const _microsecondPattern = r'(?<microsecond>\d{1,6})';
+const _microsecondPattern = r'(?<microsecond>\d+)';
 final _separatorPattern = '[${usedSeparators.reduce((v1, v2) => '$v1,$v2')}]+';
 final _s = _separatorPattern;
 // final _separatorPattern =
 // '(${usedSeparators.reduce((v1, v2) => '$v1|$v2')})+';
 // avoid const because this will be updated soon-ish
 String get _timeSep => ':';
-final _hmPattern = '$_hourPattern$_timeSep$_minutePattern';
-final _hmsPattern = '$_hmPattern$_timeSep$_secondPattern';
-final _hmsMsPattern = '$_hmsPattern.$_microsecondPattern';
+final _hmIdeal = '$_hourPattern$_timeSep$_minutePattern';
+final _hmsIdeal = '$_hmIdeal$_timeSep$_secondPattern';
+final _hmsMsIdeal = '$_hmsIdeal\\.$_microsecondPattern';
 
 /// ideal time expressions that uses only ':' as separators
 @internal
 final idealTimePatterns = [
-  _hmsMsPattern,
-  _hmsPattern, // + r'(\D|$)',
-  _hmPattern,
+  _hmsMsIdeal,
+  _hmsIdeal, // + r'(\D|$)',
+  _hmIdeal,
   _hourPattern,
 ];
 
 /// original patterns used in implementation that allows non-sense separators,
 /// such as "-"
 /// this is not entirely supported because was never tested for "h" or "min"
 /// separators
-final _timePatterns = idealTimePatterns.map((e) => e.replaceAll(_timeSep, _s));
+final _hmPattern = '$_hourPattern$_s$_minutePattern';
+final _hmsPattern = '$_hmPattern$_s$_secondPattern';
+final _hmsMsPattern = '$_hmsPattern.$_microsecondPattern';
+
+final _timePatterns = [
+  _hmsMsPattern,
+  _hmsPattern, // + r'(\D|$)',
+  _hmPattern,
+  _hourPattern,
+];
 
 /// default parsing rule from dart core
 DateTime? dateTimeTryParse(String formattedString) =>
diff --git a/lib/src/param_cleanup_rules.dart b/lib/src/param_cleanup_rules.dart
@@ -122,44 +122,65 @@ Month? _expectMonth(DateParsingParameters parameters) {
     (element) => timestamp.contains(element.name.toLowerCase()),
   );
 
-  return month.firstOrNullExtenstion ?? english.firstOrNullExtenstion;
+  return month.firstOrNullExtension ?? english.firstOrNullExtension;
 }
 
 Weekday? _expectWeekday(DateParsingParameters parameters) {
+  // TODO(gbassisp): allow weekday in any part of the string
+  // currently unsupported because some locales can have a conflict between
+  // month and weekday (e.g., "Mar" in French for Mardi and Mars)
   final timestamp = parameters.formattedString.toLowerCase();
-  final weekday = parameters.parserInfo.weekdays
+  var weekday = parameters.parserInfo.weekdays
       .where(
         (element) => timestamp.startsWith(element.name.toLowerCase()),
+        // (element) => timestamp
+        //     .contains(RegExp('\\D${element.name}', caseSensitive: false)),
       )
-      .firstOrNullExtenstion;
-  final english = allWeekdays
+      .firstOrNullExtension;
+  if (weekday != null) return weekday;
+//   weekday = parameters.parserInfo.weekdays
+//       .where((element) => timestamp.endsWith(element.name.toLowerCase()))
+//       .firstOrNullExtension;
+//   if (weekday != null) return weekday;
+
+// english
+  weekday = allWeekdays
       .where(
         (element) => timestamp.startsWith(element.name.toLowerCase()),
       )
-      .firstOrNullExtenstion;
+      .firstOrNullExtension;
+  if (weekday != null) return weekday;
 
-  return weekday ?? english;
+  return allWeekdays
+      .where(
+        (element) => timestamp.endsWith(element.name.toLowerCase()),
+      )
+      .firstOrNullExtension;
 }
 
 final _exprs = [...idealTimePatterns]..removeLast();
 final _betterTimeComponent = CleanupRule((params) {
   String padLeft(String? original) => (original ?? '').padLeft(2, '0');
   String padRight(String? original) => (original ?? '').padRight(3, '0');
+  String cleanAmpm(String? original) =>
+      original?.replaceAll(RegExp(r'(\.|-)'), '').toLowerCase() ?? '';
 
   for (final e in _exprs) {
-    final re = RegExp(e);
-    final matches = re.allMatches(params.formattedString);
+    const ampm = r'\s*(?<ampm>(a|p)(\.|-)?m(\.|-)?\W)?';
+    final re = RegExp(e + ampm, caseSensitive: false);
+    final s = '${params.formattedString} ';
+    final matches = re.allMatches(s);
     // unsure what to do if many matches
     if (matches.length == 1) {
       final m = matches.first;
-      final newString = params.formattedString.replaceAllMapped(
-        re,
-        (match) => '${padLeft(m.namedGroup('hour'))}:'
-            '${padLeft(m.tryNamedGroup('minute'))}:'
-            '${padLeft(m.tryNamedGroup('second'))}'
-            '${m.tryNamedGroup('microsecond') != null ? '.'
-                '${padRight(m.tryNamedGroup('microsecond'))}' : ''}',
-      );
+      final newTime = ' ${padLeft(m.namedGroup('hour'))}:'
+          '${padLeft(m.tryNamedGroup('minute'))}:'
+          '${padLeft(m.tryNamedGroup('second'))}'
+          '${m.tryNamedGroup('microsecond') != null ? '.'
+              '${padRight(m.tryNamedGroup('microsecond'))}' : ''} '
+          '${cleanAmpm(m.tryNamedGroup('ampm'))}';
+      final newString = s.replaceAllMapped(re, (_) => '') + newTime;
+
       params
         ..formattedString = newString
         ..simplifiedString = newString;
@@ -198,6 +219,6 @@ extension _GroupNames on RegExpMatch {
 }
 
 extension _IterableX<T> on Iterable<T> {
-  T? get firstOrNullExtenstion => isEmpty ? null : first;
+  T? get firstOrNullExtension => isEmpty ? null : first;
   // T? get lastOrNull => isEmpty ? null : last;
 }
diff --git a/pubspec.yaml b/pubspec.yaml
@@ -1,6 +1,6 @@
 name: any_date
 description: A package for parsing String into DateTime in any format. Easy way to parse a date in any format and in any language, while always respecting ISO and RFC formats.
-version: 1.0.3
+version: 1.0.4
 repository: https://github.com/gbassisp/any_date
 homepage: https://github.com/gbassisp/any_date
 
diff --git a/test/locale_based_test.dart b/test/locale_based_test.dart
@@ -103,13 +103,11 @@ Future<void> main() async {
     final englishMonths = AnyDate.defaultSettings.months;
     final englishWeekdays = AnyDate.defaultSettings.weekdays;
     final longWeekdays = englishWeekdays.sublist(0, 7);
-    final shortWeekdays = englishWeekdays.sublist(7)
-      ..removeWhere((element) => element.name == 'Sept');
+    final shortWeekdays = englishWeekdays.sublist(7);
     test('english speaking - american format', () {
       final locale = Locale.fromSubtags(languageCode: 'en', countryCode: 'US');
       final longMonths = englishMonths.sublist(0, 12);
-      final shortMonths = englishMonths.sublist(12)
-        ..removeWhere((element) => element.name == 'Sept');
+      final shortMonths = englishMonths.sublist(12);
 
       expect(locale.usesMonthFirst, isTrue);
       expect(locale.usesYearFirst, isFalse);
@@ -122,8 +120,7 @@ Future<void> main() async {
     test('english speaking - normal format', () {
       final locale = Locale.fromSubtags(languageCode: 'en', countryCode: 'AU');
       final longMonths = englishMonths.sublist(0, 12);
-      final shortMonths = englishMonths.sublist(12)
-        ..removeWhere((element) => element.name == 'Sep');
+      final shortMonths = englishMonths.sublist(12);
 
       expect(locale.usesMonthFirst, isFalse);
       expect(locale.usesYearFirst, isFalse);
@@ -135,6 +132,26 @@ Future<void> main() async {
   });
 
   group('locale specific cases', () {
+    // this is to ensure 日 is not mis-interpreted between day of the week and
+    // day of the month
+    test(
+        '2024年8月31日 on ja with format y年M月d日 resulted in null, '
+        'but expected 2024-08-31 00:00:00.000', () {
+      const locale = 'ja';
+      const formatted = '2024年8月31日';
+      const format = 'y年M月d日';
+      final formatter = DateFormat(format);
+      final expected = DateTime.parse('2024-08-31 00:00:00.000');
+      final parser = AnyDate.fromLocale(locale);
+
+      expect(
+        formatter.parseLoose(formatted),
+        equals(expected),
+        reason: 'sanity check that DateFormat $format can parse $formatted',
+      );
+      expect(parser.tryParse(formatted), equals(expected));
+    });
+
     const unambiguousEnglish = {
       'March 27, 2024',
       'March 27 2024',
diff --git a/test/nonsense_formats_test.dart b/test/nonsense_formats_test.dart
@@ -1,6 +1,9 @@
 import 'package:any_date/any_date.dart';
+import 'package:any_date/src/extensions.dart';
 import 'package:test/test.dart';
 
+import 'test_values.dart';
+
 void main() {
   group('lack of separators formats', () {
     const parser = AnyDate();
@@ -93,4 +96,62 @@ void main() {
       expect(parser.tryParse(formatted), equals(expected));
     });
   });
+
+  group('git', () {
+    final parser = parsers.first;
+    test('git default nonsense 1', () {
+      const formatted = 'Thu May 16 10:18:07 2024 +0930';
+      final expected =
+          DateTime.utc(2024, 5, 16, 10, 18, 7).copyWithOffset('+0930');
+
+      expect(parser.tryParse(formatted), equals(expected));
+    });
+    test('git default nonsense 2', () {
+      const formatted = 'Thu May 16 10:18:07 2024 -0930';
+      final expected =
+          DateTime.utc(2024, 5, 16, 10, 18, 7).copyWithOffset('-0930');
+
+      expect(parser.tryParse(formatted), equals(expected));
+    });
+    for (final am in [
+      'am',
+      'AM',
+      'a.m.',
+      'a.m',
+      'A.M',
+      ' am',
+      ' AM',
+      ' a.m.',
+      ' a.m',
+      ' A.M',
+    ]) {
+      test('git default nonsense 3 - "$am"', () {
+        final formatted = 'Thu, May 16 10:18:07$am 2024 -0930';
+        final expected =
+            DateTime.utc(2024, 5, 16, 10, 18, 7).copyWithOffset('-0930');
+
+        expect(parser.tryParse(formatted), equals(expected));
+      });
+    }
+    for (final pm in [
+      'pm',
+      'PM',
+      'p.m.',
+      'p.m',
+      'P.M',
+      ' pm',
+      ' PM',
+      ' p.m.',
+      ' p.m',
+      ' P.M',
+    ]) {
+      test('git default nonsense 4 - "$pm"', () {
+        final formatted = 'Thu, May 16 10:18:07$pm 2024 -0930';
+        final expected =
+            DateTime.utc(2024, 5, 16, 22, 18, 7).copyWithOffset('-0930');
+
+        expect(parser.tryParse(formatted), equals(expected));
+      });
+    }
+  });
 }
