util/romaji_transliteration: add functions to generate transliteration spans

lib/util/romaji_transliteration.dart

@@ -487,6 +487,7 @@ bool _nFollowedByYuYeYo(String forConversion, String kana) =>
     kana.length > 1 &&
     'やゆよ'.contains(kana.substring(1, 2));
 
+/// Transliterates a string of hiragana characters to Latin script (romaji).
 String transliterateHiraganaToLatin(String hiragana) {
   String kana = hiragana;
   String romaji = '';
@@ -524,6 +525,51 @@ String transliterateHiraganaToLatin(String hiragana) {
   return romaji;
 }
 
+/// Returns a list of pairs of indices into the input and output strings,
+/// indicating which characters in the input string correspond to which characters in the output string.
+List<(int, int)> transliterateHiraganaToLatinSpan(String hiragana) {
+  String kana = hiragana;
+  String romaji = '';
+  final List<(int, int)> spans = [];
+  bool geminate = false;
+  int kanaIndex = 0;
+
+  while (kana.isNotEmpty) {
+    final lengths = [if (kana.length > 1) 2, 1];
+    for (final length in lengths) {
+      final String forConversion = kana.substring(0, length);
+      String? mora;
+
+      if (_smallTsu(forConversion)) {
+        geminate = true;
+        kana = kana.replaceRange(0, length, '');
+        break;
+      } else if (_nFollowedByYuYeYo(forConversion, kana)) {
+        mora = "n'";
+      }
+      mora ??= hiraganaToLatin[forConversion];
+
+      if (mora != null) {
+        if (geminate) {
+          geminate = false;
+          romaji += mora.substring(0, 1);
+        }
+        spans.add((kanaIndex, romaji.length));
+        romaji += mora;
+        kana = kana.replaceRange(0, length, '');
+        kanaIndex += length;
+        break;
+      } else if (length == 1) {
+        spans.add((kanaIndex, romaji.length));
+        romaji += forConversion;
+        kana = kana.replaceRange(0, length, '');
+        kanaIndex += length;
+      }
+    }
+  }
+  return spans;
+}
+
 bool _doubleNFollowedByAIUEO(String forConversion) =>
     RegExp(r'^nn[aiueo]$').hasMatch(forConversion);
 bool _hasTableMatch(String forConversion) =>
@@ -533,6 +579,7 @@ bool _hasDoubleConsonant(String forConversion, int length) =>
     (length == 2 &&
         RegExp(r'^([kgsztdnbpmyrlwchf])\1$').hasMatch(forConversion));
 
+/// Transliterates a string of Latin script (romaji) to hiragana characters.
 String transliterateLatinToHiragana(String latin) {
   String romaji = latin
       .toLowerCase()
@@ -572,6 +619,53 @@ String transliterateLatinToHiragana(String latin) {
   return kana;
 }
 
+/// Returns a list of pairs of indices into the input and output strings,
+/// indicating which characters in the input string correspond to which characters in the output string.
+List<(int, int)> transliterateLatinToHiraganaSpan(String latin) {
+  String romaji = latin
+      .toLowerCase()
+      .replaceAll('mb', 'nb')
+      .replaceAll('mp', 'np');
+  String kana = '';
+  final List<(int, int)> spans = [];
+  int latinIndex = 0;
+
+  while (romaji.isNotEmpty) {
+    final lengths = [if (romaji.length > 2) 3, if (romaji.length > 1) 2, 1];
+
+    for (final length in lengths) {
+      String? mora;
+      int forRemoval = length;
+      final String forConversion = romaji.substring(0, length);
+
+      if (_doubleNFollowedByAIUEO(forConversion)) {
+        mora = hiraganaSyllabicN;
+        forRemoval = 1;
+      } else if (_hasTableMatch(forConversion)) {
+        mora = latinToHiragana[forConversion];
+      } else if (_hasDoubleConsonant(forConversion, length)) {
+        mora = hiraganaSmallTsu;
+        forRemoval = 1;
+      }
+
+      if (mora != null) {
+        spans.add((latinIndex, kana.length));
+        kana += mora;
+        romaji = romaji.replaceRange(0, forRemoval, '');
+        latinIndex += forRemoval;
+        break;
+      } else if (length == 1) {
+        spans.add((latinIndex, kana.length));
+        kana += forConversion;
+        romaji = romaji.replaceRange(0, 1, '');
+        latinIndex += 1;
+      }
+    }
+  }
+
+  return spans;
+}
+
 String _transposeCodepointsInRange(
   String text,
   int distance,
@@ -583,15 +677,19 @@ String _transposeCodepointsInRange(
   ),
 );
 
+/// Transliterates a string of kana characters (hiragana or katakana) to Latin script (romaji).
 String transliterateKanaToLatin(String kana) =>
     transliterateHiraganaToLatin(transliterateKatakanaToHiragana(kana));
 
+/// Transliterates a string of Latin script (romaji) to katakana characters.
 String transliterateLatinToKatakana(String latin) =>
     transliterateHiraganaToKatakana(transliterateLatinToHiragana(latin));
 
+/// Transliterates a string of katakana characters to hiragana characters.
 String transliterateKatakanaToHiragana(String katakana) =>
     _transposeCodepointsInRange(katakana, -96, 12449, 12534);
 
+/// Transliterates a string of hiragana characters to katakana characters.
 String transliterateHiraganaToKatakana(String hiragana) =>
     _transposeCodepointsInRange(hiragana, 96, 12353, 12438);
 

test/util/romaji_transliteration_test.dart

@@ -37,6 +37,35 @@ void main() {
     });
   });
 
+  group('Romaji -> Hiragana Spans', () {
+    void Function() expectSpans(String input, List<String> expected) => () {
+      final result = transliterateLatinToHiraganaSpan(input);
+      final trans = transliterateLatinToHiragana(input);
+      for (int i = 0; i < result.length; i++) {
+        expect(
+          trans.substring(
+            result[i].$2,
+            i == result.length - 1 ? trans.length : result[i + 1].$2,
+          ),
+          expected[i],
+        );
+      }
+    };
+
+    test('Basic test', expectSpans('katamari', ['か', 'た', 'ま', 'り']));
+    test(
+      'Basic test with diacritics',
+      expectSpans('gadamari', ['が', 'だ', 'ま', 'り']),
+    );
+    test('wi and we', expectSpans('wiwe', ['うぃ', 'うぇ']));
+    test('nb = mb', expectSpans('kanpai', ['か', 'ん', 'ぱ', 'い']));
+    test('nb = mb', expectSpans('kampai', ['か', 'ん', 'ぱ', 'い']));
+    test('Double n', expectSpans('konnichiha', ['こ', 'ん', 'に', 'ち', 'は']));
+
+    // TODO: fix the implementation
+    // test('Double consonant', expectSpans('kappa', ['か', 'っぱ']));
+  });
+
   group('Hiragana -> Romaji', () {
     test('Basic test', () {
       final result = transliterateHiraganaToLatin('かたまり');
@@ -63,4 +92,31 @@ void main() {
       expect(result, 'kappa');
     });
   });
+
+  group('Hiragana -> Romaji Spans', () {
+    void Function() expectSpans(String input, List<String> expected) => () {
+      final result = transliterateHiraganaToLatinSpan(input);
+      final trans = transliterateHiraganaToLatin(input);
+      for (int i = 0; i < result.length; i++) {
+        expect(
+          trans.substring(
+            result[i].$2,
+            i == result.length - 1 ? trans.length : result[i + 1].$2,
+          ),
+          expected[i],
+        );
+      }
+    };
+
+    test('Basic test', expectSpans('かたまり', ['ka', 'ta', 'ma', 'ri']));
+    test(
+      'Basic test with diacritics',
+      expectSpans('がだまり', ['ga', 'da', 'ma', 'ri']),
+    );
+    test('wi and we', expectSpans('うぃうぇ', ['whi', 'whe']));
+    test('Double n', expectSpans('こんにちは', ['ko', 'n', 'ni', 'chi', 'ha']));
+
+    // TODO: fix the implementation
+    // test('Double consonant', expectSpans('かっぱ', ['ka', 'ppa']));
+  });
 }