splitConcatenatedDictionaryWords function

String splitConcatenatedDictionaryWords(String line)

Splits concatenated words that are not in the dictionary.

When space detection misses a word boundary (e.g. "foxjumps"), the resulting token won't be a valid dictionary word. This pass tries every split point and accepts the split when exactly one partition produces two valid dictionary words, avoiding ambiguous results.

Implementation

String splitConcatenatedDictionaryWords(String line) {
  return line.replaceAllMapped(RegExp(r'[A-Za-z]+'), (Match match) {
    final String token = match.group(0)!;
    if (token.length < _splitMinHalfLength * _splitMinTotalParts) {
      return token;
    }

    final String lower = token.toLowerCase();
    if (englishWords.contains(lower)) {
      return token;
    }

    String? bestLeft;
    String? bestRight;
    int splitCount = 0;

    for (
      int i = _splitMinHalfLength;
      i <= lower.length - _splitMinHalfLength;
      i++
    ) {
      final String left = lower.substring(0, i);
      final String right = lower.substring(i);
      if (englishWords.contains(left) && englishWords.contains(right)) {
        bestLeft = token.substring(0, i);
        bestRight = token.substring(i);
        splitCount++;
        if (splitCount > 1) {
          return token; // ambiguous — don't split
        }
      }
    }

    if (splitCount == 1) {
      return '$bestLeft $bestRight';
    }
    return token;
  });
}