normalizeNumericGaps function

String normalizeNumericGaps(String line)

Repairs noisy separators and spacing in numeric expressions.

Handles cases like "1 . 23" -> "1.23" and detects digit-dominant lines to remove all whitespace.

Implementation

String normalizeNumericGaps(String line) {
  if (line.isEmpty) {
    return line;
  }

  bool hasNonDigitToken = false;
  for (int i = 0; i < line.length; i++) {
    final int code = line.codeUnitAt(i);
    if (!isDigit(code) &&
        code != spaceCodeUnit &&
        code != tabCodeUnit &&
        code != lineFeedCodeUnit &&
        code != carriageReturnCodeUnit) {
      hasNonDigitToken = true;
      break;
    }
  }

  final StringBuffer buffer = StringBuffer();
  for (int i = 0; i < line.length; i++) {
    final String ch = line[i];
    final int code = ch.codeUnitAt(0);
    final bool prevDigit = i > 0 && isDigit(line.codeUnitAt(i - 1));
    final bool nextDigit =
        i + 1 < line.length && isDigit(line.codeUnitAt(i + 1));

    if (digitNonAlnumMap.containsKey(ch) && (prevDigit || nextDigit)) {
      buffer.write(digitNonAlnumMap[ch]);
      continue;
    }

    if (code == spaceCodeUnit ||
        code == tabCodeUnit ||
        code == lineFeedCodeUnit ||
        code == carriageReturnCodeUnit) {
      buffer.write(ch);
      continue;
    }

    buffer.write(ch);
  }

  final String withMappedNonAlnum = buffer.toString();

  if (!hasNonDigitToken) {
    // Only collapse whitespace when every digit group is a single digit,
    // which suggests a fragmented number (e.g. "1 2 3 4" → "1234").
    // Multi-digit groups separated by spaces are distinct numbers
    // (e.g. "4004 5005 6006") and should keep their spaces.
    if (_allSingleDigitGroups(withMappedNonAlnum)) {
      return withMappedNonAlnum.replaceAll(RegExp(r'\s+'), '');
    }
    return withMappedNonAlnum;
  }

  return withMappedNonAlnum.replaceAllMapped(
    RegExp(r'(\d)\s+([A-Za-z])(?=\d)'),
    (Match match) {
      final String left = match.group(regexGroupFirst) ?? '';
      final String mid = match.group(regexGroupSecond) ?? '';
      final String mapped = digitConfusionMap[mid] ?? mid;
      return '$left.$mapped';
    },
  );
}