Unordered parsing?

[pratikpparikh]

is there a good example for parse in un order sequence. i.e. if i am parsing a fragment STAGE:FOO=BAR;PASS=FAIL;BOO=FOO; but the fragment may appear in any order the only rule is that it should only start with STAGE:

[pratikpparikh]

@renggli any input is appreciated.

[renggli]

I would write it as such:

final key = letter().plus();
final value = letter().plus();
final entry = key & char('=') & value & char(';');
final fragment = string('STAGE:') & entry.plus();

[pratikpparikh]

@renggli thank you this makes sense but if i want to apply any grammar rule to value then i can't for example number range, date range, string range, character range, etc... Also this also means that two keys can occur again? how can I prevent two name repeating?

Let me expand the grammar a bit more.

// no fragment should be repeated more then once but can be found or introduced in any order.
STAGE:TYPE=PAGE;VALIDATION=TRUE;ELEMENTSIZE=10;

stage = STAGE : (type | validation | elementsize) ;
type = form | page | component | element
validation = true | false
elementsize = number ; range between 1 - 30

[renggli]

This is not something that is typically specified in the grammar itself. Instead it is validated either while building your model (see example below), or in a separate pass after the model has been created:

// ...
final fragment = string('STAGE:').token() & entry.plus();
final start = fragment.map((values) => new Fragment(values[0], values[1]));

class Fragment {
  final Token startToken;
  final List<Entry> entries;

  Fragment(this.startToken, this.entries) {
    final keys = entries.map(entry => entry.key);
    if (keys.toSet().length != keys.length) {
      throw new Exception('Repeated entries at ${startToken.start}: $keys');
    }
  }
}

[pratikpparikh]

@renggli thank you this is very helpful.