bert.dart

// Bert's special parser

import "package:html5lib/parser.dart" as html;
import 'package:html5lib/dom.dart';
import "dart:convert";
import "dart:io";
import "package:csvparser/csvparser.dart";


// recursively print html node content
flatten(Element e) {
  // we only output the content of terminal nodes that have no children
  if (e.children.length == 0) {
    print(e.text);
  }
  else {
    e.children.forEach((node) => flatten(node));
  }
}


main() {

  var f = new File('/tmp/test.csv');
  var content =
      f.readAsStringSync(encoding: new Utf8Codec(allowMalformed: true));
  var c = new CsvParser(content, seperator: ';');

  while (c.moveNext()) {
    var l = c.getLineAsList(); // get line as array
    var question = l[2]; // question is 3rd field
    var answer = l[3]; // answer is 4th field
    print("question = $question"); // question is plain ascii
    // answer is html that needs to be extracted as ascii
    html.parse(answer).children.forEach((c) => flatten(c));
  }

}