!7

p {

  L<Pair<WebNode>> nodePairs = chatHistory_peepHole_nodePairs();

  print("Have " + n(nodePairs, "node pair"));

  // Analysis

  new MultiMap<S, WebNode> index; // left text to right node

  for (Pair<WebNode> p : nodePairs) {

    L<S> texts = web_texts(p.a);

    for (S text : texts)

      index.setPut(text, p.b);

  }

  print("Index size: " + index.keysSize() + " / " + index.size());

  MultiSet<S> ms = multiMapAsMultiSet(index);

  new L<Web> outWebs;

  for (S key : ms.highestFirst()) {

    L<WebNode> nodes = index.get(key);

    if (l(nodes) < 2) continue;

    print(l(nodes) + " - testing key " + key);

    L<S> common = web_commonTexts(nodes);

    LL<S> l = map web_texts(nodes);

    if (nempty(common)) {

      print("  Always: " + common);

      outWebs.add(webFromTriples2(key, ll("then", "checked " + n(nodes, "time")), common));

      for (L x : l ) x.removeAll(common);

    }

    //Pair<S, Int> p = web_mostCommonTextWithCount(nodes);

    Pair<S, Int> p = mostCommonWithCount(l);

    if (p != null && p.b > 1) {

      int percent = ratioToIntPercent(p.b, l(nodes));

      print("  Often (" + percent + "%): " + p.a + " (" + p.b + ")");

      outWebs.add(webFromTriples2(key, ll("then maybe", "(" + p.b + "/" + n(nodes, "time") + ")"), p.a));

    }

  }

  print("Made " + n(outWebs, "out web"));

  //showWeb(last(outWebs));

  if (isMain())

    showWeb(firstWebContaining(outWebs, "then maybe"));

  saveGZStructure("peephole-webs.structure.gz", outWebs);

}