!7

set flag NoComputerID.
set flag LeanMode.

sS url = "https://github.com/stefan-reich/smartbot/releases/download/1/triples.gz";

sclass InfoTriple {
  S a, b, c;     // noun, verb, noun
  S globalID;    // 16 character global ID
  bool verified; // is it reliable information?
  long created;  // unix timestamp
  S source;      // further source info
  
  toString {
    ret "[" + globalID + "] " + ai_renderTriple(a, b, c);
  }
}

static new L<InfoTriple> triples;

p {
  programID = #1012438;
  File file = getProgramFile("triples-1.gz");
  if (fileSize(file) == 0) {
    print("Downloading triples (9 MB).");
    loadBinaryPageToFile(url, file);
  } else
    print("Triples already downloaded.");
  print("Parsing.");
  time {
    // Read names
    
    Iterator<S> it = linesFromFile(file);
    new L<S> names;
    while (it.hasNext()) {
      S s = trim(it.next());
      if (empty(s)) break;
      names.add(unquote(s));
    }
    
    // Read triples
    
    triples = new L;
    while (it.hasNext()) {
      S s = it.next();
      pcall {
        addIfNotNull(triples, readTriple(s, names));
      }
    }
  }
  
  print("Have " + l(triples) + " triples");
  print("Some random triples:");
  print();
  pnl(selectRandom(triples, 10));
}

static InfoTriple readTriple(S s, L<S> names) {
  L<S> l = javaTokC(s);
  if (l(l) == 8) {
    new InfoTriple t;
    t.a = javaIntern(names.get(parseInt(l.get(0))));
    t.b = javaIntern(names.get(parseInt(l.get(1))));
    t.c = javaIntern(names.get(parseInt(l.get(2))));
    t.globalID = unquote(l.get(3));
    // t.title = unquote(l.get(4)); // unused
    t.source = javaIntern(unquote(l.get(5)));
    t.verified = eq(l.get(6), "v");
    t.created = parseLong(l.get(7));
    ret t;
  }
  null;
}