!7

static new TreeMap<U, U> wiki; // title to text (stored as compressed strings)

p { time {
  File f = userFile("Downloads/wikipedia/simplewiki-latest-pages-articles.xml");
  BufferedReader reader;
  if (f.exists())
    reader = utf8bufferedReader(f);
  else {
    f = downloadSimpleWikipedia();
    reader = utf8bufferedReader(bunzip2stream(f));
  }
  S line;
  int lines = 0, pages = 0;
  StringBuilder pageBuf = null;
  while ((line = reader.readLine()) != null) {
    /*if ((++lines % 100) == 0)
      print("Lines: " + lines);*/
    line = trim(line);
    if (eq(line, "<page>"))
      pageBuf = new StringBuilder;
    if (pageBuf != null)
      pageBuf.append(line).append("\n");
    if (eq(line, "</page>")) {
      //print("Page done. " + l(pageBuf) + " chars");
      L<S> tok = htmlTok(str(pageBuf));
      S title = join(contentsOfContainerTag(tok, "title"));
      S text = join(contentsOfContainerTag(tok, "text"));
      if (empty(text)) {
        // print("No text: " + title);
      } else wiki.put(new U(title), new U(text));
      if ((++pages % 1000) == 0) {
        fractionDone(pages/228400.0);
        print("Pages: " + pages + " (" + title + ")");
        sleep(1);
      }
      //print(title); // XXX - too much printing hangs AWT (investigate!)
      pageBuf = null;
    }
  }
  }
  fractionDone(1);
  
  // print a random entry
  U n = random(keys(wiki));
  print("== " + n + " ==");
  print();
  print(wiki.get(n));
  bot();
}

answer {
  // for (U u : keys(wiki)) if (match(str(u), s)) ...
  U u = map.floorKey(new U(s));
  if (u != null) {
    print("== " + u + " ==");
    print(wiki.get(u));
    ret str(u);
  }
}