Grab categories from Amazon aStores

Grab categories from Amazon aStores

This task use javascript sandbox with jsoup support to grab categories from Amazon aStores.

Grab categories from Amazon aStores

1. Create javascript sandbox with jsoup support
2. Create javascript as following

javascript

---

function main(env, args) {
  var astore = 'paesia';
  try {
    var categories = grabCategory(astore, env);
    var map = env.newHashMap();
    for (var i = 0; i < categories.size(); i++) {
      var cat = categories.get(i);
      map.put(cat.get('node'), cat);
    }
    for (var i = 0; i < categories.size(); i++) {
      var cat = categories.get(i);
      var node = cat.get('node');
      var title = cat.get('title');
      var parentNode = cat.get('parent');
      var parentCat = map.get(parentNode);
      var parentTitle = '';
      if (parentCat != null) {
        parentTitle = parentCat.get('title');
      }
      var line = '';
      line += '\r\nTitle: ' + title;
      line += '\r\nNode: ' + node;
      line += '\r\nParent: ' + parentNode;
      line += '\r\nParent Title: ' + parentTitle;
      env.info(line);
    }
  } catch (e) {
    env.error(e);
  }
}

function grabCategory(astore, env) {
  var tag = env.newArrayList();
  try {
    var nodelist = env.newArrayList();
    var alink = env.newURL('http://astore.amazon.com/' + astore + '-20');
    var doc = env.newJsoup().parse(alink, 60000);
    var elements = doc.select('#searchbrowse a');
    for (var i = 0; i < elements.size(); i++) {
      var element = elements.get(i);
      var title = element.text();
      var url = element.attr('href');
      var pos = url.lastIndexOf('node=');
      if (pos < 0) continue;
      var node = url.substring(pos + 5);
      pos = node.indexOf('&');
      if (pos >= 0) {
        node = node.substring(0, pos);
      }
      var item = env.newHashMap();
      item.put('title', title);
      item.put('node', node);
      item.put('parent', '');
      tag.add(item);
      nodelist.add(node);
    }
    var no = 0;
    while (no < nodelist.size()) {
      alink = env.newURL('http://astore.amazon.com/' + astore + '-20?node=' + nodelist.get(no));
      doc = env.newJsoup().parse(alink, 60000);
      elements = doc.select('#searchbrowse .indent a');
      for (var i = 0; i < elements.size(); i++) {
        var element = elements.get(i);
        var title = element.text();
        var url = element.attr('href');
        var pos = url.lastIndexOf('node=');
        if (pos < 0) continue;
        var node = url.substring(pos + 5);
        pos = node.indexOf('&');
        if (pos >= 0) {
          node = node.substring(0, pos);
        }
        if (nodelist.indexOf(node) >= 0) continue;
        var item = env.newHashMap();
        item.put('title', title);
        item.put('node', node);
        item.put('parent', nodelist.get(no));
        tag.add(item);
        nodelist.add(node);
      }
      no++;
    }
  } catch (e) {
    env.error(e);
  }
  return tag;
}
    

Java JavaScript jsoup