Google Scholar Alert (G)Mails to RSS feeds

google-scholar-to-rss.gs

var LABEL = '<your GMail label name storing Google Scholar Alert>'
var RECEIVER = '<your Blogger e-mail address>'

function get_quota() {
  var quota = MailApp.getRemainingDailyQuota()
  Logger.log(quota)
  return quota
}

function myFunction() {
  var label = GmailApp.getUserLabelByName(LABEL)

  var idx = 0
  var threads_old = null
  while (true) {
    var threads = label.getThreads(idx, 1)
    if (!threads[0]) {
      // Maybe idx reaches the end of the label. Exit
      return
    }
    if (threads == threads_old) {
      // Stuck. Exit
      return
    }
    threads_old = threads
    var messages = threads[0].getMessages()
    // check threads are Google Scholar Alert
    var from = messages[0].getFrom()
    if (!is_google_scholar(from)) {
      idx = idx + 1
      continue
    }

    for (var j = 0; j < messages.length; j++) {
      var body = messages[j].getBody()
      var draft_arr = []
      while (true) {
        // per feed
        var pos_start = body.search('<h3')
        var pos_end = body.search('</h3>')
        if (pos_start < 0 || pos_end <= pos_start) {
          break
        }
        //Logger.log(body.substring(pos_start, pos_end))
        var title_url = find_title_url(body.substring(pos_start, pos_end))
        Logger.log(title_url['title'])
        Logger.log(title_url['url'])
        body = body.substring(pos_end + 5)
//        pos_end = body.search('</div>')
//        var author_pub = find_author_pub(body.substring(0, pos_end))
        var author_pub = find_author_pub(body)
        Logger.log(author_pub)
        body = body.substring(pos_end + 6)
        var abstract = find_abstract(body)
        Logger.log(abstract)
        
        var draft = GmailApp.createDraft(RECEIVER,
                                         title_url['title'],
                                         '',
                                         {htmlBody: title_url['url'] + '<br/><br/>' + author_pub + '<br/><br/>' + abstract})
        draft_arr.push(draft)
      }
      if (draft_arr.length < get_quota()) {
        for (var k = 0; k < draft_arr.length; k++) {
          var sent = draft_arr[k].send()
          if (sent) {
            sent.moveToTrash()
          }
        }
        messages[j].moveToTrash()
      } else {
        Logger.log('Not enough quota. Try on next trigger')
        for (var k = 0; k < draft_arr.length; k++) {
          draft_arr[k].deleteDraft()
        }
        return
      } 
    }
  }
}

function is_google_scholar(from) {
  return from.indexOf('[email protected]') >= 0
}

function find_title_url(title_html) {
  var pos = title_html.search('<a href="')
  title_html = title_html.substring(pos + 9)
  pos = title_html.search('"')
  var url = title_html.substring(0, pos)
//  pos = title_html.search('<font size=3')
//  title_html = title_html.substring(pos)
  pos = title_html.search('>')
  title_html = title_html.substring(pos + 1)
//  pos = title_html.search('</font>')
  pos = title_html.search('</a>')
  title = title_html.substring(0, pos)
  return {title: title, url: url}
}

function find_author_pub(author_pub_html) {
//  var pos = author_pub_html.search('<font')
  var pos = author_pub_html.search('<div style')
  author_pub_html = author_pub_html.substring(pos)
  pos = author_pub_html.search('>')
  author_pub_html = author_pub_html.substring(pos + 1)
//  pos = author_pub_html.search('</font>')
  pos = author_pub_html.search('</div>')
  author_pub = author_pub_html.substring(0, pos)
  return author_pub
}

function find_abstract(abstract_html) {
//  var pos = abstract_html.search('<font')
  var pos = abstract_html.search('<div class')
  abstract_html = abstract_html.substring(pos)
  pos = abstract_html.search('>')
  abstract_html = abstract_html.substring(pos + 1)
//  pos = abstract_html.search('</font>')
  pos = abstract_html.search('</div>')
  abstract = abstract_html.substring(0, pos)
  return abstract
}