Delete Spam Emails from Gmail — using Google App Script.

0-README.md

A Google App Script to detele emails from the spam folder. Why? – Because that's what bored programmers do. This is a fork of https://github.com/spamzero/spamzero, with minor updates to suite my needs. In case you use this please check and update the rules. Goodluck.

1-spamzero.gs

// version 1.0.0
// Forked from https://github.com/spamzero/spamzero with some improvements.

/**
 * The threads you want to process. A thread (GmailThread) consists
 * of multiple messages (GmailMessage).
 * See: https://developers.google.com/apps-script/reference/gmail/
 */
var threads = GmailApp.getSpamThreads();

/**
 * The action(s) to take on the threads matching your rules.
 *
 * Possible values are: READ, MOVE_TO_TRASH, LABEL(name).
 * You can also use multiple actions ["READ", "LABEL(Some label)"].
 */
var actions = ['MOVE_TO_TRASH'];

var rules = [

  // This first one should be the first rule in the ruleset
  function (m, raw) {
    'No subject in the headers'
    if (!raw.headers.subject) {
      return true;
    }
  },

  function (m, raw) {
    'Goomoji + name in subject'
    return StringHelper.containsGoomoji(raw.headers.subject);
  },

  function (m, raw) {
    'From *.ru + image attachment'
    var ruSender = m.getFrom().match(/\.ru>?$/);
    var withImageAttachment = MessageHelper.hasImageAttachment(m);

    return ruSender && withImageAttachment;
  },

  function (m, raw) {
    'Profanity filter'
    var matchers = [
      /slut/i,
      /f.ck/i,
      /s.ck/i,
      /c.ck/i,
      /lady/i,
      /h..kup/i,
      /cum/i,
      /babe/i,
      /p.rn/i,
      /s[e3\*\.]xy?/i,
      /house\s?wife/i
    ];

    for (var i = 0; i < matchers.length; i++) {
      if (
        raw.headers.subject.match(matchers[i]) ||
        raw.body.match(matchers[i]) ||
        raw.headers.from.match(matchers[i])
      ) {
        return true;
      }
    }

    return false;
  },

  function (m, raw) {
    'Has link to dropbox in body'
    return raw.body.match(/dl\.dropboxusercontent\.com/i);
  },

  function (m, raw) {
    'Shortened url in body'
    var matchers = [/bit\.ly/i, /tinyurl\.com/i, /goo\.gl/];

    for (var i = 0; i < matchers.length; i++) {
      if (raw.body.match(matchers[i])) {
        return true;
      }
    }

    return false;
  },

  function (m, raw) {
    'From: *.club, *.date, *.top, etc.'
    return m.getFrom().match(/(\.club|\.date|\.top|\.xyz|\.trade|\.party)>?$/i);
  },

  function (m, raw) {
    'URL to *.club, *.date, *.top, etc.'
    return raw.body.match(
      /(https?|www).+(\.club|\.date|\.top|\.xyz|\.trade|\.party)/i
    );
  },

  function (m, raw) {
    "URL to a .php site & 'confidential' / 'broken' words in body"
    return (
      raw.body.match(/(https?|www).+\.php/i) &&
      raw.body.match(/(confidential|broken)/i)
    );
  }
];

// Debug mode (will log events, use the log viewer [View / Show logs] to view them)
var isDebug = true;

// Use these helper methods in your rules
var MessageHelper = {
  hasImageAttachment: function (message) {
    var attachments = message.getAttachments();
    for (var i = 0; i < attachments.length; i++) {
      if (attachments[i].getContentType().match(/image/)) {
        return true;
      }
    }

    return false;
  }
};

var StringHelper = {
  containsGoomoji: function (string) {
    return string.match(/=\?[Uu][Tt][Ff]-8\?[BQ]\?[^\?]+\?=/i);
  }
};

// Probably you won't need to edit the code below
var getLogger = function () {
  if (isDebug) {
    return Logger;
  } else {
    return {
      clear: function () {},
      log: function () {},
      getLog: function () {}
    };
  }
};

/**
 * Checks whether the given thread's messages matches any of your rules.
 * Only one message needs to be matched by your rules for this function to return true.
 */
var anyMessageMatchesAnyRuleInThread = function (thread) {
  var messages = thread.getMessages();

  for (var i = 0; i < messages.length; i++) {
    var raw = parseRawContent(messages[i].getRawContent());

    for (j = 0; j < rules.length; j++) {
      if (rules[j](messages[i], raw)) {
        var ruleDescription = parseRuleDescription(rules[j]);
        getLogger().log(
          '"' +
            messages[i].getSubject() +
            '" matched rule "' +
            (ruleDescription ? ruleDescription : j) +
            '"'
        );
        return true;
      }
    }
  }

  return false;
};

/**
 * Parses the raw message into an object. For the returned object's format see the rule array's description.
 */
var parseRawContent = function (rawContent) {
  var lines = rawContent.split('\n');
  var result = {};
  var headers = {};
  var body = '';

  var currentHeaderKey = null;
  var headerParsed = false;

  for (var i = 0; i < lines.length; i++) {
    if (lines[i].trim() === '') {
      if (headers.date === undefined) {
        continue;
      }

      headerParsed = true;
      continue;
    }

    if (!headerParsed) {
      var headerParts = lines[i].match(/^([-a-z]+):(.*)/i);

      if (headerParts) {
        currentHeaderKey = headerParts[1].toLowerCase();
        headers[currentHeaderKey] = headerParts[2].trim();
      } else {
        // Header continues on new line
        headers[currentHeaderKey] += ' ' + lines[i].trim();
      }
    } else {
      body += lines[i];
    }
  }

  if (headers['content-transfer-encoding'] === 'base64') {
    try {
      body = Utilities.newBlob(Utilities.base64Decode(body)).getDataAsString();
    } catch (err) {
      getLogger().log('Could not base64 decode body.');
    }
  }

  result.headers = headers;
  result.body = body;

  return result;
};

/**
 * Extracts the description from a rule.
 */
var parseRuleDescription = function (func) {
  var lines = func.toString().split('\n');

  for (var i = 0; i < lines.length; i++) {
    var matches = lines[i].trim().match(/^"([^"]+)";?$/);
    if (matches) {
      return matches[1];
    }
  }

  return '';
};

var ReadAction = function () {
  this.run = function (thread) {
    thread.markRead();
  };
};

var MoveToTrashAction = function () {
  this.run = function (thread) {
    thread.moveToTrash();
  };
};

var LabelAction = function (labelName) {
  this.run = function (thread) {
    var label = GmailApp.getUserLabelByName(labelName);

    if (!label) {
      label = GmailApp.createLabel(labelName);
    }

    thread.addLabel(label);
  };
};

var ActionFactory = {
  create: function (actionName) {
    if (actionName === 'READ') {
      return new ReadAction();
    }

    if (actionName === 'MOVE_TO_TRASH') {
      return new MoveToTrashAction();
    }

    var labelParts = actionName.match(/LABEL\((.+)\)/i);
    if (labelParts) {
      return new LabelAction(labelParts[1]);
    }

    throw 'Unknown action: ' + actionName;
  }
};

var runActionsOnThread = function (thread) {
  for (var i = 0; i < actions.length; i++) {
    var action = ActionFactory.create(actions[i]);
    getLogger().log('Running action "' + actions[i] + '"');
    action.run(thread);
  }
};

/**
 * The main function. This should be the function chosen in the execute dropdown
 * in the script editor toolbar above.
 */
function main() {
  getLogger().clear();

  for (var i = 0; i < threads.length; i++) {
    if (i > 0) {
      getLogger().log('---');
    }

    getLogger().log('Processing thread number ' + i);

    // Match the rules, or check if the thread even has a subject (seemingly a new hack from the spammers)
    if (anyMessageMatchesAnyRuleInThread(threads[i]) || !threads[i].subject) {
      runActionsOnThread(threads[i]);
    } else {
      getLogger().log(
        'No action taken because none of your rules were matching this thread.'
      );
      getLogger().log('Subject:' + threads[i].subject);
    }
  }
}