Detecting collisions in OpenMRS ID LDAP email addresses

collisions.js

var fs = require('fs');
var _ = require('lodash');
var xml2js = require('xml2js');
var async = require('async');

var emails = {};
var users = {};

var primary = fs.readFileSync('./primaryemails').toString();
var secondary = fs.readFileSync('./secondaryemails').toString();
var all = primary + '\n' + secondary;

var currentUser;

all.split('\n')
.forEach(function(line) {
  if (line.indexOf('dn: ') === 0) {
    currentUser = line.split('dn: ')[1];
    users[currentUser] = users[currentUser] || [];

  } else {
    var m;
    var write = false;
    var type;

    if (line.indexOf('mail: ') === 0) {
      type = 'primary';
      m = line.split('mail: ')[1];
      write = true;

    } else if (line.indexOf('otherMailbox: ') === 0) {
      type = 'secondary';
      m = line.split('otherMailbox: ')[1];
      write = true;
    }

    if (write) {
      emails[m] = emails[m] || [];
      emails[m].push(currentUser);
    }
  }

});

emails = _.omit(emails, function(users) {
  return users.length < 2;
});

var pairs = [];

_.forIn(emails, function(usersWithThisAddress, email) {
  usersWithThisAddress.forEach(function(u) {
    var uid = /^uid=(.+),ou=users,dc=openmrs,dc=org$/.exec(u)[1];

    pairs.push([uid, email]);
  });
});


// Write csv to stdout

process.stdout.write('Username,Email Address,JIRA profile,Confluence Profile\n');
pairs.forEach(function(p) {

  p.push('https://issues.openmrs.org/secure/ViewProfile.jspa?name='+p[0]);
  p.push('https://wiki.openmrs.org/display/~'+p[0]);

  process.stdout.write(p.join(','));
  process.stdout.write('\n');
});

return 0;

dup-in-users.js

var users = require('./users.json');
var _ = require('lodash');

var arrayOfEmailList = _.pluck(users.objList, 'emailList');

var emails = _(arrayOfEmailList)
              .flatten()
              .compact()
              .invoke('toLowerCase')
              .sort()
              .value();

var uniqEmails = _.uniq(emails);

_.reduce(emails, function(last, current) {
  if (last === current) {
    console.warn('duplicate address:', current);
  }
  return current;
});

generate.sh

#!/bin/sh

ldapsearch -x -W -D cn=admin,dc=openmrs,dc=org -b ou=users,dc=openmrs,dc=org "(mail=*)" -LLL mail modifyTimestamp > primaryemails

ldapsearch -x -W -D cn=admin,dc=openmrs,dc=org -b ou=users,dc=openmrs,dc=org "(otherMailbox=*)" -LLL otherMailbox > secondaryemails

results

Total:
   6030 email addresses
   130 email address duplicated
   142 times.
   128 of these are primary-only duplicates.
   0 of these are secondary-only duplicates.