<?php

// THIS SCRIPT REQUIRES THE PHP5 fputcsv() FUNCTION

if(isset($_SERVER['SERVER_PORT']) && is_numeric($_SERVER['SERVER_PORT'])) {
  die('# php -q contacts_import.php only');
}
// START FILTER
// Sorts

require_once './_import/common.inc';
$import = fopen('./_import/CNUContactstest.csv', 'r');
//$reject_email = fopen('./_import/CNURejectEmail.csv', 'w');
//$reject_member = fopen('./_import/CNURejectMembership.csv', 'w');
$reject_dup = fopen('./_import/CNURejectDuplicated.csv', 'w');
$export = fopen('./_import/CNUFilteredtest.csv', 'w');

set_time_limit(0);

//fgetcsv($import, 1000, ','); // skip column headers
$lines = array();

// load whole file in to memory
echo "Filtering blank emails\n";
while (($line = fgetcsv($import, 1000, ',')) !== FALSE) {
  $line = array_map('trim_data', $line);
/* can still be input in civicrm
  if(empty($line[$n['EMAIL']])) {
    // drupal requires an email address
    fputcsv($reject_email, $line);
    continue;
  } */
  $lines[] = $line;
}
//fclose($reject_email);
fclose($import);
unset($line, $import, $reject_email);
reset($lines);

/* just don't add them to the cnu membership group
echo "Filtering missing membership\n";
while(list($key, $line) = each($lines)) {
  if(empty($line[$n['MEMBERTHROUGHMONTH']]) && empty($line[$n['MEMBERTHROUGHYEAR']])) {
    // drupal requires an email address
    fputcsv($reject_member, $line);
    unset($lines[$key]);
  }
}
fclose($reject_member);
unset($line, $reject_member);
reset($lines); */

// remove duplicate names and emails
// probably not the fastest way to do this
// maybe speed it up if the array elements were sorted and
// compared to the elements on each side of each other
echo "Filtering duplicates";
while(list(, $filter_value) = each($lines)) {
  $dups = array_filter($lines, 'find_dups');
  if(count($dups) > 1) {
    // $dups should always find at least one record as the list() value finds the $lines value
    while(list($dup_key, $filter_dup) = each($dups)) {
      unset($lines[$dup_key]);
      fputcsv($reject_dup, $filter_dup);
    }
  }
  echo '.';
}
fclose($reject_dup);
unset($dups, $n);
reset($lines);

echo "\nWriting clean records";
while(list(, $line) = each($lines)) {
  fputcsv($export, $line);
}
fclose($export);

/**
 * don't think you need to strip with fgetcsv, but maybe trim
 *
 * @deprecated
 * @return string
 */
function trim_data($value) {
  return trim($value);
}

/**
 * slow...
 *
 * @param array $var a line of the csv file
 * @return bool
 */
function find_dups($var) {
  global $filter_value, $n;
  if (!empty($var[$n['EMAIL']])) {
    return !strcasecmp($var[$n['FNAME']] .' '. $var[$n['LNAME']], $filter_value[$n['FNAME']] .' '. $filter_value[$n['LNAME']]) || !strcasecmp($var[$n['EMAIL']], $filter_value[$n['EMAIL']]);
  }
  else {
    return !strcasecmp($var[$n['FNAME']] .' '. $var[$n['LNAME']], $filter_value[$n['FNAME']] .' '. $filter_value[$n['LNAME']]);
  }
//  return !strcmp($var[$n['FNAME']] .' '. $var[$n['LNAME']], $filter_value[$n['FNAME']] .' '. $filter_value[$n['LNAME']]) || !strcmp($var[$n['EMAIL']], $filter_value[$n['EMAIL']]);
//  return ($var[$n['FNAME']] .' '. $var[$n['LNAME']] == $filter_value[$n['FNAME']] .' '. $filter_value[$n['LNAME']]) || $var[$n['EMAIL']] == $filter_value[$n['EMAIL']];
}