contrib/util/dupscore.cli.php

   1 #!/usr/bin/php
   2 <?php
   3
   4 /**
   5  * CLI script to compute patient duplication scores in patient_data.dupscore.
   6  * The score is a measure of the likelihood that the patient is a duplicate of
   7  * some patient created before it. Optional arguments specifying values are:
   8  *
   9  * --webdir   The full path to the OpenEMR web directory. Defaults to the directory
  10  *            two levels above that of this script.
  11  * --site     The site ID. Defaults to "default".
  12  * --maxmins  The maximum number of minutes to run. Defaults to 60. Use 0 for no limit.
  13  *
  14  * Arguments not having a value may be:
  15  *
  16  * -q         Suppresses messages on stdout.
  17  * -c         Clears existing scores to recompute all of them; except scores of -1
  18  *            are not cleared because they are manually assigned.
  19  *
  20  * Because we are comparing every patient with every other patient, this script can
  21  * run for a very long time with a large database. Thus we want to do it offline.
  22  * If --maxmins is exceeded the script will terminate but may be run again to resume
  23  * where it left off.
  24  *
  25  * A common usage is:
  26  * php /var/www/html/openemr/contrib/util/dupscore.cli.php --maxmins=240
  27  *
  28  * Here is a sample crontab entry to automatically run up to 2 hours nightly:
  29  * 3 1 * * * root php /var/www/html/openemr/contrib/util/dupscore.cli.php -q --maxmins=120
  30  *
  31  * @package   OpenEMR
  32  * @link      http://www.open-emr.org
  33  * @author    Rod Roark <rod@sunsetsystems.com>
  34  * @copyright Copyright (c) 2021 Rod Roark <rod@sunsetsystems.com>
  35  * @license   https://github.com/openemr/openemr/blob/master/LICENSE GNU General Public License 3
  36  */
  37
  38 // The number of scores to compute between tests for time expiration.
  39 $querylimit = 1000;
  40
  41 if (php_sapi_name() !== 'cli') {
  42     die("This script must be run from the command line!\n");
  43 }
  44
  45 $args = getopt('cq', array('webdir:', 'site:', 'maxmins:'));
  46
  47 // print_r($args); // debugging
  48
  49 $args['webdir'] = $args['webdir'] ?? dirname(dirname(dirname(__FILE__)));
  50 $args['site'] = $args['site'] ?? 'default';
  51 $args['maxmins'] = floatval($args['maxmins'] ?? 60);
  52
  53 if (stripos(PHP_OS, 'WIN') === 0) {
  54     $args['webdir'] = str_replace("\\", "/", $args['webdir']);
  55 }
  56
  57 // Bring in some libraries and settings shared with web scripts.
  58 $_GET['site'] = $args['site'];
  59 $ignoreAuth = 1;
  60 require_once($args['webdir'] . "/interface/globals.php");
  61
  62 // Bring in the getDupScoreSQL() function.
  63 require_once("$srcdir/dupscore.inc.php");
  64
  65 $endtime = time() + 365 * 24 * 60 * 60; // a year from now
  66 if (!empty($args['maxmins'])) {
  67     $endtime = time() + $args['maxmins'] * 60;
  68 }
  69
  70 if (isset($args['c'])) {
  71     // Note -1 means the patient is manually flagged as not a duplicate.
  72     sqlStatementNoLog("UPDATE patient_data SET dupscore = -9 WHERE dupscore != -1");
  73     if (!isset($args['q'])) {
  74         echo xl("All scores have been cleared.") . "\n";
  75     }
  76 }
  77
  78 $count = 0;
  79 $finished = false;
  80
  81 while (!$finished && time() < $endtime) {
  82     $scores = array();
  83     $query1 = "SELECT p1.pid, MAX(" . getDupScoreSQL() . ") AS dupscore" .
  84         " FROM patient_data AS p1, patient_data AS p2" .
  85         " WHERE p1.dupscore = -9 AND p2.pid < p1.pid" .
  86         " GROUP BY p1.pid ORDER BY p1.pid LIMIT " . escape_limit($querylimit);
  87
  88     // echo "$query1\n"; // debugging
  89
  90     $res1 = sqlStatementNoLog($query1);
  91     while ($row1 = sqlFetchArray($res1)) {
  92         $scores[$row1['pid']] = $row1['dupscore'];
  93     };
  94     foreach ($scores as $pid => $score) {
  95         sqlStatementNoLog(
  96             "UPDATE patient_data SET dupscore = ? WHERE pid = ?",
  97             array($score, $pid)
  98         );
  99         ++$count;
 100     }
 101
 102     if (!isset($args['q']) && count($scores) > 0) {
 103         echo "$count... ";
 104     }
 105     if (count($scores) < $querylimit) {
 106         $finished = true;
 107     }
 108 }
 109
 110 if (!isset($args['q'])) {
 111     if (!$count) {
 112         echo xl("No patients without scores were found.");
 113     }
 114     if ($finished) {
 115         echo "\n" . xl("All done.") . "\n";
 116     } else {
 117         echo "\n" . xl("This run is incomplete due to time expiration.") . "\n";
 118     }
 119 }
 120
 121 if (!$finished) {
 122     exit(1);
 123 }