<?php
// $Id: browscap.module,v 1.6 2006-04-13 15:49:30 unconed Exp $

/**
 * @file
 * Replacement for PHP's get_browser() function
 */

/**
 * Implementation of hook_help().
 */
function browscap_help($section) {
  switch ($section) {
    case 'admin/modules#description':
      return t('Provides statistics on browsers, and a replacement for '.
        "PHP's get_browser() function.");
  }
}

function browscap_get_browser($useragent = NULL, $return_array = FALSE) {
  if (!$useragent) {
    $useragent = $_SERVER['HTTP_USER_AGENT'];
  }
  // Cache the results
  $cacheid = 'browscap:'.$useragent;
  $cache = cache_get($cacheid);
  if ((!empty($cache)) and ($cache->created > time() - 60*60*24)) {
    // Found a fresh entry in the cache
    $browserinfo = unserialize($cache->data);
  } else {
    // Note the 'backwards' use of LIKE - the useragent column contains
    // the wildcarded pattern to match against our full-length string
    // The ORDER BY chooses the most-specific matching pattern
    $browserinfo = db_fetch_object(db_query_range(
      "SELECT * from {browscap} WHERE '%s' LIKE useragent ORDER BY LENGTH(useragent) DESC",
      $useragent, 0, 1));
    // A couple of fieldnames not in our database, provided for
    // compatibility with PHP's get_browser()
    $browserinfo->browser_name_pattern = strtr($browserinfo->useragent, '%_', '*?');
    $browserinfo->tables = $browserinfo->htmltables;
    cache_set($cacheid, serialize($browserinfo));
  }

  if ($return_array) {
    return get_object_vars($browserinfo);
  } else {
    return $browserinfo;
  }
}

/**
 * Implementation of hook_cron().
 */
function browscap_cron() {
  // Has it been a week since the last (attempt to) import?
  $last_imported = variable_get('browscap_imported', 0);
  if (($last_imported + 60*60*24*7) < time()) {
    _browscap_import();
    variable_set('browscap_imported', time());
  }
}

// A numeric interpretation of browscap.csv's TRUE/FALSE/default fields
function _browscap_boolean($value) {
  switch ($value) {
    case 'TRUE':
    case 'true':
      return 1;
    case 'FALSE':
    case 'false':
    case 'default':
    default:
      return 0;
  }
}

/**
 * If there's a new version of browscap.csv, fetch it and update the
 * database.
 */
function _browscap_import() {
  // Politely check the version for updates before fetching the file
  $versionpage = drupal_http_request('http://www.garykeith.com/browsers/version.asp');
  if ($versionpage->error) {
    watchdog('browscap', "Couldn't check version: ".$versionpage->error);
    return;
  }
  $browscapversion = trim($versionpage->data);
  $oldversion = variable_get('browscap_version', 'Never fetched');
  if ($browscapversion == $oldversion) {
    // No update, nothing to do here
    watchdog('browscap', 'No new version of browscap to import');
    return;
  }

  // Fetch the new version, and dump it in the temp directory
  $server = $_SERVER['SERVER_NAME'];
  $path = variable_get('file_directory_temp', '/tmp');
  $browscapfile = "$path/browscap_$server.csv";

  // If we can, download the zipped version and extract the file
  if (is_callable('zip_open')) {
    $browscapzipfile = "$path/csv_browscap_$server.zip";
    $browscapzip = drupal_http_request('http://www.garykeith.com/browsers/stream.asp?CSV_BrowsCapZIP');
    if ($browscapzip->error or !trim($browscapzip->data)) {
      watchdog('browscap', "Couldn't retrieve updated browscap: ".$browscapzip->error);
      return;
    }
    $browscapzipfp = fopen($browscapzipfile, "w");
    fwrite($browscapzipfp, $browscapzip->data);
    fclose($browscapzipfp);
    $zip = zip_open($browscapzipfile);
    if ($zip) {
      while ($zip_entry = zip_read($zip)) {
        if (zip_entry_name($zip_entry) == 'browscap.csv') {
          if (zip_entry_open($zip, $zip_entry, 'rb')) {
            $browscapfp = fopen($browscapfile, 'w');
            fwrite($browscapfp, zip_entry_read($zip_entry, zip_entry_filesize($zip_entry)));
            fclose($browscapfp);
            break;
          }
        }
      }
      zip_close($zip);
    }
    unlink($browscapzipfile);
  // Can't handle zip, get the unzipped version
  } else {
    $browscap = drupal_http_request('http://www.garykeith.com/browsers/stream.asp?BrowsCapCSV');
    if ($browscap->error or !trim($browscap->data)) {
      watchdog('browscap', "Couldn't retrieve updated browscap: ".$browscap->error);
      return;
    }
    $browscapfp = fopen($browscapfile, "w");
    fwrite($browscapfp, $browscap->data);
    fclose($browscapfp);
  }

  // Parse the .csv file, importing each row (except the first two)
  // into the {browscap} table
  $browscapfp = fopen($browscapfile, 'r');
  if ($browscapfp) {
    // Ignore the first two rows (column headers & version info)
    fgetcsv($browscapfp, 1000); fgetcsv($browscapfp, 1000);
    while (($browserinfo = fgetcsv($browscapfp, 1000)) != FALSE) {
      // Strip brackets
      $useragent = substr($browserinfo[0], 1, -1);
      // Replace wildcards with SQL equivalents
      $useragent = strtr($useragent, '*?', '%_');
      $parent = $browserinfo[1];
      $browser = $browserinfo[2];
      $version = $browserinfo[3];
      $majorversion = $browserinfo[4];
      $minorversion = $browserinfo[5];
      $platform = $browserinfo[6];
      $authenticodeupdate = $browserinfo[7];
      if ($browserinfo[8] == 'default') {
        $cssversion = -1;
      } else {
        $cssversion = $browserinfo[8];
      }
      $frames = _browscap_boolean($browserinfo[9]);
      $iframes = _browscap_boolean($browserinfo[10]);
      $htmltables = _browscap_boolean($browserinfo[11]);
      $cookies = _browscap_boolean($browserinfo[12]);
      $backgroundsounds = _browscap_boolean($browserinfo[13]);
      $vbscript = _browscap_boolean($browserinfo[14]);
      $javascript = _browscap_boolean($browserinfo[15]);
      $javaapplets = _browscap_boolean($browserinfo[16]);
      $activexcontrols = _browscap_boolean($browserinfo[17]);
      $cdf = _browscap_boolean($browserinfo[18]);
      $aol = _browscap_boolean($browserinfo[19]);
      $beta = _browscap_boolean($browserinfo[20]);
      $win16 = _browscap_boolean($browserinfo[21]);
      $crawler = _browscap_boolean($browserinfo[22]);
      $stripper = _browscap_boolean($browserinfo[23]);
      $wap = _browscap_boolean($browserinfo[24]);
      $netclr = _browscap_boolean($browserinfo[25]);

      db_query('REPLACE {browscap} (useragent,parent,browser,version,'.
        'majorversion,minorversion,platform,authenticodeupdate,'.
        'cssversion,frames,iframes,htmltables,cookies,backgroundsounds,'.
        'vbscript,javascript,javaapplets,activexcontrols,cdf,aol,'.
        'beta,win16,crawler,stripper,wap,netclr) '.
        "VALUES('%s','%s','%s','%s','%s','%s','%s','%s',%d,%d,%d,%d,%d,".
        '%d,%d,%d,%d,%d,%d,%d,%d,%d,%d,%d,%d,%d)',
        $useragent, $parent, $browser, $version,
        $majorversion, $minorversion, $platform, $authenticodeupdate,
        $cssversion, $frames, $iframes, $htmltables, $cookies, $backgroundsounds,
        $vbscript, $javascript, $javaapplets, $activexcontrols, $cdf, $aol,
        $beta, $win16, $crawler, $stripper, $wap, $netclr);
    }
    fclose($browscapfp);
    unlink($browscapfile);

    // Phase 2 - for every entry which isn't a top-level entry, plug
    // in the data from its parent
    $query = 'SELECT useragent,parent from {browscap} WHERE useragent <> parent';
    $result = db_query($query);
    while ($destrow = db_fetch_object($result)) {
      $query = "SELECT * FROM {browscap} WHERE useragent='%s'";
      $srcrow = db_fetch_object(db_query($query, $destrow->parent));
      db_query("UPDATE {browscap} SET browser='%s',version='%s',".
        "majorversion='%s',minorversion='%s',platform='%s',".
        "authenticodeupdate='%s',cssversion=%d,frames=%d,iframes=%d,".
        "htmltables=%d,cookies=%d,backgroundsounds=%d,vbscript=%d,".
        "javascript=%d,javaapplets=%d,activexcontrols=%d,cdf=%d,".
        "aol=%d,beta=%d,win16=%d,crawler=%d,stripper=%d,wap=%d,netclr=%d ".
        "WHERE useragent='%s'",
        $srcrow->browser,$srcrow->version,
        $srcrow->majorversion,$srcrow->minorversion,$srcrow->platform,
        $srcrow->authenticodeupdate,$srcrow->cssversion,$srcrow->frames,$srcrow->iframes,
        $srcrow->htmltables,$srcrow->cookies,$srcrow->backgroundsounds,$srcrow->vbscript,
        $srcrow->javascript,$srcrow->javaapplets,$srcrow->activexcontrols,$srcrow->cdf,
        $srcrow->aol,$srcrow->beta,$srcrow->win16,$srcrow->crawler,$srcrow->stripper,$srcrow->wap,$srcrow->netclr,
        $destrow->useragent);
    }
  // All done updating the browscap info - invalidate cached data
  // from the last version, and record the version we're currently
  // using
  cache_clear_all('browscap:', TRUE);
  variable_set('browscap_version', $browscapversion);
  watchdog('browscap', "New version of browscap imported: $browscapversion");
  }
}

/**
 * Implementation of hook_exit().
 *
 * Keep tabs on browsers that visit
 */
function browscap_exit() {
  // No point if statistics aren't enabled
  if (!module_exist('statistics')) {
    return;
  }

  // If monitoring is enabled, record the browser
  if (variable_get('browscap_monitor', FALSE)) {
    $browser = browscap_get_browser();
    $browserstring = substr(trim($browser->parent), 0, 255);
    if ($browserstring == '' or $browserstring == 'Default Browser') {
      $browserstring = trim($_SERVER['HTTP_USER_AGENT']);
    }
    db_query("UPDATE {browscap_statistics} SET counter = counter + 1, is_crawler=%d ".
      "WHERE parent='%s'", $browser->crawler, $browserstring);
    // If we affected 0 rows, this is the first time we've seen this browser
    if (!db_affected_rows()) {
      // We must create a new row to store counters for the new browser.
      db_query('INSERT INTO {browscap_statistics} (parent,counter,is_crawler) '.
        "VALUES('%s', 1, %d)", $browserstring, $browser->crawler);
    }
  }
}

/*
 * Undo a recorded browser visit by request
 *
 * This function serves the statistics_filter module, enabling it
 * to ignore visits from specified roles.
 */
function browscap_unmonitor() {
  // No point if statistics aren't enabled
  if (!module_exist('statistics')) {
    return;
  }

  // If monitoring is enabled, unrecord the browser
  if (variable_get('browscap_monitor', FALSE)) {
    $browser = browscap_get_browser();
    $browserstring = trim($browser->parent);
    if ($browserstring == '' or $browserstring == 'Default Browser') {
      $browserstring = trim($_SERVER['HTTP_USER_AGENT']);
    }
    db_query("UPDATE {browscap_statistics} SET counter = counter - 1, is_crawler=%d ".
      "WHERE parent='%s'", $browser->crawler, $browserstring);
  }
}

/**
 * Implementation of hook_perm().
 *
 * @return array
 */
function browscap_perm() {
  return array('administer browscap');
}

/**
 * Turn monitor on or off
 *
 * @return array
 */
function browscap_settings() {
  // Restrict administration of this module
  if (user_access('administer browscap') === false) {
    return drupal_access_denied();
  }

  $form['browscap_monitor'] = array(
    '#type' => 'checkbox',
    '#prefix' => t('<p>Browscap data current as of %fileversion</p>',
                   array('%fileversion' => variable_get('browscap_version', 'Never fetched'))),
    '#title' => t('Monitor browsers'),
    '#default_value' => variable_get('browscap_monitor', FALSE),
    '#description' => t('Monitor all user agents visiting the site.')
  );
  return $form;
}

/**
 * Implementation of hook_menu().
 *
 * @return array
 */
function browscap_menu($may_cache) {
  $items = array();

  if ($may_cache && variable_get('browscap_monitor', FALSE)) {
    $access = user_access('access statistics');

    $items[] = array('path' => 'admin/logs/useragents',
      'title' => t('user agents'), 'callback' => 'browscap_top_useragents',
      'access' => $access, 'weight' => 5);
    $items[] = array('path' => 'admin/logs/useragents/browsers',
      'title' => t('browsers'), 'access' => $access);
    $items[] = array('path' => 'admin/logs/useragents/crawlers',
      'title' => t('crawlers'), 'access' => $access);
  }

  return $items;
}

/**
 * Menu callback; presents the user agents monitoring page.
 *
 * @param $view
 *   - "browsers": Only display "real" browsers
 *   - "crawlers": Only display search engine crawlers
 *   - "all": Display all user agents.
 */
function browscap_top_useragents($view = 'all') {
  if ($view == 'all') {
    $result = db_query('SELECT SUM(counter) FROM {browscap_statistics}');
    $total = db_result($result);
    if (!$total) $total = 1;
    $query = "SELECT parent,counter,(100*counter)/$total as percent,is_crawler FROM {browscap_statistics}";
    $query_cnt = 'SELECT COUNT(parent) FROM {browscap_statistics}';
    $title = t('Top user agents');
    $header = array(
      array('data' => t('User agent'), 'field' => 'parent'),
      array('data' => t('Count'), 'field' => 'counter', 'sort' => 'desc'),
      array('data' => t('Percent'), 'field' => 'percent'),
      array('data' => t('Crawler?'), 'field' => 'is_crawler')
    );
  }
  elseif ($view == 'browsers') {
    $result = db_query('SELECT SUM(counter) FROM {browscap_statistics} WHERE is_crawler=0');
    $total = db_result($result);
    if (!$total) $total = 1;
    $query = "SELECT parent,counter,(100*counter)/$total as percent FROM {browscap_statistics} WHERE is_crawler=0";
    $query_cnt = 'SELECT COUNT(parent) FROM {browscap_statistics} WHERE is_crawler=0';
    $title = t('Top browsers');
    $header = array(
      array('data' => t('Browser'), 'field' => 'parent'),
      array('data' => t('Count'), 'field' => 'counter', 'sort' => 'desc'),
      array('data' => t('Percent'), 'field' => 'percent')
    );
  } else {
    $result = db_query('SELECT SUM(counter) FROM {browscap_statistics} WHERE is_crawler=1');
    $total = db_result($result);
    if (!$total) $total = 1;
    $query = "SELECT parent,counter,(100*counter)/$total as percent FROM {browscap_statistics} WHERE is_crawler=1";
    $query_cnt = 'SELECT COUNT(parent) FROM {browscap_statistics} WHERE is_crawler=1';
    $title = t('Top crawlers');
    $header = array(
      array('data' => t('Crawler'), 'field' => 'parent'),
      array('data' => t('Count'), 'field' => 'counter', 'sort' => 'desc'),
      array('data' => t('Percent'), 'field' => 'percent')
    );
  }

  $query .= tablesort_sql($header);

  $result = pager_query($query, 50, 0, $query_cnt);

  while ($useragent = db_fetch_object($result)) {
    if ($view == 'all') {
      if ($useragent->is_crawler) {
        $is_crawler = 'Yes';
      } else {
        $is_crawler = 'No';
      }
      $rows[] = array($useragent->parent, $useragent->counter, $useragent->percent, $is_crawler);
    } else {
      $rows[] = array($useragent->parent, $useragent->counter, $useragent->percent);
    }
  }
  if ($pager = theme('pager', NULL, 50, 0)) {
    $rows[] = array(array('data' => $pager, 'colspan' => 2));
  }

  $output .= theme('table', $header, $rows);

  print theme('page', $output, $title);
}

?>