<?php
// $Id: $

/**
 * @file  /drupal/database/create.php
 * Creates all tables for all sites in /drupal/sites/*.
 *
 * The database connection information (username, password, etc) and
 * table name prefixes are obtained from the various
 * /drupal/sites/<sitename>/settings.php files.
 *
 * To execute this script via a web browser:
 *    http://<domain>/<path/>database/create.php?dbms=<dbms>
 *
 * To execute this script from the command line:
 *    php ./create.php <dbms>
 *
 * Where <dbms> is the Database Management System you are using
 * ('mysql' or 'pgsql').
 *
 * The command line example above assumes 'php' is in your PATH.  If it
 * is not, you need to provide the full path to the PHP executable.
 *
 * This script is configured to execute the table creation procedures
 * only when called directly.  This gives you the option to include
 * the file and use its functions elsewhere (provided the name of
 * your script is _NOT_ named 'create.php').
 */


if (basename($_SERVER['PHP_SELF']) == basename(__FILE__)) {
  if (isset($_GET) && isset($_GET['dbms'])) {
    if (create($_GET['dbms'], 'web')) {
      echo "\nYour tables were created!\n";
    }
    else {
      echo "\nNo tables were created.\n";
    }
  }
  elseif (isset($_SERVER['argv']) && isset($_SERVER['argv'][1])) {
    if (create($_SERVER['argv'][1], 'shell')) {
      echo "\nYour tables were created!\n";
    }
    else {
      echo "\nNo tables were created.\n";
    }
  }
}


/**
 * Creates all tables for all sites in /drupal/sites/*
 *
 * Calls the various functions needed to gather and implement the tables:
 *  + Gets the $db_url and $db_prefix from each settings.php file.
 *  + Gets the SQL script for the DBMS specified.
 *  + Assembles the queries for each site, prefixing table names as needed.
 *  + Bunches all of the sites' queries per db_url and removes duplicates.
 *  + Connects to each db_url and executes the queries for it.
 *
 * WARNING:  If the variable_get() function has not been defined yet,
 * this function will define a fake version of it.  This is necessary
 * to avoid having to include bootstrap.inc
 *
 * @param $dbms
 *   The string with the DBMS' name ('mysql', 'pgsql').
 * @param $interface
 *   The string saying how the user is interacting ('web', 'shell').
 *
 * @return void
 */
function create($dbms, $interface) {
  if ($interface == 'web') {
    $dbms_text = htmlspecialchars($dbms);
  }
  else {
    $dbms_text = $dbms;
  }

  $settings = get_site_settings();
  if (empty($settings)) {
    trigger_error('Found no valid settings.php files in /drupal/sites/*.',
                  E_USER_ERROR);
  }

  $queries = get_queries_from_file($dbms);
  if ($queries === false) {
    trigger_error("'$dbms_text' is an invalid/unsupported DBMS.",
                  E_USER_ERROR);
  }

  define_variable_get();
  $dbms_len = strlen($dbms);

  $success = false;
  foreach ($settings as $db_url => $sites) {
    if (substr($db_url, 0, $dbms_len) != $dbms) {
      echo "Drupal's database functions only work with one DBMS at a time.\n"
         . "Your command requested building tables for '$dbms_text'\n"
         . 'but the $db_url in sites/' . key($sites) . "/settings.php\n"
         . "specified a different DBMS.\n";
      continue;
    }
    $queries_for_db_url = get_queries_for_db_url($sites, $queries);
    execute_arrays_queries($dbms, $db_url, $queries_for_db_url);
    $success = true;
  }
  return $success;
}

/**
 * Obtains configuration settings specified in all settings.php files.
 *
 * NOTE:  This method will not return data from a settings file that has
 * its $db_url set to the default value shipped by Drupal.
 *
 * An example of an array returned by this function:
 * @verbatim
 * Array (
 *   [mysql://usr:pw@localhost/drp] => Array (
 *     [localhost2] => Array (
 *       [default] => site2_
 *       [named] => Array (
 *         [users] => share_
 *         [sessions] => share_
 *       )
 *     )
 *     [localhost3] => Array (
 *       [default] => site3_
 *       [named] => Array (
 *         [users] => share_
 *         [sessions] => share_
 *       )
 *     )
 *     [localhost5] => Array (
 *       [default] => site5_
 *       [named] => Array (
 *         [sessions] => whatever_
 *       )
 *     )
 *   )
 * )
 * @endverbatim
 *
 * @return
 *   The four dimensional associative array settings for all sites.
 *   The main array's keys is the $db_url.
 *   The sub-array's keys are the site directory names.
 *   The sub-sub-array has two keys: 'default' and 'named'.
 *   'named' is an array with table names for keys and table name prefixes
 *   for values.
 *   'default' is a string that should prefix all other tables.
 *   Returns FALSE if the sites directory can not be found.
 */
function get_site_settings() {
  $sites_dir   = '../sites';
  $default_url = 'mysql://username:password@localhost/database';
  $settings    = array();

  $dh = opendir($sites_dir);
  if (!$dh) {
      return false;
  }
  while (false !== ($site = readdir($dh))) {
    if (is_dir($sites_dir . '/' . $site)) {
      switch ($site) {
        case '.':
        case '..':
        case 'CVS':
          continue 2;
      }

      // Get the $db_url and $db_prefix variables.
      include $sites_dir . '/' . $site . '/settings.php';

      if ($db_url == $default_url) {
        continue;
      }
      if (is_string($db_prefix)) {
        $settings[$db_url][$site]['default'] = $db_prefix;
        $settings[$db_url][$site]['named'] = array();
      }
      else {
        if (array_key_exists('default', $db_prefix)) {
          $settings[$db_url][$site]['default'] = $db_prefix['default'];
          unset($db_prefix['default']);
        }
        else {
          $settings[$db_url][$site]['default'] = '';
        }
        $settings[$db_url][$site]['named'] = $db_prefix;
      }
    }
  }
  closedir($dh);
  return $settings;
}

/**
 * Grabs the contents of the query script and puts each query into
 * an array element.
 *
 * @param $dbms
 *   The name of the DBMS you are want the queries for ('mysql', 'pgsql').
 *
 * @return
 *   An array with each query in a separate element.  Returns FALSE if
 *   the database query script does not exist.
 */
function get_queries_from_file($dbms) {
  $file = './database.' . $dbms;
  if (!file_exists($file)) {
    return false;
  }
  $func     = '';
  $contents = '';
  $fh = fopen($file, 'r');
  while (!feof($fh)) {
    $line = fgets($fh, 5000);
    if (preg_match("/^\s*[a-z)']/i", $line)) {
      /*
       * Mush functions into one line because they can contain ';' at the
       * ends of lines and at the end of the query, which trips up the split
       * ';[\r\n]+' that is used to put queries into separate array elements.
       */
      if (preg_match('/^\s*CREATE FUNCTION/i', $line)) {
        // Beginning of CREATE FUNCTION statement.
        $func = trim($line);
      } elseif ($func) {
        // CREATE FUNCTION statement is still in the works.
        $line = ' ' . trim($line);
        if (preg_match("/ ' LANGUAGE '\w+';/i", $line)) {
          // End of CREATE FUNCTION statement.
          $contents .= $func . $line . "\n";
          $func = '';
        }
        else {
          // Middle of CREATE FUNCTION statement.
          $func .= $line;
        }
      } else {
        // Regular query content.
        $contents .= $line;
      }
    }
  }
  $queries = preg_split('/;[\r\n]+/', $contents);
  $last_key = count($queries) - 1;
  if (trim($queries[$last_key]) == '') {
    unset($queries[$last_key]);
  }
  return $queries;
}

/**
 * Assembles all of the queries for a db_url.
 *
 * The process includes getting the table names prefixed (via
 * prefix_queries_in_array()) and then eliminating duplicate queries.
 *
 * @param $sites
 *   The associative array of sites containing the sub-array of its
 *   table name prefixes.
 * @param $queries
 *   The enumerated array containing one query per element.
 *
 * @return
 *   The array of queries for the db_url.
 */
function get_queries_for_db_url($sites, $queries) {
  $out = array();
  foreach ($sites as $site => $prefixes) {
    $out = array_merge($out,
                       prefix_queries_in_array($queries,
                                               $prefixes['default'],
                                               $prefixes['named']));
  }
  return array_unique($out);
}

/**
 * Goes through each element of the queries array, prefixing the table
 * names as needed.
 *
 * @param $queries
 *   The enumerated array containing one query per element.
 * @param $default
 *   The string for prefixing all other tables.
 * @param $named
 *   The associative array of specific tables to prefix.
 *   Each key is a table name and the value is the prefix for that table.
 *
 * @return
 *   The array of queries with the table names prefixed as needed.
 */
function prefix_queries_in_array($queries, $default, $named) {
  $commands = 'CREATE TABLE|CREATE SEQUENCE|'
            . 'ALTER TABLE|ALTER SEQUENCE|'
            . 'INSERT INTO|UPDATE|REPLACE|DELETE FROM|'
            . 'DROP TABLE|DROP SEQUENCE|DROP INDEX';
  $out = array();

  foreach ($named as $table => $prefix) {
    foreach ($queries as $key => $query) {
      $query = ltrim($query);
      if (preg_match("/^($commands) ($table)([\s(]+.*)$/is",
                     $query,
                     $match))
      {
        $out[$key] = $match[1] . ' ' . $prefix . $match[2] . $match[3];
        unset($queries[$key]);
      }
      elseif (preg_match("/^CREATE INDEX ([a-z0-9_]+) ON ($table)([\s(]+.*)$/is",
                         $query,
                         $match))
      {
        $out[$key] = 'CREATE INDEX ' . $prefix . $match[1]
                   . ' ON ' . $prefix . $match[2] . $match[3];
        unset($queries[$key]);
      }
    }
  }
  if (trim($default) == '') {
    return array_merge($out, $queries);
  }
  foreach ($queries as $key => $query) {
    $query = ltrim($query);
    if (preg_match("/^($commands) ([a-z0-9_]+)([\s(]+.*)$/is",
                   $query,
                   $match))
    {
      $out[$key] = $match[1] . ' ' . $default . $match[2] . $match[3];
    }
    elseif (preg_match('/^CREATE INDEX ([a-z0-9_]+) ON ([a-z0-9_]+)([\s(]+.*)$/is',
                       $query,
                       $match))
    {
      $out[$key] = 'CREATE INDEX ' . $default . $match[1]
                 . ' ON ' . $default . $match[2] . $match[3];
    }
    else {
      $out[$key] = $query;
    }
  }
  ksort($out);
  return $out;
}

/**
 * Runs all queries in an array.
 *
 * @param $dbms
 *   The string with the DBMS' name ('mysql', 'pgsql').
 * @param $db_url
 *   The string holding the database connection information.
 * @param $queries
 *   The array containing one query per element.
 *
 * @return void
 */
function execute_arrays_queries($dbms, $db_url, $queries) {
  global $active_db;
  $includes_dir = '../includes';

  require_once $includes_dir . '/database.' . $dbms . '.inc';

  /*
   * Don't need to catch errors here due to die() in db_connect().
   *
   * The return value must be stored in a varialbe named $active_db
   * because that's the connection variable used in _db_query().
   */
  $active_db = db_connect($db_url);

  ksort($queries);
  foreach ($queries as $query) {
    // Don't need to catch errors here due to triger_error() in _db_query().
    _db_query($query);
  }
}

/**
 * Defines a fake version of variable_get() if it hasn't been defined yet.
 *
 * Necessary becasue we don't want to include bootstrap.inc.
 *
 * @return void
 */
function define_variable_get() {
  if (!function_exists('variable_get')) {
    function variable_get($name, $default) {
      return $default;
    }
  }
}

?>