tools/scripts/solr/createSolrXML.php

   1 <?php
   2 /*
   3  +--------------------------------------------------------------------+
   4  | Copyright CiviCRM LLC. All rights reserved.                        |
   5  |                                                                    |
   6  | This code is published under the GNU AGPLv3 license with some      |
   7  | permitted exceptions and without any warranty. For full license    |
   8  | and copyright information, see https://civicrm.org/licensing       |
   9  +--------------------------------------------------------------------+
  10 */
  11
  12 /**
  13  * Create a xml file for a set of contact ID's in a format digestible
  14  * by Solr
  15  */
  16
  17 require_once '../../civicrm.config.php';
  18 require_once 'CRM/Core/Config.php';
  19
  20 define('CHUNK_SIZE', 128);
  21
  22 /**
  23  * Split a large array of contactIDs into more manageable smaller chunks
  24  * @param $contactIDs
  25  * @return array
  26  */
  27 function &splitContactIDs(&$contactIDs) {
  28   // contactIDs could be a real large array, so we split it up into
  29   // smaller chunks and then general xml for each chunk
  30   $chunks = [];
  31   $current = 0;
  32   $chunks[$current] = [];
  33   $count = 0;
  34
  35   foreach ($contactIDs as $cid) {
  36     $chunks[$current][] = $cid;
  37     $count++;
  38
  39     if ($count == CHUNK_SIZE) {
  40       $current++;
  41       $chunks[$current] = [];
  42       $count = 0;
  43     }
  44   }
  45
  46   if (empty($chunks[$current])) {
  47     unset($chunks[$current]);
  48   }
  49
  50   return $chunks;
  51 }
  52
  53 /**
  54  * Given an array of values, generate the XML in the Solr format
  55  * @param $values
  56  * @return string
  57  */
  58 function &generateSolrXML($values) {
  59   $result = "<add>\n";
  60   foreach ($values as $cid => $tokens) {
  61     if (empty($tokens)) {
  62       continue;
  63     }
  64
  65     $result .= <<<EOT
  66   <doc>
  67     <field name="id">$cid</field>\n
  68 EOT;
  69
  70     foreach ($tokens as $t) {
  71       $result .= <<<EOT
  72     <field name="$t[0]">$t[1]</field>\n
  73 EOT;
  74     }
  75
  76     $result .= "  </doc>\n";
  77   }
  78   $result .= "</add>\n";
  79
  80
  81   return $result;
  82 }
  83
  84 /**
  85  * Given a set of contact IDs get the values
  86  * @param $contactIDs
  87  * @param $values
  88  * @return array
  89  */
  90 function getValues(&$contactIDs, &$values) {
  91   $values = [];
  92
  93   foreach ($contactIDs as $cid) {
  94     $values[$cid] = [];
  95   }
  96
  97   getContactInfo($contactIDs, $values);
  98   getLocationInfo($contactIDs, $values);
  99
 100   return $values;
 101 }
 102
 103 /**
 104  * @param $contactIDs
 105  * @param $values
 106  * @param $tableName
 107  * @param $fields
 108  * @param $whereField
 109  * @param null $additionalWhereCond
 110  */
 111 function getTableInfo(&$contactIDs, &$values, $tableName, &$fields, $whereField, $additionalWhereCond = NULL) {
 112   $selectString = implode(',', array_keys($fields));
 113   $idString = implode(',', $contactIDs);
 114
 115   $sql = "
 116 SELECT $selectString, $whereField as contact_id
 117   FROM $tableName
 118  WHERE $whereField IN ( $idString )
 119 ";
 120
 121   if ($additionalWhereCond) {
 122     $sql .= " AND $additionalWhereCond";
 123   }
 124
 125   $dao = CRM_Core_DAO::executeQuery($sql);
 126   while ($dao->fetch()) {
 127     foreach ($fields as $fld => $name) {
 128       if (empty($dao->$fld)) {
 129         continue;
 130       }
 131       if (!$name) {
 132         $name = $fld;
 133       }
 134       $values[$dao->contact_id][] = [$name, $dao->$fld];
 135     }
 136   }
 137 }
 138
 139 /**
 140  * @param $contactIDs
 141  * @param $values
 142  */
 143 function getContactInfo(&$contactIDs, &$values) {
 144   $fields = [
 145     'sort_name' => NULL,
 146     'display_name' => NULL,
 147     'contact_type' => NULL,
 148     'legal_identifier' => NULL,
 149     'external_identifier' => NULL,
 150     'source' => 'contact_source',
 151   ];
 152   getTableInfo($contactIDs, $values, 'civicrm_contact', $fields, 'id');
 153
 154   $fields = [
 155     'first_name' => NULL,
 156     'last_name' => NULL,
 157     'middle_name' => NULL,
 158     'job_title' => NULL,
 159   ];
 160   getTableInfo($contactIDs, $values, 'civicrm_individual', $fields, 'contact_id');
 161
 162   $fields = ['household_name' => NULL];
 163   getTableInfo($contactIDs, $values, 'civicrm_household', $fields, 'contact_id');
 164
 165   $fields = [
 166     'organization_name' => NULL,
 167     'legal_name' => NULL,
 168     'sic_code' => NULL,
 169   ];
 170   getTableInfo($contactIDs, $values, 'civicrm_organization', $fields, 'contact_id');
 171
 172   $fields = [
 173     'note' => 'note_body',
 174     'subject' => 'note_subject',
 175   ];
 176   getTableInfo($contactIDs, $values, 'civicrm_note', $fields, 'entity_id', "entity_table = 'civicrm_contact'");
 177 }
 178
 179 /**
 180  * @param $contactIDs
 181  * @param $values
 182  */
 183 function getLocationInfo(&$contactIDs, &$values) {
 184   $ids = implode(',', $contactIDs);
 185
 186   $sql = "
 187 SELECT
 188   l.entity_id as contact_id, l.name as location_name,
 189   a.street_address, a.supplemental_address_1, a.supplemental_address_2,
 190   a.supplemental_address_3,
 191   a.city, a.postal_code,
 192   co.name as county, s.name as state, c.name as country,
 193   e.email, p.phone, i.name as im
 194 FROM
 195   civicrm_location l
 196 LEFT JOIN civicrm_address        a  ON a.location_id       = l.id
 197 LEFT JOIN civicrm_email          e  ON e.location_id       = l.id
 198 LEFT JOIN civicrm_phone          p  ON p.location_id       = l.id
 199 LEFT JOIN civicrm_im             i  ON i.location_id       = l.id
 200 LEFT JOIN civicrm_state_province s  ON a.state_province_id = s.id
 201 LEFT JOIN civicrm_country        c  ON a.country_id        = c.id
 202 LEFT JOIN civicrm_county         co ON a.county_id         = co.id
 203 WHERE l.entity_table = 'civicrm_contact'
 204   AND l.entity_id IN ( $ids )
 205 ";
 206
 207   $fields = [
 208     'location_name',
 209     'street_address',
 210     'supplemental_address_1',
 211     'supplemental_address_2',
 212     'supplemental_address_3',
 213     'city',
 214     'postal_code',
 215     'county',
 216     'state',
 217     'country',
 218     'email',
 219     'phone',
 220     'im',
 221   ];
 222   $dao = CRM_Core_DAO::executeQuery($sql);
 223   while ($dao->fetch()) {
 224     foreach ($fields as $fld) {
 225       if (empty($dao->$fld)) {
 226         continue;
 227       }
 228       $values[$dao->contact_id][] = [$fld, $dao->$fld];
 229     }
 230   }
 231 }
 232
 233 /**
 234  * @param $contactIDs
 235  */
 236 function run(&$contactIDs) {
 237   $chunks = &splitContactIDs($contactIDs);
 238
 239   foreach ($chunks as $chunk) {
 240     $values = [];
 241     getValues($chunk, $values);
 242     $xml = &generateSolrXML($values);
 243     echo $xml;
 244   }
 245 }
 246
 247 $config = CRM_Core_Config::singleton();
 248 $config->userFramework = 'Soap';
 249 $config->userFrameworkClass = 'CRM_Utils_System_Soap';
 250 $config->userHookClass = 'CRM_Utils_Hook_Soap';
 251
 252 $sql = <<<EOT
 253 SELECT id
 254 FROM civicrm_contact
 255 EOT;
 256 $dao = CRM_Core_DAO::executeQuery($sql);
 257
 258 $contactIDs = [];
 259 while ($dao->fetch()) {
 260   $contactIDs[] = $dao->id;
 261 }
 262
 263 run($contactIDs);
 264