tools/scripts/solr/createSolrJSON.php

   1 <?php
   2 /*
   3  +--------------------------------------------------------------------+
   4  | CiviCRM version 4.6                                                |
   5  +--------------------------------------------------------------------+
   6  | Copyright CiviCRM LLC (c) 2004-2014                                |
   7  +--------------------------------------------------------------------+
   8  | This file is a part of CiviCRM.                                    |
   9  |                                                                    |
  10  | CiviCRM is free software; you can copy, modify, and distribute it  |
  11  | under the terms of the GNU Affero General Public License           |
  12  | Version 3, 19 November 2007.                                       |
  13  |                                                                    |
  14  | CiviCRM is distributed in the hope that it will be useful, but     |
  15  | WITHOUT ANY WARRANTY; without even the implied warranty of         |
  16  | MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.               |
  17  | See the GNU Affero General Public License for more details.        |
  18  |                                                                    |
  19  | You should have received a copy of the GNU Affero General Public   |
  20  | License along with this program; if not, contact CiviCRM LLC       |
  21  | at info[AT]civicrm[DOT]org. If you have questions about the        |
  22  | GNU Affero General Public License or the licensing of CiviCRM,     |
  23  | see the CiviCRM license FAQ at http://civicrm.org/licensing        |
  24  +--------------------------------------------------------------------+
  25 */
  26
  27 /**
  28  * Create a xml file for a set of contact ID's in a format digestible
  29  * by Solr
  30  */
  31
  32 require_once '../../../civicrm.settings.php';
  33 require_once 'CRM/Core/Config.php';
  34
  35 define('CHUNK_SIZE', 128);
  36
  37 /**
  38  * Split a large array of contactIDs into more manageable smaller chunks
  39  */
  40 function &splitContactIDs(&$contactIDs) {
  41   // contactIDs could be a real large array, so we split it up into
  42   // smaller chunks and then general xml for each chunk
  43   $chunks           = array();
  44   $current          = 0;
  45   $chunks[$current] = array();
  46   $count            = 0;
  47
  48   foreach ($contactIDs as $cid) {
  49     $chunks[$current][] = $cid;
  50     $count++;
  51
  52     if ($count == CHUNK_SIZE) {
  53       $current++;
  54       $chunks[$current] = array();
  55       $count = 0;
  56     }
  57   }
  58
  59   if (empty($chunks[$current])) {
  60     unset($chunks[$current]);
  61   }
  62
  63   return $chunks;
  64 }
  65
  66 /**
  67  * Given an array of values, generate the JSON in the Solr format
  68  */
  69 function &generateSolrJSON($values) {
  70   $result = "[";
  71   foreach ($values as $cid => $tokens) {
  72     if (empty($tokens)) {
  73       continue;
  74     }
  75
  76     $result .= "\n  {\n    \"contact_id\" : \"$cid\",";
  77
  78     foreach ($tokens as $n => $v) {
  79       if (is_array($v)) {
  80         $str = array();
  81         foreach ($v as $el) {
  82           $el = escapeJsonString($el);
  83           $str[] = "\"$el\"";
  84         }
  85         $string = implode(",", $str);
  86         $result .= "\n    \"{$n}\" : [$string],";
  87       }
  88       else {
  89         $v = escapeJsonString($v);
  90         $result .= "\n    \"{$n}\" : \"{$v}\",";
  91       }
  92     }
  93
  94     // remove the last comma
  95     $result = rtrim($result, ",");
  96
  97     $result .= "\n  },";
  98   }
  99   // remove the last comma
 100   $result = rtrim($result, ",");
 101
 102   $result .= "\n]\n";
 103
 104
 105   return $result;
 106 }
 107
 108 /**
 109  * @param $value
 110  *
 111  * @return mixed
 112  */
 113 function escapeJsonString($value) {
 114   $escapers = array("\\", "/", "\"", "\n", "\r", "\t", "\x08", "\x0c");
 115   $replacements = array("\\\\", "\\/", "\\\"", "\\n", "\\r", "\\t", "\\f", "\\b");
 116   return str_replace($escapers, $replacements, $value);
 117 }
 118
 119 /**
 120  * Given a set of contact IDs get the values
 121  */
 122 function getValues(&$contactIDs, &$values) {
 123   $values = array();
 124
 125   foreach ($contactIDs as $cid) {
 126     $values[$cid] = array();
 127   }
 128
 129   getContactInfo($contactIDs, $values);
 130   getAddressInfo($contactIDs, $values);
 131   getPhoneInfo($contactIDs, $values);
 132   getEmailInfo($contactIDs, $values);
 133   getNoteInfo($contactIDs, $values);
 134
 135   return $values;
 136 }
 137
 138 /**
 139  * @param $contactIDs
 140  * @param $values
 141  * @param $tableName
 142  * @param $fields
 143  * @param $whereField
 144  * @param null $additionalWhereCond
 145  */
 146 function getTableInfo(&$contactIDs, &$values, $tableName, &$fields, $whereField, $additionalWhereCond = NULL) {
 147   $selectString = implode(',', array_keys($fields));
 148   $idString = implode(',', $contactIDs);
 149
 150   $sql = "
 151 SELECT $selectString, $whereField as contact_id
 152   FROM $tableName
 153  WHERE $whereField IN ( $idString )
 154 ";
 155
 156   if ($additionalWhereCond) {
 157     $sql .= " AND $additionalWhereCond";
 158   }
 159
 160   $dao = &CRM_Core_DAO::executeQuery($sql);
 161   while ($dao->fetch()) {
 162     foreach ($fields as $fld => $name) {
 163       $name = $name ? $name : $fld;
 164       appendValue($values, $dao->contact_id, $name, $dao->$fld);
 165     }
 166   }
 167 }
 168
 169 /**
 170  * @param $contactIDs
 171  * @param $values
 172  */
 173 function getContactInfo(&$contactIDs, &$values) {
 174   $fields = array('sort_name' => NULL,
 175     'display_name' => NULL,
 176     'contact_type' => NULL,
 177     'legal_identifier' => NULL,
 178     'external_identifier' => NULL,
 179     'first_name' => NULL,
 180     'last_name' => NULL,
 181     'middle_name' => NULL,
 182     'household_name' => NULL,
 183     'organization_name' => NULL,
 184     'legal_name' => NULL,
 185     'job_title' => NULL,
 186   );
 187   getTableInfo($contactIDs, $values, 'civicrm_contact', $fields, 'id');
 188 }
 189
 190 /**
 191  * @param $contactIDs
 192  * @param $values
 193  */
 194 function getNoteInfo(&$contactIDs, &$values) {
 195   $ids = implode(',', $contactIDs);
 196
 197   $sql = "
 198 SELECT
 199       entity_id as contact_id,
 200       note as note, subject as subject
 201 FROM  civicrm_note
 202 WHERE entity_id IN ( $ids )
 203 AND   entity_table = 'civicrm_contact'
 204 ";
 205
 206   $dao = &CRM_Core_DAO::executeQuery($sql);
 207   while ($dao->fetch()) {
 208     $note = empty($dao->subject) ? '' : "{$dao->subject}: ";
 209     $note .= empty($dao->note) ? '' : $dao->note;
 210
 211     appendValue($values, $dao->contact_id, 'note', $note);
 212   }
 213 }
 214
 215 /**
 216  * @param $contactIDs
 217  * @param $values
 218  */
 219 function getPhoneInfo(&$contactIDs, &$values) {
 220   $ids = implode(',', $contactIDs);
 221
 222   $sql = "
 223 SELECT
 224   c.id as contact_id,
 225   l.name as location_type,
 226   p.phone as phone,
 227   v.label as phone_type
 228 FROM      civicrm_contact c
 229 INNER JOIN civicrm_phone          p  ON p.contact_id        = c.id
 230 LEFT  JOIN civicrm_location_type  l  ON p.location_type_id  = l.id
 231 LEFT  JOIN civicrm_option_group   g  ON g.name = 'phone_type'
 232 LEFT  JOIN civicrm_option_value   v  ON v.option_group_id = g.id AND p.phone_type_id = v.value
 233 WHERE      c.id IN ( $ids )
 234 AND        p.phone IS NOT NULL
 235 ";
 236
 237   $dao = &CRM_Core_DAO::executeQuery($sql);
 238   while ($dao->fetch()) {
 239     $phone = '';
 240
 241     if (!empty($dao->location_type)) {
 242       $phone = "{$dao->location_type}: ";
 243     }
 244
 245     $phone .= $dao->phone;
 246
 247     if (!empty($dao->phone_type)) {
 248       $phone .= " ({$dao->phone_type})";
 249     }
 250
 251     appendValue($values, $dao->contact_id, 'phone', $phone);
 252   }
 253 }
 254
 255 /**
 256  * @param $contactIDs
 257  * @param $values
 258  */
 259 function getEmailInfo(&$contactIDs, &$values) {
 260   $ids = implode(',', $contactIDs);
 261
 262   $sql = "
 263 SELECT
 264   c.id as contact_id,
 265   l.name as location_type,
 266   e.email as email
 267 FROM      civicrm_contact c
 268 INNER JOIN civicrm_email          e  ON e.contact_id        = c.id
 269 LEFT  JOIN civicrm_location_type  l  ON e.location_type_id  = l.id
 270 WHERE      c.id IN ( $ids )
 271 AND        e.email IS NOT NULL
 272 ";
 273
 274   $dao = &CRM_Core_DAO::executeQuery($sql);
 275   while ($dao->fetch()) {
 276     $email = '';
 277
 278     if (!empty($dao->location_type)) {
 279       $email = "{$dao->location_type}: ";
 280     }
 281
 282     $email .= $dao->email;
 283     appendValue($values, $dao->contact_id, 'email', $email);
 284   }
 285 }
 286
 287 /**
 288  * @param $contactIDs
 289  * @param $values
 290  */
 291 function getAddressInfo(&$contactIDs, &$values) {
 292   $ids = implode(',', $contactIDs);
 293
 294   $sql = "
 295 SELECT     c.id as contact_id, l.name as location_type,
 296            a.street_address, a.supplemental_address_1, a.supplemental_address_2,
 297            a.city, a.postal_code,
 298            s.name as state, co.name as country
 299 FROM       civicrm_contact c
 300 INNER JOIN civicrm_address        a  ON a.contact_id        = c.id
 301 LEFT  JOIN civicrm_location_type  l  ON a.location_type_id  = l.id
 302 LEFT  JOIN civicrm_state_province s  ON a.state_province_id = s.id
 303 LEFT  JOIN civicrm_country        co ON a.country_id        = co.id
 304 WHERE c.id IN ( $ids )
 305 ";
 306
 307   $fields = array('location_type', 'street_address', 'supplemental_address_1',
 308     'supplemental_address_2', 'city', 'postal_code',
 309     'state', 'country',
 310   );
 311   $dao = &CRM_Core_DAO::executeQuery($sql);
 312   while ($dao->fetch()) {
 313     $address = '';
 314     foreach ($fields as $fld) {
 315       if (empty($dao->$fld)) {
 316         continue;
 317       }
 318
 319       $address .= ($fld == 'location_type') ? "{$dao->$fld}: " : " {$dao->$fld},";
 320       appendValue($values, $dao->contact_id, $fld, $dao->$fld);
 321     }
 322
 323     if (!empty($address)) {
 324       $address = rtrim($address, ",");
 325       appendValue($values, $dao->contact_id, 'address', $address);
 326     }
 327   }
 328 }
 329
 330 /**
 331  * @param $values
 332  * @param $contactID
 333  * @param $name
 334  * @param $value
 335  */
 336 function appendValue(&$values, $contactID, $name, $value) {
 337   if (empty($value)) {
 338     return;
 339   }
 340
 341   if (!isset($values[$contactID][$name])) {
 342     $values[$contactID][$name] = $value;
 343   }
 344   else {
 345     if (!is_array($values[$contactID][$name])) {
 346       $save = $values[$contactID][$name];
 347       $values[$contactID][$name] = array();
 348       $values[$contactID][$name][] = $save;
 349     }
 350     $values[$contactID][$name][] = $value;
 351   }
 352 }
 353
 354 /**
 355  * @param $contactIDs
 356  */
 357 function run(&$contactIDs) {
 358   $chunks = &splitContactIDs($contactIDs);
 359
 360   foreach ($chunks as $chunk) {
 361     $values = array();
 362     getValues($chunk, $values);
 363     $xml = &generateSolrJSON($values);
 364     echo $xml;
 365   }
 366 }
 367
 368 $config = &CRM_Core_Config::singleton();
 369 $config->userFramework = 'Soap';
 370 $config->userFrameworkClass = 'CRM_Utils_System_Soap';
 371 $config->userHookClass = 'CRM_Utils_Hook_Soap';
 372
 373 $sql = <<<EOT
 374 SELECT id
 375 FROM civicrm_contact
 376 EOT;
 377 $dao = &CRM_Core_DAO::executeQuery($sql);
 378
 379
 380 $contactIDs = array();
 381 while ($dao->fetch()) {
 382   $contactIDs[] = $dao->id;
 383 }
 384
 385 run($contactIDs);
 386