| 1 | <?php |
| 2 | /* |
| 3 | +--------------------------------------------------------------------+ |
| 4 | | Copyright CiviCRM LLC. All rights reserved. | |
| 5 | | | |
| 6 | | This work 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 | * |
| 14 | * @package CRM |
| 15 | * @copyright CiviCRM LLC https://civicrm.org/licensing |
| 16 | */ |
| 17 | |
| 18 | /** |
| 19 | * A PHP cron script to format all the addresses in the database. Currently |
| 20 | * it only does geocoding if the geocode values are not set. At a later |
| 21 | * stage we will also handle USPS address cleanup and other formatting |
| 22 | * issues |
| 23 | */ |
| 24 | class CRM_Utils_Address_BatchUpdate { |
| 25 | |
| 26 | public $start = NULL; |
| 27 | public $end = NULL; |
| 28 | public $geocoding = 1; |
| 29 | public $parse = 1; |
| 30 | public $throttle = 0; |
| 31 | |
| 32 | public $returnMessages = []; |
| 33 | public $returnError = 0; |
| 34 | |
| 35 | /** |
| 36 | * Class constructor. |
| 37 | * |
| 38 | * @param array $params |
| 39 | */ |
| 40 | public function __construct($params) { |
| 41 | |
| 42 | foreach ($params as $name => $value) { |
| 43 | $this->$name = $value; |
| 44 | } |
| 45 | |
| 46 | // fixme: more params verification |
| 47 | } |
| 48 | |
| 49 | /** |
| 50 | * Run batch update. |
| 51 | * |
| 52 | * @return array |
| 53 | */ |
| 54 | public function run() { |
| 55 | |
| 56 | // do check for geocoding. |
| 57 | $processGeocode = FALSE; |
| 58 | if (!CRM_Utils_GeocodeProvider::getUsableClassName()) { |
| 59 | if (CRM_Utils_String::strtobool($this->geocoding) === TRUE) { |
| 60 | $this->returnMessages[] = ts('Error: You need to set a mapping provider under Administer > System Settings > Mapping and Geocoding'); |
| 61 | $this->returnError = 1; |
| 62 | $this->returnResult(); |
| 63 | } |
| 64 | } |
| 65 | else { |
| 66 | $processGeocode = TRUE; |
| 67 | // user might want to over-ride. |
| 68 | if (CRM_Utils_String::strtobool($this->geocoding) === FALSE) { |
| 69 | $processGeocode = FALSE; |
| 70 | } |
| 71 | } |
| 72 | |
| 73 | // do check for parse street address. |
| 74 | $parseAddress = FALSE; |
| 75 | $parseAddress = CRM_Utils_Array::value('street_address_parsing', |
| 76 | CRM_Core_BAO_Setting::valueOptions(CRM_Core_BAO_Setting::SYSTEM_PREFERENCES_NAME, |
| 77 | 'address_options' |
| 78 | ), |
| 79 | FALSE |
| 80 | ); |
| 81 | $parseStreetAddress = FALSE; |
| 82 | if (!$parseAddress) { |
| 83 | if (CRM_Utils_String::strtobool($this->parse) === TRUE) { |
| 84 | $this->returnMessages[] = ts('Error: You need to enable Street Address Parsing under Administer > Localization > Address Settings.'); |
| 85 | $this->returnError = 1; |
| 86 | return $this->returnResult(); |
| 87 | } |
| 88 | } |
| 89 | else { |
| 90 | $parseStreetAddress = TRUE; |
| 91 | // user might want to over-ride. |
| 92 | if (CRM_Utils_String::strtobool($this->parse) === FALSE) { |
| 93 | $parseStreetAddress = FALSE; |
| 94 | } |
| 95 | } |
| 96 | |
| 97 | // don't process. |
| 98 | if (!$parseStreetAddress && !$processGeocode) { |
| 99 | $this->returnMessages[] = ts('Error: Both Geocode mapping as well as Street Address Parsing are disabled. You must configure one or both options to use this script.'); |
| 100 | $this->returnError = 1; |
| 101 | return $this->returnResult(); |
| 102 | } |
| 103 | |
| 104 | // do check for parse street address. |
| 105 | return $this->processContacts($processGeocode, $parseStreetAddress); |
| 106 | } |
| 107 | |
| 108 | /** |
| 109 | * Process contacts. |
| 110 | * |
| 111 | * @param bool $processGeocode |
| 112 | * @param bool $parseStreetAddress |
| 113 | * |
| 114 | * @return array |
| 115 | * @throws Exception |
| 116 | */ |
| 117 | public function processContacts($processGeocode, $parseStreetAddress) { |
| 118 | // build where clause. |
| 119 | $clause = ['( c.id = a.contact_id )']; |
| 120 | $params = []; |
| 121 | if ($this->start) { |
| 122 | $clause[] = "( c.id >= %1 )"; |
| 123 | $params[1] = [$this->start, 'Integer']; |
| 124 | } |
| 125 | |
| 126 | if ($this->end) { |
| 127 | $clause[] = "( c.id <= %2 )"; |
| 128 | $params[2] = [$this->end, 'Integer']; |
| 129 | } |
| 130 | |
| 131 | if ($processGeocode) { |
| 132 | $clause[] = '( a.geo_code_1 is null OR a.geo_code_1 = 0 )'; |
| 133 | $clause[] = '( a.geo_code_2 is null OR a.geo_code_2 = 0 )'; |
| 134 | $clause[] = '( a.country_id is not null )'; |
| 135 | } |
| 136 | |
| 137 | $whereClause = implode(' AND ', $clause); |
| 138 | |
| 139 | $query = " |
| 140 | SELECT c.id, |
| 141 | a.id as address_id, |
| 142 | a.street_address, |
| 143 | a.city, |
| 144 | a.postal_code, |
| 145 | a.country_id, |
| 146 | s.name as state, |
| 147 | o.name as country |
| 148 | FROM civicrm_contact c |
| 149 | INNER JOIN civicrm_address a ON a.contact_id = c.id |
| 150 | LEFT JOIN civicrm_country o ON a.country_id = o.id |
| 151 | LEFT JOIN civicrm_state_province s ON a.state_province_id = s.id |
| 152 | WHERE {$whereClause} |
| 153 | ORDER BY a.id |
| 154 | "; |
| 155 | |
| 156 | $totalGeocoded = $totalAddresses = $totalAddressParsed = 0; |
| 157 | |
| 158 | $dao = CRM_Core_DAO::executeQuery($query, $params); |
| 159 | |
| 160 | $unparseableContactAddress = []; |
| 161 | while ($dao->fetch()) { |
| 162 | $totalAddresses++; |
| 163 | $params = [ |
| 164 | 'street_address' => $dao->street_address, |
| 165 | 'postal_code' => $dao->postal_code, |
| 166 | 'city' => $dao->city, |
| 167 | 'state_province' => $dao->state, |
| 168 | 'country' => $dao->country, |
| 169 | 'country_id' => $dao->country_id, |
| 170 | ]; |
| 171 | |
| 172 | $addressParams = []; |
| 173 | |
| 174 | // process geocode. |
| 175 | if ($processGeocode) { |
| 176 | // loop through the address removing more information |
| 177 | // so we can get some geocode for a partial address |
| 178 | // i.e. city -> state -> country |
| 179 | |
| 180 | $maxTries = 5; |
| 181 | do { |
| 182 | if ($this->throttle) { |
| 183 | usleep(5000000); |
| 184 | } |
| 185 | |
| 186 | CRM_Core_BAO_Address::addGeocoderData($params); |
| 187 | |
| 188 | // see if we got a geocode error, in this case we'll trigger a fatal |
| 189 | // CRM-13760 |
| 190 | if ( |
| 191 | isset($params['geo_code_error']) && |
| 192 | $params['geo_code_error'] == 'OVER_QUERY_LIMIT' |
| 193 | ) { |
| 194 | throw new CRM_Core_Exception('Aborting batch geocoding. Hit the over query limit on geocoder.'); |
| 195 | } |
| 196 | |
| 197 | array_shift($params); |
| 198 | $maxTries--; |
| 199 | } while ( |
| 200 | (!isset($params['geo_code_1']) || $params['geo_code_1'] == 'null') && |
| 201 | ($maxTries > 1) |
| 202 | ); |
| 203 | |
| 204 | if (isset($params['geo_code_1']) && $params['geo_code_1'] != 'null') { |
| 205 | $totalGeocoded++; |
| 206 | $addressParams = $params; |
| 207 | } |
| 208 | } |
| 209 | |
| 210 | // parse street address |
| 211 | if ($parseStreetAddress) { |
| 212 | $parsedFields = CRM_Core_BAO_Address::parseStreetAddress($dao->street_address); |
| 213 | $success = TRUE; |
| 214 | // consider address is automatically parseable, |
| 215 | // when we should found street_number and street_name |
| 216 | if (empty($parsedFields['street_name']) || empty($parsedFields['street_number'])) { |
| 217 | $success = FALSE; |
| 218 | } |
| 219 | |
| 220 | // do check for all elements. |
| 221 | if ($success) { |
| 222 | $totalAddressParsed++; |
| 223 | } |
| 224 | elseif ($dao->street_address) { |
| 225 | //build contact edit url, |
| 226 | //so that user can manually fill the street address fields if the street address is not parsed, CRM-5886 |
| 227 | $url = CRM_Utils_System::url('civicrm/contact/add', "reset=1&action=update&cid={$dao->id}"); |
| 228 | $unparseableContactAddress[] = " Contact ID: " . $dao->id . " <a href =\"$url\"> " . $dao->street_address . " </a> "; |
| 229 | // reset element values. |
| 230 | $parsedFields = array_fill_keys(array_keys($parsedFields), ''); |
| 231 | } |
| 232 | $addressParams = array_merge($addressParams, $parsedFields); |
| 233 | } |
| 234 | |
| 235 | // finally update address object. |
| 236 | if (!empty($addressParams)) { |
| 237 | $address = new CRM_Core_DAO_Address(); |
| 238 | $address->id = $dao->address_id; |
| 239 | $address->copyValues($addressParams); |
| 240 | $address->save(); |
| 241 | } |
| 242 | } |
| 243 | |
| 244 | $this->returnMessages[] = ts("Addresses Evaluated: %1", [ |
| 245 | 1 => $totalAddresses, |
| 246 | ]) . "\n"; |
| 247 | if ($processGeocode) { |
| 248 | $this->returnMessages[] = ts("Addresses Geocoded: %1", [ |
| 249 | 1 => $totalGeocoded, |
| 250 | ]) . "\n"; |
| 251 | } |
| 252 | if ($parseStreetAddress) { |
| 253 | $this->returnMessages[] = ts("Street Addresses Parsed: %1", [ |
| 254 | 1 => $totalAddressParsed, |
| 255 | ]) . "\n"; |
| 256 | if ($unparseableContactAddress) { |
| 257 | $this->returnMessages[] = "<br />\n" . ts("Following is the list of contacts whose address is not parsed:") . "<br />\n"; |
| 258 | foreach ($unparseableContactAddress as $contactLink) { |
| 259 | $this->returnMessages[] = $contactLink . "<br />\n"; |
| 260 | } |
| 261 | } |
| 262 | } |
| 263 | |
| 264 | return $this->returnResult(); |
| 265 | } |
| 266 | |
| 267 | /** |
| 268 | * Return result. |
| 269 | * |
| 270 | * @return array |
| 271 | */ |
| 272 | public function returnResult() { |
| 273 | $result = []; |
| 274 | $result['is_error'] = $this->returnError; |
| 275 | $result['messages'] = ''; |
| 276 | // Pad message size to allow for prefix added by CRM_Core_JobManager. |
| 277 | $messageSize = 255; |
| 278 | // Ensure that each message can fit in the civicrm_job_log.data column. |
| 279 | foreach ($this->returnMessages as $message) { |
| 280 | $messageSize += strlen($message); |
| 281 | if ($messageSize > CRM_Utils_Type::BLOB_SIZE) { |
| 282 | $result['messages'] .= '...'; |
| 283 | break; |
| 284 | } |
| 285 | $result['messages'] .= $message; |
| 286 | } |
| 287 | return $result; |
| 288 | } |
| 289 | |
| 290 | } |