3 +--------------------------------------------------------------------+
4 | CiviCRM version 4.3 |
5 +--------------------------------------------------------------------+
6 | Copyright CiviCRM LLC (c) 2004-2013 |
7 +--------------------------------------------------------------------+
8 | This file is a part of CiviCRM. |
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 and the CiviCRM Licensing Exception. |
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. |
19 | You should have received a copy of the GNU Affero General Public |
20 | License and the CiviCRM Licensing Exception along |
21 | with this program; if not, contact CiviCRM LLC |
22 | at info[AT]civicrm[DOT]org. If you have questions about the |
23 | GNU Affero General Public License or the licensing of CiviCRM, |
24 | see the CiviCRM license FAQ at http://civicrm.org/licensing |
25 +--------------------------------------------------------------------+
31 * @copyright CiviCRM LLC (c) 2004-2013
37 * The CiviCRM duplicate discovery engine is based on an
38 * algorithm designed by David Strauss <david@fourkitchens.com>.
40 class CRM_Dedupe_Finder
{
43 * Return a contact_id-keyed array of arrays of possible dupes
44 * (of the key contact_id) - limited to dupes of $cids if provided.
46 * @param int $rgid rule group id
47 * @param array $cids contact ids to limit the search to
49 * @return array array of (cid1, cid2, weight) dupe triples
51 static function dupes($rgid, $cids = array(
53 $rgBao = new CRM_Dedupe_BAO_RuleGroup();
55 $rgBao->contactIds
= $cids;
56 if (!$rgBao->find(TRUE)) {
58 CRM_Core_Error
::fatal("$level rule for $ctype does not exist");
62 $dao = new CRM_Core_DAO();
63 $dao->query($rgBao->thresholdQuery());
65 while ($dao->fetch()) {
66 $dupes[] = array($dao->id1
, $dao->id2
, $dao->weight
);
68 $dao->query($rgBao->tableDropQuery());
74 * Return an array of possible dupes, based on the provided array of
75 * params, using the default rule group for the given contact type and
78 * check_permission is a boolean flag to indicate if permission should be considered.
79 * default is to always check permissioning but public pages for example might not want
80 * permission to be checked for anonymous users. Refer CRM-6211. We might be beaking
81 * Multi-Site dedupe for public pages.
83 * @param array $params array of params of the form $params[$table][$field] == $value
84 * @param string $ctype contact type to match against
85 * @param string $used dedupe rule group usage ('Unsupervised' or 'Supervised' or 'General')
86 * @param array $except array of contacts that shouldn't be considered dupes
87 * @param int $ruleGroupID the id of the dedupe rule we should be using
89 * @return array matching contact ids
91 static function dupesByParams($params,
93 $used = 'Unsupervised',
97 // If $params is empty there is zero reason to proceed.
104 $rgBao = new CRM_Dedupe_BAO_RuleGroup();
105 $rgBao->id
= $ruleGroupID;
106 $rgBao->contact_type
= $ctype;
107 if ($rgBao->find(TRUE)) {
113 $rgBao = new CRM_Dedupe_BAO_RuleGroup();
114 $rgBao->contact_type
= $ctype;
115 $rgBao->used
= $used;
116 if (!$rgBao->find(TRUE)) {
117 CRM_Core_Error
::fatal("$used rule for $ctype does not exist");
120 $params['check_permission'] = CRM_Utils_Array
::value('check_permission', $params, TRUE);
122 $rgBao->params
= $params;
124 $dao = new CRM_Core_DAO();
125 $dao->query($rgBao->thresholdQuery($params['check_permission']));
127 while ($dao->fetch()) {
128 if (isset($dao->id
) && $dao->id
) {
132 $dao->query($rgBao->tableDropQuery());
133 return array_diff($dupes, $except);
137 * Return a contact_id-keyed array of arrays of possible dupes in the given group.
139 * @param int $rgid rule group id
140 * @param int $gid contact group id (currently, works only with non-smart groups)
142 * @return array array of (cid1, cid2, weight) dupe triples
144 static function dupesInGroup($rgid, $gid) {
145 $cids = array_keys(CRM_Contact_BAO_Group
::getMember($gid));
146 if ( !empty($cids) ) {
147 return self
::dupes($rgid, $cids);
153 * Return dupes of a given contact, using the default rule group (of a provided usage).
155 * @param int $cid contact id of the given contact
156 * @param string $used dedupe rule group usage ('Unsupervised' or 'Supervised' or 'General')
157 * @param string $ctype contact type of the given contact
159 * @return array array of dupe contact_ids
161 static function dupesOfContact($cid, $used = 'Unsupervised', $ctype = NULL) {
162 // if not provided, fetch the contact type from the database
164 $dao = new CRM_Contact_DAO_Contact();
166 if (!$dao->find(TRUE)) {
167 CRM_Core_Error
::fatal("contact id of $cid does not exist");
169 $ctype = $dao->contact_type
;
171 $rgBao = new CRM_Dedupe_BAO_RuleGroup();
172 $rgBao->used
= $used;
173 $rgBao->contact_type
= $ctype;
174 if (!$rgBao->find(TRUE)) {
175 CRM_Core_Error
::fatal("$used rule for $ctype does not exist");
177 $dupes = self
::dupes($rgBao->id
, array($cid));
179 // get the dupes for this cid
181 foreach ($dupes as $dupe) {
182 if ($dupe[0] == $cid) {
183 $result[] = $dupe[1];
185 elseif ($dupe[1] == $cid) {
186 $result[] = $dupe[0];
193 * A hackish function needed to massage CRM_Contact_Form_$ctype::formRule()
194 * object into a valid $params array for dedupe
196 * @param array $fields contact structure from formRule()
197 * @param string $ctype contact type of the given contact
199 * @return array valid $params array for dedupe
201 static function formatParams($fields, $ctype) {
203 CRM_Utils_Array
::flatten($fields, $flat);
205 $replace_these = array(
206 'individual_prefix' => 'prefix_id',
207 'individual_suffix' => 'suffix_id',
208 'gender' => 'gender_id',
210 //handle for individual_suffix, individual_prefix, gender
212 'individual_suffix', 'individual_prefix', 'gender') as $name) {
213 if (CRM_Utils_Array
::value($name, $fields)) {
214 $flat[$replace_these[$name]] = $flat[$name];
219 // handle {birth,deceased}_date
221 'birth_date', 'deceased_date') as $date) {
222 if (CRM_Utils_Array
::value($date, $fields)) {
223 $flat[$date] = $fields[$date];
224 if (is_array($flat[$date])) {
225 $flat[$date] = CRM_Utils_Date
::format($flat[$date]);
227 $flat[$date] = CRM_Utils_Date
::processDate($flat[$date]);
231 if (CRM_Utils_Array
::value('contact_source', $flat)) {
232 $flat['source'] = $flat['contact_source'];
233 unset($flat['contact_source']);
236 // handle preferred_communication_method
237 if (array_key_exists('preferred_communication_method', $fields)) {
238 $methods = array_intersect($fields['preferred_communication_method'], array('1'));
239 $methods = array_keys($methods);
242 $flat['preferred_communication_method'] = CRM_Core_DAO
::VALUE_SEPARATOR
. implode(CRM_Core_DAO
::VALUE_SEPARATOR
, $methods) . CRM_Core_DAO
::VALUE_SEPARATOR
;
246 // handle custom data
247 $tree = CRM_Core_BAO_CustomGroup
::getTree($ctype, CRM_Core_DAO
::$_nullObject, NULL, -1);
248 CRM_Core_BAO_CustomGroup
::postProcess($tree, $fields, TRUE);
249 foreach ($tree as $key => $cg) {
253 foreach ($cg['fields'] as $cf) {
254 $flat[$cf['column_name']] = CRM_Utils_Array
::value('data', $cf['customValue']);
258 // if the key is dotted, keep just the last part of it
259 foreach ($flat as $key => $value) {
260 if (substr_count($key, '.')) {
261 $last = explode('.', $key);
262 $last = array_pop($last);
263 // make sure the first occurence is kept, not the last
264 if (!isset($flat[$last])) {
265 $flat[$last] = $value;
271 // drop the -digit (and -Primary, for CRM-3902) postfixes (so event registration's $flat['email-5'] becomes $flat['email'])
272 // FIXME: CRM-5026 should be fixed here; the below clobbers all address info; we should split off address fields and match
273 // the -digit to civicrm_address.location_type_id and -Primary to civicrm_address.is_primary
274 foreach ($flat as $key => $value) {
276 if (preg_match('/(.*)-(\d+|Primary)$/', $key, $matches)) {
277 $flat[$matches[1]] = $value;
283 $supportedFields = CRM_Dedupe_BAO_RuleGroup
::supportedFields($ctype);
284 if (is_array($supportedFields)) {
285 foreach ($supportedFields as $table => $fields) {
286 if ($table == 'civicrm_address') {
287 // for matching on civicrm_address fields, we also need the location_type_id
288 $fields['location_type_id'] = '';
289 // FIXME: we also need to do some hacking for id and name fields, see CRM-3902’s comments
291 'address_name' => 'name', 'country' => 'country_id',
292 'state_province' => 'state_province_id', 'county' => 'county_id',
294 foreach ($fixes as $orig => $target) {
295 if (CRM_Utils_Array
::value($orig, $flat)) {
296 $params[$table][$target] = $flat[$orig];
300 foreach ($fields as $field => $title) {
301 if (CRM_Utils_Array
::value($field, $flat)) {
302 $params[$table][$field] = $flat[$field];