3 +--------------------------------------------------------------------+
4 | Copyright CiviCRM LLC. All rights reserved. |
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 +--------------------------------------------------------------------+
15 * @copyright CiviCRM LLC https://civicrm.org/licensing
19 * The CiviCRM duplicate discovery engine is based on an
20 * algorithm designed by David Strauss <david@fourkitchens.com>.
22 class CRM_Dedupe_BAO_DedupeRuleGroup
extends CRM_Dedupe_DAO_DedupeRuleGroup
{
25 * Ids of the contacts to limit the SQL queries (whole-database queries otherwise)
28 public $contactIds = [];
31 * Set the contact IDs to restrict the dedupe to.
33 * @param array $contactIds
35 public function setContactIds($contactIds) {
36 $this->contactIds
= $contactIds;
40 * Params to dedupe against (queries against the whole contact set otherwise)
46 * If there are no rules in rule group.
49 public $noRules = FALSE;
51 protected $temporaryTables = [];
54 * Return a structure holding the supported tables, fields and their titles
56 * @param string $requestedType
57 * The requested contact type.
60 * a table-keyed array of field-keyed arrays holding supported fields' titles
62 public static function supportedFields($requestedType) {
63 static $fields = NULL;
65 // this is needed, as we're piggy-backing importableFields() below
67 'civicrm_country.name' => 'civicrm_address.country_id',
68 'civicrm_county.name' => 'civicrm_address.county_id',
69 'civicrm_state_province.name' => 'civicrm_address.state_province_id',
70 'gender.label' => 'civicrm_contact.gender_id',
71 'individual_prefix.label' => 'civicrm_contact.prefix_id',
72 'individual_suffix.label' => 'civicrm_contact.suffix_id',
73 'addressee.label' => 'civicrm_contact.addressee_id',
74 'email_greeting.label' => 'civicrm_contact.email_greeting_id',
75 'postal_greeting.label' => 'civicrm_contact.postal_greeting_id',
76 'civicrm_phone.phone' => 'civicrm_phone.phone_numeric',
78 // the table names we support in dedupe rules - a filter for importableFields()
89 foreach (['Individual', 'Organization', 'Household'] as $ctype) {
90 // take the table.field pairs and their titles from importableFields() if the table is supported
91 foreach (CRM_Contact_BAO_Contact
::importableFields($ctype) as $iField) {
92 if (isset($iField['where'])) {
93 $where = $iField['where'];
94 if (isset($replacements[$where])) {
95 $where = $replacements[$where];
97 list($table, $field) = explode('.', $where);
98 if (!in_array($table, $supportedTables)) {
101 $fields[$ctype][$table][$field] = $iField['title'];
104 // Note that most of the fields available come from 'importable fields' -
105 // I thought about making this field 'importable' but it felt like there might be unknown consequences
106 // so I opted for just adding it in & securing it with a unit test.
107 /// Example usage of sort_name - It is possible to alter sort name via hook so 2 organization names might differ as in
108 // Justice League vs The Justice League but these could have the same sort_name if 'the the'
109 // exension is installed (https://github.com/eileenmcnaughton/org.wikimedia.thethe)
110 $fields[$ctype]['civicrm_contact']['sort_name'] = ts('Sort Name');
111 // add custom data fields
112 foreach (CRM_Core_BAO_CustomGroup
::getTree($ctype, NULL, NULL, -1) as $key => $cg) {
116 foreach ($cg['fields'] as $cf) {
117 $fields[$ctype][$cg['table_name']][$cf['column_name']] = $cf['label'];
122 CRM_Utils_Hook
::dupeQuery(CRM_Core_DAO
::$_nullObject, 'supportedFields', $fields);
123 return !empty($fields[$requestedType]) ?
$fields[$requestedType] : [];
127 * Return the SQL query for dropping the temporary table.
129 public function tableDropQuery() {
130 return 'DROP TEMPORARY TABLE IF EXISTS dedupe';
134 * Return a set of SQL queries whose cummulative weights will mark matched
135 * records for the RuleGroup::threasholdQuery() to retrieve.
137 public function tableQuery() {
138 // make sure we've got a fetched dbrecord, not sure if this is enforced
139 if (!$this->name
== NULL ||
$this->is_reserved
== NULL) {
143 // Reserved Rule Groups can optionally get special treatment by
144 // implementing an optimization class and returning a query array.
145 if ($this->is_reserved
&&
146 CRM_Utils_File
::isIncludable("CRM/Dedupe/BAO/QueryBuilder/{$this->name}.php")
148 $command = empty($this->params
) ?
'internal' : 'record';
149 $queries = call_user_func(["CRM_Dedupe_BAO_QueryBuilder_{$this->name}", $command], $this);
152 // All other rule groups have queries generated by the member dedupe
153 // rules defined in the administrative interface.
155 // Find all rules contained by this script sorted by weight so that
156 // their execution can be short circuited on RuleGroup::fillTable()
157 $bao = new CRM_Dedupe_BAO_DedupeRule();
158 $bao->dedupe_rule_group_id
= $this->id
;
159 $bao->orderBy('rule_weight DESC');
162 // Generate a SQL query for each rule in the rule group that is
163 // tailored to respect the param and contactId options provided.
165 while ($bao->fetch()) {
166 $bao->contactIds
= $this->contactIds
;
167 $bao->params
= $this->params
;
169 // Skipping empty rules? Empty rules shouldn't exist; why check?
170 if ($query = $bao->sql()) {
171 $queries["{$bao->rule_table}.{$bao->rule_field}.{$bao->rule_weight}"] = $query;
176 // if there are no rules in this rule group
177 // add an empty query fulfilling the pattern
179 $this->noRules
= TRUE;
186 public function fillTable() {
187 // get the list of queries handy
188 $tableQueries = $this->tableQuery();
190 if ($this->params
&& !$this->noRules
) {
191 $this->temporaryTables
['dedupe'] = CRM_Utils_SQL_TempTable
::build()
192 ->setCategory('dedupe')
193 ->createWithColumns("id1 int, weight int, UNIQUE UI_id1 (id1)")->getName();
194 $dedupeCopyTemporaryTableObject = CRM_Utils_SQL_TempTable
::build()
195 ->setCategory('dedupe');
196 $this->temporaryTables
['dedupe_copy'] = $dedupeCopyTemporaryTableObject->getName();
197 $insertClause = "INSERT INTO {$this->temporaryTables['dedupe']} (id1, weight)";
198 $groupByClause = "GROUP BY id1, weight";
199 $dupeCopyJoin = " JOIN {$this->temporaryTables['dedupe_copy']} ON {$this->temporaryTables['dedupe_copy']}.id1 = t1.column WHERE ";
202 $this->temporaryTables
['dedupe'] = CRM_Utils_SQL_TempTable
::build()
203 ->setCategory('dedupe')
204 ->createWithColumns("id1 int, id2 int, weight int, UNIQUE UI_id1_id2 (id1, id2)")->getName();
205 $dedupeCopyTemporaryTableObject = CRM_Utils_SQL_TempTable
::build()
206 ->setCategory('dedupe');
207 $this->temporaryTables
['dedupe_copy'] = $dedupeCopyTemporaryTableObject->getName();
208 $insertClause = "INSERT INTO {$this->temporaryTables['dedupe']} (id1, id2, weight)";
209 $groupByClause = "GROUP BY id1, id2, weight";
210 $dupeCopyJoin = " JOIN {$this->temporaryTables['dedupe_copy']} ON {$this->temporaryTables['dedupe_copy']}.id1 = t1.column AND {$this->temporaryTables['dedupe_copy']}.id2 = t2.column WHERE ";
212 $patternColumn = '/t1.(\w+)/';
215 CRM_Utils_Hook
::dupeQuery($this, 'table', $tableQueries);
217 while (!empty($tableQueries)) {
218 list($isInclusive, $isDie) = self
::isQuerySetInclusive($tableQueries, $this->threshold
, $exclWeightSum);
221 // order queries by table count
222 self
::orderByTableCount($tableQueries);
224 $weightSum = array_sum($exclWeightSum);
225 $searchWithinDupes = !empty($exclWeightSum) ?
1 : 0;
227 while (!empty($tableQueries)) {
228 // extract the next query ( and weight ) to be executed
229 $fieldWeight = array_keys($tableQueries);
230 $fieldWeight = $fieldWeight[0];
231 $query = array_shift($tableQueries);
233 if ($searchWithinDupes) {
234 // drop dedupe_copy table just in case if its already there.
235 $dedupeCopyTemporaryTableObject->drop();
236 // get prepared to search within already found dupes if $searchWithinDupes flag is set
237 $dedupeCopyTemporaryTableObject->createWithQuery("SELECT * FROM {$this->temporaryTables['dedupe']} WHERE weight >= {$weightSum}");
239 preg_match($patternColumn, $query, $matches);
240 $query = str_replace(' WHERE ', str_replace('column', $matches[1], $dupeCopyJoin), $query);
242 // CRM-19612: If there's a union, there will be two WHEREs, and you
243 // can't use the temp table twice.
244 if (preg_match('/' . $this->temporaryTables
['dedupe_copy'] . '[\S\s]*(union)[\S\s]*' . $this->temporaryTables
['dedupe_copy'] . '/i', $query, $matches, PREG_OFFSET_CAPTURE
)) {
245 // Make a second temp table:
246 $this->temporaryTables
['dedupe_copy_2'] = CRM_Utils_SQL_TempTable
::build()
247 ->setCategory('dedupe')
248 ->createWithQuery("SELECT * FROM {$this->temporaryTables['dedupe']} WHERE weight >= {$weightSum}")
250 // After the union, use that new temp table:
251 $part1 = substr($query, 0, $matches[1][1]);
252 $query = $part1 . str_replace($this->temporaryTables
['dedupe_copy'], $this->temporaryTables
['dedupe_copy_2'], substr($query, $matches[1][1]));
255 $searchWithinDupes = 1;
257 // construct and execute the intermediate query
258 $query = "{$insertClause} {$query} {$groupByClause} ON DUPLICATE KEY UPDATE weight = weight + VALUES(weight)";
259 $dao = CRM_Core_DAO
::executeQuery($query);
261 // FIXME: we need to be more acurate with affected rows, especially for insert vs duplicate insert.
262 // And that will help optimize further.
263 $affectedRows = $dao->affectedRows();
265 // In an inclusive situation, failure of any query means no further processing -
266 if ($affectedRows == 0) {
267 // reset to make sure no further execution is done.
271 $weightSum = substr($fieldWeight, strrpos($fieldWeight, '.') +
1) +
$weightSum;
273 // An exclusive situation -
276 // since queries are already sorted by weights, we can continue as is
277 $fieldWeight = array_keys($tableQueries);
278 $fieldWeight = $fieldWeight[0];
279 $query = array_shift($tableQueries);
280 $query = "{$insertClause} {$query} {$groupByClause} ON DUPLICATE KEY UPDATE weight = weight + VALUES(weight)";
281 $dao = CRM_Core_DAO
::executeQuery($query);
282 if ($dao->affectedRows() >= 1) {
283 $exclWeightSum[] = substr($fieldWeight, strrpos($fieldWeight, '.') +
1);
287 // its a die situation
294 * Function to determine if a given query set contains inclusive or exclusive set of weights.
295 * The function assumes that the query set is already ordered by weight in desc order.
296 * @param $tableQueries
298 * @param array $exclWeightSum
302 public static function isQuerySetInclusive($tableQueries, $threshold, $exclWeightSum = []) {
304 foreach ($tableQueries as $key => $query) {
305 $input[] = substr($key, strrpos($key, '.') +
1);
308 if (!empty($exclWeightSum)) {
309 $input = array_merge($input, $exclWeightSum);
313 if (count($input) == 1) {
314 return [FALSE, $input[0] < $threshold];
317 $totalCombinations = 0;
318 for ($i = 0; $i < count($input); $i++
) {
319 $combination = [$input[$i]];
320 if (array_sum($combination) >= $threshold) {
321 $totalCombinations++
;
324 for ($j = $i +
1; $j < count($input); $j++
) {
325 $combination[] = $input[$j];
326 if (array_sum($combination) >= $threshold) {
327 $totalCombinations++
;
331 return [$totalCombinations == 1, $totalCombinations <= 0];
335 * sort queries by number of records for the table associated with them.
336 * @param $tableQueries
338 public static function orderByTableCount(&$tableQueries) {
339 static $tableCount = [];
342 foreach ($tableQueries as $key => $query) {
343 $table = explode(".", $key);
345 if (!array_key_exists($table, $tableCount)) {
346 $query = "SELECT COUNT(*) FROM {$table}";
347 $tableCount[$table] = CRM_Core_DAO
::singleValueQuery($query);
349 $tempArray[$key] = $tableCount[$table];
353 foreach ($tempArray as $key => $count) {
354 $tempArray[$key] = $tableQueries[$key];
356 $tableQueries = $tempArray;
360 * Return the SQL query for getting only the interesting results out of the dedupe table.
362 * @$checkPermission boolean $params a flag to indicate if permission should be considered.
363 * default is to always check permissioning but public pages for example might not want
364 * permission to be checked for anonymous users. Refer CRM-6211. We might be beaking
365 * Multi-Site dedupe for public pages.
367 * @param bool $checkPermission
371 public function thresholdQuery($checkPermission = TRUE) {
372 $this->_aclFrom
= '';
375 if ($this->params
&& !$this->noRules
) {
376 if ($checkPermission) {
377 list($this->_aclFrom
, $aclWhere) = CRM_Contact_BAO_Contact_Permission
::cacheClause('civicrm_contact');
378 $aclWhere = $aclWhere ?
"AND {$aclWhere}" : '';
380 $query = "SELECT {$this->temporaryTables['dedupe']}.id1 as id
381 FROM {$this->temporaryTables['dedupe']} JOIN civicrm_contact ON {$this->temporaryTables['dedupe']}.id1 = civicrm_contact.id {$this->_aclFrom}
382 WHERE contact_type = '{$this->contact_type}' AND is_deleted = 0 $aclWhere
383 AND weight >= {$this->threshold}";
387 if ($checkPermission) {
388 list($this->_aclFrom
, $aclWhere) = CRM_Contact_BAO_Contact_Permission
::cacheClause(['c1', 'c2']);
389 $aclWhere = $aclWhere ?
"AND {$aclWhere}" : '';
391 $query = "SELECT IF({$this->temporaryTables['dedupe']}.id1 < {$this->temporaryTables['dedupe']}.id2, {$this->temporaryTables['dedupe']}.id1, {$this->temporaryTables['dedupe']}.id2) as id1,
392 IF({$this->temporaryTables['dedupe']}.id1 < {$this->temporaryTables['dedupe']}.id2, {$this->temporaryTables['dedupe']}.id2, {$this->temporaryTables['dedupe']}.id1) as id2, {$this->temporaryTables['dedupe']}.weight
393 FROM {$this->temporaryTables['dedupe']} JOIN civicrm_contact c1 ON {$this->temporaryTables['dedupe']}.id1 = c1.id
394 JOIN civicrm_contact c2 ON {$this->temporaryTables['dedupe']}.id2 = c2.id {$this->_aclFrom}
395 LEFT JOIN civicrm_dedupe_exception exc ON {$this->temporaryTables['dedupe']}.id1 = exc.contact_id1 AND {$this->temporaryTables['dedupe']}.id2 = exc.contact_id2
396 WHERE c1.contact_type = '{$this->contact_type}' AND
397 c2.contact_type = '{$this->contact_type}'
398 AND c1.is_deleted = 0 AND c2.is_deleted = 0
400 AND weight >= {$this->threshold} AND exc.contact_id1 IS NULL";
403 CRM_Utils_Hook
::dupeQuery($this, 'threshold', $query);
408 * find fields related to a rule group.
410 * @param array $params
413 * (rule field => weight) array and threshold associated to rule group
415 public static function dedupeRuleFieldsWeight($params) {
416 $rgBao = new CRM_Dedupe_BAO_DedupeRuleGroup();
417 $rgBao->contact_type
= $params['contact_type'];
418 if (!empty($params['id'])) {
419 // accept an ID if provided
420 $rgBao->id
= $params['id'];
423 $rgBao->used
= $params['used'];
427 $ruleBao = new CRM_Dedupe_BAO_DedupeRule();
428 $ruleBao->dedupe_rule_group_id
= $rgBao->id
;
431 while ($ruleBao->fetch()) {
432 $field_name = $ruleBao->rule_field
;
433 if ($field_name == 'phone_numeric') {
434 $field_name = 'phone';
436 $ruleFields[$field_name] = $ruleBao->rule_weight
;
439 return [$ruleFields, $rgBao->threshold
];
443 * Get all of the combinations of fields that would work with a rule.
445 * @param array $rgFields
446 * @param int $threshold
447 * @param array $combos
448 * @param array $running
450 public static function combos($rgFields, $threshold, &$combos, $running = []) {
451 foreach ($rgFields as $rgField => $weight) {
452 unset($rgFields[$rgField]);
453 $diff = $threshold - $weight;
454 $runningnow = $running;
455 $runningnow[] = $rgField;
457 self
::combos($rgFields, $diff, $combos, $runningnow);
460 $combos[] = $runningnow;
466 * Get an array of rule group id to rule group name
467 * for all th groups for that contactType. If contactType
468 * not specified, do it for all
470 * @param string $contactType
471 * Individual, Household or Organization.
475 * id => "nice name" of rule group
477 public static function getByType($contactType = NULL) {
478 $dao = new CRM_Dedupe_DAO_DedupeRuleGroup();
481 $dao->contact_type
= $contactType;
486 while ($dao->fetch()) {
487 $title = !empty($dao->title
) ?
$dao->title
: (!empty($dao->name
) ?
$dao->name
: $dao->contact_type
);
489 $name = "$title - {$dao->used}";
490 $result[$dao->id
] = $name;
496 * Get the cached contact type for a particular rule group.
498 * @param int $rule_group_id
502 public static function getContactTypeForRuleGroup($rule_group_id) {
503 if (!isset(\Civi
::$statics[__CLASS__
]) ||
!isset(\Civi
::$statics[__CLASS__
]['rule_groups'])) {
504 \Civi
::$statics[__CLASS__
]['rule_groups'] = [];
506 if (empty(\Civi
::$statics[__CLASS__
]['rule_groups'][$rule_group_id])) {
507 \Civi
::$statics[__CLASS__
]['rule_groups'][$rule_group_id]['contact_type'] = CRM_Core_DAO
::getFieldValue(
508 'CRM_Dedupe_DAO_DedupeRuleGroup',
514 return \Civi
::$statics[__CLASS__
]['rule_groups'][$rule_group_id]['contact_type'];