Commit | Line | Data |
---|---|---|
6a488035 TO |
1 | <?php |
2 | /* | |
3 | +--------------------------------------------------------------------+ | |
bc77d7c0 | 4 | | Copyright CiviCRM LLC. All rights reserved. | |
6a488035 | 5 | | | |
bc77d7c0 TO |
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 | | |
6a488035 | 9 | +--------------------------------------------------------------------+ |
d25dd0ee | 10 | */ |
6a488035 TO |
11 | |
12 | /** | |
13 | * | |
14 | * @package CRM | |
ca5cec67 | 15 | * @copyright CiviCRM LLC https://civicrm.org/licensing |
6a488035 TO |
16 | * $Id$ |
17 | * | |
18 | */ | |
19 | ||
20 | /** | |
21 | * The CiviCRM duplicate discovery engine is based on an | |
22 | * algorithm designed by David Strauss <david@fourkitchens.com>. | |
23 | */ | |
24 | class CRM_Dedupe_BAO_Rule extends CRM_Dedupe_DAO_Rule { | |
25 | ||
26 | /** | |
100fef9d | 27 | * Ids of the contacts to limit the SQL queries (whole-database queries otherwise) |
518fa0ee | 28 | * @var array |
6a488035 | 29 | */ |
518fa0ee | 30 | public $contactIds = []; |
6a488035 TO |
31 | |
32 | /** | |
100fef9d | 33 | * Params to dedupe against (queries against the whole contact set otherwise) |
518fa0ee | 34 | * @var array |
6a488035 | 35 | */ |
518fa0ee | 36 | public $params = []; |
6a488035 TO |
37 | |
38 | /** | |
39 | * Return the SQL query for the given rule - either for finding matching | |
40 | * pairs of contacts, or for matching against the $params variable (if set). | |
41 | * | |
a6c01b45 CW |
42 | * @return string |
43 | * SQL query performing the search | |
2cbe6e87 | 44 | * |
45 | * @throws \CRM_Core_Exception | |
46 | * @throws \CiviCRM_API3_Exception | |
6a488035 | 47 | */ |
00be9182 | 48 | public function sql() { |
6a488035 TO |
49 | if ($this->params && |
50 | (!array_key_exists($this->rule_table, $this->params) || | |
51 | !array_key_exists($this->rule_field, $this->params[$this->rule_table]) | |
52 | ) | |
53 | ) { | |
54 | // if params is present and doesn't have an entry for a field, don't construct the clause. | |
55 | return NULL; | |
56 | } | |
57 | ||
58 | // we need to initialise WHERE, ON and USING here, as some table types | |
59 | // extend them; $where is an array of required conditions, $on and | |
60 | // $using are arrays of required field matchings (for substring and | |
61 | // full matches, respectively) | |
be2fb01f CW |
62 | $where = []; |
63 | $on = ["SUBSTR(t1.{$this->rule_field}, 1, {$this->rule_length}) = SUBSTR(t2.{$this->rule_field}, 1, {$this->rule_length})"]; | |
7965a7f1 PN |
64 | |
65 | $innerJoinClauses = [ | |
66 | "t1.{$this->rule_field} IS NOT NULL", | |
67 | "t2.{$this->rule_field} IS NOT NULL", | |
518fa0ee | 68 | "t1.{$this->rule_field} = t2.{$this->rule_field}", |
7965a7f1 | 69 | ]; |
692e041d | 70 | if ($this->getFieldType($this->rule_field) === CRM_Utils_Type::T_DATE) { |
7965a7f1 PN |
71 | $innerJoinClauses[] = "t1.{$this->rule_field} > '1000-01-01'"; |
72 | $innerJoinClauses[] = "t2.{$this->rule_field} > '1000-01-01'"; | |
73 | } | |
74 | else { | |
75 | $innerJoinClauses[] = "t1.{$this->rule_field} <> ''"; | |
76 | $innerJoinClauses[] = "t2.{$this->rule_field} <> ''"; | |
77 | } | |
6a488035 TO |
78 | |
79 | switch ($this->rule_table) { | |
80 | case 'civicrm_contact': | |
81 | $id = 'id'; | |
82 | //we should restrict by contact type in the first step | |
83 | $sql = "SELECT contact_type FROM civicrm_dedupe_rule_group WHERE id = {$this->dedupe_rule_group_id};"; | |
84 | $ct = CRM_Core_DAO::singleValueQuery($sql); | |
85 | if ($this->params) { | |
86 | $where[] = "t1.contact_type = '{$ct}'"; | |
87 | } | |
88 | else { | |
89 | $where[] = "t1.contact_type = '{$ct}'"; | |
90 | $where[] = "t2.contact_type = '{$ct}'"; | |
91 | } | |
92 | break; | |
93 | ||
94 | case 'civicrm_address': | |
353ffa53 TO |
95 | $id = 'contact_id'; |
96 | $on[] = 't1.location_type_id = t2.location_type_id'; | |
2984eeb8 D |
97 | $innerJoinClauses[] = 't1.location_type_id = t2.location_type_id'; |
98 | if (!empty($this->params['civicrm_address']['location_type_id'])) { | |
6a488035 TO |
99 | $locTypeId = CRM_Utils_Type::escape($this->params['civicrm_address']['location_type_id'], 'Integer', FALSE); |
100 | if ($locTypeId) { | |
101 | $where[] = "t1.location_type_id = $locTypeId"; | |
102 | } | |
103 | } | |
104 | break; | |
105 | ||
106 | case 'civicrm_email': | |
107 | case 'civicrm_im': | |
108 | case 'civicrm_openid': | |
109 | case 'civicrm_phone': | |
110 | $id = 'contact_id'; | |
111 | break; | |
112 | ||
113 | case 'civicrm_note': | |
114 | $id = 'entity_id'; | |
115 | if ($this->params) { | |
116 | $where[] = "t1.entity_table = 'civicrm_contact'"; | |
117 | } | |
118 | else { | |
119 | $where[] = "t1.entity_table = 'civicrm_contact'"; | |
120 | $where[] = "t2.entity_table = 'civicrm_contact'"; | |
121 | } | |
122 | break; | |
123 | ||
124 | default: | |
125 | // custom data tables | |
126 | if (preg_match('/^civicrm_value_/', $this->rule_table) || preg_match('/^custom_value_/', $this->rule_table)) { | |
127 | $id = 'entity_id'; | |
128 | } | |
129 | else { | |
2cbe6e87 | 130 | throw new CRM_Core_Exception("Unsupported rule_table for civicrm_dedupe_rule.id of {$this->id}"); |
6a488035 TO |
131 | } |
132 | break; | |
133 | } | |
134 | ||
135 | // build SELECT based on the field names containing contact ids | |
136 | // if there are params provided, id1 should be 0 | |
137 | if ($this->params) { | |
138 | $select = "t1.$id id1, {$this->rule_weight} weight"; | |
0226a9f3 | 139 | $subSelect = 'id1, weight'; |
6a488035 TO |
140 | } |
141 | else { | |
142 | $select = "t1.$id id1, t2.$id id2, {$this->rule_weight} weight"; | |
0226a9f3 | 143 | $subSelect = 'id1, id2, weight'; |
6a488035 TO |
144 | } |
145 | ||
146 | // build FROM (and WHERE, if it's a parametrised search) | |
147 | // based on whether the rule is about substrings or not | |
148 | if ($this->params) { | |
149 | $from = "{$this->rule_table} t1"; | |
150 | $str = 'NULL'; | |
151 | if (isset($this->params[$this->rule_table][$this->rule_field])) { | |
152 | $str = CRM_Utils_Type::escape($this->params[$this->rule_table][$this->rule_field], 'String'); | |
153 | } | |
154 | if ($this->rule_length) { | |
155 | $where[] = "SUBSTR(t1.{$this->rule_field}, 1, {$this->rule_length}) = SUBSTR('$str', 1, {$this->rule_length})"; | |
156 | $where[] = "t1.{$this->rule_field} IS NOT NULL"; | |
157 | } | |
158 | else { | |
159 | $where[] = "t1.{$this->rule_field} = '$str'"; | |
160 | } | |
161 | } | |
162 | else { | |
163 | if ($this->rule_length) { | |
164 | $from = "{$this->rule_table} t1 JOIN {$this->rule_table} t2 ON (" . implode(' AND ', $on) . ")"; | |
165 | } | |
166 | else { | |
7965a7f1 | 167 | $from = "{$this->rule_table} t1 INNER JOIN {$this->rule_table} t2 ON (" . implode(' AND ', $innerJoinClauses) . ")"; |
6a488035 TO |
168 | } |
169 | } | |
170 | ||
171 | // finish building WHERE, also limit the results if requested | |
172 | if (!$this->params) { | |
173 | $where[] = "t1.$id < t2.$id"; | |
6a488035 | 174 | } |
be61083d | 175 | $query = "SELECT $select FROM $from WHERE " . implode(' AND ', $where); |
6a488035 | 176 | if ($this->contactIds) { |
be2fb01f | 177 | $cids = []; |
6a488035 TO |
178 | foreach ($this->contactIds as $cid) { |
179 | $cids[] = CRM_Utils_Type::escape($cid, 'Integer'); | |
180 | } | |
181 | if (count($cids) == 1) { | |
be61083d | 182 | $query .= " AND (t1.$id = {$cids[0]}) UNION $query AND t2.$id = {$cids[0]}"; |
6a488035 TO |
183 | } |
184 | else { | |
be61083d | 185 | $query .= " AND t1.$id IN (" . implode(',', $cids) . ") |
186 | UNION $query AND t2.$id IN (" . implode(',', $cids) . ")"; | |
6a488035 | 187 | } |
0226a9f3 AH |
188 | // The `weight` is ambiguous in the context of the union; put the whole |
189 | // thing in a subquery. | |
190 | $query = "SELECT $subSelect FROM ($query) subunion"; | |
6a488035 TO |
191 | } |
192 | ||
be61083d | 193 | return $query; |
6a488035 TO |
194 | } |
195 | ||
196 | /** | |
dc195289 | 197 | * find fields related to a rule group. |
6a488035 | 198 | * |
389bcebf | 199 | * @param array $params contains the rule group property to identify rule group |
6a488035 | 200 | * |
72b3a70c CW |
201 | * @return array |
202 | * rule fields array associated to rule group | |
6a488035 | 203 | */ |
00be9182 | 204 | public static function dedupeRuleFields($params) { |
353ffa53 TO |
205 | $rgBao = new CRM_Dedupe_BAO_RuleGroup(); |
206 | $rgBao->used = $params['used']; | |
6a488035 TO |
207 | $rgBao->contact_type = $params['contact_type']; |
208 | $rgBao->find(TRUE); | |
209 | ||
210 | $ruleBao = new CRM_Dedupe_BAO_Rule(); | |
211 | $ruleBao->dedupe_rule_group_id = $rgBao->id; | |
212 | $ruleBao->find(); | |
be2fb01f | 213 | $ruleFields = []; |
6a488035 | 214 | while ($ruleBao->fetch()) { |
0ca748fb JM |
215 | $field_name = $ruleBao->rule_field; |
216 | if ($field_name == 'phone_numeric') { | |
217 | $field_name = 'phone'; | |
218 | } | |
219 | $ruleFields[] = $field_name; | |
6a488035 TO |
220 | } |
221 | return $ruleFields; | |
222 | } | |
223 | ||
e0ef6999 | 224 | /** |
100fef9d CW |
225 | * @param int $cid |
226 | * @param int $oid | |
e0ef6999 EM |
227 | * |
228 | * @return bool | |
229 | */ | |
00be9182 | 230 | public static function validateContacts($cid, $oid) { |
6a488035 | 231 | if (!$cid || !$oid) { |
389bcebf | 232 | return NULL; |
6a488035 TO |
233 | } |
234 | $exception = new CRM_Dedupe_DAO_Exception(); | |
235 | $exception->contact_id1 = $cid; | |
236 | $exception->contact_id2 = $oid; | |
237 | //make sure contact2 > contact1. | |
238 | if ($cid > $oid) { | |
239 | $exception->contact_id1 = $oid; | |
240 | $exception->contact_id2 = $cid; | |
241 | } | |
242 | ||
91768280 | 243 | return !$exception->find(TRUE); |
6a488035 | 244 | } |
96025800 | 245 | |
692e041d | 246 | /** |
247 | * Get the specification for the given field. | |
248 | * | |
249 | * @param string $fieldName | |
250 | * | |
251 | * @return array | |
252 | * @throws \CiviCRM_API3_Exception | |
253 | */ | |
254 | public function getFieldType($fieldName) { | |
255 | $entity = CRM_Core_DAO_AllCoreTables::getBriefName(CRM_Core_DAO_AllCoreTables::getClassForTable($this->rule_table)); | |
256 | if (!$entity) { | |
257 | // This means we have stored a custom field rather than an entity name in rule_table, figure out the entity. | |
258 | $entity = civicrm_api3('CustomGroup', 'getvalue', ['table_name' => $this->rule_table, 'return' => 'extends']); | |
259 | if (in_array($entity, ['Individual', 'Household', 'Organization'])) { | |
260 | $entity = 'Contact'; | |
261 | } | |
262 | $fieldName = 'custom_' . civicrm_api3('CustomField', 'getvalue', ['column_name' => $fieldName, 'return' => 'id']); | |
263 | } | |
264 | $fields = civicrm_api3($entity, 'getfields', ['action' => 'create'])['values']; | |
265 | return $fields[$fieldName]['type']; | |
266 | } | |
267 | ||
6a488035 | 268 | } |