Merge branch 'm41_MDL-74912_Fix_RegExp_Word_Boundaries_On_MySQL8' of https://github.com/scara/moodle

lib/dml/moodle_database.php

@@ -2515,6 +2515,30 @@ abstract class moodle_database {
         return '';
     }
 
+    /**
+     * Returns the word-beginning boundary marker if this database driver supports regex syntax when searching.
+     * @return string The word-beginning boundary marker. Otherwise, an empty string.
+     */
+    public function sql_regex_get_word_beginning_boundary_marker() {
+        if ($this->sql_regex_supported()) {
+            return '[[:<:]]';
+        }
+
+        return '';
+    }
+
+    /**
+     * Returns the word-end boundary marker if this database driver supports regex syntax when searching.
+     * @return string The word-end boundary marker. Otherwise, an empty string.
+     */
+    public function sql_regex_get_word_end_boundary_marker() {
+        if ($this->sql_regex_supported()) {
+            return '[[:>:]]';
+        }
+
+        return '';
+    }
+
     /**
      * Returns the SQL that allows to find intersection of two or more queries
      *

lib/dml/mysqli_native_moodle_database.php

@@ -244,9 +244,9 @@ class mysqli_native_moodle_database extends moodle_database {
     /**
      * Set 'dbcollation' option
      *
-     * @return string $dbcollation
+     * @return string|null $dbcollation
      */
-    private function detect_collation(): string {
+    private function detect_collation(): ?string {
         if ($this->external) {
             return null;
         }
@@ -1889,6 +1889,38 @@ class mysqli_native_moodle_database extends moodle_database {
         return $collation . ($positivematch ? 'REGEXP' : 'NOT REGEXP');
     }
 
+    /**
+     * Returns the word-beginning boundary marker based on MySQL version.
+     * @return string The word-beginning boundary marker.
+     */
+    public function sql_regex_get_word_beginning_boundary_marker() {
+        $ismysql = ($this->get_dbtype() == 'mysqli' || $this->get_dbtype() == 'auroramysql');
+        $ismysqlge8d0d4 = ($ismysql && version_compare($this->get_server_info()['version'], '8.0.4', '>='));
+        if ($ismysqlge8d0d4) {
+            return '\\b';
+        }
+        // Prior to MySQL 8.0.4, MySQL used the Henry Spencer regular expression library to support regular expression operations,
+        // rather than International Components for Unicode (ICU).
+        // MariaDB still supports the "old marker" (MDEV-5357).
+        return '[[:<:]]';
+    }
+
+    /**
+     * Returns the word-end boundary marker based on MySQL version.
+     * @return string The word-end boundary marker.
+     */
+    public function sql_regex_get_word_end_boundary_marker() {
+        $ismysql = ($this->get_dbtype() == 'mysqli' || $this->get_dbtype() == 'auroramysql');
+        $ismysqlge8d0d4 = ($ismysql && version_compare($this->get_server_info()['version'], '8.0.4', '>='));
+        if ($ismysqlge8d0d4) {
+            return '\\b';
+        }
+        // Prior to MySQL 8.0.4, MySQL used the Henry Spencer regular expression library to support regular expression operations,
+        // rather than International Components for Unicode (ICU).
+        // MariaDB still supports the "old marker" (MDEV-5357).
+        return '[[:>:]]';
+    }
+
     /**
      * Returns the SQL to be used in order to an UNSIGNED INTEGER column to SIGNED.
      *

lib/searchlib.php

@@ -418,6 +418,8 @@ function search_generate_SQL($parsetree, $datafield, $metafield, $mainidfield, $
     if ($DB->sql_regex_supported()) {
         $REGEXP    = $DB->sql_regex(true);
         $NOTREGEXP = $DB->sql_regex(false);
+        $regexwordbegin = $DB->sql_regex_get_word_beginning_boundary_marker();
+        $regexwordend = $DB->sql_regex_get_word_end_boundary_marker();
     }
 
     $params = array();
@@ -451,40 +453,40 @@ function search_generate_SQL($parsetree, $datafield, $metafield, $mainidfield, $
         switch($type){
             case TOKEN_STRING:
                 $SQLString .= "((".$DB->sql_like($datafield, ":$name1", false).") OR (".$DB->sql_like($metafield, ":$name2", false)."))";
-                $params[$name1] =  "%$value%";
-                $params[$name2] =  "%$value%";
+                $params[$name1] = "%$value%";
+                $params[$name2] = "%$value%";
                 break;
             case TOKEN_EXACT:
                 $SQLString .= "(($datafield $REGEXP :$name1) OR ($metafield $REGEXP :$name2))";
-                $params[$name1] =  "[[:<:]]".$value."[[:>:]]";
-                $params[$name2] =  "[[:<:]]".$value."[[:>:]]";
+                $params[$name1] = $regexwordbegin.$value.$regexwordend;
+                $params[$name2] = $regexwordbegin.$value.$regexwordend;
                 break;
             case TOKEN_META:
                 if ($metafield != '') {
                     $SQLString .= "(".$DB->sql_like($metafield, ":$name1", false).")";
-                    $params[$name1] =  "%$value%";
+                    $params[$name1] = "%$value%";
                 }
                 break;
             case TOKEN_USER:
                 $SQLString .= "(($mainidfield = $useridfield) AND ((".$DB->sql_like($userfirstnamefield, ":$name1", false).") OR (".$DB->sql_like($userlastnamefield, ":$name2", false).")))";
-                $params[$name1] =  "%$value%";
-                $params[$name2] =  "%$value%";
+                $params[$name1] = "%$value%";
+                $params[$name2] = "%$value%";
                 break;
             case TOKEN_USERID:
                 $SQLString .= "($useridfield = :$name1)";
-                $params[$name1] =  $value;
+                $params[$name1] = $value;
                 break;
             case TOKEN_INSTANCE:
                 $SQLString .= "($instancefield = :$name1)";
-                $params[$name1] =  $value;
+                $params[$name1] = $value;
                 break;
             case TOKEN_DATETO:
                 $SQLString .= "($timefield <= :$name1)";
-                $params[$name1] =  $value;
+                $params[$name1] = $value;
                 break;
             case TOKEN_DATEFROM:
                 $SQLString .= "($timefield >= :$name1)";
-                $params[$name1] =  $value;
+                $params[$name1] = $value;
                 break;
             case TOKEN_TAGS:
                 $sqlstrings = [];
@@ -504,8 +506,8 @@ function search_generate_SQL($parsetree, $datafield, $metafield, $mainidfield, $
                 break;
             case TOKEN_NEGATE:
                 $SQLString .= "(NOT ((".$DB->sql_like($datafield, ":$name1", false).") OR (".$DB->sql_like($metafield, ":$name2", false).")))";
-                $params[$name1] =  "%$value%";
-                $params[$name2] =  "%$value%";
+                $params[$name1] = "%$value%";
+                $params[$name2] = "%$value%";
                 break;
             default:
                 return '';

lib/upgrade.txt

@@ -60,6 +60,8 @@ Declaration is as follow:
 * New function set_additional_classes() has been implemented to add additional classes to action_menu.
 * Most Behat functionality for the Moodle App has been removed from core, refer to the documentation in order to upgrade your testing
   setup: https://moodledev.io/general/app/development/testing/acceptance-testing#upgrading-tests-from-an-older-version
+* New DML driver methods `$DB->sql_regex_get_word_beginning_boundary_marker` and `$DB->sql_regex_get_word_end_boundary_marker`
+  for managing word boundary markers in a database driver supporting regex syntax when searching.
 
 === 4.0 ===