[TikiWiki-commits] [Git][tikiwiki/tiki][29.x] [FIX] Manticore: handling wildcards in regex

Victor Emanouilov pushed to branch 29.x at Tiki Wiki CMS Groupware / Tiki


Commits:
98859c1e by Victor Emanouilov at 2026-04-30T10:53:23+03:00
[FIX] Manticore: handling wildcards in regex
---
* [FIX] Manticore: handling wildcards in regex resulted in parse error or incorrect regex syntax when fulltext was scaled down to regex due to missing fulltext field index

See merge request tikiwiki/tiki!10112

- - - - -


2 changed files:

- lib/core/Search/Manticore/QueryBuilder.php
- + lib/test/core/Search/Manticore/WildcardTest.php


Changes:

=====================================
lib/core/Search/Manticore/QueryBuilder.php
=====================================
@@ -226,7 +226,7 @@ class QueryBuilder
         } elseif ($node instanceof Initial) {
             $field = $this->getField($node);
             Index::addSearchedField($node->getField(), 'others');
-            $value = $this->getQuoted($node, '(?i)^');
+            $value = $this->quoteRegex($node, '(?i)^');
             $key = 'tf_' . uniqid();
             $this->select[$key] = "REGEX(TO_STRING({$field}), $value)";
             return "$key = 1";
@@ -295,7 +295,7 @@ class QueryBuilder
                 $this->select[$key] = "TO_STRING({$field}) = $value";
                 return "$key = 1";
             } else {
-                $value = $this->getQuoted($node, '(?i)');
+                $value = $this->quoteRegex($node, '(?i)');
                 $key = 'tf_' . uniqid();
                 $this->select[$key] = "REGEX(TO_STRING({$field}), $value)";
                 return "$key = 1";
@@ -310,17 +310,18 @@ class QueryBuilder
             $value = $this->getQuoted($node);
             return "{$field} = $value";
         } else {
-            $value = $this->getQuoted($node, '(?i)');
-            if (is_array($value)) {
+            $rawValue = $this->getRaw($node);
+            if (is_array($rawValue)) {
                 return '(' . implode(' OR ', array_filter(array_map(function ($v) use ($field) {
                     if (is_scalar($v)) {
-                        $v = $this->pdo_client->quote('(?i)' . strval($v));
+                        $v = $this->pdo_client->quote('(?i)' . $this->wildcardToRe2(strval($v)));
                     } else {
                         return null;
                     }
                     return "REGEX({$field}, $v)";
-                }, $value))) . ')';
+                }, $rawValue))) . ')';
             } else {
+                $value = $this->quoteRegex($node, '(?i)');
                 return "REGEX({$field}, $value)";
             }
         }
@@ -393,6 +394,20 @@ class QueryBuilder
         }
     }
 
+    private function quoteRegex($node, $prefix)
+    {
+        return $this->pdo_client->quote($prefix . $this->wildcardToRe2(strval($this->getRaw($node))));
+    }
+
+    private function wildcardToRe2($value)
+    {
+        if ($value === '') {
+            return $value;
+        }
+        $escaped = preg_quote($value, '/');
+        return str_replace('\\*', '.*', $escaped);
+    }
+
     private function getRaw($node, $forceType = null)
     {
         $value = $node->getValue($this->factory);


=====================================
lib/test/core/Search/Manticore/WildcardTest.php
=====================================
@@ -0,0 +1,172 @@
+<?php
+
+// (c) Copyright by authors of the Tiki Wiki CMS Groupware Project
+//
+// All Rights Reserved. See copyright.txt for details and a complete list of authors.
+// Licensed under the GNU LESSER GENERAL PUBLIC LICENSE. See license.txt for details.
+namespace Search\Manticore;
+
+use ReflectionMethod;
+use Search_Expr_Token;
+use Search_Query;
+
+/**
+ * Verifies how the Manticore QueryBuilder turns wildcard ("*foo*") and
+ * regex-special user input into RE2 patterns inside REGEX() calls.
+ *
+ * The object selector (and similar callers) wrap user input with "*" to
+ * emulate substring search. Because Manticore evaluates these patterns
+ * with RE2, a leading "*" is invalid and previously caused a
+ * "no argument for repetition operator: *" parse error.
+ */
+class WildcardTest extends \PHPUnit\Framework\TestCase
+{
+    use IndexBuilder;
+
+    public $index;
+    private $old_prefs;
+
+    protected function setUp(): void
+    {
+        global $prefs;
+        $this->old_prefs = $prefs;
+
+        $this->index = $this->createIndex('_wildcard');
+        $this->index->destroy();
+
+        $typeFactory = $this->index->getTypeFactory();
+        $this->index->addDocument(
+            [
+                'object_type' => $typeFactory->identifier('trackeritem'),
+                'object_id' => $typeFactory->identifier('1'),
+                'title' => $typeFactory->plaintext('Michaela Smith'),
+                'tracker_field_name' => $typeFactory->sortable('Michaela Smith'),
+            ]
+        );
+        $this->index->addDocument(
+            [
+                'object_type' => $typeFactory->identifier('trackeritem'),
+                'object_id' => $typeFactory->identifier('2'),
+                'title' => $typeFactory->plaintext('John Doe'),
+                'tracker_field_name' => $typeFactory->sortable('John Doe'),
+            ]
+        );
+        $this->index->addDocument(
+            [
+                'object_type' => $typeFactory->identifier('trackeritem'),
+                'object_id' => $typeFactory->identifier('3'),
+                'title' => $typeFactory->plaintext('first.middle last'),
+                'tracker_field_name' => $typeFactory->sortable('first.middle last'),
+            ]
+        );
+        $this->index->addDocument(
+            [
+                'object_type' => $typeFactory->identifier('trackeritem'),
+                'object_id' => $typeFactory->identifier('4'),
+                'title' => $typeFactory->plaintext('firstXmiddle last'),
+                'tracker_field_name' => $typeFactory->sortable('firstXmiddle last'),
+            ]
+        );
+    }
+
+    protected function tearDown(): void
+    {
+        global $prefs;
+        $prefs = $this->old_prefs;
+
+        if ($this->index) {
+            $this->index->destroy();
+        }
+    }
+
+    public function testLeadingAndTrailingWildcardsAreAccepted()
+    {
+        $query = $this->newQuery();
+        $query->filterContent('*michaela*', 'tracker_field_name');
+        $this->assertCount(1, $query->search($this->index));
+    }
+
+    public function testWildcardOnlyMatchesAnything()
+    {
+        $query = $this->newQuery();
+        $query->filterContent('*', 'tracker_field_name');
+        $this->assertCount(4, $query->search($this->index));
+    }
+
+    public function testInnerWildcardActsAsGlob()
+    {
+        $query = $this->newQuery();
+        $query->filterContent('first*last', 'tracker_field_name');
+        // Should match both "first.middle last" and "firstXmiddle last".
+        $this->assertCount(2, $query->search($this->index));
+    }
+
+    public function testRegexMetacharactersAreTreatedLiterally()
+    {
+        $query = $this->newQuery();
+        $query->filterContent('first.middle', 'tracker_field_name');
+        $this->assertCount(1, $query->search($this->index));
+    }
+
+    public function testWildcardToRe2EscapesMetacharacters()
+    {
+        $convert = new ReflectionMethod(QueryBuilder::class, 'wildcardToRe2');
+        $convert->setAccessible(true);
+
+        $builder = new QueryBuilder($this->index);
+
+        $cases = [
+            // input              => expected RE2 body
+            ''                    => '',
+            'michaela'            => 'michaela',
+            '*michaela*'          => '.*michaela.*',
+            'foo*bar'             => 'foo.*bar',
+            '*'                   => '.*',
+            'foo.bar'             => 'foo\\.bar',
+            'a+b'                 => 'a\\+b',
+            'a(b)c'               => 'a\\(b\\)c',
+        ];
+
+        foreach ($cases as $input => $expected) {
+            $this->assertSame(
+                $expected,
+                $convert->invoke($builder, $input),
+                'wildcardToRe2 for input ' . var_export($input, true)
+            );
+        }
+    }
+
+    public function testQuoteRegexBuildsExpectedPattern()
+    {
+        $quote = new ReflectionMethod(QueryBuilder::class, 'quoteRegex');
+        $quote->setAccessible(true);
+
+        $builder = new QueryBuilder($this->index);
+
+        $token = new Search_Expr_Token('*michaela*');
+        $token->setType('plaintext');
+        $token->setField('tracker_field_name');
+
+        $this->assertSame(
+            "'(?i).*michaela.*'",
+            $quote->invoke($builder, $token, '(?i)')
+        );
+
+        $token = new Search_Expr_Token('mich');
+        $token->setType('plaintext');
+        $token->setField('tracker_field_name');
+
+        $this->assertSame(
+            "'(?i)^mich'",
+            $quote->invoke($builder, $token, '(?i)^')
+        );
+    }
+
+    private function newQuery(): Search_Query
+    {
+        $query = new Search_Query();
+        \TikiLib::lib('unifiedsearch')->initQuery($query);
+        $query->filterType('trackeritem');
+        return $query;
+    }
+}



View it on GitLab: https://gitlab.com/tikiwiki/tiki/-/commit/98859c1e00edcc6d81ae44ef4699fb0318666acb

-- 
View it on GitLab: https://gitlab.com/tikiwiki/tiki/-/commit/98859c1e00edcc6d81ae44ef4699fb0318666acb
You're receiving this email because of your account on gitlab.com. Manage all notifications: https://gitlab.com/-/profile/notifications | Help: https://gitlab.com/help

_______________________________________________
TikiWiki-cvs mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/tikiwiki-cvs