xref: /dokuwiki/_test/tests/Search/IntegrityTest.php (revision 21fbd01b3c3eea88b767376b7b158f31f0f63127) 
1*21fbd01bSAndreas Gohr<?php
2*21fbd01bSAndreas Gohr
3*21fbd01bSAndreas Gohrnamespace dokuwiki\test\Search;
4*21fbd01bSAndreas Gohr
5*21fbd01bSAndreas Gohruse dokuwiki\Search\Collection\PageFulltextCollection;
6*21fbd01bSAndreas Gohruse dokuwiki\Search\Collection\PageTitleCollection;
7*21fbd01bSAndreas Gohruse dokuwiki\Search\Exception\IndexIntegrityException;
8*21fbd01bSAndreas Gohruse dokuwiki\Search\Indexer;
9*21fbd01bSAndreas Gohr
10*21fbd01bSAndreas Gohr/**
11*21fbd01bSAndreas Gohr * Tests the index integrity checking
12*21fbd01bSAndreas Gohr */
13*21fbd01bSAndreas Gohrclass IntegrityTest extends \DokuWikiTest
14*21fbd01bSAndreas Gohr{
15*21fbd01bSAndreas Gohr    /**
16*21fbd01bSAndreas Gohr     * Clear the index directory before each test
17*21fbd01bSAndreas Gohr     */
18*21fbd01bSAndreas Gohr    public function setUp(): void
19*21fbd01bSAndreas Gohr    {
20*21fbd01bSAndreas Gohr        parent::setUp();
21*21fbd01bSAndreas Gohr        global $conf;
22*21fbd01bSAndreas Gohr        $files = glob($conf['indexdir'] . '/*.idx');
23*21fbd01bSAndreas Gohr        foreach ($files as $file) {
24*21fbd01bSAndreas Gohr            @unlink($file);
25*21fbd01bSAndreas Gohr        }
26*21fbd01bSAndreas Gohr        \dokuwiki\Search\Index\Lock::releaseAll();
27*21fbd01bSAndreas Gohr    }
28*21fbd01bSAndreas Gohr
29*21fbd01bSAndreas Gohr    /**
30*21fbd01bSAndreas Gohr     * Index a page so we have data to check
31*21fbd01bSAndreas Gohr     */
32*21fbd01bSAndreas Gohr    protected function indexTestPage(): void
33*21fbd01bSAndreas Gohr    {
34*21fbd01bSAndreas Gohr        saveWikiText('integritytest', 'Hello world testing integrity check.', 'Test');
35*21fbd01bSAndreas Gohr        $indexer = new Indexer();
36*21fbd01bSAndreas Gohr        $indexer->addPage('integritytest');
37*21fbd01bSAndreas Gohr    }
38*21fbd01bSAndreas Gohr
39*21fbd01bSAndreas Gohr    /**
40*21fbd01bSAndreas Gohr     * A healthy index should not throw
41*21fbd01bSAndreas Gohr     */
42*21fbd01bSAndreas Gohr    public function testHealthyIndex()
43*21fbd01bSAndreas Gohr    {
44*21fbd01bSAndreas Gohr        $this->indexTestPage();
45*21fbd01bSAndreas Gohr
46*21fbd01bSAndreas Gohr        $indexer = new Indexer();
47*21fbd01bSAndreas Gohr        $indexer->checkIntegrity();
48*21fbd01bSAndreas Gohr        $this->assertFalse($indexer->isIndexEmpty());
49*21fbd01bSAndreas Gohr    }
50*21fbd01bSAndreas Gohr
51*21fbd01bSAndreas Gohr    /**
52*21fbd01bSAndreas Gohr     * An empty index should not throw
53*21fbd01bSAndreas Gohr     */
54*21fbd01bSAndreas Gohr    public function testEmptyIndex()
55*21fbd01bSAndreas Gohr    {
56*21fbd01bSAndreas Gohr        $indexer = new Indexer();
57*21fbd01bSAndreas Gohr        $indexer->checkIntegrity();
58*21fbd01bSAndreas Gohr        $this->assertTrue($indexer->isIndexEmpty());
59*21fbd01bSAndreas Gohr    }
60*21fbd01bSAndreas Gohr
61*21fbd01bSAndreas Gohr    /**
62*21fbd01bSAndreas Gohr     * Corrupted fulltext index (token/frequency mismatch) should throw
63*21fbd01bSAndreas Gohr     */
64*21fbd01bSAndreas Gohr    public function testCorruptedFulltextTokenFrequency()
65*21fbd01bSAndreas Gohr    {
66*21fbd01bSAndreas Gohr        global $conf;
67*21fbd01bSAndreas Gohr        $this->indexTestPage();
68*21fbd01bSAndreas Gohr
69*21fbd01bSAndreas Gohr        // Append an extra line to a token index to create a mismatch
70*21fbd01bSAndreas Gohr        $collection = new PageFulltextCollection();
71*21fbd01bSAndreas Gohr        $max = $collection->getTokenIndexMaximum();
72*21fbd01bSAndreas Gohr        $this->assertGreaterThan(0, $max);
73*21fbd01bSAndreas Gohr
74*21fbd01bSAndreas Gohr        $tokenFile = $conf['indexdir'] . '/w' . $max . '.idx';
75*21fbd01bSAndreas Gohr        $this->assertFileExists($tokenFile);
76*21fbd01bSAndreas Gohr        file_put_contents($tokenFile, "corruptedentry\n", FILE_APPEND);
77*21fbd01bSAndreas Gohr
78*21fbd01bSAndreas Gohr        $this->expectException(IndexIntegrityException::class);
79*21fbd01bSAndreas Gohr        (new PageFulltextCollection())->checkIntegrity();
80*21fbd01bSAndreas Gohr    }
81*21fbd01bSAndreas Gohr
82*21fbd01bSAndreas Gohr    /**
83*21fbd01bSAndreas Gohr     * Corrupted fulltext index (entity/reverse mismatch) should throw
84*21fbd01bSAndreas Gohr     */
85*21fbd01bSAndreas Gohr    public function testCorruptedFulltextEntityReverse()
86*21fbd01bSAndreas Gohr    {
87*21fbd01bSAndreas Gohr        global $conf;
88*21fbd01bSAndreas Gohr        $this->indexTestPage();
89*21fbd01bSAndreas Gohr
90*21fbd01bSAndreas Gohr        $reverseFile = $conf['indexdir'] . '/pageword.idx';
91*21fbd01bSAndreas Gohr        $this->assertFileExists($reverseFile);
92*21fbd01bSAndreas Gohr        file_put_contents($reverseFile, "0\n", FILE_APPEND);
93*21fbd01bSAndreas Gohr
94*21fbd01bSAndreas Gohr        $this->expectException(IndexIntegrityException::class);
95*21fbd01bSAndreas Gohr        (new PageFulltextCollection())->checkIntegrity();
96*21fbd01bSAndreas Gohr    }
97*21fbd01bSAndreas Gohr
98*21fbd01bSAndreas Gohr    /**
99*21fbd01bSAndreas Gohr     * Corrupted title index (entity/token mismatch) should throw
100*21fbd01bSAndreas Gohr     */
101*21fbd01bSAndreas Gohr    public function testCorruptedTitleIndex()
102*21fbd01bSAndreas Gohr    {
103*21fbd01bSAndreas Gohr        global $conf;
104*21fbd01bSAndreas Gohr        $this->indexTestPage();
105*21fbd01bSAndreas Gohr
106*21fbd01bSAndreas Gohr        $titleFile = $conf['indexdir'] . '/title.idx';
107*21fbd01bSAndreas Gohr        $this->assertFileExists($titleFile);
108*21fbd01bSAndreas Gohr        file_put_contents($titleFile, "extra title\n", FILE_APPEND);
109*21fbd01bSAndreas Gohr
110*21fbd01bSAndreas Gohr        $this->expectException(IndexIntegrityException::class);
111*21fbd01bSAndreas Gohr        (new PageTitleCollection())->checkIntegrity();
112*21fbd01bSAndreas Gohr    }
113*21fbd01bSAndreas Gohr
114*21fbd01bSAndreas Gohr    /**
115*21fbd01bSAndreas Gohr     * Indexer.checkIntegrity aggregates all collection checks
116*21fbd01bSAndreas Gohr     */
117*21fbd01bSAndreas Gohr    public function testIndexerCheckIntegrityDetectsCorruption()
118*21fbd01bSAndreas Gohr    {
119*21fbd01bSAndreas Gohr        global $conf;
120*21fbd01bSAndreas Gohr        $this->indexTestPage();
121*21fbd01bSAndreas Gohr
122*21fbd01bSAndreas Gohr        // Corrupt title index
123*21fbd01bSAndreas Gohr        $titleFile = $conf['indexdir'] . '/title.idx';
124*21fbd01bSAndreas Gohr        file_put_contents($titleFile, "extra title\n", FILE_APPEND);
125*21fbd01bSAndreas Gohr
126*21fbd01bSAndreas Gohr        $this->expectException(IndexIntegrityException::class);
127*21fbd01bSAndreas Gohr        (new Indexer())->checkIntegrity();
128*21fbd01bSAndreas Gohr    }
129*21fbd01bSAndreas Gohr}
130