BookStack Code Mirror - bookstack/blob - tests/Search/SearchIndexingTest.php

   1 <?php
   2
   3 namespace Tests\Search;
   4
   5 use Tests\TestCase;
   6
   7 class SearchIndexingTest extends TestCase
   8 {
   9     public function test_terms_in_headers_have_an_adjusted_index_score()
  10     {
  11         $page = $this->entities->newPage(['name' => 'Test page A', 'html' => '
  12             <p>TermA</p>
  13             <h1>TermB <strong>TermNested</strong></h1>
  14             <h2>TermC</h2>
  15             <h3>TermD</h3>
  16             <h4>TermE</h4>
  17             <h5>TermF</h5>
  18             <h6>TermG</h6>
  19         ']);
  20
  21         $scoreByTerm = $page->searchTerms()->pluck('score', 'term');
  22
  23         $this->assertEquals(1, $scoreByTerm->get('TermA'));
  24         $this->assertEquals(10, $scoreByTerm->get('TermB'));
  25         $this->assertEquals(10, $scoreByTerm->get('TermNested'));
  26         $this->assertEquals(5, $scoreByTerm->get('TermC'));
  27         $this->assertEquals(4, $scoreByTerm->get('TermD'));
  28         $this->assertEquals(3, $scoreByTerm->get('TermE'));
  29         $this->assertEquals(2, $scoreByTerm->get('TermF'));
  30         // Is 1.5 but stored as integer, rounding up
  31         $this->assertEquals(2, $scoreByTerm->get('TermG'));
  32     }
  33
  34     public function test_indexing_works_as_expected_for_page_with_lots_of_terms()
  35     {
  36         $this->markTestSkipped('Time consuming test');
  37
  38         $count = 100000;
  39         $text = '';
  40         $chars = 'abcdefghijklmnopqrstuvwxyzABCDEFGHIJKLMNOPQRSTUVWXYZ0123456789-_#';
  41         for ($i = 0; $i < $count; $i++) {
  42             $text .= substr(str_shuffle($chars), 0, 5) . ' ';
  43         }
  44
  45         $page = $this->entities->newPage(['name' => 'Test page A', 'html' => '<p>' . $text . '</p>']);
  46
  47         $termCount = $page->searchTerms()->count();
  48
  49         // Expect at least 90% unique rate
  50         $this->assertGreaterThan($count * 0.9, $termCount);
  51     }
  52
  53     public function test_name_and_content_terms_are_merged_to_single_score()
  54     {
  55         $page = $this->entities->newPage(['name' => 'TermA', 'html' => '
  56             <p>TermA</p>
  57         ']);
  58
  59         $scoreByTerm = $page->searchTerms()->pluck('score', 'term');
  60
  61         // Scores 40 for being in the name then 1 for being in the content
  62         $this->assertEquals(41, $scoreByTerm->get('TermA'));
  63     }
  64
  65     public function test_tag_names_and_values_are_indexed_for_search()
  66     {
  67         $page = $this->entities->newPage(['name' => 'PageA', 'html' => '<p>content</p>', 'tags' => [
  68             ['name' => 'Animal', 'value' => 'MeowieCat'],
  69             ['name' => 'SuperImportant'],
  70         ]]);
  71
  72         $scoreByTerm = $page->searchTerms()->pluck('score', 'term');
  73         $this->assertEquals(5, $scoreByTerm->get('MeowieCat'));
  74         $this->assertEquals(3, $scoreByTerm->get('Animal'));
  75         $this->assertEquals(3, $scoreByTerm->get('SuperImportant'));
  76     }
  77
  78     public function test_terms_containing_guillemets_handled()
  79     {
  80         $page = $this->entities->newPage(['html' => '<p>«Hello there» and « there »</p>']);
  81
  82         $scoreByTerm = $page->searchTerms()->pluck('score', 'term');
  83         $expected = ['Hello', 'there', 'and'];
  84         foreach ($expected as $term) {
  85             $this->assertNotNull($scoreByTerm->get($term), "Failed asserting that \"$term\" is indexed");
  86         }
  87
  88         $nonExpected = ['«', '»'];
  89         foreach ($nonExpected as $term) {
  90             $this->assertNull($scoreByTerm->get($term), "Failed asserting that \"$term\" is not indexed");
  91         }
  92     }
  93
  94     public function test_terms_containing_punctuation_within_retain_original_form_and_split_form_in_index()
  95     {
  96         $page = $this->entities->newPage(['html' => '<p>super.duper awesome-beans big- barry cheese.</p><p>biscuits</p><p>a-bs</p>']);
  97
  98         $scoreByTerm = $page->searchTerms()->pluck('score', 'term');
  99         $expected = ['super', 'duper', 'super.duper', 'awesome-beans', 'awesome', 'beans', 'big', 'barry', 'cheese', 'biscuits', 'a-bs', 'a', 'bs'];
 100         foreach ($expected as $term) {
 101             $this->assertNotNull($scoreByTerm->get($term), "Failed asserting that \"$term\" is indexed");
 102         }
 103
 104         $nonExpected = ['big-', 'big-barry', 'cheese.', 'cheese.biscuits'];
 105         foreach ($nonExpected as $term) {
 106             $this->assertNull($scoreByTerm->get($term), "Failed asserting that \"$term\" is not indexed");
 107         }
 108     }
 109 }