Code Coverage |
||||||||||
Lines |
Functions and Methods |
Classes and Traits |
||||||||
Total | |
100.00% |
28 / 28 |
|
100.00% |
5 / 5 |
CRAP | |
100.00% |
1 / 1 |
HighlightingTrait | |
100.00% |
28 / 28 |
|
100.00% |
5 / 5 |
11 | |
100.00% |
1 / 1 |
escapeHighlightedText | |
100.00% |
8 / 8 |
|
100.00% |
1 / 1 |
1 | |||
findRedirectTitle | |
100.00% |
14 / 14 |
|
100.00% |
1 / 1 |
7 | |||
containsMatches | |
100.00% |
1 / 1 |
|
100.00% |
1 / 1 |
1 | |||
stripHighlighting | |
100.00% |
2 / 2 |
|
100.00% |
1 / 1 |
1 | |||
findSectionTitle | |
100.00% |
3 / 3 |
|
100.00% |
1 / 1 |
1 | |||
getTitleHelper | n/a |
0 / 0 |
n/a |
0 / 0 |
0 |
1 | <?php |
2 | |
3 | namespace CirrusSearch\Search\Fetch; |
4 | |
5 | use CirrusSearch\Search\TitleHelper; |
6 | use CirrusSearch\Searcher; |
7 | use MediaWiki\Logger\LoggerFactory; |
8 | use MediaWiki\Title\Title; |
9 | |
10 | trait HighlightingTrait { |
11 | /** |
12 | * Escape highlighted text coming back from Elasticsearch. |
13 | * |
14 | * @param string $snippet highlighted snippet returned from elasticsearch |
15 | * @return string $snippet with html escaped _except_ highlighting pre and post tags |
16 | */ |
17 | protected function escapeHighlightedText( $snippet ) { |
18 | /** |
19 | * \p{M} matches any combining Unicode character |
20 | * \P{M} matches any non-combining Unicode character |
21 | * |
22 | * For HIGHLIGHT_PRE_MARKER, move the marker earlier if it occurs before a |
23 | * combining character, and there is a non-combining character (and zero |
24 | * or more combining characters) directly before it. |
25 | * |
26 | * For HIGHLIGHT_POST_MARKER, move the marker later if it occurs before |
27 | * one or more combining characters. |
28 | */ |
29 | $snippet = preg_replace( '/(\P{M}\p{M}*)(' . Searcher::HIGHLIGHT_PRE_MARKER . |
30 | ')(\p{M}+)/u', '$2$1$3', $snippet ); |
31 | $snippet = preg_replace( '/(' . Searcher::HIGHLIGHT_POST_MARKER . ')(\p{M}+)/u', |
32 | '$2$1', $snippet ); |
33 | return strtr( htmlspecialchars( $snippet ), [ |
34 | Searcher::HIGHLIGHT_PRE_MARKER => Searcher::HIGHLIGHT_PRE, |
35 | Searcher::HIGHLIGHT_POST_MARKER => Searcher::HIGHLIGHT_POST |
36 | ] ); |
37 | } |
38 | |
39 | /** |
40 | * Build the redirect title from the highlighted redirect snippet. |
41 | * |
42 | * @param \Elastica\Result $result |
43 | * @param string $snippet Highlighted redirect snippet |
44 | * @return Title|null object representing the redirect |
45 | */ |
46 | protected function findRedirectTitle( \Elastica\Result $result, $snippet ) { |
47 | $title = $this->stripHighlighting( $snippet ); |
48 | // Grab the redirect that matches the highlighted title with the lowest namespace. |
49 | $redirects = $result->redirect; |
50 | // That is pretty arbitrary but it prioritizes 0 over others. |
51 | $best = null; |
52 | if ( $redirects !== null ) { |
53 | foreach ( $redirects as $redirect ) { |
54 | if ( $redirect[ 'title' ] === $title && ( $best === null || $best[ 'namespace' ] > $redirect['namespace'] ) ) { |
55 | $best = $redirect; |
56 | } |
57 | } |
58 | } |
59 | if ( $best === null ) { |
60 | LoggerFactory::getInstance( 'CirrusSearch' )->warning( |
61 | "Search backend highlighted a redirect ({title}) but didn't return it.", |
62 | [ 'title' => $title ] |
63 | ); |
64 | return null; |
65 | } |
66 | return $this->getTitleHelper()->makeRedirectTitle( $result, $best['title'], $best['namespace'] ); |
67 | } |
68 | |
69 | /** |
70 | * Checks if a snippet contains matches by looking for HIGHLIGHT_PRE. |
71 | * |
72 | * @param string $snippet highlighted snippet returned from elasticsearch |
73 | * @return bool true if $snippet contains matches, false otherwise |
74 | */ |
75 | protected function containsMatches( $snippet ) { |
76 | return strpos( $snippet, Searcher::HIGHLIGHT_PRE_MARKER ) !== false; |
77 | } |
78 | |
79 | /** |
80 | * @param string $highlighted |
81 | * @return string |
82 | */ |
83 | protected function stripHighlighting( $highlighted ) { |
84 | $markers = [ Searcher::HIGHLIGHT_PRE_MARKER, Searcher::HIGHLIGHT_POST_MARKER ]; |
85 | return str_replace( $markers, '', $highlighted ); |
86 | } |
87 | |
88 | /** |
89 | * @param string $highlighted |
90 | * @param Title $title |
91 | * @return Title |
92 | */ |
93 | protected function findSectionTitle( $highlighted, Title $title ) { |
94 | return $title->createFragmentTarget( $this->getTitleHelper()->sanitizeSectionFragment( |
95 | $this->stripHighlighting( $highlighted ) |
96 | ) ); |
97 | } |
98 | |
99 | /** |
100 | * @return TitleHelper |
101 | */ |
102 | abstract protected function getTitleHelper(): TitleHelper; |
103 | } |