Code Coverage |
||||||||||
Lines |
Functions and Methods |
Classes and Traits |
||||||||
Total | |
0.00% |
0 / 36 |
|
0.00% |
0 / 8 |
CRAP | |
0.00% |
0 / 1 |
NaiveSubphrasesSuggestionsBuilder | |
0.00% |
0 / 36 |
|
0.00% |
0 / 8 |
272 | |
0.00% |
0 / 1 |
__construct | |
0.00% |
0 / 2 |
|
0.00% |
0 / 1 |
2 | |||
create | |
0.00% |
0 / 6 |
|
0.00% |
0 / 1 |
6 | |||
getCharRange | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
2 | |||
getExtraFields | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
2 | |||
getRequiredFields | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
2 | |||
build | |
0.00% |
0 / 11 |
|
0.00% |
0 / 1 |
30 | |||
splitTranslatedPage | |
0.00% |
0 / 5 |
|
0.00% |
0 / 1 |
12 | |||
tokenize | |
0.00% |
0 / 9 |
|
0.00% |
0 / 1 |
6 |
1 | <?php |
2 | |
3 | namespace CirrusSearch\BuildDocument\Completion; |
4 | |
5 | /** |
6 | * Simple class for SuggestionsBuilder that needs to munge the title |
7 | * into a list of "subphrases" suggestions. |
8 | * Subphrases are only generated for title, redirects are not yet supported. |
9 | * A set of new fields is used to insert these suggestions 'suggest-extra' |
10 | * is used by default but can be overridden with string[] getExtraFields(). |
11 | */ |
12 | class NaiveSubphrasesSuggestionsBuilder implements ExtraSuggestionsBuilder { |
13 | /** @const string */ |
14 | private const LANG_FIELD = 'language'; |
15 | |
16 | /** @const int */ |
17 | private const MAX_SUBPHRASES = 10; |
18 | |
19 | /** @const string subpage type */ |
20 | public const SUBPAGE_TYPE = 'subpage'; |
21 | |
22 | /** @const string subpage type */ |
23 | public const STARTS_WITH_ANY_WORDS_TYPE = 'anywords'; |
24 | |
25 | /** |
26 | * @var string[] list of regex char ranges indexed by type |
27 | */ |
28 | private static $RANGES_BY_TYPE = [ |
29 | self::SUBPAGE_TYPE => '\/', |
30 | self::STARTS_WITH_ANY_WORDS_TYPE => '\/\s', |
31 | ]; |
32 | |
33 | /** @var int */ |
34 | private $maxSubPhrases; |
35 | |
36 | /** |
37 | * @var string regex character range, this value must be a valid char |
38 | * range and will be used to build a regular expression like |
39 | * '[' . $charRange . ']' |
40 | */ |
41 | private $charRange; |
42 | |
43 | /** |
44 | * @param string $charRange character range used to split subphrases |
45 | * @param int $maxSubPhrases defaults to MAX_SUBPHRASES |
46 | */ |
47 | public function __construct( $charRange, $maxSubPhrases = self::MAX_SUBPHRASES ) { |
48 | $this->charRange = $charRange; |
49 | $this->maxSubPhrases = $maxSubPhrases; |
50 | } |
51 | |
52 | public static function create( array $config ) { |
53 | $limit = $config['limit'] ?? self::MAX_SUBPHRASES; |
54 | if ( !isset( self::$RANGES_BY_TYPE[$config['type']] ) ) { |
55 | throw new \Exception( "Unsupported NaiveSubphrasesSuggestionsBuilder type " . |
56 | $config['type'] ); |
57 | } |
58 | $cr = self::$RANGES_BY_TYPE[$config['type']]; |
59 | return new self( $cr, $limit ); |
60 | } |
61 | |
62 | /** |
63 | * Get the char range used by this builder |
64 | * to split and generate subphrase suggestions |
65 | * @return string a valid regex char range that will be inserted inside |
66 | * square brackets. |
67 | */ |
68 | protected function getCharRange() { |
69 | return $this->charRange; |
70 | } |
71 | |
72 | /** |
73 | * List of FST fields where the subphrase suggestions |
74 | * will be added. |
75 | * @return string[] |
76 | */ |
77 | protected function getExtraFields() { |
78 | return [ 'suggest-subphrases' ]; |
79 | } |
80 | |
81 | /** |
82 | * @inheritDoc |
83 | */ |
84 | public function getRequiredFields() { |
85 | // This builder needs the language field |
86 | // to exclude subpages generated by the translate |
87 | // extension |
88 | return [ self::LANG_FIELD ]; |
89 | } |
90 | |
91 | /** |
92 | * @param mixed[] $inputDoc |
93 | * @param string $suggestType (title or redirect) |
94 | * @param int $score |
95 | * @param \Elastica\Document $suggestDoc suggestion type (title or redirect) |
96 | * @param int $targetNamespace |
97 | */ |
98 | public function build( array $inputDoc, $suggestType, $score, \Elastica\Document $suggestDoc, $targetNamespace ) { |
99 | if ( $suggestType === SuggestBuilder::REDIRECT_SUGGESTION ) { |
100 | // It's unclear howto support redirects here. |
101 | // It seems hard to retrieve the best redirect if |
102 | // we destroy it with this builder. We would have to |
103 | // add a special code at search time and apply the |
104 | // same splitting strategy on retrieved redirects. |
105 | return; |
106 | } |
107 | |
108 | $language = ""; |
109 | if ( isset( $inputDoc[self::LANG_FIELD] ) ) { |
110 | $language = $inputDoc[self::LANG_FIELD]; |
111 | } |
112 | |
113 | $subPages = $this->tokenize( $inputDoc['title'], $language ); |
114 | if ( $subPages ) { |
115 | $suggest = $suggestDoc->get( 'suggest' ); |
116 | $suggest['input'] = $subPages; |
117 | foreach ( $this->getExtraFields() as $field ) { |
118 | $suggestDoc->set( $field, $suggest ); |
119 | } |
120 | } |
121 | } |
122 | |
123 | /** |
124 | * Split a translated page title into an array |
125 | * with the title at offset 0 and the language |
126 | * subpage at offset 1. |
127 | * |
128 | * e.g. splitTranslatedPage("Hello/en", "en") |
129 | * - will output [ "Hello", "/en" ] |
130 | * e.g. splitTranslatedPage("Hello/test", "en") |
131 | * - will output [ "Hello/test", "" ] |
132 | * |
133 | * @param string $title |
134 | * @param string $language |
135 | * @return string[] |
136 | */ |
137 | public function splitTranslatedPage( $title, $language ) { |
138 | $langSubPage = '/' . $language; |
139 | if ( strlen( $langSubPage ) < strlen( $title ) && |
140 | substr_compare( $title, $langSubPage, -strlen( $langSubPage ) ) == 0 |
141 | ) { |
142 | return [ substr( $title, 0, -strlen( $langSubPage ) ), $langSubPage ]; |
143 | } else { |
144 | return [ $title, "" ]; |
145 | } |
146 | } |
147 | |
148 | /** |
149 | * Tokenize the input $title by generating phrases suited |
150 | * for completion search. |
151 | * e.g. : |
152 | * $title = "Hello Beautifull Word/en"; |
153 | * $builder->tokenize( $title, "en", "\\s" ); |
154 | * will generate the following array: |
155 | * [ "Beautifull Word/en", "Word/en" ] |
156 | * |
157 | * @param string $title |
158 | * @param string $language |
159 | * @return string[] tokenized phrasal suggestions |
160 | */ |
161 | public function tokenize( $title, $language ) { |
162 | list( $title, $langSubPage ) = $this->splitTranslatedPage( $title, $language ); |
163 | |
164 | $cr = $this->getCharRange(); |
165 | $matches = preg_split( "/[$cr]+/", $title, $this->maxSubPhrases + 1, |
166 | PREG_SPLIT_OFFSET_CAPTURE | PREG_SPLIT_NO_EMPTY ); |
167 | // Remove the first one because it's the whole title |
168 | array_shift( $matches ); |
169 | $subphrases = []; |
170 | foreach ( $matches as $m ) { |
171 | $subphrases[] = substr( $title, (int)$m[1] ) . $langSubPage; |
172 | } |
173 | return $subphrases; |
174 | } |
175 | } |