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 | use UnexpectedValueException; |
| 6 | |
| 7 | /** |
| 8 | * Simple class for SuggestionsBuilder that needs to munge the title |
| 9 | * into a list of "subphrases" suggestions. |
| 10 | * Subphrases are only generated for title, redirects are not yet supported. |
| 11 | * A set of new fields is used to insert these suggestions 'suggest-extra' |
| 12 | * is used by default but can be overridden with string[] getExtraFields(). |
| 13 | */ |
| 14 | class NaiveSubphrasesSuggestionsBuilder implements ExtraSuggestionsBuilder { |
| 15 | /** @const string */ |
| 16 | private const LANG_FIELD = 'language'; |
| 17 | |
| 18 | /** @const int */ |
| 19 | private const MAX_SUBPHRASES = 10; |
| 20 | |
| 21 | /** @const string subpage type */ |
| 22 | public const SUBPAGE_TYPE = 'subpage'; |
| 23 | |
| 24 | /** @const string subpage type */ |
| 25 | public const STARTS_WITH_ANY_WORDS_TYPE = 'anywords'; |
| 26 | |
| 27 | /** |
| 28 | * @var string[] list of regex char ranges indexed by type |
| 29 | */ |
| 30 | private static $RANGES_BY_TYPE = [ |
| 31 | self::SUBPAGE_TYPE => '\/', |
| 32 | self::STARTS_WITH_ANY_WORDS_TYPE => '\/\s', |
| 33 | ]; |
| 34 | |
| 35 | /** @var int */ |
| 36 | private $maxSubPhrases; |
| 37 | |
| 38 | /** |
| 39 | * @var string regex character range, this value must be a valid char |
| 40 | * range and will be used to build a regular expression like |
| 41 | * '[' . $charRange . ']' |
| 42 | */ |
| 43 | private $charRange; |
| 44 | |
| 45 | /** |
| 46 | * @param string $charRange character range used to split subphrases |
| 47 | * @param int $maxSubPhrases defaults to MAX_SUBPHRASES |
| 48 | */ |
| 49 | public function __construct( $charRange, $maxSubPhrases = self::MAX_SUBPHRASES ) { |
| 50 | $this->charRange = $charRange; |
| 51 | $this->maxSubPhrases = $maxSubPhrases; |
| 52 | } |
| 53 | |
| 54 | public static function create( array $config ) { |
| 55 | $limit = $config['limit'] ?? self::MAX_SUBPHRASES; |
| 56 | if ( !isset( self::$RANGES_BY_TYPE[$config['type']] ) ) { |
| 57 | throw new UnexpectedValueException( "Unsupported NaiveSubphrasesSuggestionsBuilder type " . |
| 58 | $config['type'] ); |
| 59 | } |
| 60 | $cr = self::$RANGES_BY_TYPE[$config['type']]; |
| 61 | return new self( $cr, $limit ); |
| 62 | } |
| 63 | |
| 64 | /** |
| 65 | * Get the char range used by this builder |
| 66 | * to split and generate subphrase suggestions |
| 67 | * @return string a valid regex char range that will be inserted inside |
| 68 | * square brackets. |
| 69 | */ |
| 70 | protected function getCharRange() { |
| 71 | return $this->charRange; |
| 72 | } |
| 73 | |
| 74 | /** |
| 75 | * List of FST fields where the subphrase suggestions |
| 76 | * will be added. |
| 77 | * @return string[] |
| 78 | */ |
| 79 | protected function getExtraFields() { |
| 80 | return [ 'suggest-subphrases' ]; |
| 81 | } |
| 82 | |
| 83 | /** |
| 84 | * @inheritDoc |
| 85 | */ |
| 86 | public function getRequiredFields() { |
| 87 | // This builder needs the language field |
| 88 | // to exclude subpages generated by the translate |
| 89 | // extension |
| 90 | return [ self::LANG_FIELD ]; |
| 91 | } |
| 92 | |
| 93 | /** |
| 94 | * @param mixed[] $inputDoc |
| 95 | * @param string $suggestType (title or redirect) |
| 96 | * @param int $score |
| 97 | * @param \Elastica\Document $suggestDoc suggestion type (title or redirect) |
| 98 | * @param int $targetNamespace |
| 99 | */ |
| 100 | public function build( array $inputDoc, $suggestType, $score, \Elastica\Document $suggestDoc, $targetNamespace ) { |
| 101 | if ( $suggestType === SuggestBuilder::REDIRECT_SUGGESTION ) { |
| 102 | // It's unclear howto support redirects here. |
| 103 | // It seems hard to retrieve the best redirect if |
| 104 | // we destroy it with this builder. We would have to |
| 105 | // add a special code at search time and apply the |
| 106 | // same splitting strategy on retrieved redirects. |
| 107 | return; |
| 108 | } |
| 109 | |
| 110 | $language = ""; |
| 111 | if ( isset( $inputDoc[self::LANG_FIELD] ) ) { |
| 112 | $language = $inputDoc[self::LANG_FIELD]; |
| 113 | } |
| 114 | |
| 115 | $subPages = $this->tokenize( $inputDoc['title'], $language ); |
| 116 | if ( $subPages ) { |
| 117 | $suggest = $suggestDoc->get( 'suggest' ); |
| 118 | $suggest['input'] = $subPages; |
| 119 | foreach ( $this->getExtraFields() as $field ) { |
| 120 | $suggestDoc->set( $field, $suggest ); |
| 121 | } |
| 122 | } |
| 123 | } |
| 124 | |
| 125 | /** |
| 126 | * Split a translated page title into an array |
| 127 | * with the title at offset 0 and the language |
| 128 | * subpage at offset 1. |
| 129 | * |
| 130 | * e.g. splitTranslatedPage("Hello/en", "en") |
| 131 | * - will output [ "Hello", "/en" ] |
| 132 | * e.g. splitTranslatedPage("Hello/test", "en") |
| 133 | * - will output [ "Hello/test", "" ] |
| 134 | * |
| 135 | * @param string $title |
| 136 | * @param string $language |
| 137 | * @return string[] |
| 138 | */ |
| 139 | public function splitTranslatedPage( $title, $language ) { |
| 140 | $langSubPage = '/' . $language; |
| 141 | if ( strlen( $langSubPage ) < strlen( $title ) && |
| 142 | substr_compare( $title, $langSubPage, -strlen( $langSubPage ) ) == 0 |
| 143 | ) { |
| 144 | return [ substr( $title, 0, -strlen( $langSubPage ) ), $langSubPage ]; |
| 145 | } else { |
| 146 | return [ $title, "" ]; |
| 147 | } |
| 148 | } |
| 149 | |
| 150 | /** |
| 151 | * Tokenize the input $title by generating phrases suited |
| 152 | * for completion search. |
| 153 | * e.g. : |
| 154 | * $title = "Hello Beautifull Word/en"; |
| 155 | * $builder->tokenize( $title, "en", "\\s" ); |
| 156 | * will generate the following array: |
| 157 | * [ "Beautifull Word/en", "Word/en" ] |
| 158 | * |
| 159 | * @param string $title |
| 160 | * @param string $language |
| 161 | * @return string[] tokenized phrasal suggestions |
| 162 | */ |
| 163 | public function tokenize( $title, $language ) { |
| 164 | [ $title, $langSubPage ] = $this->splitTranslatedPage( $title, $language ); |
| 165 | |
| 166 | $cr = $this->getCharRange(); |
| 167 | $matches = preg_split( "/[$cr]+/", $title, $this->maxSubPhrases + 1, |
| 168 | PREG_SPLIT_OFFSET_CAPTURE | PREG_SPLIT_NO_EMPTY ); |
| 169 | // Remove the first one because it's the whole title |
| 170 | array_shift( $matches ); |
| 171 | $subphrases = []; |
| 172 | foreach ( $matches as $m ) { |
| 173 | $subphrases[] = substr( $title, (int)$m[1] ) . $langSubPage; |
| 174 | } |
| 175 | return $subphrases; |
| 176 | } |
| 177 | } |