Code Coverage |
||||||||||
Lines |
Functions and Methods |
Classes and Traits |
||||||||
| Total | |
98.57% |
69 / 70 |
|
80.00% |
4 / 5 |
CRAP | |
0.00% |
0 / 1 |
| LicenseParser | |
98.57% |
69 / 70 |
|
80.00% |
4 / 5 |
28 | |
0.00% |
0 / 1 |
| parseLicenseString | |
100.00% |
2 / 2 |
|
100.00% |
1 / 1 |
2 | |||
| sortDataByLicensePriority | |
100.00% |
14 / 14 |
|
100.00% |
1 / 1 |
3 | |||
| parseCreativeCommonsLicenseString | |
97.30% |
36 / 37 |
|
0.00% |
0 / 1 |
13 | |||
| parsePublicDomainLicenseString | |
100.00% |
6 / 6 |
|
100.00% |
1 / 1 |
3 | |||
| getLicensePriority | |
100.00% |
11 / 11 |
|
100.00% |
1 / 1 |
7 | |||
| 1 | <?php |
| 2 | |
| 3 | namespace CommonsMetadata; |
| 4 | |
| 5 | /** |
| 6 | * Takes a license name string, and splits it up into various license elements (version, etc). |
| 7 | * The string is typically a Commons category name, or template name, or license shortname |
| 8 | * (see {@link https://commons.wikimedia.org/wiki/Commons:Machine-readable_data}) |
| 9 | */ |
| 10 | class LicenseParser { |
| 11 | |
| 12 | /** |
| 13 | * @var string[] Nonstandard license name patterns used in categories/templates/shortnames |
| 14 | */ |
| 15 | public static $licenseAliases = [ |
| 16 | 'cc-by-sa-3.0-migrated' => 'cc-by-sa-3.0', |
| 17 | 'cc-by-sa-3.0-migrated-with-disclaimers' => 'cc-by-sa-3.0', |
| 18 | 'cc-by-sa-3.0-2.5-2.0-1.0' => 'cc-by-sa-3.0', |
| 19 | 'cc-by-sa-2.5-2.0-1.0' => 'cc-by-sa-2.5', |
| 20 | 'cc-by-2.0-stma' => 'cc-by-2.0', |
| 21 | 'cc-by-sa-1.0+' => 'cc-by-sa-3.0', |
| 22 | ]; |
| 23 | |
| 24 | /** |
| 25 | * Takes a CC license string (could be a category name, template name etc) |
| 26 | * and returns template information (or null if the license was not recognized). |
| 27 | * The returned array can have the following keys: |
| 28 | * - family: e.g. cc, gfdl |
| 29 | * - type: e.g. cc-by-sa |
| 30 | * - version: e.g. 2.5 |
| 31 | * - region: e.g. nl |
| 32 | * - name: all the above put together, e.g. cc-by-sa-2.5-nl |
| 33 | * Only name is required. |
| 34 | * @param string $str |
| 35 | * @return array|null |
| 36 | */ |
| 37 | public function parseLicenseString( $str ) { |
| 38 | return $this->parseCreativeCommonsLicenseString( $str ) |
| 39 | ?: $this->parsePublicDomainLicenseString( $str ); |
| 40 | } |
| 41 | |
| 42 | /** |
| 43 | * Takes an array width license data and sorts it by license priority. |
| 44 | * The sort is stable, and the input array is not changed. |
| 45 | * The method will call $getLicenseStringCallback( $data[$key], $key ) and expect a license name |
| 46 | * @param string[] $data |
| 47 | * @param callable $getLicenseStringCallback |
| 48 | * @return array |
| 49 | */ |
| 50 | public function sortDataByLicensePriority( array $data, $getLicenseStringCallback ) { |
| 51 | $licensePriorities = []; |
| 52 | $i = 0; |
| 53 | foreach ( $data as $key => $value ) { |
| 54 | $license = $getLicenseStringCallback( $value, $key ); |
| 55 | $licenseDetails = $this->parseLicenseString( $license ); |
| 56 | $priority = $this->getLicensePriority( $licenseDetails ); |
| 57 | $licensePriorities[$key] = [ $priority, $i ]; |
| 58 | $i++; |
| 59 | } |
| 60 | |
| 61 | uksort( $data, static function ( $a, $b ) use ( $licensePriorities ) { |
| 62 | // equal priority, keep original order |
| 63 | if ( $licensePriorities[$a][0] === $licensePriorities[$b][0] ) { |
| 64 | return $licensePriorities[$a][1] - $licensePriorities[$b][1]; |
| 65 | } else { |
| 66 | // higher priority means smaller wrt sorting |
| 67 | return $licensePriorities[$b][0] - $licensePriorities[$a][0]; |
| 68 | } |
| 69 | } ); |
| 70 | |
| 71 | return $data; |
| 72 | } |
| 73 | |
| 74 | /** |
| 75 | * Takes a CC license string and returns template information. |
| 76 | * @see parseLicenceString() |
| 77 | * @param string $str |
| 78 | * @return array|null |
| 79 | */ |
| 80 | protected function parseCreativeCommonsLicenseString( $str ) { |
| 81 | $data = [ |
| 82 | 'family' => 'cc', |
| 83 | 'type' => null, |
| 84 | 'version' => null, |
| 85 | 'region' => null, |
| 86 | 'name' => null, |
| 87 | ]; |
| 88 | |
| 89 | $str = strtolower( trim( $str ) ); |
| 90 | if ( isset( self::$licenseAliases[$str] ) ) { |
| 91 | $str = self::$licenseAliases[$str]; |
| 92 | } |
| 93 | |
| 94 | // some special cases first |
| 95 | if ( in_array( $str, [ 'cc0', 'cc-pd' ], true ) ) { |
| 96 | $data['type'] = $data['name'] = $str; |
| 97 | return $data; |
| 98 | } |
| 99 | |
| 100 | $parts = preg_split( '/[- ]/', $str ); |
| 101 | if ( $parts[0] != 'cc' ) { |
| 102 | return null; |
| 103 | } |
| 104 | |
| 105 | $countParts = count( $parts ); |
| 106 | for ( $i = 1; $i < $countParts; $i++ ) { |
| 107 | if ( !in_array( $parts[$i], [ 'by', 'sa', 'nc', 'nd' ] ) ) { |
| 108 | break; |
| 109 | } |
| 110 | if ( in_array( $parts[$i], [ 'nc', 'nd' ] ) ) { |
| 111 | // ignore non-free licenses |
| 112 | return null; |
| 113 | } |
| 114 | } |
| 115 | $data['type'] = implode( '-', array_slice( $parts, 0, $i ) ); |
| 116 | |
| 117 | if ( $i < $countParts && is_numeric( $parts[$i] ) ) { |
| 118 | $data['version'] = $parts[$i]; |
| 119 | $i++; |
| 120 | } else { |
| 121 | return null; |
| 122 | } |
| 123 | |
| 124 | if ( $i < $countParts && ( |
| 125 | preg_match( '/^\w\w$/', $parts[$i] ) |
| 126 | || $parts[$i] == 'scotland' |
| 127 | ) |
| 128 | ) { |
| 129 | $data['region'] = $parts[$i]; |
| 130 | $i++; |
| 131 | } |
| 132 | |
| 133 | if ( $i != $countParts ) { |
| 134 | return null; |
| 135 | } |
| 136 | |
| 137 | $data['name'] = implode( '-', |
| 138 | array_filter( [ $data['type'], $data['version'], $data['region'] ] ) ); |
| 139 | return $data; |
| 140 | } |
| 141 | |
| 142 | /** |
| 143 | * Takes a PD license string and returns template information. |
| 144 | * @see parseLicenceString() |
| 145 | * @param string $str |
| 146 | * @return array|null |
| 147 | */ |
| 148 | protected function parsePublicDomainLicenseString( $str ) { |
| 149 | // A very simple approach, but should work most of the time with licence shortnames. |
| 150 | if ( strtolower( $str ) === 'public domain' || strtolower( $str ) === 'pd' ) { |
| 151 | return [ |
| 152 | 'family' => 'pd', |
| 153 | 'name' => 'pd', |
| 154 | ]; |
| 155 | } |
| 156 | return null; |
| 157 | } |
| 158 | |
| 159 | /** |
| 160 | * Returns a priority value for this license. The license with the highest priority will be |
| 161 | * returned in the GetExtendedMetadata hook. |
| 162 | * @param array $licenseData data from LicenseParser::parseLicenseString() |
| 163 | * @return int |
| 164 | */ |
| 165 | protected function getLicensePriority( $licenseData ) { |
| 166 | $priority = 0; |
| 167 | if ( isset( $licenseData[ 'family' ] ) ) { |
| 168 | if ( $licenseData[ 'family' ] === 'pd' ) { |
| 169 | $priority = 2000; |
| 170 | } elseif ( $licenseData[ 'family' ] === 'cc' ) { |
| 171 | // ignore non-free CC licenses for now; an image with such a license probably |
| 172 | // won't have a better license anyway |
| 173 | $priority += 1000; |
| 174 | if ( isset( $licenseData['type'] ) && $licenseData['type'] === 'cc-by' ) { |
| 175 | // prefer the less restrictive CC-BY over CC-BY-SA |
| 176 | $priority += 100; |
| 177 | } |
| 178 | if ( isset( $licenseData['version'] ) ) { |
| 179 | // prefer newer licenses |
| 180 | $priority += (int)( 10 * (float)$licenseData['version'] ); |
| 181 | } |
| 182 | } |
| 183 | } |
| 184 | return $priority; |
| 185 | } |
| 186 | |
| 187 | } |