Code Coverage |
||||||||||
Lines |
Functions and Methods |
Classes and Traits |
||||||||
| Total | |
96.72% |
59 / 61 |
|
84.62% |
11 / 13 |
CRAP | |
0.00% |
0 / 1 |
| WikitextConversions | |
96.72% |
59 / 61 |
|
84.62% |
11 / 13 |
28 | |
0.00% |
0 / 1 |
| __construct | |
100.00% |
24 / 24 |
|
100.00% |
1 / 1 |
9 | |||
| swapHeading | |
100.00% |
1 / 1 |
|
100.00% |
1 / 1 |
1 | |||
| isTemplateGood | |
100.00% |
2 / 2 |
|
100.00% |
1 / 1 |
1 | |||
| hasGoodTemplates | |
100.00% |
1 / 1 |
|
100.00% |
1 / 1 |
1 | |||
| isTemplateBad | |
100.00% |
2 / 2 |
|
100.00% |
1 / 1 |
1 | |||
| isCategoryBad | |
100.00% |
2 / 2 |
|
100.00% |
1 / 1 |
1 | |||
| isObsoleteTemplate | |
100.00% |
1 / 1 |
|
100.00% |
1 / 1 |
1 | |||
| swapTemplate | |
100.00% |
2 / 2 |
|
100.00% |
1 / 1 |
1 | |||
| getTemplateParameters | |
92.86% |
13 / 14 |
|
0.00% |
0 / 1 |
5.01 | |||
| getRequiredTemplateParameters | |
88.89% |
8 / 9 |
|
0.00% |
0 / 1 |
4.02 | |||
| lowercasePageName | |
100.00% |
1 / 1 |
|
100.00% |
1 / 1 |
1 | |||
| normalizePageName | |
100.00% |
1 / 1 |
|
100.00% |
1 / 1 |
1 | |||
| removeNamespace | |
100.00% |
1 / 1 |
|
100.00% |
1 / 1 |
1 | |||
| 1 | <?php |
| 2 | |
| 3 | namespace FileImporter\Data; |
| 4 | |
| 5 | /** |
| 6 | * Class holding validation and cleanup rules for the file description wikitext. This class is not |
| 7 | * aware of the source of these rules. They can be extracted from CommonsHelper2-compatible |
| 8 | * configuration files or other, yet to be defined sources. |
| 9 | * |
| 10 | * @license GPL-2.0-or-later |
| 11 | */ |
| 12 | class WikitextConversions { |
| 13 | |
| 14 | public const REQUIRED_TEMPLATES = 'requiredTemplates'; |
| 15 | public const FORBIDDEN_TEMPLATES = 'forbiddenTemplates'; |
| 16 | public const OBSOLETE_TEMPLATES = 'obsoleteTemplates'; |
| 17 | public const TEMPLATE_TRANSFORMATIONS = 'templateTransformations'; |
| 18 | public const FORBIDDEN_CATEGORIES = 'forbiddenCategories'; |
| 19 | public const HEADING_REPLACEMENTS = 'headingReplacements'; |
| 20 | |
| 21 | /** @var array<string,string> */ |
| 22 | private array $headingReplacements = []; |
| 23 | /** @var array<string,true> A string => true map for performance reasons */ |
| 24 | private array $goodTemplates = []; |
| 25 | /** @var array<string,true> A string => true map for performance reasons */ |
| 26 | private array $badTemplates = []; |
| 27 | /** @var array<string,true> A string => true map for performance reasons */ |
| 28 | private array $badCategories = []; |
| 29 | /** @var array<string,true> A string => true map for performance reasons */ |
| 30 | private array $obsoleteTemplates = []; |
| 31 | /** @var array<string,array{targetTemplate: string, parameters: array[]}> */ |
| 32 | private array $transferTemplates = []; |
| 33 | |
| 34 | /** |
| 35 | * @param array[] $conversions A nested array structure in the following format: |
| 36 | * [ |
| 37 | * self::REQUIRED_TEMPLATES => string[] List of case-insensitive page names without |
| 38 | * namespace prefix |
| 39 | * self::FORBIDDEN_TEMPLATES => string[] List of case-insensitive page names without |
| 40 | * namespace prefix |
| 41 | * self::OBSOLETE_TEMPLATES => string[] List of case-insensitive page names without |
| 42 | * namespace prefix |
| 43 | * self::TEMPLATE_TRANSFORMATIONS => array[] List mapping source template names without |
| 44 | * namespace prefix to replacement rules in the following format: |
| 45 | * string $sourceTemplate => [ |
| 46 | * 'targetTemplate' => string |
| 47 | * 'parameters' => [ |
| 48 | * string $targetParameter => [ |
| 49 | * 'addIfMissing' => bool |
| 50 | * 'addLanguageTemplate' => bool |
| 51 | * 'sourceParameters' => string[]|string |
| 52 | * ], |
| 53 | * … |
| 54 | * ] |
| 55 | * ], |
| 56 | * … |
| 57 | * self::FORBIDDEN_CATEGORIES => string[] List of case-insensitive page names without |
| 58 | * namespace prefix |
| 59 | * self::HEADING_REPLACEMENTS => string[] Straight 1:1 mapping of source to target headings |
| 60 | * without any `==` syntax |
| 61 | * ] |
| 62 | * |
| 63 | * @throws \InvalidArgumentException if the input format misses expected fields. This should be |
| 64 | * unreachable, as the only provider is the CommonsHelperConfigParser. |
| 65 | */ |
| 66 | public function __construct( array $conversions ) { |
| 67 | $goodTemplates = $conversions[self::REQUIRED_TEMPLATES] ?? []; |
| 68 | $badTemplates = $conversions[self::FORBIDDEN_TEMPLATES] ?? []; |
| 69 | $obsoleteTemplates = $conversions[self::OBSOLETE_TEMPLATES] ?? []; |
| 70 | $transferTemplates = $conversions[self::TEMPLATE_TRANSFORMATIONS] ?? []; |
| 71 | $badCategories = $conversions[self::FORBIDDEN_CATEGORIES] ?? []; |
| 72 | $this->headingReplacements = $conversions[self::HEADING_REPLACEMENTS] ?? []; |
| 73 | |
| 74 | foreach ( $goodTemplates as $pageName ) { |
| 75 | $this->goodTemplates[$this->lowercasePageName( $pageName )] = true; |
| 76 | } |
| 77 | |
| 78 | foreach ( $badTemplates as $pageName ) { |
| 79 | $this->badTemplates[$this->lowercasePageName( $pageName )] = true; |
| 80 | } |
| 81 | |
| 82 | foreach ( $badCategories as $pageName ) { |
| 83 | $this->badCategories[$this->lowercasePageName( $pageName )] = true; |
| 84 | } |
| 85 | |
| 86 | foreach ( $obsoleteTemplates as $pageName ) { |
| 87 | $this->obsoleteTemplates[$this->lowercasePageName( $pageName )] = true; |
| 88 | } |
| 89 | |
| 90 | foreach ( $transferTemplates as $from => $to ) { |
| 91 | // TODO: Accepts strings for backwards-compatibility; remove if not needed any more |
| 92 | if ( is_string( $to ) ) { |
| 93 | $to = [ 'targetTemplate' => $to, 'parameters' => [] ]; |
| 94 | } |
| 95 | |
| 96 | if ( empty( $to['targetTemplate'] ) ) { |
| 97 | throw new \InvalidArgumentException( "$from transfer rule misses targetTemplate" ); |
| 98 | } |
| 99 | if ( !isset( $to['parameters'] ) ) { |
| 100 | throw new \InvalidArgumentException( "$from transfer rule misses parameters" ); |
| 101 | } |
| 102 | |
| 103 | $from = $this->lowercasePageName( $from ); |
| 104 | $to['targetTemplate'] = $this->normalizePageName( $to['targetTemplate'] ); |
| 105 | $this->transferTemplates[$from] = $to; |
| 106 | } |
| 107 | } |
| 108 | |
| 109 | public function swapHeading( string $heading ): string { |
| 110 | return $this->headingReplacements[$heading] ?? $heading; |
| 111 | } |
| 112 | |
| 113 | /** |
| 114 | * @param string $pageName Case-insensitive page name. The namespace is ignored. Titles like |
| 115 | * "Template:A" and "User:A" are considered equal. |
| 116 | */ |
| 117 | public function isTemplateGood( string $pageName ): bool { |
| 118 | $pageName = $this->removeNamespace( $pageName ); |
| 119 | return array_key_exists( $this->lowercasePageName( $pageName ), $this->goodTemplates ); |
| 120 | } |
| 121 | |
| 122 | public function hasGoodTemplates(): bool { |
| 123 | return $this->goodTemplates !== []; |
| 124 | } |
| 125 | |
| 126 | /** |
| 127 | * @param string $pageName Case-insensitive page name. The namespace is ignored. Titles like |
| 128 | * "Template:A" and "User:A" are considered equal. |
| 129 | */ |
| 130 | public function isTemplateBad( string $pageName ): bool { |
| 131 | $pageName = $this->removeNamespace( $pageName ); |
| 132 | return array_key_exists( $this->lowercasePageName( $pageName ), $this->badTemplates ); |
| 133 | } |
| 134 | |
| 135 | /** |
| 136 | * @param string $pageName Case-insensitive page name. The namespace is ignored. Titles like |
| 137 | * "Category:A" and "User:A" are considered equal. |
| 138 | */ |
| 139 | public function isCategoryBad( string $pageName ): bool { |
| 140 | $pageName = $this->removeNamespace( $pageName ); |
| 141 | return array_key_exists( $this->lowercasePageName( $pageName ), $this->badCategories ); |
| 142 | } |
| 143 | |
| 144 | /** |
| 145 | * @param string $pageName Case-insensitive page name. Prefixes are significant. |
| 146 | */ |
| 147 | public function isObsoleteTemplate( string $pageName ): bool { |
| 148 | return array_key_exists( $this->lowercasePageName( $pageName ), $this->obsoleteTemplates ); |
| 149 | } |
| 150 | |
| 151 | /** |
| 152 | * @param string $templateName Case-insensitive page name. Prefixes are significant. |
| 153 | */ |
| 154 | public function swapTemplate( string $templateName ): ?string { |
| 155 | $templateName = $this->lowercasePageName( $templateName ); |
| 156 | return $this->transferTemplates[$templateName]['targetTemplate'] ?? null; |
| 157 | } |
| 158 | |
| 159 | /** |
| 160 | * @param string $templateName Case-insensitive page name. Prefixes are significant. |
| 161 | * |
| 162 | * @return array[] Array mapping source to target parameters: |
| 163 | * [ |
| 164 | * string $source => [ |
| 165 | * 'target' => string Target parameter name |
| 166 | * 'addLanguageTemplate' => bool Whether or not to add a template like {{de|…}} |
| 167 | * ], |
| 168 | * … |
| 169 | * ] |
| 170 | */ |
| 171 | public function getTemplateParameters( string $templateName ): array { |
| 172 | $templateName = $this->lowercasePageName( $templateName ); |
| 173 | if ( !isset( $this->transferTemplates[$templateName] ) ) { |
| 174 | return []; |
| 175 | } |
| 176 | |
| 177 | $replacements = []; |
| 178 | foreach ( $this->transferTemplates[$templateName]['parameters'] as $targetParameter => $opt ) { |
| 179 | $sourceParameters = (array)( $opt['sourceParameters'] ?? [] ); |
| 180 | $addLanguageTemplate = (bool)( $opt['addLanguageTemplate'] ?? false ); |
| 181 | |
| 182 | foreach ( $sourceParameters as $sourceParameter ) { |
| 183 | if ( $sourceParameter !== '' ) { |
| 184 | $replacements[$sourceParameter] = [ |
| 185 | 'target' => $targetParameter, |
| 186 | 'addLanguageTemplate' => $addLanguageTemplate |
| 187 | ]; |
| 188 | } |
| 189 | } |
| 190 | } |
| 191 | return $replacements; |
| 192 | } |
| 193 | |
| 194 | /** |
| 195 | * @param string $templateName Case-insensitive page name. Prefixes are significant. |
| 196 | * |
| 197 | * @return string[] Array mapping required target parameter names to static string values. |
| 198 | */ |
| 199 | public function getRequiredTemplateParameters( string $templateName ): array { |
| 200 | $templateName = $this->lowercasePageName( $templateName ); |
| 201 | if ( !isset( $this->transferTemplates[$templateName] ) ) { |
| 202 | return []; |
| 203 | } |
| 204 | |
| 205 | $additions = []; |
| 206 | foreach ( $this->transferTemplates[$templateName]['parameters'] as $targetParameter => $opt ) { |
| 207 | $addIfMissing = $opt['addIfMissing'] ?? false; |
| 208 | if ( $addIfMissing ) { |
| 209 | $additions[$targetParameter] = $opt['value'] ?? ''; |
| 210 | } |
| 211 | } |
| 212 | return $additions; |
| 213 | } |
| 214 | |
| 215 | private function lowercasePageName( string $pageName ): string { |
| 216 | return mb_convert_case( $this->normalizePageName( $pageName ), MB_CASE_LOWER ); |
| 217 | } |
| 218 | |
| 219 | private function normalizePageName( string $pageName ): string { |
| 220 | return trim( str_replace( '_', ' ', $pageName ) ); |
| 221 | } |
| 222 | |
| 223 | private function removeNamespace( string $title ): string { |
| 224 | return array_last( explode( ':', $title, 2 ) ); |
| 225 | } |
| 226 | |
| 227 | } |