Code Coverage |
||||||||||
Lines |
Functions and Methods |
Classes and Traits |
||||||||
| Total | |
73.29% |
118 / 161 |
|
17.65% |
3 / 17 |
CRAP | |
0.00% |
0 / 1 |
| Title | |
73.29% |
118 / 161 |
|
17.65% |
3 / 17 |
170.76 | |
0.00% |
0 / 1 |
| __construct | |
100.00% |
5 / 5 |
|
100.00% |
1 / 1 |
1 | |||
| newFromText | |
97.00% |
97 / 100 |
|
0.00% |
0 / 1 |
42 | |||
| getInterwiki | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
2 | |||
| getKey | |
66.67% |
2 / 3 |
|
0.00% |
0 / 1 |
2.15 | |||
| getDBkey | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
2 | |||
| getPrefixedDBKey | |
85.71% |
6 / 7 |
|
0.00% |
0 / 1 |
4.05 | |||
| getPrefixedText | |
85.71% |
6 / 7 |
|
0.00% |
0 / 1 |
4.05 | |||
| getFullText | |
0.00% |
0 / 4 |
|
0.00% |
0 / 1 |
6 | |||
| getFullDBKey | |
0.00% |
0 / 4 |
|
0.00% |
0 / 1 |
6 | |||
| getNamespace | |
100.00% |
1 / 1 |
|
100.00% |
1 / 1 |
1 | |||
| getNamespaceName | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
2 | |||
| getFragment | |
100.00% |
1 / 1 |
|
100.00% |
1 / 1 |
1 | |||
| equals | |
0.00% |
0 / 3 |
|
0.00% |
0 / 1 |
12 | |||
| isSpecialPage | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
2 | |||
| fixSpecialName | |
0.00% |
0 / 6 |
|
0.00% |
0 / 1 |
6 | |||
| createFragmentTarget | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
6 | |||
| newFromLinkTarget | |
0.00% |
0 / 15 |
|
0.00% |
0 / 1 |
6 | |||
| 1 | <?php |
| 2 | declare( strict_types = 1 ); |
| 3 | |
| 4 | namespace Wikimedia\Parsoid\Utils; |
| 5 | |
| 6 | use Wikimedia\Assert\Assert; |
| 7 | use Wikimedia\IPUtils; |
| 8 | use Wikimedia\Parsoid\Config\SiteConfig; |
| 9 | use Wikimedia\Parsoid\Core\LinkTarget; |
| 10 | use Wikimedia\Parsoid\Core\LinkTargetTrait; |
| 11 | |
| 12 | class Title implements LinkTarget { |
| 13 | use LinkTargetTrait; |
| 14 | |
| 15 | /** @var string */ |
| 16 | private $interwiki; |
| 17 | |
| 18 | /** @var int */ |
| 19 | private $namespaceId; |
| 20 | |
| 21 | /** @var string */ |
| 22 | private $namespaceName; |
| 23 | |
| 24 | /** @var string */ |
| 25 | private $dbkey; |
| 26 | |
| 27 | /** @var string */ |
| 28 | private $fragment; |
| 29 | |
| 30 | // cached values of prefixed title/key |
| 31 | private ?string $prefixedDBKey = null; |
| 32 | private ?string $prefixedText = null; |
| 33 | |
| 34 | /** |
| 35 | * @param string $interwiki Interwiki prefix, or empty string if none |
| 36 | * @param string $key Page DBkey (with underscores, not spaces) |
| 37 | * @param int $namespaceId |
| 38 | * @param string $namespaceName (with spaces, not underscores) |
| 39 | * @param ?string $fragment |
| 40 | */ |
| 41 | private function __construct( |
| 42 | string $interwiki, string $key, int $namespaceId, string $namespaceName, ?string $fragment = null |
| 43 | ) { |
| 44 | $this->interwiki = $interwiki; |
| 45 | $this->dbkey = $key; |
| 46 | $this->namespaceId = $namespaceId; |
| 47 | $this->namespaceName = $namespaceName; |
| 48 | $this->fragment = $fragment ?? ''; |
| 49 | } |
| 50 | |
| 51 | public static function newFromText( |
| 52 | string $title, SiteConfig $siteConfig, ?int $defaultNs = null |
| 53 | ): Title { |
| 54 | if ( $defaultNs === null ) { |
| 55 | $defaultNs = 0; |
| 56 | } |
| 57 | $origTitle = $title; |
| 58 | |
| 59 | if ( !mb_check_encoding( $title, 'UTF-8' ) ) { |
| 60 | throw new TitleException( "Bad UTF-8 in title \"$origTitle\"", 'title-invalid-utf8', $origTitle ); |
| 61 | } |
| 62 | |
| 63 | // Strip Unicode bidi override characters. |
| 64 | $title = preg_replace( '/[\x{200E}\x{200F}\x{202A}-\x{202E}]+/u', '', $title ); |
| 65 | if ( $title === null ) { |
| 66 | throw new TitleException( "Bad UTF-8 in title \"$origTitle\"", 'title-invalid-utf8', $origTitle ); |
| 67 | } |
| 68 | |
| 69 | // Clean up whitespace |
| 70 | $title = preg_replace( |
| 71 | '/[ _\x{00A0}\x{1680}\x{180E}\x{2000}-\x{200A}\x{2028}\x{2029}\x{202F}\x{205F}\x{3000}]+/u', |
| 72 | '_', $title |
| 73 | ); |
| 74 | // Trim _ from beginning and end |
| 75 | $title = trim( $title, '_' ); |
| 76 | |
| 77 | if ( str_contains( $title, \UtfNormal\Constants::UTF8_REPLACEMENT ) ) { |
| 78 | throw new TitleException( "Bad UTF-8 in title \"$title\"", 'title-invalid-utf8', $title ); |
| 79 | } |
| 80 | |
| 81 | // Initial colon indicates main namespace rather than specified default |
| 82 | // but should not create invalid {ns,title} pairs such as {0,Project:Foo} |
| 83 | if ( $title !== '' && $title[0] === ':' ) { |
| 84 | $title = ltrim( substr( $title, 1 ), '_' ); |
| 85 | $defaultNs = 0; |
| 86 | } |
| 87 | |
| 88 | if ( $title === '' ) { |
| 89 | throw new TitleException( 'Empty title', 'title-invalid-empty', $title ); |
| 90 | } |
| 91 | |
| 92 | $ns = $defaultNs; |
| 93 | $interwiki = null; |
| 94 | |
| 95 | # Namespace or interwiki prefix |
| 96 | $prefixRegexp = "/^(.+?)_*:_*(.*)$/S"; |
| 97 | // MediaWikiTitleCodec::splitTitleString wraps a loop around the |
| 98 | // next section, to allow it to repeat this prefix processing if |
| 99 | // an interwiki prefix is found which points at the local wiki. |
| 100 | $m = []; |
| 101 | if ( preg_match( $prefixRegexp, $title, $m ) ) { |
| 102 | $p = $m[1]; |
| 103 | $pLower = mb_strtolower( $p ); |
| 104 | $nsId = $siteConfig->canonicalNamespaceId( $pLower ) ?? |
| 105 | $siteConfig->namespaceId( $pLower ); |
| 106 | if ( $nsId !== null ) { |
| 107 | $title = $m[2]; |
| 108 | $ns = $nsId; |
| 109 | # For Talk:X pages, check if X has a "namespace" prefix |
| 110 | if ( |
| 111 | $nsId === $siteConfig->canonicalNamespaceId( 'talk' ) && |
| 112 | preg_match( $prefixRegexp, $title, $x ) |
| 113 | ) { |
| 114 | $xLower = mb_strtolower( $x[1] ); |
| 115 | if ( $siteConfig->namespaceId( $xLower ) ) { |
| 116 | // Disallow Talk:File:x type titles. |
| 117 | throw new TitleException( |
| 118 | "Invalid Talk namespace title \"$origTitle\"", 'title-invalid-talk-namespace', $title |
| 119 | ); |
| 120 | } elseif ( $siteConfig->interwikiMapNoNamespaces()[$xLower] ?? null ) { |
| 121 | // Disallow Talk:Interwiki:x type titles. |
| 122 | throw new TitleException( |
| 123 | "Invalid Talk namespace title \"$origTitle\"", 'title-invalid-talk-namespace', $title |
| 124 | ); |
| 125 | } |
| 126 | } |
| 127 | } elseif ( $siteConfig->interwikiMapNoNamespaces()[$pLower] ?? null ) { |
| 128 | # Interwiki link |
| 129 | $title = $m[2]; |
| 130 | $interwiki = $pLower; |
| 131 | |
| 132 | # We don't check for a redundant interwiki prefix to the |
| 133 | # local wiki, like core does here in |
| 134 | # MediaWikiTitleCodec::splitTitleString; |
| 135 | # core then does a `continue` to repeat the processing |
| 136 | |
| 137 | // If there's an initial colon after the interwiki, that also |
| 138 | // resets the default namespace |
| 139 | if ( $title !== '' && $title[0] === ':' ) { |
| 140 | $title = trim( substr( $title, 1 ), '_' ); |
| 141 | $ns = 0; |
| 142 | } |
| 143 | } |
| 144 | # If there's no recognized interwiki or namespace, |
| 145 | # then let the colon expression be part of the title |
| 146 | } |
| 147 | |
| 148 | $fragment = null; |
| 149 | $fragmentIndex = strpos( $title, '#' ); |
| 150 | if ( $fragmentIndex !== false ) { |
| 151 | $fragment = substr( $title, $fragmentIndex + 1 ); |
| 152 | $title = rtrim( substr( $title, 0, $fragmentIndex ), '_' ); |
| 153 | } |
| 154 | |
| 155 | $illegalCharsRe = '/[^' . $siteConfig->legalTitleChars() . ']' |
| 156 | // URL percent encoding sequences interfere with the ability |
| 157 | // to round-trip titles -- you can't link to them consistently. |
| 158 | . '|%[0-9A-Fa-f]{2}' |
| 159 | // XML/HTML character references produce similar issues. |
| 160 | . '|&[A-Za-z0-9\x80-\xff]+;/S'; |
| 161 | if ( preg_match( $illegalCharsRe, $title ) ) { |
| 162 | throw new TitleException( |
| 163 | "Invalid characters in title \"$origTitle\"", 'title-invalid-characters', $title |
| 164 | ); |
| 165 | } |
| 166 | |
| 167 | // Pages with "/./" or "/../" appearing in the URLs will often be |
| 168 | // unreachable due to the way web browsers deal with 'relative' URLs. |
| 169 | // Also, they conflict with subpage syntax. Forbid them explicitly. |
| 170 | if ( str_contains( $title, '.' ) && ( |
| 171 | $title === '.' || $title === '..' || |
| 172 | str_starts_with( $title, './' ) || |
| 173 | str_starts_with( $title, '../' ) || |
| 174 | str_contains( $title, '/./' ) || |
| 175 | str_contains( $title, '/../' ) || |
| 176 | str_ends_with( $title, '/.' ) || |
| 177 | str_ends_with( $title, '/..' ) |
| 178 | ) ) { |
| 179 | throw new TitleException( |
| 180 | "Title \"$origTitle\" contains relative path components", 'title-invalid-relative', $title |
| 181 | ); |
| 182 | } |
| 183 | |
| 184 | // Magic tilde sequences? Nu-uh! |
| 185 | if ( str_contains( $title, '~~~' ) ) { |
| 186 | throw new TitleException( |
| 187 | "Title \"$origTitle\" contains ~~~", 'title-invalid-magic-tilde', $title |
| 188 | ); |
| 189 | } |
| 190 | |
| 191 | $maxLength = $ns === $siteConfig->canonicalNamespaceId( 'special' ) ? 512 : 255; |
| 192 | if ( strlen( $title ) > $maxLength ) { |
| 193 | throw new TitleException( |
| 194 | "Title \"$origTitle\" is too long", 'title-invalid-too-long', $title |
| 195 | ); |
| 196 | } |
| 197 | |
| 198 | if ( $interwiki === null && $siteConfig->namespaceCase( $ns ) === 'first-letter' ) { |
| 199 | $title = $siteConfig->ucfirst( $title ); |
| 200 | } |
| 201 | |
| 202 | # Can't make a link to a namespace alone... "empty" local links can only be |
| 203 | # self-links with a fragment identifier. |