Code Coverage |
||||||||||
Lines |
Functions and Methods |
Classes and Traits |
||||||||
| Total | |
0.00% |
0 / 143 |
|
0.00% |
0 / 9 |
CRAP | |
0.00% |
0 / 1 |
| EditCheckFileParser | |
0.00% |
0 / 143 |
|
0.00% |
0 / 9 |
3192 | |
0.00% |
0 / 1 |
| __construct | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
2 | |||
| extractStaticValue | |
0.00% |
0 / 6 |
|
0.00% |
0 / 1 |
12 | |||
| extractDefaultConfig | |
0.00% |
0 / 8 |
|
0.00% |
0 / 1 |
12 | |||
| tryJsonDecodeObjectString | |
0.00% |
0 / 19 |
|
0.00% |
0 / 1 |
6 | |||
| parseMessage | |
0.00% |
0 / 14 |
|
0.00% |
0 / 1 |
72 | |||
| extractStaticAssignmentExpression | |
0.00% |
0 / 9 |
|
0.00% |
0 / 1 |
12 | |||
| findTopLevelDelimiter | |
0.00% |
0 / 45 |
|
0.00% |
0 / 1 |
462 | |||
| getMessageExpressionPattern | |
0.00% |
0 / 12 |
|
0.00% |
0 / 1 |
12 | |||
| convertSingleQuotedToDoubleQuoted | |
0.00% |
0 / 29 |
|
0.00% |
0 / 1 |
156 | |||
| 1 | <?php |
| 2 | /** |
| 3 | * Parser for reading metadata from edit check JavaScript source files. |
| 4 | */ |
| 5 | |
| 6 | namespace MediaWiki\Extension\VisualEditor; |
| 7 | |
| 8 | use MediaWiki\Language\MessageLocalizer; |
| 9 | |
| 10 | /** |
| 11 | * Extracts structured metadata from edit check JavaScript source files. |
| 12 | * |
| 13 | * Handles these kinds of value expression that appear in the JS: |
| 14 | * - string literals ('foo' / "foo") |
| 15 | * - structured arrays / objects (decoded via tryJsonDecodeObjectString) |
| 16 | * - message-function calls (ve.msg, mw.msg, OO.ui.deferMsg, …) inside arrays / objects |
| 17 | */ |
| 18 | class EditCheckFileParser { |
| 19 | |
| 20 | public function __construct( |
| 21 | private readonly MessageLocalizer $localizer |
| 22 | ) { |
| 23 | } |
| 24 | |
| 25 | /** |
| 26 | * Extract the value of a static property from a JS source file. |
| 27 | * |
| 28 | * Supports string literals, and arrays or objects which tryJsonDecodeObjectString can decode. |
| 29 | * |
| 30 | * @param string $src Source code |
| 31 | * @param string $prop Property name (e.g. 'name', 'allowedContentLanguages') |
| 32 | * @return mixed|null Value, or null if not found or not decodable |
| 33 | */ |
| 34 | public function extractStaticValue( string $src, string $prop ) { |
| 35 | $expr = $this->extractStaticAssignmentExpression( $src, $prop ); |
| 36 | if ( $expr === null ) { |
| 37 | return null; |
| 38 | } |
| 39 | |
| 40 | if ( preg_match( '/^([\"\\\'])(.*?)\1$/', $expr, $mm ) ) { |
| 41 | return $mm[2]; |
| 42 | } |
| 43 | |
| 44 | return $this->tryJsonDecodeObjectString( $expr ); |
| 45 | } |
| 46 | |
| 47 | /** |
| 48 | * Extract the defaultConfig object literal from the source code. |
| 49 | * |
| 50 | * @param string $src Source code |
| 51 | * @return string Raw JS object literal, or empty string if not found |
| 52 | */ |
| 53 | public function extractDefaultConfig( string $src ): string { |
| 54 | // Capture object literal used as overrides in ve.extendObject(..., {...}) or a direct object. |
| 55 | if ( |
| 56 | preg_match( |
| 57 | '/static\s*\.\s*defaultConfig\s*=' . |
| 58 | '\s*ve\.extendObject\s*\(\s*\{.*?\}\s*,\s*[^,]+,\s*(\{[\s\S]*?\})\s*\)\s*;/', |
| 59 | $src, $m ) |
| 60 | ) { |
| 61 | return $m[ 1 ]; |
| 62 | } |
| 63 | if ( preg_match( '/static\s*\.\s*defaultConfig\s*=\s*(\{[\s\S]*?\})\s*;/', $src, $m ) ) { |
| 64 | return $m[ 1 ]; |
| 65 | } |
| 66 | return ''; |
| 67 | } |
| 68 | |
| 69 | /** |
| 70 | * Attempt to decode a JS object/array literal as PHP data. |
| 71 | * |
| 72 | * Handles JS-specific syntax: single-quoted strings, unquoted keys, trailing commas, |
| 73 | * `undefined`, inline comments, and message-function calls. |
| 74 | * |
| 75 | * @param string $js JS object literal |
| 76 | * @return mixed|null PHP value, or null on parse failure |
| 77 | */ |
| 78 | public function tryJsonDecodeObjectString( string $js ) { |
| 79 | $src = trim( $js ); |
| 80 | // Remove comments |
| 81 | $src = preg_replace( '/\/\/.*?(?=\n|$)/', '', $src ); |
| 82 | $src = preg_replace( '/\/\*[\s\S]*?\*\//', '', $src ); |
| 83 | // Remove trailing commas before } or ] |
| 84 | $src = preg_replace( '/,\s*(\}|\])/', '$1', $src ); |
| 85 | // Replace undefined with null |
| 86 | $src = preg_replace( '/\bundefined\b/', 'null', $src ); |
| 87 | // Quote unquoted keys: { key: ... } or , key: ... |
| 88 | $src = preg_replace( '/([\{,]\s*)([A-Za-z_$][A-Za-z0-9_$]*)\s*:/', '$1"$2":', $src ); |
| 89 | // Message expressions, e.g. ve.msg('...'), OO.ui.deferMsg('...'), mw.msg('...') |
| 90 | $messageExprPattern = $this->getMessageExpressionPattern( false ); |
| 91 | $src = preg_replace_callback( |
| 92 | $messageExprPattern, |
| 93 | function ( $mm ) { |
| 94 | return json_encode( (string)$this->parseMessage( $mm[0] ) ); |
| 95 | }, |
| 96 | $src |
| 97 | ); |
| 98 | |
| 99 | $src = $this->convertSingleQuotedToDoubleQuoted( $src ); |
| 100 | |
| 101 | // Try decode |
| 102 | $data = json_decode( $src, true ); |
| 103 | if ( json_last_error() !== JSON_ERROR_NONE ) { |
| 104 | return null; |
| 105 | } |
| 106 | return $data; |
| 107 | } |
| 108 | |
| 109 | /** |
| 110 | * Parse a JS message expression and return the rendered message. |
| 111 | * |
| 112 | * Supported call forms: |
| 113 | * - ve.msg(...) |
| 114 | * - ve.htmlMsg(...) |
| 115 | * - ve.deferHtmlMsg(...) |
| 116 | * - ve.deferJQueryMsg(...) |
| 117 | * - mw.msg(...) |
| 118 | * - OO.ui.deferMsg(...) |
| 119 | * |
| 120 | * @param string $expr JS expression |
| 121 | * @return string|\OOUI\HtmlSnippet Empty string if the expression doesn't match. |
| 122 | */ |
| 123 | public function parseMessage( string $expr ) { |
| 124 | if ( preg_match( $this->getMessageExpressionPattern( true ), $expr, $mm ) ) { |
| 125 | $argsStr = $mm[4]; |
| 126 | $args = []; |
| 127 | if ( preg_match_all( '/,\s*([\"\\\'])(.*?)\1/', $argsStr, $am, PREG_SET_ORDER ) ) { |
| 128 | foreach ( $am as $a ) { |
| 129 | $args[] = $a[2]; |
| 130 | } |
| 131 | } |
| 132 | $msg = $this->localizer->msg( $mm[3], ...$args ); |
| 133 | switch ( $mm[1] ) { |
| 134 | case 've.htmlMsg': |
| 135 | case 've.deferHtmlMsg': |
| 136 | case 've.deferJQueryMsg': |
| 137 | return new \OOUI\HtmlSnippet( $msg->parse() ); |
| 138 | default: |
| 139 | return $msg->text(); |
| 140 | } |
| 141 | } |
| 142 | |
| 143 | return ''; |
| 144 | } |
| 145 | |
| 146 | /** |
| 147 | * Extract the full RHS expression of a `static.<prop> = ...;` assignment. |
| 148 | * |
| 149 | * @param string $src Source code |
| 150 | * @param string $prop Property name |
| 151 | * @return string|null RHS expression, or null if not found |
| 152 | */ |
| 153 | public function extractStaticAssignmentExpression( string $src, string $prop ): ?string { |
| 154 | $pattern = '/static\s*\.\s*' . preg_quote( $prop, '/' ) . '\s*=\s*/'; |
| 155 | if ( !preg_match( $pattern, $src, $m, PREG_OFFSET_CAPTURE ) ) { |
| 156 | return null; |
| 157 | } |
| 158 | $start = $m[ 0 ][ 1 ] + strlen( $m[ 0 ][ 0 ] ); |
| 159 | $end = $this->findTopLevelDelimiter( $src, $start, [ ';' ] ); |
| 160 | return trim( $end === null ? |
| 161 | substr( $src, $start ) : |
| 162 | substr( $src, $start, $end - $start ) |
| 163 | ); |
| 164 | } |
| 165 | |
| 166 | /** |
| 167 | * Find the offset of the first character from $delimiters that appears at the |
| 168 | * top level of a JS expression (not inside a string or nested brackets/parens/braces). |
| 169 | * |
| 170 | * @param string $src Source string |
| 171 | * @param int $start Start offset |
| 172 | * @param string[] $delimiters Single-character stop characters |
| 173 | * @return int|null Offset of the delimiter, or null if not found before end of string |
| 174 | */ |
| 175 | private function findTopLevelDelimiter( string $src, int $start, array $delimiters ): ?int { |
| 176 | $len = strlen( $src ); |
| 177 | $depthParen = 0; |
| 178 | $depthBracket = 0; |
| 179 | $depthBrace = 0; |
| 180 | $inSingle = false; |
| 181 | $inDouble = false; |
| 182 | $escape = false; |
| 183 | |
| 184 | for ( $i = $start; $i < $len; $i++ ) { |
| 185 | $ch = $src[$i]; |
| 186 | if ( $escape ) { |
| 187 | $escape = false; |
| 188 | continue; |
| 189 | } |
| 190 | if ( $ch === '\\' ) { |
| 191 | $escape = true; |
| 192 | continue; |
| 193 | } |
| 194 | if ( !$inDouble && $ch === '\'' ) { |
| 195 | $inSingle = !$inSingle; |
| 196 | continue; |
| 197 | } |
| 198 | if ( !$inSingle && $ch === '"' ) { |
| 199 | $inDouble = !$inDouble; |
| 200 | continue; |
| 201 | } |
| 202 | if ( $inSingle || $inDouble ) { |
| 203 | continue; |
| 204 | } |
| 205 | switch ( $ch ) { |
| 206 | case '(': |
| 207 | $depthParen++; |
| 208 | break; |
| 209 | case ')': |
| 210 | $depthParen = max( 0, $depthParen - 1 ); |
| 211 | break; |
| 212 | case '[': |
| 213 | $depthBracket++; |
| 214 | break; |
| 215 | case ']': |
| 216 | $depthBracket = max( 0, $depthBracket - 1 ); |
| 217 | break; |
| 218 | case '{': |
| 219 | $depthBrace++; |
| 220 | break; |
| 221 | case '}': |
| 222 | $depthBrace = max( 0, $depthBrace - 1 ); |
| 223 | break; |
| 224 | default: |
| 225 | if ( $depthParen === 0 && $depthBracket === 0 && $depthBrace === 0 |
| 226 | && in_array( $ch, $delimiters, true ) |
| 227 | ) { |
| 228 | return $i; |
| 229 | } |
| 230 | } |
| 231 | } |
| 232 | return null; |
| 233 | } |
| 234 | |
| 235 | /** |
| 236 | * Build a regex for matching supported message-call expressions. |
| 237 | * |
| 238 | * @param bool $anchored Whether to anchor the pattern to the whole string |
| 239 | * @return string |
| 240 | */ |
| 241 | private function getMessageExpressionPattern( bool $anchored ): string { |
| 242 | $start = $anchored ? '^' : '\b'; |
| 243 | $end = $anchored ? '$' : ''; |
| 244 | $pattern = |
| 245 | '/' . |
| 246 | $start . |
| 247 | '(ve\.msg|ve\.htmlMsg|ve\.deferHtmlMsg|ve\.deferJQueryMsg|mw\.msg|OO\.ui\.deferMsg)' . |
| 248 | "\\s*\\(\\s*(\"|')" . |
| 249 | "([^\"']+)" . |
| 250 | "\\2(.*?)\\)" . |
| 251 | $end . |
| 252 | '/s'; |
| 253 | return $pattern; |
| 254 | } |
| 255 | |
| 256 | /** |
| 257 | * Convert all JS single-quoted string literals to double-quoted, handling escapes. |
| 258 | * |
| 259 | * @param string $code JS source |
| 260 | * @return string |
| 261 | */ |
| 262 | private function convertSingleQuotedToDoubleQuoted( string $code ): string { |
| 263 | $out = ''; |
| 264 | $len = strlen( $code ); |
| 265 | $inSingle = false; |
| 266 | $inDouble = false; |
| 267 | $escape = false; |
| 268 | for ( $i = 0; $i < $len; $i++ ) { |
| 269 | $ch = $code[$i]; |
| 270 | if ( $escape ) { |
| 271 | // Preserve escaped char, but if we are in a single-quoted string |
| 272 | // and the escaped char is a single quote, drop the escape |
| 273 | if ( $inSingle && $ch === '\'' ) { |
| 274 | $out .= '\''; |
| 275 | } else { |
| 276 | $out .= '\\' . $ch; |
| 277 | } |
| 278 | $escape = false; |
| 279 | continue; |
| 280 | } |
| 281 | if ( $ch === '\\' ) { |
| 282 | $escape = true; |
| 283 | continue; |
| 284 | } |
| 285 | if ( !$inDouble && $ch === '\'' ) { |
| 286 | // Toggle single-quoted string; replace quote with double quote |
| 287 | $inSingle = !$inSingle; |
| 288 | $out .= '"'; |
| 289 | continue; |
| 290 | } |
| 291 | if ( !$inSingle && $ch === '"' ) { |
| 292 | // Track double quotes to avoid interfering while inside |
| 293 | $inDouble = !$inDouble; |
| 294 | $out .= '"'; |
| 295 | continue; |
| 296 | } |
| 297 | // Inside single-quoted string: ensure double quotes are escaped |
| 298 | if ( $inSingle && $ch === '"' ) { |
| 299 | $out .= '\\"'; |
| 300 | continue; |
| 301 | } |
| 302 | $out .= $ch; |
| 303 | } |
| 304 | return $out; |
| 305 | } |
| 306 | } |