Code Coverage |
||||||||||
Lines |
Functions and Methods |
Classes and Traits |
||||||||
| Total | |
0.00% |
0 / 282 |
|
0.00% |
0 / 6 |
CRAP | |
0.00% |
0 / 1 |
| MhchemPatterns | |
0.00% |
0 / 282 |
|
0.00% |
0 / 6 |
1806 | |
0.00% |
0 / 1 |
| getPatterns | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
2 | |||
| findObserveGroups | |
0.00% |
0 / 28 |
|
0.00% |
0 / 1 |
132 | |||
| matchObsGrInner | |
0.00% |
0 / 6 |
|
0.00% |
0 / 1 |
30 | |||
| findObserveGroupsInner | |
0.00% |
0 / 15 |
|
0.00% |
0 / 1 |
56 | |||
| __construct | |
0.00% |
0 / 212 |
|
0.00% |
0 / 1 |
132 | |||
| match | |
0.00% |
0 / 20 |
|
0.00% |
0 / 1 |
56 | |||
| 1 | <?php |
| 2 | /** |
| 3 | * Copyright (c) 2023 Johannes Stegmüller |
| 4 | * |
| 5 | * This file is a port of mhchemParser originally authored by Martin Hensel in javascript/typescript. |
| 6 | * The original license for this software can be found in the accompanying LICENSE.mhchemParser-ts.txt file. |
| 7 | */ |
| 8 | |
| 9 | declare( strict_types = 1 ); |
| 10 | |
| 11 | namespace MediaWiki\Extension\Math\WikiTexVC\Mhchem; |
| 12 | |
| 13 | use MediaWiki\Extension\Math\WikiTexVC\Mhchem\MhchemRegExp as Reg; |
| 14 | use RuntimeException; |
| 15 | |
| 16 | /** |
| 17 | * Contains all matching regex patterns and match functions for mhchemParser in PHP. |
| 18 | * |
| 19 | * corresponds mostly to the 'patterns' array in line ~207 in mhchemParser.js by Martin Hensel |
| 20 | * |
| 21 | * @author Johannes Stegmüller |
| 22 | * @license GPL-2.0-or-later |
| 23 | */ |
| 24 | class MhchemPatterns { |
| 25 | |
| 26 | /** @var array */ |
| 27 | private array $patterns; |
| 28 | |
| 29 | /** |
| 30 | * Matching patterns |
| 31 | * either regexes or function that return null or {match_:"a", remainder:"bc"} |
| 32 | * @return array |
| 33 | */ |
| 34 | public function getPatterns(): array { |
| 35 | return $this->patterns; |
| 36 | } |
| 37 | |
| 38 | /** |
| 39 | * @param string $input |
| 40 | * @param string|Reg $begExcl |
| 41 | * @param string|Reg $begIncl |
| 42 | * @param mixed|null $endIncl |
| 43 | * @param mixed|null $endExcl |
| 44 | * @param string|Reg|null $beg2Excl |
| 45 | * @param string|Reg|null $beg2Incl |
| 46 | * @param mixed|null $end2Incl |
| 47 | * @param mixed|null $end2Excl |
| 48 | * @param bool|null $combine |
| 49 | */ |
| 50 | public function findObserveGroups( $input, $begExcl, $begIncl, $endIncl, |
| 51 | $endExcl = null, $beg2Excl = null, $beg2Incl = null, |
| 52 | $end2Incl = null, $end2Excl = null, $combine = null ): ?array { |
| 53 | $match = $this->matchObsGrInner( $input, $begExcl ); |
| 54 | if ( $match === null ) { |
| 55 | return null; |
| 56 | } |
| 57 | $input = substr( $input, strlen( $match ) ); |
| 58 | $match = $this->matchObsGrInner( $input, $begIncl ); |
| 59 | if ( $match === null ) { |
| 60 | return null; |
| 61 | } |
| 62 | |
| 63 | if ( $endIncl === "0" ) { |
| 64 | throw new RuntimeException( "error in condition, check next loc " ); |
| 65 | } |
| 66 | $e = $this->findObserveGroupsInner( $input, strlen( $match ), |
| 67 | MhchemUtil::issetJS( $endIncl ) ? $endIncl : $endExcl ); |
| 68 | if ( $e === null ) { |
| 69 | return null; |
| 70 | } |
| 71 | $match1 = substr( $input, 0, ( $endIncl ? $e["endMatchEnd"] : $e["endMatchBegin"] ) ); |
| 72 | |
| 73 | if ( !( MhchemUtil::issetJS( $beg2Excl ) || MhchemUtil::issetJS( $beg2Incl ) ) ) { |
| 74 | return [ |
| 75 | "match_" => $match1, |
| 76 | "remainder" => substr( $input, $e["endMatchEnd"] ) |
| 77 | ]; |
| 78 | } else { |
| 79 | $group2 = $this->findObserveGroups( substr( $input, $e["endMatchEnd"] ), |
| 80 | $beg2Excl, $beg2Incl, $end2Incl, $end2Excl ); |
| 81 | if ( $group2 === null ) { |
| 82 | return null; |
| 83 | } |
| 84 | $matchRet = [ $match1, $group2["match_"] ]; |
| 85 | return [ |
| 86 | "match_" => ( $combine ? implode( "", $matchRet ) : $matchRet ), |
| 87 | "remainder" => $group2["remainder"] |
| 88 | ]; |
| 89 | } |
| 90 | } |
| 91 | |
| 92 | /** |
| 93 | * @param string $input |
| 94 | * @param string|Reg $pattern |
| 95 | * @return string|null |
| 96 | */ |
| 97 | private function matchObsGrInner( string $input, $pattern ) { |
| 98 | /** |
| 99 | * In javascript this is checking if the incoming pattern is a string, |
| 100 | * if not the assumption is that it is of regex type. Since PHP has |
| 101 | * strings here. |
| 102 | */ |
| 103 | if ( !$pattern instanceof Reg ) { |
| 104 | // Added this if to catch empty needle for strpos input in PHP |
| 105 | if ( !MhchemUtil::issetJS( $pattern ) || str_starts_with( $input, $pattern ) ) { |
| 106 | return $pattern; |
| 107 | } |
| 108 | } elseif ( preg_match( $pattern->getRegExp(), $input, $matches ) ) { |
| 109 | return $matches[0]; |
| 110 | } |
| 111 | return null; |
| 112 | } |
| 113 | |
| 114 | /** |
| 115 | * @param string $input |
| 116 | * @param int $i |
| 117 | * @param string $endChars |
| 118 | * @return array|null |
| 119 | */ |
| 120 | private function findObserveGroupsInner( string $input, $i, $endChars ): ?array { |
| 121 | $braces = 0; |
| 122 | while ( $i < strlen( $input ) ) { |
| 123 | $a = $input[$i]; |
| 124 | $match = $this->matchObsGrInner( substr( $input, $i ), $endChars ); |
| 125 | if ( $match !== null && $braces === 0 ) { |
| 126 | return [ "endMatchBegin" => $i, "endMatchEnd" => $i + strlen( $match ) ]; |
| 127 | } elseif ( $a === "{" ) { |
| 128 | $braces++; |
| 129 | } elseif ( $a === "}" ) { |
| 130 | if ( $braces === 0 ) { |
| 131 | // Unexpected character |
| 132 | throw new RuntimeException( |
| 133 | "ExtraCloseMissingOpen: Extra close brace or missing open brace" ); |
| 134 | } else { |
| 135 | $braces--; |
| 136 | } |
| 137 | } |
| 138 | $i++; |
| 139 | } |
| 140 | return null; |
| 141 | } |
| 142 | |
| 143 | public function __construct() { |
| 144 | $this->patterns = [ |
| 145 | 'empty' => new Reg( "/^$/" ), |
| 146 | 'else' => new Reg( "/^./" ), |
| 147 | 'else2' => new Reg( "/^./" ), |
| 148 | 'space' => new Reg( "/^\s/" ), |
| 149 | 'space A' => new Reg( "/^\s(?=[A-Z\\\\$])/" ), |
| 150 | 'space$' => new Reg( "/^\s$/" ), |
| 151 | 'a-z' => new Reg( "/^[a-z]/" ), |
| 152 | 'x' => new Reg( "/^x/" ), |
| 153 | 'x$' => new Reg( "/^x$/" ), |
| 154 | 'i$' => new Reg( "/^i$/" ), |
| 155 | 'letters' => new Reg( |
| 156 | "/^(?:[a-zA-Z\x{03B1}-\x{03C9}\x{0391}-\x{03A9}?@]|(?:\\\\(?:alpha|beta|gamma|delta|epsilon" |
| 157 | . "|zeta|eta|theta|iota|kappa|lambda|mu|nu|xi|omicron|pi|rho|sigma|tau|upsilon|phi|chi|psi|omega|Gamma" |
| 158 | . "|Delta|Theta|Lambda|Xi|Pi|Sigma|Upsilon|Phi|Psi|Omega)(?:\s+|\{\}|(?![a-zA-Z]))))+/u" ), |
| 159 | '\\greek' => new Reg( |
| 160 | "/^\\\\(?:alpha|beta|gamma|delta|epsilon|zeta|eta|theta|iota|kappa|lambda|mu|nu|xi|omicron|pi" |
| 161 | . "|rho|sigma|tau|upsilon|phi|chi|psi|omega|Gamma|Delta|Theta|Lambda|Xi|Pi|Sigma|Upsilon|Phi|Psi|Omega)" |
| 162 | . "(?:\s+|\{\}|(?![a-zA-Z]))/" ), |
| 163 | 'one lowercase latin letter $' => new Reg( "/^(?:([a-z])(?:$|[^a-zA-Z]))$/" ), |
| 164 | '$one lowercase latin letter$ $' => new Reg( "/^\\\$(?:([a-z])(?:$|[^a-zA-Z]))\\\$$/" ), |
| 165 | 'one lowercase greek letter $' => new Reg( |
| 166 | "/^(?:\\\$?[\x{003B1}-\x{0003C9}]\\\$?|\\\$?\\\\(?:alpha|beta|gamma|" . |
| 167 | "delta|epsilon|zeta|eta|theta|iota|kappa|lambda|mu|nu|xi|omicron|pi|rho|sigma|tau|upsilon|" . |
| 168 | "phi|chi|psi|omega)\s*\\\$?)(?:\s+|\{\}|(?![a-zA-Z]))$/u" ), |
| 169 | 'digits' => new Reg( "/^[0-9]+/" ), |
| 170 | '-9.,9' => new Reg( "/^[+\-]?(?:[0-9]+(?:[,.][0-9]+)?|[0-9]*(?:\.[0-9]+))/" ), |
| 171 | '-9.,9 no missing 0' => new Reg( "/^[+\-]?[0-9]+(?:[.,][0-9]+)?/" ), |
| 172 | '(-)(9.,9)(e)(99)' => static function ( $input ) { |
| 173 | $matches = []; |
| 174 | $match = preg_match( "/^(\+\-|\+\/\-|\+|\-|\\\\pm\s?)?([0-9]+(?:[,.][0-9]+)?|" . |
| 175 | "[0-9]*(?:\.[0-9]+))?(\((?:[0-9]+(?:[,.][0-9]+)?|[0-9]*(?:\.[0-9]+))\))?(?:(?:([eE])" . |
| 176 | "|\s*(\*|x|\\\\times|\x{00D7})\s*10\^)([+\-]?[0-9]+|\{[+\-]?[0-9]+\}))?/u", $input, $matches ); |
| 177 | if ( $match && $matches[0] ) { |
| 178 | // could also match "" |
| 179 | return [ "match_" => array_slice( $matches, 1 ), |
| 180 | "remainder" => substr( $input, strlen( $matches[0] ) ) ]; |
| 181 | } |
| 182 | return null; |
| 183 | }, |
| 184 | '(-)(9)^(-9)' => new Reg( "/^(\+\-|\+\/\-|\+|\-|\\\\pm\s?)?([0-9]+(?:[,.][0-9]+)?|" |
| 185 | . "[0-9]*(?:\.[0-9]+)?)\^([+\-]?[0-9]+|\{[+\-]?[0-9]+\})/" ), |
| 186 | 'state of aggregation $' => function ( $input ) { |
| 187 | // ... or crystal system |
| 188 | $a = $this->findObserveGroups( $input, "", |
| 189 | new Reg( "/^\([a-z]{1,3}(?=[\),])/" ), ")", "" ); |
| 190 | if ( $a && preg_match( "/^($|[\s,;\)\]\}])/", $a["remainder"] ) ) { |
| 191 | return $a; |
| 192 | } |
| 193 | $matches = []; |
| 194 | $match = preg_match( "/^(?:\((?:\\\\ca\s?)?\\\$[amothc]\\\$\))/", $input, $matches ); |
| 195 | if ( $match ) { |
| 196 | return [ "match_" => $matches[0], "remainder" => substr( $input, strlen( $matches[0] ) ) ]; |
| 197 | } |
| 198 | return null; |
| 199 | }, |
| 200 | '_{(state of aggregation)}$' => new Reg( "/^_\{(\([a-z]{1,3}\))\}/" ), |
| 201 | '{[(' => new Reg( "/^(?:\\\{|\[|\()/" ), |
| 202 | ')]}' => new Reg( "/^(?:\)|\]|\\\})/" ), |
| 203 | ', ' => new Reg( "/^[,;]\s*/" ), |
| 204 | ',' => new Reg( "/^[,;]/" ), |
| 205 | '.' => new Reg( "/^[.]/" ), |
| 206 | '. __* ' => new Reg( "/^([.\x{22C5}\x{00B7}\x{2022}]|[*])\s*/u" ), |
| 207 | '...' => new Reg( "/^\.\.\.(?=$|[^.])/" ), |
| 208 | '^{(...)}' => function ( $input ) { |
| 209 | return $this->findObserveGroups( $input, "^{", "", "", "}" ); |
| 210 | }, |
| 211 | '^($...$)' => function ( $input ) { |
| 212 | return $this->findObserveGroups( $input, "^", "$", "$", "" ); |
| 213 | }, |
| 214 | '^a' => new Reg( "/^\^([0-9]+|[^\\\_])/u" ), |
| 215 | '^\\x{}{}' => function ( $input ) { |
| 216 | return $this->findObserveGroups( $input, "^", |
| 217 | new Reg( "/^\\\\[a-zA-Z]+\{/" ), "}", "", "", |
| 218 | "{", "}", "", true ); |
| 219 | }, |
| 220 | '^\\x{}' => function ( $input ) { |
| 221 | return $this->findObserveGroups( $input, "^", |
| 222 | new Reg( "/^\\\\[a-zA-Z]+\{/" ), "}", "" ); |
| 223 | }, |
| 224 | '^\\x' => new Reg( "/^\^(\\\\[a-zA-Z]+)\s*/" ), |
| 225 | '^(-1)' => new Reg( "/^\^(-?\d+)/" ), |
| 226 | '\'' => new Reg( "/^'/" ), |
| 227 | '_{(...)}' => function ( $input ) { |
| 228 | return $this->findObserveGroups( $input, "_{", "", "", "}" ); |
| 229 | }, |
| 230 | '_($...$)' => function ( $input ) { |
| 231 | return $this->findObserveGroups( $input, "_", "$", "$", "" ); |
| 232 | }, |
| 233 | '_9' => new Reg( "/^_([+\-]?[0-9]+|[^\\\\])/" ), |
| 234 | '_\\x{}{}' => function ( $input ) { |
| 235 | return $this->findObserveGroups( $input, "_", new Reg( "/^\\\\[a-zA-Z]+\{/" ), "}", |
| 236 | "", "", "{", "}", "", true ); |
| 237 | }, |
| 238 | '_\\x{}' => function ( $input ) { |
| 239 | return $this->findObserveGroups( $input, "_", |
| 240 | new Reg( "/^\\\\[a-zA-Z]+\{/" ), "}", "" ); |
| 241 | }, |
| 242 | '_\\x' => new Reg( "/^_(\\\\[a-zA-Z]+)\s*/" ), |
| 243 | '^_' => new Reg( "/^(?:\^(?=_)|\_(?=\^)|[\^_]$)/" ), |
| 244 | '{}^' => new Reg( "/^\{\}(?=\^)/" ), |
| 245 | '{}' => new Reg( "/^\{\}/" ), |
| 246 | '{...}' => function ( $input ) { |
| 247 | return $this->findObserveGroups( $input, "", "{", "}", "" ); |
| 248 | }, |
| 249 | '{(...)}' => function ( $input ) { |
| 250 | return $this->findObserveGroups( $input, "{", "", "", "}" ); |
| 251 | }, |
| 252 | '$...$' => function ( $input ) { |
| 253 | return $this->findObserveGroups( $input, "", "\$", "\$", "" ); |
| 254 | }, |
| 255 | '${(...)}$__$(...)$' => function ( $input ) { |
| 256 | return $this->findObserveGroups( $input, "\${", "", "", "}\$" ) |
| 257 | ?? $this->findObserveGroups( $input, "\$", "", "", "\$" ); |
| 258 | }, |
| 259 | '=<>' => new Reg( "/^[=<>]/" ), |
| 260 | '#' => new Reg( "/^[#\x{2261}]/u" ), |
| 261 | '+' => new Reg( "/^\+/" ), |
| 262 | // -space -, -; -] -/ -$ -state-of-aggregation orig: "/^-(?=[\s_},;\]/]|$|\([a-z]+\))/" |
| 263 | '-$' => new Reg( "/^-(?=[\s_},;\]\/]|$|\([a-z]+\))/u" ), |
| 264 | '-9' => new Reg( "/^-(?=[0-9])/" ), |
| 265 | '- orbital overlap' => new Reg( "/^-(?=(?:[spd]|sp)(?:$|[\s,;\)\]\}]))/" ), |
| 266 | '-' => new Reg( "/^-/" ), |
| 267 | 'pm-operator' => new Reg( "/^(?:\\\\pm|\\\$\\\\pm\\\$|\+-|\+\/-)/" ), |
| 268 | 'operator' => new Reg( "/^(?:\+|(?:[\-=<>]|<<|>>|\\\\approx|\\\$\\\\approx\\\$)(?=\s|$|-?[0-9]))/" ), |
| 269 | 'arrowUpDown' => new Reg( "/^(?:v|\(v\)|\^|\(\^\))(?=$|[\s,;\)\]\}])/" ), |
| 270 | '\\bond{(...)}' => function ( $input ) { |
| 271 | return $this->findObserveGroups( $input, "\\bond{", "", "", "}" ); |
| 272 | }, |
| 273 | '->' => new Reg( '/^(?:<->|<-->|->|<-|<=>>|<<=>|<=>|[\x{2192}\x{27F6}\x{21CC}])/u' ), |
| 274 | 'CMT' => new Reg( "/^[CMT](?=\[)/" ), |
| 275 | '[(...)]' => function ( $input ) { return $this->findObserveGroups( $input, "[", "", |
| 276 | "", "]" ); |
| 277 | }, |
| 278 | '1st-level escape' => new Reg( "/^(&|\\\\\\\\|\\\\hline)\s*/" ), |
| 279 | // \\x - but output no space before |
| 280 | '\\,' => new Reg( "/^(?:\\\\[,\ ;:])/" ), |
| 281 | '\\x{}{}' => function ( $input ) { |
| 282 | return $this->findObserveGroups( $input, "", new Reg( "/^\\\\[a-zA-Z]+\{/" ), "}", |
| 283 | "", "", "{", "}", "", true ); |
| 284 | }, |
| 285 | '\\x{}' => function ( $input ) { |
| 286 | return $this->findObserveGroups( $input, "", new Reg( "/^\\\\[a-zA-Z]+\{/" ), "}", |
| 287 | "" ); |
| 288 | }, |
| 289 | '\\ca' => new Reg( "/^\\\\ca(?:\s+|(?![a-zA-Z]))/" ), |
| 290 | '\\x' => new Reg( "/^(?:\\\\[a-zA-Z]+\s*|\\\\[_&{}%])/" ), |
| 291 | // only those with numbers in front, because the others will be formatted correctly anyway |
| 292 | 'orbital' => new Reg( "/^(?:[0-9]{1,2}[spdfgh]|[0-9]{0,2}sp)(?=$|[^a-zA-Z])/" ), |
| 293 | 'others' => new Reg( "/^[\/~|]/" ), |
| 294 | '\\frac{(...)}' => function ( $input ) { |
| 295 | return $this->findObserveGroups( $input, "\\frac{", "", |
| 296 | "", "}", "{", "", "", "}" ); |
| 297 | }, |
| 298 | '\\overset{(...)}' => function ( $input ) { |
| 299 | return $this->findObserveGroups( $input, "\\overset{", "", |
| 300 | "", "}", "{", "", "", "}" ); |
| 301 | }, |
| 302 | '\\underset{(...)}' => function ( $input ) { |
| 303 | return $this->findObserveGroups( $input, "\\underset{", "", |
| 304 | "", "}", "{", "", "", "}" ); |
| 305 | }, |
| 306 | '\\underbrace{(...)}' => function ( $input ) { |
| 307 | return $this->findObserveGroups( $input, "\\underbrace{", "", |
| 308 | "", "}_", "{", "", "", "}" ); |
| 309 | }, |
| 310 | '\\color{(...)}' => function ( $input ) { |
| 311 | return $this->findObserveGroups( $input, "\\color{", "", "", "}" ); |
| 312 | }, |
| 313 | '\\color{(...)}{(...)}' => function ( $input ) { |
| 314 | // ?? instead of || |
| 315 | return $this->findObserveGroups( $input, "\\color{", "", |
| 316 | "", "}", "{", "", "", "}" ) ?? |
| 317 | $this->findObserveGroups( $input, "\\color", "\\", "", |
| 318 | new Reg( "/^(?=\{)/" ), "{", "", "", "}" ); |
| 319 | }, |
| 320 | '\\ce{(...)}' => function ( $input ) { |
| 321 | return $this->findObserveGroups( $input, "\\ce{", "", "", "}" ); |
| 322 | }, |
| 323 | '\\pu{(...)}' => function ( $input ) { return $this->findObserveGroups( $input, |
| 324 | "\\pu{", "", "", "}" ); |
| 325 | }, |
| 326 | 'oxidation$' => new Reg( "/^(?:[+-][IVX]+|(?:\\\\pm|\\\$\\\\pm\\\$|\+-|\+\/-)\s*0)$/" ), |
| 327 | 'd-oxidation$' => new Reg( "/^(?:[+-]?[IVX]+|(?:\\\\pm|\\\$\\\\pm\\\$|\+-|\+\/-)\s*0)$/" ), |
| 328 | '1/2$' => new Reg( "/^[+\-]?(?:[0-9]+|\\\$[a-z]\\\$|[a-z])\/[0-9]+(?:\\\$[a-z]\\\$|[a-z])?$/" ), |
| 329 | 'amount' => function ( $input ) { |
| 330 | $matches = []; |
| 331 | // e.g. 2, 0.5, 1/2, -2, n/2, +; $a$ could be added later in parsing |
| 332 | $match = preg_match( "/^(?:(?:(?:\([+\-]?[0-9]+\/[0-9]+\)|[+\-]?(?:[0-9]+|\\\$[a-z]\\\$" . |
| 333 | "|[a-z])\/[0-9]+|[+\-]?[0-9]+[.,][0-9]+|[+\-]?\.[0-9]+|[+\-]?[0-9]+)(?:[a-z](?=\s*[A-Z]))?)" . |
| 334 | "|[+\-]?[a-z](?=\s*[A-Z])|\+(?!\s))/", $input, $matches ); |
| 335 | |
| 336 | if ( $match ) { |
| 337 | return [ "match_" => $matches[0], "remainder" => substr( $input, strlen( $matches[0] ) ) ]; |
| 338 | } |
| 339 | $a = $this->findObserveGroups( $input, "", "$", "$", "" ); |
| 340 | // e.g. $2n-1$, $-$ |
| 341 | if ( MhchemUtil::issetJS( $a ) ) { |
| 342 | $matchesI = []; |
| 343 | |
| 344 | $match = preg_match( "/^\\\$(?:\(?[+\-]?(?:[0-9]*[a-z]?[+\-])" . |
| 345 | "?[0-9]*[a-z](?:[+\-][0-9]*[a-z]?)?\)?|\+|-)\\\$$/", $a["match_"] ?? "", |
| 346 | $matchesI ); |
| 347 | if ( $match ) { |
| 348 | return [ "match_" => $matchesI[0], "remainder" => substr( $input, strlen( $matchesI[0] ) ) ]; |
| 349 | } |
| 350 | } |
| 351 | return null; |
| 352 | }, |
| 353 | 'amount2' => function ( $input ) { |
| 354 | /* @phan-suppress-next-line PhanInfiniteRecursion, PhanUndeclaredInvokeInCallable */ |
| 355 | return $this->patterns['amount']( $input ); |
| 356 | }, |
| 357 | '(KV letters),' => new Reg( "/^(?:[A-Z][a-z]{0,2}|i)(?=,)/" ), |
| 358 | 'formula$' => static function ( $input ) { |
| 359 | if ( preg_match( "/^\([a-z]+\)$/", $input ) ) { |
| 360 | // state of aggregation = no formula |
| 361 | return null; |
| 362 | } |
| 363 | $matches = []; |
| 364 | $match = preg_match( "/^(?:[a-z]|(?:[0-9\ \+\-\,\.\(\)]+[a-z])+[0-9\ \+\-\,\.\(\)]*|" |
| 365 | . "(?:[a-z][0-9\ \+\-\,\.\(\)]+)+[a-z]?)$/", $input, $matches ); |
| 366 | if ( $match ) { |
| 367 | return [ "match_" => $matches[0], "remainder" => substr( $input, strlen( $matches[0] ) ) ]; |
| 368 | } |
| 369 | return null; |
| 370 | }, |
| 371 | 'uprightEntities' => new Reg( "/^(?:pH|pOH|pC|pK|iPr|iBu)(?=$|[^a-zA-Z])/" ), |
| 372 | '/' => new Reg( "/^\s*(\/)\s*/" ), |
| 373 | '//' => new Reg( "/^\s*(\/\/)\s*/" ), |
| 374 | '*' => new Reg( "/^\s*[*.]\s*/" ) |
| 375 | ]; |
| 376 | } |
| 377 | |
| 378 | /** |
| 379 | * Matching function |
| 380 | * e.g. match("a", input) will look for the regexp called "a" and see if it matches |
| 381 | * returns null or {match_:"a", remainder:"bc"} |
| 382 | * @param string $m key for fetching a pattern |
| 383 | * @param string $input string to check |
| 384 | * @return array|mixed|null information about the match |
| 385 | */ |
| 386 | public function match( string $m, string $input ) { |
| 387 | $pattern = $this->patterns[$m] ?? null; |
| 388 | if ( !$pattern ) { |
| 389 | // Trying to use non-existing pattern |
| 390 | throw new RuntimeException( "MhchemBugP: mhchem bug P. Please report. (" . $m . ")" ); |
| 391 | } elseif ( $pattern instanceof Reg ) { |
| 392 | $matches = []; |
| 393 | $match = preg_match( $pattern->getRegExp(), $input, $matches ); |
| 394 | if ( $match ) { |
| 395 | if ( count( $matches ) > 2 ) { |
| 396 | return [ |
| 397 | "match_" => array_slice( $matches, 1 ), |
| 398 | "remainder" => substr( $input, strlen( $matches[0] ) ) |
| 399 | ]; |
| 400 | |
| 401 | } else { |
| 402 | return [ |
| 403 | "match_" => MhchemUtil::issetJS( $matches[1] ?? null ) ? $matches[1] : $matches[0], |
| 404 | "remainder" => substr( $input, strlen( $matches[0] ) ) |
| 405 | ]; |
| 406 | } |
| 407 | } |
| 408 | return null; |
| 409 | } elseif ( is_callable( $pattern ) ) { |
| 410 | // $pattern cannot be an instance of MhchemRegExp here, which causes this warning. |
| 411 | /* @phan-suppress-next-line PhanUndeclaredInvokeInCallable */ |
| 412 | return $this->patterns[$m]( $input ); |
| 413 | } else { |
| 414 | return null; |
| 415 | } |
| 416 | } |
| 417 | } |