Code Coverage |
||||||||||
Lines |
Functions and Methods |
Classes and Traits |
||||||||
| Total | |
0.00% |
0 / 1076 |
|
0.00% |
0 / 3 |
CRAP | |
0.00% |
0 / 1 |
| MhchemStateMachines | |
0.00% |
0 / 1076 |
|
0.00% |
0 / 3 |
16002 | |
0.00% |
0 / 1 |
| mhchemCreateTransitions | |
0.00% |
0 / 25 |
|
0.00% |
0 / 1 |
182 | |||
| getGenericActions | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
2 | |||
| __construct | |
0.00% |
0 / 1050 |
|
0.00% |
0 / 1 |
12656 | |||
| 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 Closure; |
| 14 | use MediaWiki\Extension\Math\WikiTexVC\MHChem\MhchemUtil as MU; |
| 15 | use RuntimeException; |
| 16 | |
| 17 | /** |
| 18 | * Contains all state machines (~l.506) and genericActions (~l.465) as well as the mhchemCreateTransitions (~l.47) |
| 19 | * function. |
| 20 | * These can be found in the mentioned lines in mhchemParser.js by Martin Hensel. |
| 21 | * |
| 22 | * Notes: |
| 23 | * PhanParamTooMany and PhanParamTooFew warnings are suppressed in some cases. |
| 24 | * These are known false positive warnings for closure call from array with constant keys. |
| 25 | * https://github.com/phan/phan/issues/4579 |
| 26 | * |
| 27 | * @author Johannes Stegmüller |
| 28 | * @license GPL-2.0-or-later |
| 29 | */ |
| 30 | class MhchemStateMachines { |
| 31 | |
| 32 | /** @var array|array[] */ |
| 33 | public array $stateMachines; |
| 34 | |
| 35 | /** |
| 36 | * @var array|Closure[] |
| 37 | */ |
| 38 | private array $genericActions; |
| 39 | |
| 40 | /** @var MhchemParser */ |
| 41 | private MhchemParser $mhchemParser; |
| 42 | |
| 43 | private static function mhchemCreateTransitions( array $o ): array { |
| 44 | $transitions = []; |
| 45 | // 1. Collect all states |
| 46 | foreach ( $o as $pattern => $d1 ) { |
| 47 | foreach ( $d1 as $state => $d2 ) { |
| 48 | $stateArray = preg_split( "/\|/", strval( $state ), -1, PREG_SPLIT_NO_EMPTY ); |
| 49 | $o[$pattern][$state]["stateArray"] = $stateArray; |
| 50 | for ( $i = 0; $i < count( $stateArray ); $i++ ) { |
| 51 | $transitions[$stateArray[$i]] = []; |
| 52 | } |
| 53 | } |
| 54 | } |
| 55 | |
| 56 | // 2. Fill states |
| 57 | foreach ( $o as $pattern => $d1 ) { |
| 58 | foreach ( $d1 as $d2 ) { |
| 59 | $stateArray = $d2["stateArray"] ?? []; |
| 60 | |
| 61 | for ( $i = 0; $i < count( $stateArray ); $i++ ) { |
| 62 | // 2a. Normalize actions into array: 'text=' ==> [{type_:'text='}] |
| 63 | $p = $d2; |
| 64 | if ( is_string( $p["action_"] ) ) { |
| 65 | $p["action_"] = [ $p["action_"] ]; |
| 66 | } |
| 67 | $p["action_"] = array_merge( [], $p["action_"] ); |
| 68 | |
| 69 | foreach ( $p["action_"] as $key => $action ) { |
| 70 | if ( is_string( $action ) ) { |
| 71 | $p["action_"][$key] = [ "type_" => $p["action_"][$key] ]; |
| 72 | } |
| 73 | } |
| 74 | |
| 75 | // 2.b Multi-insert |
| 76 | $patternArray = preg_split( "/\|/", strval( $pattern ), -1, PREG_SPLIT_NO_EMPTY ); |
| 77 | for ( $j = 0; $j < count( $patternArray ); $j++ ) { |
| 78 | if ( $stateArray[$i] === '*' ) { |
| 79 | // insert into all |
| 80 | foreach ( $transitions as $t => $dEmpty ) { |
| 81 | $transitions[$t][] = [ "pattern" => $patternArray[$j], "task" => $p ]; |
| 82 | } |
| 83 | } else { |
| 84 | $transitions[$stateArray[$i]][] = [ "pattern" => $patternArray[$j], "task" => $p ]; |
| 85 | } |
| 86 | } |
| 87 | } |
| 88 | } |
| 89 | } |
| 90 | |
| 91 | return $transitions; |
| 92 | } |
| 93 | |
| 94 | /** |
| 95 | * @return array |
| 96 | */ |
| 97 | public function getGenericActions(): array { |
| 98 | return $this->genericActions; |
| 99 | } |
| 100 | |
| 101 | /** |
| 102 | * Initialize arrays for genericActions and StateMachines with mhchemCreateTransitions. |
| 103 | * @param-taint $mhchemParser none |
| 104 | */ |
| 105 | public function __construct( MhchemParser $mhchemParser ) { |
| 106 | $this->mhchemParser = $mhchemParser; |
| 107 | $this->genericActions = [ |
| 108 | 'a=' => static function ( &$buffer, $m ) { |
| 109 | $buffer["a"] = ( $buffer["a"] ?? "" ) . $m; |
| 110 | return null; |
| 111 | }, |
| 112 | 'b=' => static function ( &$buffer, $m ) { |
| 113 | $buffer["b"] = ( $buffer["b"] ?? "" ) . $m; |
| 114 | return null; |
| 115 | }, |
| 116 | 'p=' => static function ( &$buffer, $m ) { |
| 117 | $buffer["p"] = ( $buffer["p"] ?? "" ) . $m; |
| 118 | return null; |
| 119 | }, |
| 120 | 'o=' => static function ( &$buffer, $m ) { |
| 121 | $buffer["o"] = ( $buffer["o"] ?? "" ) . $m; |
| 122 | return null; |
| 123 | }, |
| 124 | 'o=+p1' => static function ( &$buffer, $m, $a ) { |
| 125 | $buffer["o"] = ( $buffer["o"] ?? "" ) . $a; |
| 126 | return null; |
| 127 | }, |
| 128 | 'q=' => static function ( &$buffer, $m ) { |
| 129 | $buffer["q"] = ( $buffer["q"] ?? "" ) . $m; |
| 130 | return null; |
| 131 | }, |
| 132 | 'd=' => static function ( &$buffer, $m ) { |
| 133 | $buffer["d"] = ( $buffer["d"] ?? "" ) . $m; |
| 134 | return null; |
| 135 | }, |
| 136 | 'rm=' => static function ( &$buffer, $m ) { |
| 137 | $buffer["rm"] = ( $buffer["rm"] ?? "" ) . $m; |
| 138 | return null; |
| 139 | }, |
| 140 | 'text=' => static function ( &$buffer, $m ) { |
| 141 | $buffer["text_"] = ( $buffer["text_"] ?? "" ) . $m; |
| 142 | return null; |
| 143 | }, |
| 144 | 'insert' => static function ( &$_buffer, $_m, string $a ) { |
| 145 | return [ "type_" => $a ]; |
| 146 | }, |
| 147 | 'insert+p1' => static function ( &$_buffer, $m, $a ) { |
| 148 | return [ "type_" => $a, "p1" => $m ]; |
| 149 | }, |
| 150 | 'insert+p1+p2' => static function ( &$_buffer, $m, $a ) { |
| 151 | return [ "type_" => $a, "p1" => $m[0], "p2" => $m[1] ]; |
| 152 | }, |
| 153 | 'copy' => static function ( &$_buffer, $m ) { |
| 154 | return $m; |
| 155 | }, |
| 156 | 'write' => static function ( &$_buffer, $_m, string $a ) { |
| 157 | return $a; |
| 158 | }, |
| 159 | 'rm' => static function ( &$_buffer, $m ) { |
| 160 | return [ "type_" => 'rm', "p1" => $m ]; |
| 161 | }, |
| 162 | 'text' => function ( &$_buffer, $m ) { |
| 163 | return $this->mhchemParser->go( $m, 'text' ); |
| 164 | }, |
| 165 | 'tex-math' => function ( &$_buffer, $m ) { |
| 166 | return $this->mhchemParser->go( $m, 'tex-math' ); |
| 167 | }, |
| 168 | 'tex-math tight' => function ( &$_buffer, $m ) { |
| 169 | return $this->mhchemParser->go( $m, 'tex-math tight' ); |
| 170 | }, |
| 171 | 'bond' => static function ( &$_buffer, $m, $k ) { |
| 172 | // ?? ok ? |
| 173 | return [ "type_" => 'bond', "kind_" => $k ?? $m ]; |
| 174 | }, |
| 175 | 'color0-output' => static function ( &$_buffer, $m ) { |
| 176 | return [ "type_" => 'color0', "color" => $m ]; |
| 177 | }, |
| 178 | 'ce' => function ( &$_buffer, $m ) { |
| 179 | return $this->mhchemParser->go( $m, 'ce' ); |
| 180 | }, |
| 181 | 'pu' => function ( &$_buffer, $m ) { |
| 182 | return $this->mhchemParser->go( $m, 'pu' ); |
| 183 | }, |
| 184 | '1/2' => static function ( &$_buffer, $m ) { |
| 185 | $ret = []; |
| 186 | if ( preg_match( "/^[+\-]/", $m ) ) { |
| 187 | $ret[] = substr( $m, 0, 1 ); |
| 188 | $m = substr( $m, 1 ); |
| 189 | } |
| 190 | $matches = []; |
| 191 | $n = preg_match( "/^([0-9]+|\\\$[a-z]\\\$|[a-z])\/([0-9]+)(\\\$[a-z]\\\$|[a-z])?$/", |
| 192 | $m, $matches ); |
| 193 | if ( !$n || count( $matches ) < 3 ) { |
| 194 | throw new RuntimeException( "No Result by regex in '1/2' genericAction" ); |
| 195 | } |
| 196 | $matches[1] = preg_replace( "/\\\$/", "", $matches[1] ); |
| 197 | |
| 198 | $ret[] = [ "type_" => 'frac', "p1" => $matches[1], "p2" => $matches[2] ]; |
| 199 | |
| 200 | if ( isset( $matches[3] ) ) { |
| 201 | $matches[3] = preg_replace( "/\\\$/", "", $matches[3] ); |
| 202 | $ret[] = [ "type_" => 'tex-math', "p1" => $matches[3] ]; |
| 203 | } |
| 204 | return $ret; |
| 205 | }, |
| 206 | '9,9' => function ( &$_buffer, $m ) { |
| 207 | return $this->mhchemParser->go( $m, '9,9' ); |
| 208 | } |
| 209 | ]; |
| 210 | |
| 211 | $this->stateMachines = [ |
| 212 | "tex" => [ |
| 213 | "transitions" => self::mhchemCreateTransitions( [ |
| 214 | "empty" => [ "0" => [ "action_" => [ "copy" ] ] ], |
| 215 | "\\ce{(...)}" => [ "0" => [ "action_" => [ [ "type_" => "write", "option" => "{" ], |
| 216 | "ce", [ "type_" => "write", "option" => "}" ] ] ] ], |
| 217 | "\\pu{(...)}" => [ "0" => [ "action_" => [ [ "type_" => "write", "option" => "{" ], |
| 218 | "pu", [ "type_" => "write", "option" => "}" ] ] ] ], |
| 219 | "else" => [ "0" => [ "action_" => [ "copy" ] ] ] |
| 220 | ] ), |
| 221 | "actions" => [] |
| 222 | ], |
| 223 | "ce" => [ |
| 224 | "transitions" => self::mhchemCreateTransitions( [ |
| 225 | "empty" => [ "*" => [ "action_" => "output" ] ], |
| 226 | "else" => [ "0|1|2" => [ "action_" => "beginsWithBond=false", |
| 227 | "revisit" => true, "toContinue" => true ] ], |
| 228 | "oxidation$" => [ "0" => [ "action_" => 'oxidation-output' ] ], |
| 229 | "CMT" => [ 'r' => [ "action_" => "rdt=", "nextState" => "rt" ], |
| 230 | "rd" => [ "action_" => "rqt=", "nextState" => "rdt" ] ], |
| 231 | "arrowUpDown" => [ '0|1|2|as' => |
| 232 | [ "action_" => [ 'sb=false', 'output', 'operator' ], "nextState" => '1' ] ], |
| 233 | "uprightEntities" => [ "0|1|2" => [ "action_" => [ 'o=', 'output' ], "nextState" => "1" ] ], |
| 234 | "orbital" => [ "0|1|2|3" => [ "action_" => "o=", "nextState" => "o" ] ], |
| 235 | "->" => [ "0|1|2|3" => [ "action_" => "r=", "nextState" => "r" ], |
| 236 | "a|as" => [ "action_" => [ 'output', 'r=' ], "nextState" => "r" ], |
| 237 | "*" => [ "action_" => [ 'output', 'r=' ], "nextState" => "r" ] ], |
| 238 | "+" => [ |
| 239 | "o" => [ "action_" => "d= kv", "nextState" => "d" ], |
| 240 | "d|D" => [ "action_" => 'd=', "nextState" => 'd' ], |
| 241 | "q" => [ "action_" => 'd=', "nextState" => 'qd' ], |
| 242 | "qd|qD" => [ "action_" => 'd=', "nextState" => 'qd' ], |
| 243 | "dq" => [ "action_" => [ 'output', 'd=' ], "nextState" => 'd' ], |
| 244 | "3" => [ "action_" => [ 'sb=false', 'output', 'operator' ], "nextState" => '0' ], |
| 245 | ], |
| 246 | "amount" => [ "0|2" => [ "action_" => "a=", "nextState" => "a" ] ], |
| 247 | "pm-operator" => [ "0|1|2|a|as" => [ "action_" => [ 'sb=false', 'output', |
| 248 | [ "type_" => 'operator', "option" => '\\pm' ] ], "nextState" => '0' ] ], |
| 249 | "operator" => [ "0|1|2|a|as" => |
| 250 | [ "action_" => [ 'sb=false', 'output', 'operator' ], "nextState" => '0' ] ], |
| 251 | "-$" => [ |
| 252 | "o|q" => [ "action_" => [ 'charge or bond', 'output' ], "nextState" => 'qd' ], |
| 253 | "d" => [ "action_" => 'd=', "nextState" => 'd' ], |
| 254 | "D" => [ "action_" => [ 'output', [ "type_" => "bond", |
| 255 | "option" => "-" ] ], "nextState" => '3' ], |
| 256 | "q" => [ "action_" => 'd=', "nextState" => 'qd' ], |
| 257 | "qd" => [ "action_" => 'd=', "nextState" => 'qd' ], |
| 258 | "qD|dq" => [ "action_" => [ 'output', |
| 259 | [ "type_" => "bond", "option" => "-" ] ], "nextState" => '3' ] |
| 260 | ], |
| 261 | "-9" => [ "3|o" => [ "action_" => |
| 262 | [ 'output', [ "type_" => "insert", "option" => "hyphen" ] ], "nextState" => '3' ] ], |
| 263 | '- orbital overlap' => [ |
| 264 | 'o' => [ "action_" => [ 'output', |
| 265 | [ "type_" => 'insert', "option" => 'hyphen' ] ], "nextState" => '2' ], |
| 266 | 'd' => [ "action_" => [ 'output', |
| 267 | [ "type_" => 'insert', "option" => 'hyphen' ] ], "nextState" => '2' ] ], |
| 268 | '-' => [ |
| 269 | '0|1|2' => [ "action_" => [ [ "type_" => 'output', "option" => 1 ], |
| 270 | 'beginsWithBond=true', [ "type_" => 'bond', "option" => "-" ] ], "nextState" => '3' ], |
| 271 | '3' => [ "action_" => [ [ "type_" => 'bond', "option" => "-" ] ] ], |
| 272 | 'a' => [ "action_" => [ 'output', |
| 273 | [ "type_" => 'insert', "option" => 'hyphen' ] ], "nextState" => '2' ], |
| 274 | 'as' => [ "action_" => [ [ "type_" => 'output', "option" => 2 ], |
| 275 | [ "type_" => 'bond', "option" => "-" ] ], "nextState" => '3' ], |
| 276 | 'b' => [ "action_" => 'b=' ], |
| 277 | 'o' => [ "action_" => [ [ "type_" => '- after o/d', "option" => false ] ], "nextState" => '2' ], |
| 278 | 'q' => [ "action_" => [ [ "type_" => '- after o/d', "option" => false ] ], "nextState" => '2' ], |
| 279 | 'd|qd|dq' => [ "action_" => |
| 280 | [ [ "type_" => '- after o/d', "option" => true ] ], "nextState" => '2' ], |
| 281 | 'D|qD|p' => [ "action_" => |
| 282 | [ 'output', [ "type_" => 'bond', "option" => "-" ] ], "nextState" => '3' ] ], |
| 283 | 'amount2' => [ |
| 284 | '1|3' => [ "action_" => 'a=', "nextState" => 'a' ] ], |
| 285 | 'letters' => [ |
| 286 | '0|1|2|3|a|as|b|p|bp|o' => [ "action_" => 'o=', "nextState" => 'o' ], |
| 287 | 'q|dq' => [ "action_" => [ 'output', 'o=' ], "nextState" => 'o' ], |
| 288 | 'd|D|qd|qD' => [ "action_" => 'o after d', "nextState" => 'o' ] ], |
| 289 | 'digits' => [ |
| 290 | 'o' => [ "action_" => 'q=', "nextState" => 'q' ], |
| 291 | 'd|D' => [ "action_" => 'q=', "nextState" => 'dq' ], |
| 292 | 'q' => [ "action_" => [ 'output', 'o=' ], "nextState" => 'o' ], |
| 293 | 'a' => [ "action_" => 'o=', "nextState" => 'o' ] ], |
| 294 | 'space A' => [ |
| 295 | 'b|p|bp' => [ "action_" => [] ] ], |
| 296 | 'space' => [ |
| 297 | 'a' => [ "action_" => [], "nextState" => 'as' ], |
| 298 | '0' => [ "action_" => 'sb=false' ], |
| 299 | '1|2' => [ "action_" => 'sb=true' ], |
| 300 | 'r|rt|rd|rdt|rdq' => [ "action_" => 'output', "nextState" => '0' ], |
| 301 | '*' => [ "action_" => [ 'output', 'sb=true' ], "nextState" => '1' ] ], |
| 302 | '1st-level escape' => [ |
| 303 | '1|2' => [ "action_" => [ 'output', |
| 304 | [ "type_" => 'insert+p1', "option" => '1st-level escape' ] ] ], |
| 305 | '*' => [ "action_" => [ 'output', |
| 306 | [ "type_" => 'insert+p1', "option" => '1st-level escape' ] ], "nextState" => '0' ] ], |
| 307 | '[(...)]' => [ |
| 308 | 'r|rt' => [ "action_" => 'rd=', "nextState" => 'rd' ], |
| 309 | 'rd|rdt' => [ "action_" => 'rq=', "nextState" => 'rdq' ] ], |
| 310 | '...' => [ |
| 311 | 'o|d|D|dq|qd|qD' => [ "action_" => |
| 312 | [ 'output', [ "type_" => 'bond', "option" => "..." ] ], "nextState" => '3' ], |
| 313 | '*' => [ "action_" => [ [ "type_" => 'output', "option" => 1 ], |
| 314 | [ "type_" => 'insert', "option" => 'ellipsis' ] ], "nextState" => '1' ] ], |
| 315 | '. __* ' => [ |
| 316 | '*' => [ "action_" => [ 'output', |
| 317 | [ "type_" => 'insert', "option" => 'addition compound' ] ], "nextState" => '1' ] ], |
| 318 | 'state of aggregation $' => [ |
| 319 | '*' => [ "action_" => [ 'output', 'state of aggregation' ], "nextState" => '1' ] ], |
| 320 | '{[(' => [ |
| 321 | 'a|as|o' => [ "action_" => [ 'o=', 'output', 'parenthesisLevel++' ], "nextState" => '2' ], |
| 322 | '0|1|2|3' => [ "action_" => [ 'o=', 'output', 'parenthesisLevel++' ], "nextState" => '2' ], |
| 323 | '*' => [ "action_" => |
| 324 | [ 'output', 'o=', 'output', 'parenthesisLevel++' ], "nextState" => '2' ] ], |
| 325 | ')]}' => [ |
| 326 | '0|1|2|3|b|p|bp|o' => [ "action_" => [ 'o=', 'parenthesisLevel--' ], "nextState" => 'o' ], |
| 327 | 'a|as|d|D|q|qd|qD|dq' => |
| 328 | [ "action_" => [ 'output', 'o=', 'parenthesisLevel--' ], "nextState" => 'o' ] ], |
| 329 | ', ' => [ |
| 330 | '*' => [ "action_" => [ 'output', 'comma' ], "nextState" => '0' ] ], |
| 331 | '^_' => [ |
| 332 | '*' => [ "action_" => [] ] ], |
| 333 | '^{(...)}|^($...$)' => [ |
| 334 | '0|1|2|as' => [ "action_" => 'b=', "nextState" => 'b' ], |
| 335 | 'p' => [ "action_" => 'b=', "nextState" => 'bp' ], |
| 336 | '3|o' => [ "action_" => 'd= kv', "nextState" => 'D' ], |
| 337 | 'q' => [ "action_" => 'd=', "nextState" => 'qD' ], |
| 338 | 'd|D|qd|qD|dq' => [ "action_" => [ 'output', 'd=' ], "nextState" => 'D' ] ], |
| 339 | '^a|^\\x{}{}|^\\x{}|^\\x|\'' => [ |
| 340 | '0|1|2|as' => [ "action_" => 'b=', "nextState" => 'b' ], |
| 341 | 'p' => [ "action_" => 'b=', "nextState" => 'bp' ], |
| 342 | '3|o' => [ "action_" => 'd= kv', "nextState" => 'd' ], |
| 343 | 'q' => [ "action_" => 'd=', "nextState" => 'qd' ], |
| 344 | 'd|qd|D|qD' => [ "action_" => 'd=' ], |
| 345 | 'dq' => [ "action_" => [ 'output', 'd=' ], "nextState" => 'd' ] ], |
| 346 | '_{(state of aggregation)}$' => [ |
| 347 | 'd|D|q|qd|qD|dq' => [ "action_" => [ 'output', 'q=' ], "nextState" => 'q' ] ], |
| 348 | '_{(...)}|_($...$)|_9|_\\x{}{}|_\\x{}|_\\x' => [ |
| 349 | '0|1|2|as' => [ "action_" => 'p=', "nextState" => 'p' ], |
| 350 | 'b' => [ "action_" => 'p=', "nextState" => 'bp' ], |
| 351 | '3|o' => [ "action_" => 'q=', "nextState" => 'q' ], |
| 352 | 'd|D' => [ "action_" => 'q=', "nextState" => 'dq' ], |
| 353 | 'q|qd|qD|dq' => [ "action_" => [ 'output', 'q=' ], "nextState" => 'q' ] ], |
| 354 | '=<>' => [ |
| 355 | '0|1|2|3|a|as|o|q|d|D|qd|qD|dq' => [ "action_" => |
| 356 | [ [ "type_" => 'output', "option" => 2 ], 'bond' ], "nextState" => '3' ] ], |
| 357 | '#' => [ |
| 358 | '0|1|2|3|a|as|o' => [ "action_" => [ [ "type_" => 'output', "option" => 2 ], |
| 359 | [ "type_" => 'bond', "option" => "#" ] ], "nextState" => '3' ] ], |
| 360 | '{}^' => [ |
| 361 | '*' => [ "action_" => [ [ "type_" => 'output', "option" => 1 ], |
| 362 | [ "type_" => 'insert', "option" => 'tinySkip' ] ], "nextState" => '1' ] ], |
| 363 | '{}' => [ |
| 364 | '*' => [ "action_" => [ [ "type_" => 'output', "option" => 1 ] ], "nextState" => '1' ] ], |
| 365 | '{...}' => [ |
| 366 | '0|1|2|3|a|as|b|p|bp' => [ "action_" => 'o=', "nextState" => 'o' ], |
| 367 | 'o|d|D|q|qd|qD|dq' => [ "action_" => [ 'output', 'o=' ], "nextState" => 'o' ] ], |
| 368 | '$...$' => [ |
| 369 | 'a' => [ "action_" => 'a=' ], |
| 370 | '0|1|2|3|as|b|p|bp|o' => [ "action_" => 'o=', "nextState" => 'o' ], |
| 371 | 'as|o' => [ "action_" => 'o=' ], |
| 372 | 'q|d|D|qd|qD|dq' => [ "action_" => [ 'output', 'o=' ], "nextState" => 'o' ] ], |
| 373 | '\\bond{(...)}' => [ |
| 374 | '*' => [ "action_" => [ |
| 375 | [ "type_" => 'output', "option" => 2 ], 'bond' ], "nextState" => "3" ] ], |
| 376 | '\\frac{(...)}' => [ |
| 377 | '*' => [ "action_" => [ |
| 378 | [ "type_" => 'output', "option" => 1 ], 'frac-output' ], "nextState" => '3' ] ], |
| 379 | '\\overset{(...)}' => [ |
| 380 | '*' => [ "action_" => [ |
| 381 | [ "type_" => 'output', "option" => 2 ], 'overset-output' ], "nextState" => '3' ] ], |
| 382 | '\\underset{(...)}' => [ |
| 383 | '*' => [ "action_" => [ |
| 384 | [ "type_" => 'output', "option" => 2 ], 'underset-output' ], "nextState" => '3' ] ], |
| 385 | '\\underbrace{(...)}' => [ |
| 386 | '*' => [ "action_" => [ |
| 387 | [ "type_" => 'output', "option" => 2 ], 'underbrace-output' ], "nextState" => '3' ] ], |
| 388 | '\\color{(...)}{(...)}' => [ |
| 389 | '*' => [ "action_" => [ |
| 390 | [ "type_" => 'output', "option" => 2 ], 'color-output' ], "nextState" => '3' ] ], |
| 391 | '\\color{(...)}' => [ |
| 392 | '*' => [ "action_" => [ [ "type_" => 'output', "option" => 2 ], 'color0-output' ] ] ], |
| 393 | '\\ce{(...)}' => [ |
| 394 | '*' => [ "action_" => [ [ "type_" => 'output', "option" => 2 ], 'ce' ], "nextState" => '3' ] ], |
| 395 | '\\,' => [ |
| 396 | '*' => [ "action_" => [ |
| 397 | [ "type_" => 'output', "option" => 1 ], 'copy' ], "nextState" => '1' ] ], |
| 398 | '\\pu{(...)}' => [ |
| 399 | '*' => [ "action_" => [ 'output', [ "type_" => 'write', "option" => "{" ], |
| 400 | 'pu', [ "type_" => 'write', "option" => "}" ] ], "nextState" => '3' ] ], |
| 401 | '\\x{}{}|\\x{}|\\x' => |
| 402 | [ '0|1|2|3|a|as|b|p|bp|o|c0' => [ "action_" => [ 'o=', 'output' ], "nextState" => '3' ], |
| 403 | '*' => [ "action_" => [ 'output', 'o=', 'output' ], "nextState" => '3' ] ], |
| 404 | 'others' => [ '*' => [ "action_" => |
| 405 | [ [ "type_" => 'output', "option" => 1 ], 'copy' ], "nextState" => '3' ] ], |
| 406 | 'else2' => [ 'a' => [ "action_" => 'a to o', "nextState" => 'o', "revisit" => true ], |
| 407 | 'as' => [ "action_" => [ 'output', 'sb=true' ], "nextState" => '1', "revisit" => true ], |
| 408 | 'r|rt|rd|rdt|rdq' => [ "action_" => [ 'output' ], "nextState" => '0', "revisit" => true ], |
| 409 | '*' => [ "action_" => [ 'output', 'copy' ], "nextState" => '3' ] ] |
| 410 | ] ), |
| 411 | "actions" => [ |
| 412 | 'o after d' => function ( &$buffer, $m ) { |
| 413 | if ( preg_match( "/^[1-9][0-9]*$/", $buffer['d'] ?? "" ) ) { |
| 414 | $tmp = $buffer["d"]; |
| 415 | $buffer["d"] = null; |
| 416 | $ret = $this->stateMachines["ce"]["actions"]["output"]( $buffer, null, null ); |
| 417 | $ret[] = [ "type_" => "tinySkip" ]; |
| 418 | $buffer["b"] = $tmp; |
| 419 | } else { |
| 420 | $ret = $this->stateMachines["ce"]["actions"]["output"]( $buffer, null, null ); |
| 421 | } |
| 422 | |
| 423 | /** @phan-suppress-next-line PhanParamTooFew */ |
| 424 | $this->genericActions['o=']( $buffer, $m ); |
| 425 | return $ret; |
| 426 | }, |
| 427 | 'd= kv' => static function ( &$buffer, $m ) { |
| 428 | $buffer["d"] = $m; |
| 429 | $buffer["dType"] = "kv"; |
| 430 | return null; |
| 431 | }, |
| 432 | 'charge or bond' => function ( &$buffer, $m ) { |
| 433 | if ( MhchemUtil::issetJS( $buffer['beginsWithBond'] ?? null ) ) { |
| 434 | $ret = []; |
| 435 | $im = $this->stateMachines["ce"]["actions"]["output"]( $buffer, null, null ); |
| 436 | MhchemUtil::concatArray( $ret, $im ); |
| 437 | |
| 438 | /** @phan-suppress-next-line PhanParamTooMany */ |
| 439 | MhchemUtil::concatArray( $ret, $this->genericActions['bond']( $buffer, $m, "-" ) ); |
| 440 | return $ret; |
| 441 | } else { |
| 442 | $buffer["d"] = $m; |
| 443 | return null; |
| 444 | } |
| 445 | }, |
| 446 | '- after o/d' => function ( &$buffer, $m, $isAfterD ) { |
| 447 | $c1 = $this->mhchemParser->getPatterns()->match( 'orbital', $buffer["o"] ?? "" ); |
| 448 | $c2 = $this->mhchemParser->getPatterns()->match( 'one lowercase greek letter $', |
| 449 | $buffer["o"] ?? "" ); |
| 450 | $c3 = $this->mhchemParser->getPatterns()->match( 'one lowercase latin letter $', |
| 451 | $buffer["o"] ?? "" ); |
| 452 | $c4 = $this->mhchemParser->getPatterns()->match( '$one lowercase latin letter$ $', |
| 453 | $buffer["o"] ?? "" ); |
| 454 | $hyphenFollows = $m === "-" && ( ( isset( $c1["remainder"] ) && $c1["remainder"] === "" ) |
| 455 | || $c2 !== null || $c3 !== null || $c4 !== null ); |
| 456 | if ( $hyphenFollows && !isset( $buffer["a"] ) && !isset( $buffer["b"] ) |
| 457 | && !isset( $buffer["p"] ) && !isset( $buffer["d"] ) |
| 458 | && !isset( $buffer["q"] ) && !$c1 && $c3 ) { |
| 459 | $buffer["o"] = '$' . $buffer["o"] . '$'; |
| 460 | } |
| 461 | $ret = []; |
| 462 | if ( $hyphenFollows ) { |
| 463 | $im = $this->stateMachines["ce"]["actions"]["output"]( $buffer, null, null ); |
| 464 | MhchemUtil::concatArray( $ret, $im ); |
| 465 | $ret[] = [ "type_" => 'hyphen' ]; |
| 466 | } else { |
| 467 | $c1 = $this->mhchemParser->getPatterns()->match( 'digits', $buffer["d"] ?? "" ); |
| 468 | if ( $isAfterD && isset( $c1["remainder"] ) && $c1["remainder"] === '' ) { |
| 469 | /** @phan-suppress-next-line PhanParamTooFew */ |
| 470 | MhchemUtil::concatArray( $ret, $this->genericActions['d=']( $buffer, $m ) ); |
| 471 | $im = $this->stateMachines["ce"]["actions"]["output"]( $buffer, null, null ); |
| 472 | MhchemUtil::concatArray( $ret, $im ); |
| 473 | } else { |
| 474 | $im = $this->stateMachines["ce"]["actions"]["output"]( $buffer, null, null ); |
| 475 | MhchemUtil::concatArray( $ret, $im ); |
| 476 | /** @phan-suppress-next-line PhanParamTooMany */ |
| 477 | MhchemUtil::concatArray( $ret, $this->genericActions['bond']( $buffer, $m, "-" ) ); |
| 478 | } |
| 479 | } |
| 480 | return $ret; |
| 481 | }, |
| 482 | 'a to o' => static function ( &$buffer ) { |
| 483 | $buffer["o"] = $buffer["a"]; |
| 484 | $buffer["a"] = null; |
| 485 | return null; |
| 486 | }, |
| 487 | 'sb=true' => static function ( &$buffer ) { |
| 488 | $buffer["sb"] = true; |
| 489 | return null; |
| 490 | }, |
| 491 | 'sb=false' => static function ( &$buffer ) { |
| 492 | $buffer["sb"] = false; |
| 493 | return null; |
| 494 | }, |
| 495 | 'beginsWithBond=true' => static function ( &$buffer ) { |
| 496 | $buffer['beginsWithBond'] = true; |
| 497 | return null; |
| 498 | }, |
| 499 | 'beginsWithBond=false' => static function ( &$buffer ) { |
| 500 | $buffer['beginsWithBond'] = false; |
| 501 | return null; |
| 502 | }, |
| 503 | 'parenthesisLevel++' => static function ( &$buffer ) { |
| 504 | $buffer['parenthesisLevel']++; |
| 505 | return null; |
| 506 | }, |
| 507 | 'parenthesisLevel--' => static function ( &$buffer ) { |
| 508 | $buffer['parenthesisLevel']--; |
| 509 | return null; |
| 510 | }, |
| 511 | 'state of aggregation' => function ( $_buffer, $m ) { |
| 512 | return [ "type_" => 'state of aggregation', |
| 513 | "p1" => $this->mhchemParser->go( $m, 'o' ) ]; |
| 514 | }, |
| 515 | 'comma' => static function ( $buffer, $m ) { |
| 516 | // $a = preg_replace('/\s*$/', '', $m); tbd: final check if using rtrim is ok |
| 517 | $a = rtrim( $m ); |
| 518 | $withSpace = ( $a !== $m ); |
| 519 | if ( $withSpace && $buffer['parenthesisLevel'] === 0 ) { |
| 520 | return [ "type_" => 'comma enumeration L', "p1" => $a ]; |
| 521 | } else { |
| 522 | return [ "type_" => 'comma enumeration M', "p1" => $a ]; |
| 523 | } |
| 524 | }, |
| 525 | 'output' => function ( &$buffer, $_m, $entityFollows ) { |
| 526 | if ( !isset( $buffer["r"] ) ) { |
| 527 | $ret = []; |
| 528 | if ( !isset( $buffer["a"] ) && !isset( $buffer["b"] ) && !isset( $buffer["p"] ) |
| 529 | && !isset( $buffer["o"] ) && !isset( $buffer["q"] ) |
| 530 | && !isset( $buffer["d"] ) && !$entityFollows ) { |
| 531 | // do nothing. |
| 532 | } else { |
| 533 | if ( MhchemUtil::issetJS( $buffer["sb"] ?? null ) ) { |
| 534 | $ret[] = [ "type_" => 'entitySkip' ]; |
| 535 | } |
| 536 | if ( !isset( $buffer["o"] ) && !isset( $buffer["q"] ) && !isset( $buffer["d"] ) |
| 537 | && !isset( $buffer["b"] ) && !isset( $buffer["p"] ) && $entityFollows !== 2 ) { |
| 538 | $buffer["o"] = $buffer["a"] ?? null; |
| 539 | $buffer["a"] = null; |
| 540 | } elseif ( !isset( $buffer["o"] ) && !isset( $buffer["q"] ) && !isset( $buffer["d"] ) |
| 541 | && ( isset( $buffer["b"] ) || isset( $buffer["p"] ) ) ) { |
| 542 | $buffer["o"] = $buffer["a"] ?? null; |
| 543 | $buffer["d"] = $buffer["b"] ?? null; |
| 544 | $buffer["q"] = $buffer["p"] ?? null; |
| 545 | $buffer["a"] = $buffer["b"] = $buffer["p"] = null; |
| 546 | } else { |
| 547 | if ( isset( $buffer["o"] ) && isset( $buffer["dType"] ) && $buffer["dType"] === 'kv' |
| 548 | && $this->mhchemParser->getPatterns()->match( 'd-oxidation$', |
| 549 | $buffer["d"] ?? "" ) ) { |
| 550 | $buffer["dType"] = 'oxidation'; |
| 551 | } elseif ( isset( $buffer["o"] ) && isset( $buffer["dType"] ) |
| 552 | && $buffer["dType"] === 'kv' && !isset( $buffer["q"] ) ) { |
| 553 | $buffer["dType"] = null; |
| 554 | } |
| 555 | } |
| 556 | |
| 557 | $retIm = [ |
| 558 | "type_" => 'chemfive', |
| 559 | "a" => $this->mhchemParser->go( $buffer["a"] ?? null, 'a' ), |
| 560 | "b" => $this->mhchemParser->go( $buffer["b"] ?? null, 'bd' ), |
| 561 | "p" => $this->mhchemParser->go( $buffer["p"] ?? null, 'pq' ), |
| 562 | "o" => $this->mhchemParser->go( $buffer["o"] ?? null, 'o' ), |
| 563 | "q" => $this->mhchemParser->go( $buffer["q"] ?? null, 'pq' ), |
| 564 | "d" => $this->mhchemParser->go( $buffer["d"] ?? null, |
| 565 | ( isset( $buffer["dType"] ) |
| 566 | && $buffer["dType"] === 'oxidation' ? 'oxidation' : 'bd' ) ), |
| 567 | ]; |
| 568 | |
| 569 | if ( isset( $buffer["dType"] ) ) { |
| 570 | $retIm["dType"] = $buffer["dType"]; |
| 571 | } |
| 572 | |
| 573 | $ret[] = $retIm; |
| 574 | |
| 575 | } |
| 576 | } else { |
| 577 | if ( isset( $buffer["rdt"] ) && $buffer["rdt"] === 'M' ) { |
| 578 | $rd = $this->mhchemParser->go( $buffer["rd"], 'tex-math' ); |
| 579 | } elseif ( isset( $buffer["rdt"] ) && $buffer["rdt"] === 'T' ) { |
| 580 | // tbd double array ok ? |
| 581 | $rd = [ [ "type_" => 'text', "p1" => $buffer["rd"] ?? "" ] ]; |
| 582 | } else { |
| 583 | $rd = $this->mhchemParser->go( $buffer["rd"] ?? null, 'ce' ); |
| 584 | } |
| 585 | |
| 586 | if ( isset( $buffer["rqt"] ) && $buffer["rqt"] === 'M' ) { |
| 587 | $rq = $this->mhchemParser->go( $buffer["rq"], 'tex-math' ); |
| 588 | } elseif ( isset( $buffer["rqt"] ) && $buffer["rqt"] === 'T' ) { |
| 589 | $rq = [ [ "type_" => 'text', "p1" => $buffer["rq"] ?? "" ] ]; |
| 590 | } else { |
| 591 | $rq = $this->mhchemParser->go( $buffer["rq"] ?? null, 'ce' ); |
| 592 | } |
| 593 | $ret = [ |
| 594 | "type_" => 'arrow', |
| 595 | "r" => $buffer["r"], |
| 596 | "rd" => $rd, |
| 597 | "rq" => $rq |
| 598 | ]; |
| 599 | } |
| 600 | foreach ( $buffer as $key => $value ) { |
| 601 | if ( $key !== 'parenthesisLevel' && $key !== 'beginsWithBond' ) { |
| 602 | unset( $buffer[$key] ); |
| 603 | } |
| 604 | } |
| 605 | return $ret; |
| 606 | }, |
| 607 | 'oxidation-output' => function ( $_buffer, $m ) { |
| 608 | $ret = [ "{" ]; |
| 609 | MhchemUtil::concatArray( $ret, $this->mhchemParser->go( $m, 'oxidation' ) ); |
| 610 | $ret[] = "}"; |
| 611 | return $ret; |
| 612 | }, |
| 613 | 'frac-output' => function ( $_buffer, $m ) { |
| 614 | return [ "type_" => 'frac-ce', |
| 615 | "p1" => $this->mhchemParser->go( $m[0] ?? null, 'ce' ), |
| 616 | "p2" => $this->mhchemParser->go( $m[1] ?? null, 'ce' ) ]; |
| 617 | }, |
| 618 | 'overset-output' => function ( $_buffer, $m ) { |
| 619 | return [ "type_" => 'overset', |
| 620 | "p1" => $this->mhchemParser->go( $m[0] ?? null, 'ce' ), |
| 621 | "p2" => $this->mhchemParser->go( $m[1] ?? null, 'ce' ) ]; |
| 622 | }, |
| 623 | 'underset-output' => function ( $_buffer, $m ) { |
| 624 | return [ "type_" => 'underset', |
| 625 | "p1" => $this->mhchemParser->go( $m[0] ?? null, 'ce' ), |
| 626 | "p2" => $this->mhchemParser->go( $m[1] ?? null, 'ce' ) ]; |
| 627 | }, |
| 628 | 'underbrace-output' => function ( $_buffer, $m ) { |
| 629 | return [ "type_" => 'underbrace', |
| 630 | "p1" => $this->mhchemParser->go( $m[0] ?? null, 'ce' ), |
| 631 | "p2" => $this->mhchemParser->go( $m[1] ?? null, 'ce' ) ]; |
| 632 | }, |
| 633 | 'color-output' => function ( $_buffer, $m ) { |
| 634 | return [ "type_" => 'color', "color1" => $m[0] ?? null, |
| 635 | "color2" => $this->mhchemParser->go( $m[1] ?? null, 'ce' ) ]; |
| 636 | }, |
| 637 | 'r=' => static function ( &$buffer, $m ) { |
| 638 | $buffer["r"] = $m; |
| 639 | return null; |
| 640 | }, |
| 641 | 'rdt=' => static function ( &$buffer, $m ) { |
| 642 | $buffer["rdt"] = $m; |
| 643 | return null; |
| 644 | }, |
| 645 | 'rd=' => static function ( &$buffer, $m ) { |
| 646 | $buffer["rd"] = $m; |
| 647 | return null; |
| 648 | }, |
| 649 | 'rqt=' => static function ( &$buffer, $m ) { |
| 650 | $buffer["rqt"] = $m; |
| 651 | return null; |
| 652 | }, |
| 653 | 'rq=' => static function ( &$buffer, $m ) { |
| 654 | $buffer["rq"] = $m; |
| 655 | return null; |
| 656 | }, |
| 657 | 'operator' => static function ( &$_buffer, $m, $p1 ) { |
| 658 | return [ "type_" => 'operator', "kind_" => ( $p1 ?: $m ) ]; |
| 659 | }, |
| 660 | ], |
| 661 | ], |
| 662 | 'a' => [ |
| 663 | "transitions" => static::mhchemCreateTransitions( [ |
| 664 | 'empty' => [ |
| 665 | '*' => [ "action_" => [] ] |
| 666 | ], |
| 667 | '1/2$' => [ |
| 668 | '0' => [ "action_" => '1/2' ] |
| 669 | ], |
| 670 | 'else' => [ |
| 671 | '0' => [ "action_" => [], "nextState" => '1', "revisit" => true ] |
| 672 | ], |
| 673 | '${(...)}$__$(...)$' => [ |
| 674 | '*' => [ "action_" => 'tex-math tight', "nextState" => '1' ] |
| 675 | ], |
| 676 | ',' => [ |
| 677 | '*' => [ "action_" => [ [ "type_" => 'insert', "option" => 'commaDecimal' ] ] ] |
| 678 | ], |
| 679 | 'else2' => [ |
| 680 | '*' => [ "action_" => 'copy' ] |
| 681 | ] |
| 682 | ] ), |
| 683 | "actions" => [] |
| 684 | ], |
| 685 | 'o' => [ |
| 686 | "transitions" => static::mhchemCreateTransitions( [ |
| 687 | 'empty' => [ |
| 688 | '*' => [ "action_" => [] ] |
| 689 | ], |
| 690 | '1/2$' => [ |
| 691 | '0' => [ "action_" => '1/2' ] |
| 692 | ], |
| 693 | 'else' => [ |
| 694 | '0' => [ "action_" => [], "nextState" => '1', "revisit" => true ] |
| 695 | ], |
| 696 | 'letters' => [ |
| 697 | '*' => [ "action_" => 'rm' ] |
| 698 | ], |
| 699 | '\\ca' => [ |
| 700 | '*' => [ "action_" => [ [ "type_" => 'insert', "option" => 'circa' ] ] ] |
| 701 | ], |
| 702 | '\\pu{(...)}' => [ |
| 703 | '*' => [ "action_" => [ [ "type_" => 'write', "option" => "{" ], |
| 704 | 'pu', [ "type_" => 'write', "option" => "}" ] ] ] |
| 705 | ], |
| 706 | '\\x{}{}|\\x{}|\\x' => [ |
| 707 | '*' => [ "action_" => 'copy' ] |
| 708 | ], |
| 709 | '${(...)}$__$(...)$' => [ |
| 710 | '*' => [ "action_" => 'tex-math' ] |
| 711 | ], |
| 712 | '{(...)}' => [ |
| 713 | '*' => [ "action_" => [ [ "type_" => 'write', "option" => "{" ], |
| 714 | 'text', [ "type_" => 'write', "option" => "}" ] ] ] |
| 715 | ], |
| 716 | 'else2' => [ |
| 717 | '*' => [ "action_" => 'copy' ] |
| 718 | ] |
| 719 | ] ), |
| 720 | "actions" => [] |
| 721 | ], |
| 722 | 'text' => [ |
| 723 | "transitions" => static::mhchemCreateTransitions( [ |
| 724 | 'empty' => [ |
| 725 | '*' => [ "action_" => 'output' ] |
| 726 | ], |
| 727 | '{...}' => [ |
| 728 | '*' => [ "action_" => 'text=' ] |
| 729 | ], |
| 730 | '${(...)}$__$(...)$' => [ |
| 731 | '*' => [ "action_" => 'tex-math' ] |
| 732 | ], |
| 733 | '\\greek' => [ |
| 734 | '*' => [ "action_" => [ 'output', 'rm' ] ] |
| 735 | ], |
| 736 | '\\pu{(...)}' => [ |
| 737 | '*' => [ "action_" => [ 'output', [ "type_" => 'write', "option" => "{" ], |
| 738 | 'pu', [ "type_" => 'write', "option" => "}" ] ] ] |
| 739 | ], |
| 740 | '\\,|\\x{}{}|\\x{}|\\x' => [ |
| 741 | '*' => [ "action_" => [ 'output', 'copy' ] ] |
| 742 | ], |
| 743 | 'else' => [ |
| 744 | '*' => [ "action_" => 'text=' ] |
| 745 | ], |
| 746 | ] ), |
| 747 | "actions" => [ |
| 748 | 'output' => static function ( &$buffer ) { |
| 749 | if ( isset( $buffer["text_"] ) ) { |
| 750 | $ret = [ "type_" => 'text', "p1" => $buffer["text_"] ]; |
| 751 | foreach ( $buffer as $key => $value ) { |
| 752 | unset( $buffer[$key] ); |
| 753 | } |
| 754 | return $ret; |
| 755 | } |
| 756 | return null; |
| 757 | } |
| 758 | ] |
| 759 | ], |
| 760 | 'pq' => [ |
| 761 | "transitions" => static::mhchemCreateTransitions( [ |
| 762 | 'empty' => [ |
| 763 | '*' => [ "action_" => [] ] |
| 764 | ], |
| 765 | 'state of aggregation $' => [ |
| 766 | '*' => [ "action_" => 'state of aggregation' ] |
| 767 | ], |
| 768 | 'i$' => [ |
| 769 | '0' => [ "action_" => [], "nextState" => '!f', "revisit" => true ] |
| 770 | ], |
| 771 | '(KV letters),' => [ |
| 772 | '0' => [ "action_" => 'rm', "nextState" => '0' ] |
| 773 | ], |
| 774 | 'formula$' => [ |
| 775 | '0' => [ "action_" => [], "nextState" => 'f', "revisit" => true ] |
| 776 | ], |
| 777 | '1/2$' => [ |
| 778 | '0' => [ "action_" => '1/2' ] |
| 779 | ], |
| 780 | 'else' => [ |
| 781 | '0' => [ "action_" => [], "nextState" => '!f', "revisit" => true ] |
| 782 | ], |
| 783 | '${(...)}$__$(...)$' => [ |
| 784 | '*' => [ "action_" => 'tex-math' ] |
| 785 | ], |
| 786 | '{(...)}' => [ |
| 787 | '*' => [ "action_" => 'text' ] |
| 788 | ], |
| 789 | 'a-z' => [ |
| 790 | 'f' => [ "action_" => 'tex-math' ] |
| 791 | ], |
| 792 | 'letters' => [ |
| 793 | '*' => [ "action_" => 'rm' ] |
| 794 | ], |
| 795 | '-9.,9' => [ |
| 796 | '*' => [ "action_" => '9,9' ] |
| 797 | ], |
| 798 | ',' => [ |
| 799 | '*' => [ "action_" => [ [ "type_" => 'insert+p1', "option" => 'comma enumeration S' ] ] ] |
| 800 | ], |
| 801 | '\\color{(...)}{(...)}' => [ |
| 802 | '*' => [ "action_" => 'color-output' ] |
| 803 | ], |
| 804 | '\\color{(...)}' => [ |
| 805 | '*' => [ "action_" => 'color0-output' ] |
| 806 | ], |
| 807 | '\\ce{(...)}' => [ |
| 808 | '*' => [ "action_" => 'ce' ] |
| 809 | ], |
| 810 | '\\pu{(...)}' => [ |
| 811 | '*' => [ "action_" => [ [ "type_" => 'write', "option" => "{" ], |
| 812 | 'pu', [ "type_" => 'write', "option" => "}" ] ] ] |
| 813 | ], |
| 814 | '\\,|\\x{}{}|\\x{}|\\x' => [ |
| 815 | '*' => [ "action_" => 'copy' ] |
| 816 | ], |
| 817 | 'else2' => [ |
| 818 | '*' => [ "action_" => 'copy' ] |
| 819 | ] |
| 820 | ] ), |
| 821 | "actions" => [ |
| 822 | 'state of aggregation' => function ( $_buffer, $m ) { |
| 823 | return [ "type_" => 'state of aggregation subscript', |
| 824 | "p1" => $this->mhchemParser->go( $m, 'o' ) ]; |
| 825 | }, |
| 826 | 'color-output' => function ( $_buffer, $m ) { |
| 827 | return [ "type_" => 'color', "color1" => $m[0] ?? null, |
| 828 | "color2" => $this->mhchemParser->go( $m[1] ?? null, 'pq' ) ]; |
| 829 | } |
| 830 | ] |
| 831 | ], |
| 832 | 'bd' => [ |
| 833 | "transitions" => static::mhchemCreateTransitions( [ |
| 834 | 'empty' => [ |
| 835 | '*' => [ "action_" => [] ] |
| 836 | ], |
| 837 | 'x$' => [ |
| 838 | '0' => [ "action_" => [], "nextState" => '!f', "revisit" => true ] |
| 839 | ], |
| 840 | 'formula$' => [ |
| 841 | '0' => [ "action_" => [], "nextState" => 'f', "revisit" => true ] |
| 842 | ], |
| 843 | 'else' => [ |
| 844 | '0' => [ "action_" => [], "nextState" => '!f', "revisit" => true ] |
| 845 | ], |
| 846 | '-9.,9 no missing 0' => [ |
| 847 | '*' => [ "action_" => '9,9' ] |
| 848 | ], |
| 849 | '.' => [ |
| 850 | '*' => [ "action_" => [ [ "type_" => 'insert', "option" => 'electron dot' ] ] ] |
| 851 | ], |
| 852 | 'a-z' => [ |
| 853 | 'f' => [ "action_" => 'tex-math' ] |
| 854 | ], |
| 855 | 'x' => [ |
| 856 | '*' => [ "action_" => [ [ "type_" => 'insert', "option" => 'KV x' ] ] ] |
| 857 | ], |
| 858 | 'letters' => [ |
| 859 | '*' => [ "action_" => 'rm' ] |
| 860 | ], |
| 861 | '\'' => [ |
| 862 | '*' => [ "action_" => [ [ "type_" => 'insert', "option" => 'prime' ] ] ] |
| 863 | ], |
| 864 | '${(...)}$__$(...)$' => [ |
| 865 | '*' => [ "action_" => 'tex-math' ] |
| 866 | ], |
| 867 | '{(...)}' => [ |
| 868 | '*' => [ "action_" => 'text' ] |
| 869 | ], |
| 870 | '\\color{(...)}{(...)}' => [ |
| 871 | '*' => [ "action_" => 'color-output' ] |
| 872 | ], |
| 873 | '\\color{(...)}' => [ |
| 874 | '*' => [ "action_" => 'color0-output' ] |
| 875 | ], |
| 876 | '\\ce{(...)}' => [ |
| 877 | '*' => [ "action_" => 'ce' ] |
| 878 | ], |
| 879 | '\\pu{(...)}' => [ |
| 880 | '*' => [ "action_" => [ [ "type_" => 'write', "option" => "{" ], |
| 881 | 'pu', [ "type_" => 'write', "option" => "}" ] ] ] |
| 882 | ], |
| 883 | '\\,|\\x{}{}|\\x{}|\\x' => [ |
| 884 | '*' => [ "action_" => 'copy' ] |
| 885 | ], |
| 886 | 'else2' => [ |
| 887 | '*' => [ "action_" => 'copy' ] |
| 888 | ] |
| 889 | ] ), |
| 890 | "actions" => [ |
| 891 | 'color-output' => function ( $_buffer, $m ) { |
| 892 | return [ "type_" => 'color', "color1" => $m[0] ?? null, |
| 893 | "color2" => $this->mhchemParser->go( $m[1] ?? null, 'bd' ) ]; |
| 894 | } |
| 895 | ] |
| 896 | ], |
| 897 | 'oxidation' => [ |
| 898 | "transitions" => static::mhchemCreateTransitions( [ |
| 899 | 'empty' => [ |
| 900 | '*' => [ "action_" => 'roman-numeral' ] |
| 901 | ], |
| 902 | 'pm-operator' => [ |
| 903 | '*' => [ "action_" => [ [ "type_" => 'o=+p1', "option" => '\\pm' ] ] ] |
| 904 | ], |
| 905 | 'else' => [ |
| 906 | '*' => [ "action_" => 'o=' ] |
| 907 | ] |
| 908 | ] ), |
| 909 | "actions" => [ |
| 910 | 'roman-numeral' => static function ( $buffer ) { |
| 911 | return [ "type_" => 'roman numeral', "p1" => $buffer["o"] ?? "" ]; |
| 912 | } |
| 913 | ] |
| 914 | ], |
| 915 | 'tex-math' => [ |
| 916 | "transitions" => static::mhchemCreateTransitions( [ |
| 917 | 'empty' => [ |
| 918 | '*' => [ "action_" => 'output' ] |
| 919 | ], |
| 920 | '\\ce{(...)}' => [ |
| 921 | '*' => [ "action_" => [ 'output', 'ce' ] ] |
| 922 | ], |
| 923 | '\\pu{(...)}' => [ |
| 924 | '*' => [ "action_" => [ 'output', [ "type_" => 'write', "option" => "{" ], |
| 925 | 'pu', [ "type_" => 'write', "option" => "}" ] ] ] |
| 926 | ], |
| 927 | '{...}|\\,|\\x{}{}|\\x{}|\\x' => [ |
| 928 | '*' => [ "action_" => 'o=' ] |
| 929 | ], |
| 930 | 'else' => [ |
| 931 | '*' => [ "action_" => 'o=' ] |
| 932 | ] |
| 933 | ] ), |
| 934 | "actions" => [ |
| 935 | 'output' => static function ( &$buffer ) { |
| 936 | if ( isset( $buffer["o"] ) ) { |
| 937 | $ret = [ "type_" => 'tex-math', "p1" => $buffer["o"] ]; |
| 938 | foreach ( $buffer as $key => $value ) { |
| 939 | unset( $buffer[$key] ); |
| 940 | } |
| 941 | return $ret; |
| 942 | } |
| 943 | return null; |
| 944 | } |
| 945 | ] |
| 946 | ], |
| 947 | 'tex-math tight' => [ |
| 948 | "transitions" => static::mhchemCreateTransitions( [ |
| 949 | 'empty' => [ |
| 950 | '*' => [ "action_" => 'output' ] |
| 951 | ], |
| 952 | '\\ce{(...)}' => [ |
| 953 | '*' => [ "action_" => [ 'output', 'ce' ] ] |
| 954 | ], |
| 955 | '\\pu{(...)}' => [ |
| 956 | '*' => [ "action_" => [ 'output', [ "type_" => 'write', "option" => "{" ], |
| 957 | 'pu', [ "type_" => 'write', "option" => "}" ] ] ] |
| 958 | ], |
| 959 | '{...}|\\,|\\x{}{}|\\x{}|\\x' => [ |
| 960 | '*' => [ "action_" => 'o=' ] |
| 961 | ], |
| 962 | '-|+' => [ |
| 963 | '*' => [ "action_" => 'tight operator' ] |
| 964 | ], |
| 965 | 'else' => [ |
| 966 | '*' => [ "action_" => 'o=' ] |
| 967 | ] |
| 968 | ] ), |
| 969 | "actions" => [ |
| 970 | 'tight operator' => static function ( &$buffer, $m ) { |
| 971 | $buffer["o"] = ( $buffer["o"] ?? "" ) . "{" . $m . "}"; |
| 972 | return null; |
| 973 | }, |
| 974 | 'output' => static function ( &$buffer ) { |
| 975 | if ( $buffer["o"] ) { |
| 976 | $ret = [ "type_" => 'tex-math', "p1" => $buffer["o"] ]; |
| 977 | foreach ( $buffer as $key => $value ) { |
| 978 | unset( $buffer[$key] ); |
| 979 | } |
| 980 | return $ret; |
| 981 | } |
| 982 | return null; |
| 983 | } |
| 984 | ] |
| 985 | ], |
| 986 | '9,9' => [ |
| 987 | "transitions" => static::mhchemCreateTransitions( [ |
| 988 | 'empty' => [ |
| 989 | '*' => [ "action_" => [] ] |
| 990 | ], |
| 991 | ',' => [ |
| 992 | '*' => [ "action_" => 'comma' ] |
| 993 | ], |
| 994 | 'else' => [ |
| 995 | '*' => [ "action_" => 'copy' ] |
| 996 | ] |
| 997 | ] ), |
| 998 | "actions" => [ |
| 999 | 'comma' => static function () { |
| 1000 | return [ "type_" => 'commaDecimal' ]; |
| 1001 | } |
| 1002 | ] |
| 1003 | ], |
| 1004 | 'pu' => [ |
| 1005 | "transitions" => static::mhchemCreateTransitions( [ |
| 1006 | 'empty' => [ |
| 1007 | '*' => [ "action_" => 'output' ] |
| 1008 | ], |
| 1009 | 'space$' => [ |
| 1010 | '*' => [ "action_" => [ 'output', 'space' ] ] |
| 1011 | ], |
| 1012 | '{[(|)]}' => [ |
| 1013 | '0|a' => [ "action_" => 'copy' ] |
| 1014 | ], |
| 1015 | '(-)(9)^(-9)' => [ |
| 1016 | '0' => [ "action_" => 'number^', "nextState" => 'a' ] |
| 1017 | ], |
| 1018 | '(-)(9.,9)(e)(99)' => [ |
| 1019 | '0' => [ "action_" => 'enumber', "nextState" => 'a' ] |
| 1020 | ], |
| 1021 | 'space' => [ |
| 1022 | '0|a' => [ "action_" => [] ] |
| 1023 | ], |
| 1024 | 'pm-operator' => [ |
| 1025 | '0|a' => [ "action_" => [ [ "type_" => 'operator', "option" => '\\pm' ] ], "nextState" => '0' ] |
| 1026 | ], |
| 1027 | 'operator' => [ |
| 1028 | '0|a' => [ "action_" => 'copy', "nextState" => '0' ] |
| 1029 | ], |
| 1030 | '//' => [ |
| 1031 | 'd' => [ "action_" => 'o=', "nextState" => '/' ] |
| 1032 | ], |
| 1033 | '/' => [ |
| 1034 | 'd' => [ "action_" => 'o=', "nextState" => '/' ] |
| 1035 | ], |
| 1036 | '{...}|else' => [ |
| 1037 | '0|d' => [ "action_" => 'd=', "nextState" => 'd' ], |
| 1038 | 'a' => [ "action_" => [ 'space', 'd=' ], "nextState" => 'd' ], |
| 1039 | '/|q' => [ "action_" => 'q=', "nextState" => 'q' ] |
| 1040 | ] |
| 1041 | ] ), |
| 1042 | "actions" => [ |
| 1043 | 'enumber' => function ( $_buffer, $m ) { |
| 1044 | $ret = []; |
| 1045 | if ( MU::issetJS( $m[0] ?? null ) && ( $m[0] === "+-" || $m[0] === "+/-" ) ) { |
| 1046 | $ret[] = "\\pm "; |
| 1047 | } elseif ( MU::issetJS( $m[0] ?? null ) ) { |
| 1048 | $ret[] = $m[0]; |
| 1049 | } |
| 1050 | if ( MU::issetJS( $m[1] ?? null ) ) { |
| 1051 | MhchemUtil::concatArray( $ret, $this->mhchemParser->go( $m[1], 'pu-9,9' ) ); |
| 1052 | |
| 1053 | if ( MU::issetJS( ( $m[2] ?? null ) ) ) { |
| 1054 | if ( preg_match( "/[,.]/", $m[2] ) ) { |
| 1055 | MhchemUtil::concatArray( $ret, |
| 1056 | $this->mhchemParser->go( $m[2], 'pu-9,9' ) ); |
| 1057 | } else { |
| 1058 | $ret[] = $m[2]; |
| 1059 | } |
| 1060 | } |
| 1061 | |
| 1062 | if ( MU::issetJS( $m[3] ?? null ) || MU::issetJS( $m[4] ?? null ) ) { |
| 1063 | if ( $m[3] === "e" || $m[4] === "*" ) { |
| 1064 | $ret[] = [ "type_" => 'cdot' ]; |
| 1065 | } else { |
| 1066 | $ret[] = [ "type_" => 'times' ]; |
| 1067 | } |
| 1068 | } |
| 1069 | } |
| 1070 | |
| 1071 | if ( MU::issetJS( $m[5] ?? null ) ) { |
| 1072 | $ret[] = "10^{" . $m[5] . "}"; |
| 1073 | } |
| 1074 | |
| 1075 | return $ret; |
| 1076 | }, |
| 1077 | 'number^' => function ( $_buffer, $m ) { |
| 1078 | $ret = []; |
| 1079 | if ( isset( $m[0] ) && ( $m[0] === "+-" || $m[0] === "+/-" ) ) { |
| 1080 | $ret[] = "\\pm "; |
| 1081 | } elseif ( isset( $m[0] ) ) { |
| 1082 | $ret[] = $m[0]; |
| 1083 | } |
| 1084 | MhchemUtil::concatArray( $ret, |
| 1085 | $this->mhchemParser->go( $m[1] ?? null, 'pu-9,9' ) ); |
| 1086 | $ret[] = "^{" . ( $m[2] ?? "" ) . "}"; |
| 1087 | return $ret; |
| 1088 | }, |
| 1089 | 'operator' => static function ( $_buffer, $m, $p1 ) { |
| 1090 | return [ "type_" => 'operator', "kind_" => $p1 ?? $m ]; |
| 1091 | }, |
| 1092 | 'space' => static function () { |
| 1093 | return [ "type_" => 'pu-space-1' ]; |
| 1094 | }, |
| 1095 | 'output' => function ( &$buffer ) { |
| 1096 | $md = $this->mhchemParser->getPatterns()->match( '{(...)}', $buffer["d"] ?? "" ); |
| 1097 | if ( $md && $md["remainder"] === '' ) { |
| 1098 | $buffer["d"] = $md["match_"]; |
| 1099 | } |
| 1100 | $mq = $this->mhchemParser->getPatterns()->match( '{(...)}', $buffer["q"] ?? "" ); |
| 1101 | if ( $mq && $mq["remainder"] === '' ) { |
| 1102 | $buffer["q"] = $mq["match_"]; |
| 1103 | } |
| 1104 | if ( isset( $buffer["d"] ) ) { |
| 1105 | // tbd: g modifiers necessary in regexes ? |
| 1106 | $buffer["d"] = preg_replace( '/\x{00B0}C|\^oC|\^{o}C/u', |
| 1107 | "{}^{\\circ}C", $buffer["d"] ); |
| 1108 | $buffer["d"] = preg_replace( '/\x{00B0}F|\^oF|\^{o}F/u', |
| 1109 | "{}^{\\circ}C", $buffer["d"] ); |
| 1110 | } |
| 1111 | if ( isset( $buffer["q"] ) ) { |
| 1112 | $buffer["q"] = preg_replace( "/\x{00B0}C|\^oC|\^{o}C/u", |
| 1113 | "{}^{\\circ}C", $buffer["q"] ); |
| 1114 | $buffer["q"] = preg_replace( "/\x{00B0}F|\^oF|\^{o}F/u", |
| 1115 | "{}^{\\circ}F", $buffer["q"] ); |
| 1116 | $b5 = [ |
| 1117 | "d" => $this->mhchemParser->go( $buffer["d"] ?? "", 'pu' ), |
| 1118 | "q" => $this->mhchemParser->go( $buffer["q"], 'pu' ) |
| 1119 | ]; |
| 1120 | if ( $buffer["o"] === '//' ) { |
| 1121 | $ret = [ "type_" => 'pu-frac', "p1" => $b5["d"], "p2" => $b5["q"] ]; |
| 1122 | } else { |
| 1123 | $ret = $b5["d"]; |
| 1124 | if ( count( $b5["d"] ) > 1 || count( $b5["q"] ) > 1 ) { |
| 1125 | $ret[] = [ "type_" => ' / ' ]; |
| 1126 | } else { |
| 1127 | $ret[] = [ "type_" => '/' ]; |
| 1128 | } |
| 1129 | MhchemUtil::concatArray( $ret, $b5["q"] ); |
| 1130 | } |
| 1131 | } else { |
| 1132 | $ret = $this->mhchemParser->go( $buffer["d"] ?? null, 'pu-2' ); |
| 1133 | } |
| 1134 | |
| 1135 | foreach ( $buffer as $key => $value ) { |
| 1136 | unset( $buffer[$key] ); |
| 1137 | } |
| 1138 | return $ret; |
| 1139 | } |
| 1140 | ], |
| 1141 | ], |
| 1142 | 'pu-2' => [ |
| 1143 | "transitions" => static::mhchemCreateTransitions( [ |
| 1144 | 'empty' => [ |
| 1145 | '*' => [ "action_" => 'output' ], |
| 1146 | ], |
| 1147 | '*' => [ |
| 1148 | '*' => [ "action_" => [ 'output', 'cdot' ], "nextState" => '0' ], |
| 1149 | ], |
| 1150 | '\\x' => [ '*' => [ "action_" => 'rm=' ] ], |
| 1151 | 'space' => [ '*' => [ "action_" => [ 'output', 'space' ], "nextState" => '0' ] ], |
| 1152 | |
| 1153 | '^{(...)}|^(-1)' => [ '1' => [ "action_" => '^(-1)' ] ], |
| 1154 | '-9.,9' => [ |
| 1155 | '0' => [ "action_" => 'rm=', "nextState" => '0' ], |
| 1156 | '1' => [ "action_" => '^(-1)', "nextState" => '0' ] |
| 1157 | ], |
| 1158 | '{...}|else' => [ '*' => [ "action_" => 'rm=', "nextState" => '1' ] ] |
| 1159 | ] ), |
| 1160 | "actions" => [ |
| 1161 | 'cdot' => static function () { |
| 1162 | return [ "type_" => 'tight cdot' ]; |
| 1163 | }, |
| 1164 | '^(-1)' => static function ( &$buffer, $m ) { |
| 1165 | $buffer["rm"] .= "^{" . $m . "}"; |
| 1166 | return null; |
| 1167 | }, |
| 1168 | 'space' => static function () { |
| 1169 | return [ "type_" => 'pu-space-2' ]; |
| 1170 | }, |
| 1171 | |
| 1172 | 'output' => function ( &$buffer ){ |
| 1173 | $ret = []; |
| 1174 | if ( MU::issetJS( $buffer["rm"] ) ) { |
| 1175 | $mrm = $this->mhchemParser->getPatterns()->match( '{(...)}', $buffer["rm"] ?? "" ); |
| 1176 | if ( isset( $mrm["remainder"] ) && $mrm["remainder"] === '' ) { |
| 1177 | $ret = $this->mhchemParser->go( $mrm["match_"], 'pu' ); |
| 1178 | } else { |
| 1179 | $ret = [ "type_" => 'rm', "p1" => $buffer["rm"] ]; |
| 1180 | } |
| 1181 | } |
| 1182 | foreach ( $buffer as $key => $value ) { |
| 1183 | unset( $buffer[$key] ); |
| 1184 | } |
| 1185 | return $ret; |
| 1186 | } |
| 1187 | ] |
| 1188 | ], |
| 1189 | 'pu-9,9' => [ |
| 1190 | "transitions" => static::mhchemCreateTransitions( [ |
| 1191 | 'empty' => [ |
| 1192 | '0' => [ "action_" => 'output-0' ], |
| 1193 | 'o' => [ "action_" => 'output-o' ] |
| 1194 | ], |
| 1195 | ',' => [ |
| 1196 | '0' => [ "action_" => [ 'output-0', 'comma' ], "nextState" => 'o' ] |
| 1197 | ], |
| 1198 | '.' => [ |
| 1199 | '0' => [ "action_" => [ 'output-0', 'copy' ], "nextState" => 'o' ] |
| 1200 | ], |
| 1201 | 'else' => [ '*' => [ "action_" => 'text=' ] ] |
| 1202 | ] ), |
| 1203 | "actions" => [ |
| 1204 | 'comma' => static function () { |
| 1205 | return [ "type_" => 'commaDecimal' ]; |
| 1206 | }, |
| 1207 | 'output-0' => static function ( &$buffer ) { |
| 1208 | $ret = []; |
| 1209 | $buffer["text_"] ??= ""; |
| 1210 | if ( strlen( $buffer["text_"] ) > 4 ) { |
| 1211 | $a = strlen( $buffer["text_"] ) % 3; |
| 1212 | if ( $a === 0 ) { |
| 1213 | $a = 3; |
| 1214 | } |
| 1215 | for ( $i = strlen( $buffer["text_"] ) - 3; $i > 0; $i -= 3 ) { |
| 1216 | $ret[] = substr( $buffer["text_"], $i, 3 ); |
| 1217 | $ret[] = [ "type_" => '1000 separator' ]; |
| 1218 | } |
| 1219 | $ret[] = substr( $buffer["text_"], 0, $a ); |
| 1220 | $ret = array_reverse( $ret ); |
| 1221 | } else { |
| 1222 | $ret[] = $buffer["text_"]; |
| 1223 | } |
| 1224 | foreach ( $buffer as $key => $value ) { |
| 1225 | unset( $buffer[$key] ); |
| 1226 | } |
| 1227 | return $ret; |
| 1228 | }, |
| 1229 | 'output-o' => static function ( &$buffer ) { |
| 1230 | $ret = []; |
| 1231 | $buffer["text_"] ??= ""; |
| 1232 | if ( strlen( $buffer["text_"] ) > 4 ) { |
| 1233 | $a = strlen( $buffer["text_"] ) - 3; |
| 1234 | for ( $i = 0; $i < $a; $i += 3 ) { |
| 1235 | $ret[] = substr( $buffer["text_"], $i, 3 ); |
| 1236 | $ret[] = [ "type_" => '1000 separator' ]; |
| 1237 | } |
| 1238 | $ret[] = substr( $buffer["text_"], $i ); |
| 1239 | } else { |
| 1240 | $ret[] = $buffer["text_"]; |
| 1241 | } |
| 1242 | foreach ( $buffer as $key => $value ) { |
| 1243 | unset( $buffer[$key] ); |
| 1244 | } |
| 1245 | return $ret; |
| 1246 | } |
| 1247 | ] |
| 1248 | ] |
| 1249 | |
| 1250 | ]; |
| 1251 | } |
| 1252 | |
| 1253 | } |