extensions/LocalisationUpdate/QuickArrayReader.php

   1 <?php
   2
   3 /**
   4  * Quickie parser class that can happily read the subset of PHP we need
   5  * for our localization arrays safely.
   6  *
   7  * Still an order of magnitude slower than eval().
   8  */
   9 class QuickArrayReader {
  10         private $vars = [];
  11
  12         /**
  13          * @param $string string
  14          */
  15         function __construct( $string ) {
  16                 $scalarTypes = [
  17                         T_LNUMBER => true,
  18                         T_DNUMBER => true,
  19                         T_STRING => true,
  20                         T_CONSTANT_ENCAPSED_STRING => true,
  21                 ];
  22                 $skipTypes = [
  23                         T_WHITESPACE => true,
  24                         T_COMMENT => true,
  25                         T_DOC_COMMENT => true,
  26                 ];
  27                 $tokens = token_get_all( $string );
  28                 $count = count( $tokens );
  29                 for ( $i = 0; $i < $count; ) {
  30                         while ( isset( $skipTypes[$tokens[$i][0]] ) ) {
  31                                 $i++;
  32                         }
  33                         switch ( $tokens[$i][0] ) {
  34                                 case T_OPEN_TAG:
  35                                         $i++;
  36                                         continue;
  37                                 case T_VARIABLE:
  38                                         // '$messages' -> 'messages'
  39                                         $varname = trim( substr( $tokens[$i][1], 1 ) );
  40                                         $varindex = null;
  41
  42                                         while ( isset( $skipTypes[$tokens[++$i][0]] ) ) {
  43                                         }
  44
  45                                         if ( $tokens[$i] === '[' ) {
  46                                                 while ( isset( $skipTypes[$tokens[++$i][0]] ) ) {
  47                                                 }
  48
  49                                                 if ( isset( $scalarTypes[$tokens[$i][0]] ) ) {
  50                                                         $varindex = $this->parseScalar( $tokens[$i] );
  51                                                 } else {
  52                                                         throw $this->except( $tokens[$i], 'scalar index' );
  53                                                 }
  54                                                 while ( isset( $skipTypes[$tokens[++$i][0]] ) ) {
  55                                                 }
  56
  57                                                 if ( $tokens[$i] !== ']' ) {
  58                                                         throw $this->except( $tokens[$i], ']' );
  59                                                 }
  60                                                 while ( isset( $skipTypes[$tokens[++$i][0]] ) ) {
  61                                                 }
  62                                         }
  63
  64                                         if ( $tokens[$i] !== '=' ) {
  65                                                 throw $this->except( $tokens[$i], '=' );
  66                                         }
  67                                         while ( isset( $skipTypes[$tokens[++$i][0]] ) ) {
  68                                         }
  69
  70                                         if ( isset( $scalarTypes[$tokens[$i][0]] ) ) {
  71                                                 $buildval = $this->parseScalar( $tokens[$i] );
  72                                         } elseif ( $tokens[$i][0] === T_ARRAY ) {
  73                                                 while ( isset( $skipTypes[$tokens[++$i][0]] ) ) {
  74                                                 }
  75                                                 if ( $tokens[$i] !== '(' ) {
  76                                                         throw $this->except( $tokens[$i], '(' );
  77                                                 }
  78                                                 $buildval = [];
  79                                                 do {
  80                                                         while ( isset( $skipTypes[$tokens[++$i][0]] ) ) {
  81                                                         }
  82
  83                                                         if ( $tokens[$i] === ')' ) {
  84                                                                 break;
  85                                                         }
  86                                                         if ( isset( $scalarTypes[$tokens[$i][0]] ) ) {
  87                                                                 $key = $this->parseScalar( $tokens[$i] );
  88                                                         }
  89                                                         while ( isset( $skipTypes[$tokens[++$i][0]] ) ) {
  90                                                         }
  91
  92                                                         if ( $tokens[$i][0] !== T_DOUBLE_ARROW ) {
  93                                                                 throw $this->except( $tokens[$i], '=>' );
  94                                                         }
  95                                                         while ( isset( $skipTypes[$tokens[++$i][0]] ) ) {
  96                                                         }
  97
  98                                                         if ( isset( $scalarTypes[$tokens[$i][0]] ) ) {
  99                                                                 $val = $this->parseScalar( $tokens[$i] );
 100                                                         }
 101                                                         wfSuppressWarnings();
 102                                                         $buildval[$key] = $val;
 103                                                         wfRestoreWarnings();
 104                                                         while ( isset( $skipTypes[$tokens[++$i][0]] ) ) {
 105                                                         }
 106
 107                                                         if ( $tokens[$i] === ',' ) {
 108                                                                 continue;
 109                                                         } elseif ( $tokens[$i] === ')' ) {
 110                                                                 break;
 111                                                         } else {
 112                                                                 throw $this->except( $tokens[$i], ', or )' );
 113                                                         }
 114                                                 } while ( true );
 115                                         } else {
 116                                                 throw $this->except( $tokens[$i], 'scalar or array' );
 117                                         }
 118                                         if ( is_null( $varindex ) ) {
 119                                                 $this->vars[$varname] = $buildval;
 120                                         } else {
 121                                                 wfSuppressWarnings();
 122                                                 $this->vars[$varname][$varindex] = $buildval;
 123                                                 wfRestoreWarnings();
 124                                         }
 125                                         while ( isset( $skipTypes[$tokens[++$i][0]] ) ) {
 126                                         }
 127                                         if ( $tokens[$i] !== ';' ) {
 128                                                 throw $this->except( $tokens[$i], ';' );
 129                                         }
 130                                         $i++;
 131                                         break;
 132                                 default:
 133                                         throw $this->except( $tokens[$i], 'open tag, whitespace, or variable.' );
 134                         }
 135                 }
 136         }
 137
 138         /**
 139          * @param $got string
 140          * @param $expected string
 141          * @return Exception
 142          */
 143         private function except( $got, $expected ) {
 144                 if ( is_array( $got ) ) {
 145                         $got = token_name( $got[0] ) . " ('" . $got[1] . "')";
 146                 } else {
 147                         $got = "'" . $got . "'";
 148                 }
 149
 150                 return new Exception( "Expected $expected, got $got" );
 151         }
 152
 153         /**
 154          * Parse a scalar value in PHP
 155          *
 156          * @param $token string
 157          *
 158          * @return mixed Parsed value
 159          */
 160         function parseScalar( $token ) {
 161                 if ( is_array( $token ) ) {
 162                         $str = $token[1];
 163                 } else {
 164                         $str = $token;
 165                 }
 166                 if ( $str !== '' && $str[0] == '\'' ) {
 167                         // Single-quoted string
 168                         // @fixme trim() call is due to mystery bug where whitespace gets
 169                         // appended to the token; without it we ended up reading in the
 170                         // extra quote on the end!
 171                         return strtr( substr( trim( $str ), 1, -1 ),
 172                                 [ '\\\'' => '\'', '\\\\' => '\\' ] );
 173                 }
 174
 175                 wfSuppressWarnings();
 176                 if ( $str !== '' && $str[0] == '"' ) {
 177                         // Double-quoted string
 178                         // @fixme trim() call is due to mystery bug where whitespace gets
 179                         // appended to the token; without it we ended up reading in the
 180                         // extra quote on the end!
 181                         wfRestoreWarnings();
 182                         return stripcslashes( substr( trim( $str ), 1, -1 ) );
 183                 }
 184                 wfRestoreWarnings();
 185
 186                 if ( substr( $str, 0, 4 ) === 'true' ) {
 187                         return true;
 188                 }
 189
 190                 if ( substr( $str, 0, 5 ) === 'false' ) {
 191                         return false;
 192                 }
 193
 194                 if ( substr( $str, 0, 4 ) === 'null' ) {
 195                         return null;
 196                 }
 197
 198                 // Must be some kind of numeric value, so let PHP's weak typing
 199                 // be useful for a change
 200                 return $str;
 201         }
 202
 203         /**
 204          * @param $varname string
 205          * @return null|string
 206          */
 207         function getVar( $varname ) {
 208                 if ( isset( $this->vars[$varname] ) ) {
 209                         return $this->vars[$varname];
 210                 } else {
 211                         return null;
 212                 }
 213         }
 214 }