libraries/string.lib.php

   1 <?php
   2 /* $Id$ */
   3 // vim: expandtab sw=4 ts=4 sts=4:
   4
   5 /** Specialized String Functions for phpMyAdmin
   6  *
   7  * Copyright 2002 Robin Johnson <robbat2@users.sourceforge.net>
   8  * http://www.orbis-terrarum.net/?l=people.robbat2
   9  *
  10  * Defines a set of function callbacks that have a pure C version available if
  11  * the "ctype" extension is available, but otherwise have PHP versions to use
  12  * (that are slower).
  13  *
  14  * The SQL Parser code relies heavily on these functions.
  15  */
  16
  17 /* Try to load mbstring, unless we're using buggy php version */
  18 if (PMA_PHP_INT_VERSION != 40203) {
  19     if (!@extension_loaded('mbstring')) {
  20         PMA_dl('mbstring');
  21     }
  22 }
  23
  24 /* windows-* and tis-620 are not supported and are not multibyte,
  25  * others can be ignored as they're not multibyte */
  26 $GLOBALS['using_mb_charset'] =
  27     substr($GLOBALS['charset'], 0, 8) != 'windows-' &&
  28     substr($GLOBALS['charset'], 0, 9) != 'iso-8859-' &&
  29     substr($GLOBALS['charset'], 0, 3) != 'cp-' &&
  30     $GLOBALS['charset'] != 'koi8-r' &&
  31     $GLOBALS['charset'] != 'tis-620';
  32
  33 $GLOBALS['PMA_allow_mbstr'] = @function_exists('mb_strlen') && $GLOBALS['using_mb_charset'];
  34
  35 if ($GLOBALS['PMA_allow_mbstr']) {
  36     // the hebrew lang file uses iso-8859-8-i, encoded RTL,
  37     // but mb_internal_encoding only supports iso-8859-8
  38     if ($GLOBALS['charset'] == 'iso-8859-8-i'){
  39         mb_internal_encoding('iso-8859-8');
  40     } else {
  41         mb_internal_encoding($GLOBALS['charset']);
  42     }
  43 }
  44
  45 // This is for handling input better
  46 if (defined('PMA_MULTIBYTE_ENCODING') || $GLOBALS['PMA_allow_mbstr']) {
  47     $GLOBALS['PMA_strpos']  = 'mb_strpos';
  48     $GLOBALS['PMA_strrpos'] = 'mb_strrpos';
  49 } else {
  50     $GLOBALS['PMA_strpos']  = 'strpos';
  51     $GLOBALS['PMA_strrpos'] = 'strrpos';
  52 }
  53
  54 /**
  55  * Returns length of string depending on current charset.
  56  *
  57  * @param   string   string to count
  58  *
  59  * @return  int      string length
  60  *
  61  * @access  public
  62  *
  63  * @author  nijel
  64  */
  65 function PMA_strlen($string)
  66 {
  67     // windows-* charsets are not multibyte and not supported by mb_*
  68     if (defined('PMA_MULTIBYTE_ENCODING') || $GLOBALS['PMA_allow_mbstr']) {
  69         return mb_strlen($string);
  70     } else {
  71         return strlen($string);
  72     }
  73 }
  74
  75 /**
  76  * Returns substring from string, works depending on current charset.
  77  *
  78  * @param   string   string to count
  79  * @param   int      start of substring
  80  * @param   int      length of substring
  81  *
  82  * @return  int      substring
  83  *
  84  * @access  public
  85  *
  86  * @author  nijel
  87  */
  88 function PMA_substr($string, $start, $length = 2147483647)
  89 {
  90     if (defined('PMA_MULTIBYTE_ENCODING') || $GLOBALS['PMA_allow_mbstr']) {
  91         return mb_substr($string, $start, $length);
  92     } else {
  93         return substr($string, $start, $length);
  94     }
  95 }
  96
  97
  98 /**
  99  * This checks if a string actually exists inside another string
 100  * We try to do it in a PHP3-portable way.
 101  * We don't care about the position it is in.
 102  *
 103  * @param   string   string to search for
 104  * @param   string   string to search in
 105  *
 106  * @return  boolean  whether the needle is in the haystack or not
 107  */
 108 function PMA_STR_strInStr($needle, $haystack)
 109 {
 110     // $GLOBALS['PMA_strpos']($haystack, $needle) !== FALSE
 111     // return (is_integer($GLOBALS['PMA_strpos']($haystack, $needle)));
 112     return $GLOBALS['PMA_strpos'](' ' . $haystack, $needle);
 113 } // end of the "PMA_STR_strInStr()" function
 114
 115
 116 /**
 117  * Checks if a given character position in the string is escaped or not
 118  *
 119  * @param   string   string to check for
 120  * @param   integer  the character to check for
 121  * @param   integer  starting position in the string
 122  *
 123  * @return  boolean  whether the character is escaped or not
 124  */
 125 function PMA_STR_charIsEscaped($string, $pos, $start = 0)
 126 {
 127     $len = PMA_strlen($string);
 128     // Base case:
 129     // Check for string length or invalid input or special case of input
 130     // (pos == $start)
 131     if (($pos == $start) || ($len <= $pos)) {
 132         return FALSE;
 133     }
 134
 135     $p           = $pos - 1;
 136     $escaped     = FALSE;
 137     while (($p >= $start) && (PMA_substr($string, $p, 1) == '\\')) {
 138         $escaped = !$escaped;
 139         $p--;
 140     } // end while
 141
 142     if ($pos < $start) {
 143         // throw error about strings
 144     }
 145
 146     return $escaped;
 147 } // end of the "PMA_STR_charIsEscaped()" function
 148
 149
 150 /**
 151  * Checks if a number is in a range
 152  *
 153  * @param   integer  number to check for
 154  * @param   integer  lower bound
 155  * @param   integer  upper bound
 156  *
 157  * @return  boolean  whether the number is in the range or not
 158  */
 159 function PMA_STR_numberInRangeInclusive($num, $lower, $upper)
 160 {
 161     return (($num >= $lower) && ($num <= $upper));
 162 } // end of the "PMA_STR_numberInRangeInclusive()" function
 163
 164
 165 /**
 166  * Checks if a character is a digit
 167  *
 168  * @param   string   character to check for
 169  *
 170  * @return  boolean  whether the character is a digit or not
 171  *
 172  * @see     PMA_STR_numberInRangeInclusive()
 173  */
 174 function PMA_STR_isDigit($c)
 175 {
 176     $ord_zero = 48; //ord('0');
 177     $ord_nine = 57; //ord('9');
 178     $ord_c    = ord($c);
 179
 180     return PMA_STR_numberInRangeInclusive($ord_c, $ord_zero, $ord_nine);
 181 } // end of the "PMA_STR_isDigit()" function
 182
 183
 184 /**
 185  * Checks if a character is an hexadecimal digit
 186  *
 187  * @param   string   character to check for
 188  *
 189  * @return  boolean  whether the character is an hexadecimal digit or not
 190  *
 191  * @see     PMA_STR_numberInRangeInclusive()
 192  */
 193 function PMA_STR_isHexDigit($c)
 194 {
 195     $ord_Aupper = 65;  //ord('A');
 196     $ord_Fupper = 70;  //ord('F');
 197     $ord_Alower = 97;  //ord('a');
 198     $ord_Flower = 102; //ord('f');
 199     $ord_zero   = 48;  //ord('0');
 200     $ord_nine   = 57;  //ord('9');
 201     $ord_c      = ord($c);
 202
 203     return (PMA_STR_numberInRangeInclusive($ord_c, $ord_zero, $ord_nine)
 204             || PMA_STR_numberInRangeInclusive($ord_c, $ord_Aupper, $ord_Fupper)
 205             || PMA_STR_numberInRangeInclusive($ord_c, $ord_Alower, $ord_Flower));
 206 } // end of the "PMA_STR_isHexDigit()" function
 207
 208
 209 /**
 210  * Checks if a character is an upper alphabetic one
 211  *
 212  * @param   string   character to check for
 213  *
 214  * @return  boolean  whether the character is an upper alphabetic one or
 215  *                   not
 216  *
 217  * @see     PMA_STR_numberInRangeInclusive()
 218  */
 219 function PMA_STR_isUpper($c)
 220 {
 221     $ord_zero = 65; //ord('A');
 222     $ord_nine = 90; //ord('Z');
 223     $ord_c    = ord($c);
 224
 225     return PMA_STR_numberInRangeInclusive($ord_c, $ord_zero, $ord_nine);
 226 } // end of the "PMA_STR_isUpper()" function
 227
 228
 229 /**
 230  * Checks if a character is a lower alphabetic one
 231  *
 232  * @param   string   character to check for
 233  *
 234  * @return  boolean  whether the character is a lower alphabetic one or
 235  *                   not
 236  *
 237  * @see     PMA_STR_numberInRangeInclusive()
 238  */
 239 function PMA_STR_isLower($c)
 240 {
 241     $ord_zero = 97;  //ord('a');
 242     $ord_nine = 122; //ord('z');
 243     $ord_c    = ord($c);
 244
 245     return PMA_STR_numberInRangeInclusive($ord_c, $ord_zero, $ord_nine);
 246 } // end of the "PMA_STR_isLower()" function
 247
 248
 249 /**
 250  * Checks if a character is an alphabetic one
 251  *
 252  * @param   string   character to check for
 253  *
 254  * @return  boolean  whether the character is an alphabetic one or not
 255  *
 256  * @see     PMA_STR_isUpper()
 257  * @see     PMA_STR_isLower()
 258  */
 259 function PMA_STR_isAlpha($c)
 260 {
 261     return (PMA_STR_isUpper($c) || PMA_STR_isLower($c));
 262 } // end of the "PMA_STR_isAlpha()" function
 263
 264
 265 /**
 266  * Checks if a character is an alphanumeric one
 267  *
 268  * @param   string   character to check for
 269  *
 270  * @return  boolean  whether the character is an alphanumeric one or not
 271  *
 272  * @see     PMA_STR_isUpper()
 273  * @see     PMA_STR_isLower()
 274  * @see     PMA_STR_isDigit()
 275  */
 276 function PMA_STR_isAlnum($c)
 277 {
 278     return (PMA_STR_isUpper($c) || PMA_STR_isLower($c) || PMA_STR_isDigit($c));
 279 } // end of the "PMA_STR_isAlnum()" function
 280
 281
 282 /**
 283  * Checks if a character is a space one
 284  *
 285  * @param   string   character to check for
 286  *
 287  * @return  boolean  whether the character is a space one or not
 288  *
 289  * @see     PMA_STR_numberInRangeInclusive()
 290  */
 291 function PMA_STR_isSpace($c)
 292 {
 293     $ord_space = 32; //ord(' ')
 294     $ord_tab   = 9; //ord('\t')
 295     $ord_CR    = 13; //ord('\n')
 296     $ord_NOBR  = 160; //ord('U+00A0);
 297     $ord_c     = ord($c);
 298
 299     return (($ord_c == $ord_space)
 300             || ($ord_c == $ord_NOBR)
 301             || PMA_STR_numberInRangeInclusive($ord_c, $ord_tab, $ord_CR));
 302 } // end of the "PMA_STR_isSpace()" function
 303
 304
 305 /**
 306  * Checks if a character is an accented character
 307  *
 308  * @note    Presently this only works for some character sets. More work
 309  *          may be needed to fix it.
 310  *
 311  * @param   string   character to check for
 312  *
 313  * @return  boolean  whether the character is an accented one or not
 314  *
 315  * @see     PMA_STR_numberInRangeInclusive()
 316  */
 317 function PMA_STR_isAccented($c)
 318 {
 319     $ord_min1 = 192; //ord('A');
 320     $ord_max1 = 214; //ord('Z');
 321     $ord_min2 = 216; //ord('A');
 322     $ord_max2 = 246; //ord('Z');
 323     $ord_min3 = 248; //ord('A');
 324     $ord_max3 = 255; //ord('Z');
 325
 326     $ord_c    = ord($c);
 327
 328     return PMA_STR_numberInRangeInclusive($ord_c, $ord_min1, $ord_max1)
 329            || PMA_STR_numberInRangeInclusive($ord_c, $ord_min2, $ord_max2)
 330            || PMA_STR_numberInRangeInclusive($ord_c, $ord_min2, $ord_max2);
 331 } // end of the "PMA_STR_isAccented()" function
 332
 333
 334 /**
 335  * Checks if a character is an SQL identifier
 336  *
 337  * @param   string   character to check for
 338  * @param   boolean  whether the dot character is valid or not
 339  *
 340  * @return  boolean  whether the character is an SQL identifier or not
 341  *
 342  * @see     PMA_STR_isAlnum()
 343  */
 344 function PMA_STR_isSqlIdentifier($c, $dot_is_valid = FALSE)
 345 {
 346     return (PMA_STR_isAlnum($c)
 347             || PMA_STR_isAccented($c)
 348             || ($c == '_') || ($c == '$')
 349             || (($dot_is_valid != FALSE) && ($c == '.')));
 350 } // end of the "PMA_STR_isSqlIdentifier()" function
 351
 352
 353 /**
 354  * Binary search of a value in a sorted array
 355  *
 356  * @param   string   string to search for
 357  * @param   array    sorted array to search into
 358  * @param   integer  size of sorted array to search into
 359  *
 360  * @return  boolean  whether the string has been found or not
 361  */
 362 function PMA_STR_binarySearchInArr($str, $arr, $arrsize)
 363 {
 364     // $arr MUST be sorted, due to binary search
 365     $top    = $arrsize - 1;
 366     $bottom = 0;
 367     $found  = FALSE;
 368
 369     while (($top >= $bottom) && ($found == FALSE)) {
 370         $mid        = intval(($top + $bottom) / 2);
 371         $res        = strcmp($str, $arr[$mid]);
 372         if ($res == 0) {
 373             $found  = TRUE;
 374         } elseif ($res < 0) {
 375             $top    = $mid - 1;
 376         } else {
 377             $bottom = $mid + 1;
 378         }
 379     } // end while
 380
 381     return $found;
 382 } // end of the "PMA_STR_binarySearchInArr()" function
 383
 384 ?>