Use charset aware substr and strlen functions (bugs #854755 and #910575).

ChangeLog

@@ -10,6 +10,11 @@ $Source$
       conversion.
     * libraries/dbi/mysql.dbi.lib.php, libraries/dbi/mysqli.dbi.lib.php: Fix
       PMA_mysql_fetch_array return value for last result (bug #905223).
+    * browse_foreigners.php, libraries/common.lib.php,
+      libraries/display_tbl.lib.php, libraries/relation.lib.php,
+      libraries/sqlparser.lib.php, libraries/string.lib.php,
+      libraries/transformations/text_plain__substr.inc.php: Use charset aware
+      substr and strlen functions (bugs #854755 and #910575).
 
 2004-03-05 Marc Delisle  <lem9@users.sourceforge.net>
     * libraries/sqlparser.lib.php: bug 909752, floating point digit

browse_foreigners.php

@@ -186,7 +186,7 @@ if (isset($disp_row) && is_array($disp_row)) {
             $vtitle = '';
         } else {
             $vtitle = htmlspecialchars($val);
-            $value  = htmlspecialchars(substr($val, 0, $cfg['LimitChars']) . '...');
+            $value  = htmlspecialchars(PMA_substr($val, 0, $cfg['LimitChars']) . '...');
         }
 
         $key_equals_data = isset($data) && $key == $data;

libraries/common.lib.php

@@ -251,9 +251,7 @@ if ($is_minimum_common == FALSE) {
      * Charset conversion.
      */
     require_once('./libraries/charset_conversion.lib.php');
-}
 
-if ($is_minimum_common == FALSE) {
     /**
      * String handling
      */

libraries/display_tbl.lib.php

@@ -1362,8 +1362,8 @@ function PMA_displayTableBody(&$dt_result, &$is_display, $map, $analyzed_sql)
                         $vertical_display['data'][$row_no][$i] = '    <td valign="top" ' . $column_style . ' bgcolor="' . $bgcolor . '"><i>NULL</i></td>' . "\n";
                     } else if ($row[$pointer] != '') {
                         // garvin: if a transform function for blob is set, none of these replacements will be made
-                        if (strlen($row[$pointer]) > $GLOBALS['cfg']['LimitChars'] && ($dontlimitchars != 1)) {
+                        if (PMA_strlen($row[$pointer]) > $GLOBALS['cfg']['LimitChars'] && ($dontlimitchars != 1)) {
-                            $row[$pointer] = substr($row[$pointer], 0, $GLOBALS['cfg']['LimitChars']) . '...';
+                            $row[$pointer] = PMA_substr($row[$pointer], 0, $GLOBALS['cfg']['LimitChars']) . '...';
                         }
                         // loic1: displays all space characters, 4 space
                         // characters for tabulations and <cr>/<lf>
@@ -1384,8 +1384,8 @@ function PMA_displayTableBody(&$dt_result, &$is_display, $map, $analyzed_sql)
                     $relation_id = $row[$pointer];
 
                     // nijel: Cut all fields to $cfg['LimitChars']
-                    if (strlen($row[$pointer]) > $GLOBALS['cfg']['LimitChars'] && ($dontlimitchars != 1)) {
+                    if (PMA_strlen($row[$pointer]) > $GLOBALS['cfg']['LimitChars'] && ($dontlimitchars != 1)) {
-                        $row[$pointer] = substr($row[$pointer], 0, $GLOBALS['cfg']['LimitChars']) . '...';
+                        $row[$pointer] = PMA_substr($row[$pointer], 0, $GLOBALS['cfg']['LimitChars']) . '...';
                     }
 
                     // loic1: displays special characters from binaries

libraries/relation.lib.php

@@ -604,7 +604,7 @@ function PMA_foreignDropdown($disp, $foreign_field, $foreign_display, $data, $ma
     
     foreach($disp AS $disp_key => $relrow) {
         $key   = $relrow[$foreign_field];
-        if (strlen($relrow[$foreign_display]) <= $cfg['LimitChars']) {
+        if (PMA_strlen($relrow[$foreign_display]) <= $cfg['LimitChars']) {
             $value  = (($foreign_display != FALSE) ? htmlspecialchars($relrow[$foreign_display]) : '');
             $vtitle = '';
         } else {

libraries/sqlparser.lib.php

@@ -211,7 +211,7 @@ if ($is_minimum_common == FALSE) {
         $sql = str_replace("\r\n", "\n", $sql);
         $sql = str_replace("\r", "\n", $sql);
 
-        $len = $GLOBALS['PMA_strlen']($sql);
+        $len = PMA_strlen($sql);
         if ($len == 0) {
             return array();
         }
@@ -288,7 +288,7 @@ if ($is_minimum_common == FALSE) {
                         break;
                 } // end switch
                 $count2 = ($pos < $count2) ? $len : $pos;
-                $str    = $GLOBALS['PMA_substr']($sql, $count1, $count2 - $count1);
+                $str    = PMA_substr($sql, $count1, $count2 - $count1);
                 PMA_SQP_arrayAdd($sql_array, 'comment_' . $type, $str, $arraysize);
                 continue;
             } // end if
@@ -348,7 +348,7 @@ if ($is_minimum_common == FALSE) {
                     default:
                         break;
                 } // end switch
-                $data = $GLOBALS['PMA_substr']($sql, $count1, $count2 - $count1);
+                $data = PMA_substr($sql, $count1, $count2 - $count1);
                 PMA_SQP_arrayAdd($sql_array, $type, $data, $arraysize);
                 continue;
             }
@@ -387,7 +387,7 @@ if ($is_minimum_common == FALSE) {
                 if ($l == 1) {
                     $punct_data = $c;
                 } else {
-                    $punct_data = $GLOBALS['PMA_substr']($sql, $count1, $l);
+                    $punct_data = PMA_substr($sql, $count1, $l);
                 }
 
                 // Special case, sometimes, althought two characters are
@@ -424,10 +424,10 @@ if ($is_minimum_common == FALSE) {
                         $punct_data = $first;
                     } else if (($last2 == '/*') || ($last2 == '--')) {
                         $count2     -= 2;
-                        $punct_data = $GLOBALS['PMA_substr']($sql, $count1, $count2 - $count1);
+                        $punct_data = PMA_substr($sql, $count1, $count2 - $count1);
                     } else if (($last == '-') || ($last == '+') || ($last == '!')) {
                         $count2--;
-                        $punct_data = $GLOBALS['PMA_substr']($sql, $count1, $count2 - $count1);
+                        $punct_data = PMA_substr($sql, $count1, $count2 - $count1);
                     // TODO: for negation operator, split in 2 tokens ?
                     // "select x&~1 from t"
                     // becomes "select x & ~ 1 from t" ?
@@ -475,7 +475,7 @@ if ($is_minimum_common == FALSE) {
                             continue;
                         } else {
                             $debugstr = $GLOBALS['strSQPBugInvalidIdentifer'] . ' @ ' . ($count1+1) . "\n"
-                                      . 'STR: ' . $GLOBALS['PMA_substr']($sql, $count1, $count2 - $count1);
+                                      . 'STR: ' . PMA_substr($sql, $count1, $count2 - $count1);
                             PMA_SQP_throwError($debugstr, $sql);
                             return $sql;
                         }
@@ -503,7 +503,7 @@ if ($is_minimum_common == FALSE) {
                 } // end while
 
                 $l    = $count2 - $count1;
-                $str  = $GLOBALS['PMA_substr']($sql, $count1, $l);
+                $str  = PMA_substr($sql, $count1, $l);
 
                 $type = '';
                 if ($is_digit) {
@@ -532,7 +532,7 @@ if ($is_minimum_common == FALSE) {
             $count2++;
 
             $debugstr = 'C1 C2 LEN: ' . $count1 . ' ' . $count2 . ' ' . $len .  "\n"
-                      . 'STR: ' . $GLOBALS['PMA_substr']($sql, $count1, $count2 - $count1) . "\n";
+                      . 'STR: ' . PMA_substr($sql, $count1, $count2 - $count1) . "\n";
             PMA_SQP_bug($debugstr, $sql);
             return $sql;
 
@@ -1567,7 +1567,7 @@ if ($is_minimum_common == FALSE) {
         $i         = $GLOBALS['PMA_strpos']($arr['type'], '_');
         $class     = '';
         if ($i > 0) {
-            $class = 'syntax_' . $GLOBALS['PMA_substr']($arr['type'], 0, $i) . ' ';
+            $class = 'syntax_' . PMA_substr($arr['type'], 0, $i) . ' ';
         }
 
         $class     .= 'syntax_' . $arr['type'];

libraries/string.lib.php

@@ -15,18 +15,61 @@
  * The SQL Parser code relies heavily on these functions.
  */
 
-
 // This is for handling input better
 if (defined('PMA_MULTIBYTE_ENCODING')) {
-    $GLOBALS['PMA_strlen']  = 'mb_strlen';
     $GLOBALS['PMA_strpos']  = 'mb_strpos';
     $GLOBALS['PMA_strrpos'] = 'mb_strrpos';
-    $GLOBALS['PMA_substr']  = 'mb_substr';
 } else {
-    $GLOBALS['PMA_strlen']  = 'strlen';
     $GLOBALS['PMA_strpos']  = 'strpos';
     $GLOBALS['PMA_strrpos'] = 'strrpos';
-    $GLOBALS['PMA_substr']  = 'substr';
+}
+
+$GLOBALS['PMA_has_mbstr'] = @function_exists('mb_strlen');
+
+/**
+ * Returns length of string depending on current charset.
+ *
+ * @param   string   string to count
+ *
+ * @return  int      string length
+ *
+ * @access  public
+ *
+ * @author  nijel
+ */
+function PMA_strlen($string)
+{
+    if (defined('PMA_MULTIBYTE_ENCODING')) {
+        return mb_strlen($string);
+    } elseif($GLOBALS['PMA_has_mbstr']) {
+        return mb_strlen($string, $GLOBALS['charset']);
+    } else {
+        return strlen($string);
+    }
+}
+
+/**
+ * Returns substring from string, works depending on current charset.
+ *
+ * @param   string   string to count
+ * @param   int      start of substring
+ * @param   int      length of substring
+ *
+ * @return  int      string length
+ *
+ * @access  public
+ *
+ * @author  nijel
+ */
+function PMA_substr($string, $start, $length = 2147483647)
+{
+    if (defined('PMA_MULTIBYTE_ENCODING')) {
+        return mb_substr($string, $start, $length);
+    } elseif($GLOBALS['PMA_has_mbstr']) {
+        return mb_substr($string, $start, $length, $GLOBALS['charset']);
+    } else {
+        return substr($string, $start, $length);
+    }
 }
 
 

libraries/transformations/text_plain__substr.inc.php

@@ -21,9 +21,9 @@ function PMA_transformation_text_plain__substr($buffer, $options = array(), $met
 
     $newtext = '';
     if ($options[1] != 'all') {
-        $newtext = substr($buffer, $options[0], $options[1]);
+        $newtext = PMA_substr($buffer, $options[0], $options[1]);
     } else {
-        $newtext = substr($buffer, $options[0]);
+        $newtext = PMA_substr($buffer, $options[0]);
     }
 
     $length = strlen($newtext);