sqlparser.lib.php

一个用PHP编写的

                    } elseif (($pos + 1 < $len) && (PMA_substr($sql, $pos, 1) == $quotetype) && (PMA_substr($sql, $pos + 1, 1) == $quotetype)) {
                        $pos = $pos + 2;
                        continue;
                    } else {
                        break;
                    }
                } while ($len > $pos); // end do

                $count2       = $pos;
                $count2++;
                $type         = 'quote_';
                switch ($quotetype) {
                    case '\'':
                        $type .= 'single';
                        break;
                    case '"':
                        $type .= 'double';
                        break;
                    case '`':
                        $type .= 'backtick';
                        break;
                    default:
                        break;
                } // end switch
                $data = PMA_substr($sql, $count1, $count2 - $count1);
                PMA_SQP_arrayAdd($sql_array, $type, $data, $arraysize);
                continue;
            }

            // Checks for brackets
            if (PMA_STR_strInStr($c, $bracket_list)) {
                // All bracket tokens are only one item long
                $count2++;
                $type_type     = '';
                if (PMA_STR_strInStr($c, '([{')) {
                    $type_type = 'open';
                } else {
                    $type_type = 'close';
                }

                $type_style     = '';
                if (PMA_STR_strInStr($c, '()')) {
                    $type_style = 'round';
                } elseif (PMA_STR_strInStr($c, '[]')) {
                    $type_style = 'square';
                } else {
                    $type_style = 'curly';
                }

                $type = 'punct_bracket_' . $type_type . '_' . $type_style;
                PMA_SQP_arrayAdd($sql_array, $type, $c, $arraysize);
                continue;
            }

            // Checks for identifier (alpha or numeric)
            if (PMA_STR_isSqlIdentifier($c, FALSE) || ($c == '@') || ($c == '.' && PMA_STR_isDigit(PMA_substr($sql, $count2 + 1, 1)))) {
                $count2 ++;

                //TODO: a @ can also be present in expressions like
                // FROM 'user'@'%'
                // or  TO 'user'@'%'
                // in this case, the @ is wrongly marked as alpha_variable

                $is_sql_variable         = ($c == '@');
                $is_digit                = (!$is_sql_variable) && PMA_STR_isDigit($c);
                $is_hex_digit            = ($is_digit) && ($c == '.') && ($c == '0') && ($count2 < $len) && (PMA_substr($sql, $count2, 1) == 'x');
                $is_float_digit          = $c == '.';
                $is_float_digit_exponent = FALSE;

                // Nijel: Fast skip is especially needed for huge BLOB data, requires PHP at least 4.3.0:
                if (PMA_PHP_INT_VERSION >= 40300) {
                    if ($is_hex_digit) {
                        $count2++;
                        $pos = strspn($sql, '0123456789abcdefABCDEF', $count2);
                        if ($pos > $count2) {
                            $count2 = $pos;
                        }
                        unset($pos);
                    } elseif ($is_digit) {
                        $pos = strspn($sql, '0123456789', $count2);
                        if ($pos > $count2) {
                            $count2 = $pos;
                        }
                        unset($pos);
                    }
                }

                while (($count2 < $len) && PMA_STR_isSqlIdentifier(PMA_substr($sql, $count2, 1), ($is_sql_variable || $is_digit))) {
                    $c2 = PMA_substr($sql, $count2, 1);
                    if ($is_sql_variable && ($c2 == '.')) {
                        $count2++;
                        continue;
                    }
                    if ($is_digit && (!$is_hex_digit) && ($c2 == '.')) {
                        $count2++;
                        if (!$is_float_digit) {
                            $is_float_digit = TRUE;
                            continue;
                        } else {
                            $debugstr = $GLOBALS['strSQPBugInvalidIdentifer'] . ' @ ' . ($count1+1) . "\n"
                                      . 'STR: ' . htmlspecialchars(PMA_substr($sql, $count1, $count2 - $count1));
                            PMA_SQP_throwError($debugstr, $sql);
                            return $sql;
                        }
                    }
                    if ($is_digit && (!$is_hex_digit) && (($c2 == 'e') || ($c2 == 'E'))) {
                        if (!$is_float_digit_exponent) {
                            $is_float_digit_exponent = TRUE;
                            $is_float_digit          = TRUE;
                            $count2++;
                            continue;
                        } else {
                            $is_digit                = FALSE;
                            $is_float_digit          = FALSE;
                        }
                    }
                    if (($is_hex_digit && PMA_STR_isHexDigit($c2)) || ($is_digit && PMA_STR_isDigit($c2))) {
                        $count2++;
                        continue;
                    } else {
                        $is_digit     = FALSE;
                        $is_hex_digit = FALSE;
                    }

                    $count2++;
                } // end while

                $l    = $count2 - $count1;
                $str  = PMA_substr($sql, $count1, $l);

                $type = '';
                if ($is_digit) {
                    $type     = 'digit';
                    if ($is_float_digit) {
                        $type .= '_float';
                    } elseif ($is_hex_digit) {
                        $type .= '_hex';
                    } else {
                        $type .= '_integer';
                    }
                } else {
                    if ($is_sql_variable != FALSE) {
                        $type = 'alpha_variable';
                    } else {
                        $type = 'alpha';
                    }
                } // end if... else....
                PMA_SQP_arrayAdd($sql_array, $type, $str, $arraysize);

                continue;
            }

            // Checks for punct
            if (PMA_STR_strInStr($c, $allpunct_list)) {
                while (($count2 < $len) && PMA_STR_strInStr(PMA_substr($sql, $count2, 1), $allpunct_list)) {
                    $count2++;
                }
                $l = $count2 - $count1;
                if ($l == 1) {
                    $punct_data = $c;
                } else {
                    $punct_data = PMA_substr($sql, $count1, $l);
                }

                // Special case, sometimes, althought two characters are
                // adjectent directly, they ACTUALLY need to be seperate
                if ($l == 1) {
                    $t_suffix         = '';
                    switch ($punct_data) {
                        case $punct_queryend:
                            $t_suffix = '_queryend';
                            break;
                        case $punct_qualifier:
                            $t_suffix = '_qualifier';
                            break;
                        case $punct_listsep:
                            $t_suffix = '_listsep';
                            break;
                        default:
                            break;
                    }
                    PMA_SQP_arrayAdd($sql_array, 'punct' . $t_suffix, $punct_data, $arraysize);
                } elseif (PMA_STR_binarySearchInArr($punct_data, $allpunct_list_pair, $allpunct_list_pair_size)) {
                    // Ok, we have one of the valid combined punct expressions
                    PMA_SQP_arrayAdd($sql_array, 'punct', $punct_data, $arraysize);
                } else {
                    // Bad luck, lets split it up more
                    $first  = $punct_data[0];
                    $first2 = $punct_data[0] . $punct_data[1];
                    $last2  = $punct_data[$l - 2] . $punct_data[$l - 1];
                    $last   = $punct_data[$l - 1];
                    if (($first == ',') || ($first == ';') || ($first == '.') || ($first == '*')) {
                        $count2     = $count1 + 1;
                        $punct_data = $first;
                    } elseif (($last2 == '/*') || (($last2 == '--') && ($count2 == $len || PMA_substr($sql, $count2, 1) <= ' ') )) {
                        $count2     -= 2;
                        $punct_data = PMA_substr($sql, $count1, $count2 - $count1);
                    } elseif (($last == '-') || ($last == '+') || ($last == '!')) {
                        $count2--;
                        $punct_data = PMA_substr($sql, $count1, $count2 - $count1);
                    // TODO: for negation operator, split in 2 tokens ?
                    // "select x&~1 from t"
                    // becomes "select x & ~ 1 from t" ?

                    } elseif ($last != '~') {
                        $debugstr =  $GLOBALS['strSQPBugUnknownPunctuation'] . ' @ ' . ($count1+1) . "\n"
                                  . 'STR: ' . htmlspecialchars($punct_data);
                        PMA_SQP_throwError($debugstr, $sql);
                        return $sql;
                    }
                    PMA_SQP_arrayAdd($sql_array, 'punct', $punct_data, $arraysize);
                    continue;
                } // end if... elseif... else
                continue;
            }

            // DEBUG
            $count2++;

            $debugstr = 'C1 C2 LEN: ' . $count1 . ' ' . $count2 . ' ' . $len .  "\n"
                      . 'STR: ' . PMA_substr($sql, $count1, $count2 - $count1) . "\n";
            PMA_SQP_bug($debugstr, $sql);
            return $sql;

        } // end while ($count2 < $len)


        if ($arraysize > 0) {
            $t_next           = $sql_array[0]['type'];
            $t_prev           = '';
            $t_bef_prev       = '';
            $t_cur            = '';
            $d_next           = $sql_array[0]['data'];
            $d_prev           = '';
            $d_bef_prev       = '';
            $d_cur            = '';
            $d_next_upper     = $t_next == 'alpha' ? strtoupper($d_next) : $d_next;
            $d_prev_upper     = '';
            $d_bef_prev_upper = '';
            $d_cur_upper      = '';
        }

        for ($i = 0; $i < $arraysize; $i++) {
            $t_bef_prev       = $t_prev;
            $t_prev           = $t_cur;
            $t_cur            = $t_next;
            $d_bef_prev       = $d_prev;
            $d_prev           = $d_cur;
            $d_cur            = $d_next;
            $d_bef_prev_upper = $d_prev_upper;
            $d_prev_upper     = $d_cur_upper;
            $d_cur_upper      = $d_next_upper;
            if (($i + 1) < $arraysize) {
                $t_next = $sql_array[$i + 1]['type'];
                $d_next = $sql_array[$i + 1]['data'];
                $d_next_upper = $t_next == 'alpha' ? strtoupper($d_next) : $d_next;
            } else {
                $t_next       = '';
                $d_next       = '';
                $d_next_upper = '';
            }

            //DEBUG echo "[prev: <b>".$d_prev."</b> ".$t_prev."][cur: <b>".$d_cur."</b> ".$t_cur."][next: <b>".$d_next."</b> ".$t_next."]<br />";

            if ($t_cur == 'alpha') {
                $t_suffix     = '_identifier';
                if (($t_next == 'punct_qualifier') || ($t_prev == 'punct_qualifier')) {
                    $t_suffix = '_identifier';
                } elseif (($t_next == 'punct_bracket_open_round')
                  && PMA_STR_binarySearchInArr($d_cur_upper, $PMA_SQPdata_function_name, $PMA_SQPdata_function_name_cnt)) {
                    // FIXME-2005-10-16: in the case of a CREATE TABLE containing a TIMESTAMP,
                    // since TIMESTAMP() is also a function, it's found here and
                    // the token is wrongly marked as alpha_functionName. But we
                    // compensate for this when analysing for timestamp_not_null
                    // later in this script.
                    $t_suffix = '_functionName';
                } elseif (PMA_STR_binarySearchInArr($d_cur_upper, $PMA_SQPdata_column_type, $PMA_SQPdata_column_type_cnt)) {
                    $t_suffix = '_columnType';
    
                    // Temporary fix for BUG #621357
                    //TODO FIX PROPERLY NEEDS OVERHAUL OF SQL TOKENIZER
                    if ($d_cur_upper == 'SET' && $t_next != 'punct_bracket_open_round') {
                        $t_suffix = '_reservedWord';
                    }
                    //END OF TEMPORARY FIX
    
                    // CHARACTER is a synonym for CHAR, but can also be meant as
                    // CHARACTER SET. In this case, we have a reserved word.
                    if ($d_cur_upper == 'CHARACTER' && $d_next_upper == 'SET') {
                        $t_suffix = '_reservedWord';
                    }
    
                    // experimental
                    // current is a column type, so previous must not be
                    // a reserved word but an identifier
                    // CREATE TABLE SG_Persons (first varchar(64))
    
                    //if ($sql_array[$i-1]['type'] =='alpha_reservedWord') {
                    //    $sql_array[$i-1]['type'] = 'alpha_identifier';
                    //}
    
                } elseif (PMA_STR_binarySearchInArr($d_cur_upper, $PMA_SQPdata_reserved_word, $PMA_SQPdata_reserved_word_cnt)) {
                    $t_suffix = '_reservedWord';
                } elseif (PMA_STR_binarySearchInArr($d_cur_upper, $PMA_SQPdata_column_attrib, $PMA_SQPdata_column_attrib_cnt)) {
                    $t_suffix = '_columnAttrib';
                    // INNODB is a MySQL table type, but in "SHOW INNODB STATUS",
                    // it should be regarded as a reserved word.
                    if ($d_cur_upper == 'INNODB' && $d_prev_upper == 'SHOW' && $d_next_upper == 'STATUS') {
                        $t_suffix = '_reservedWord';
                    }
    
                    if ($d_cur_upper == 'DEFAULT' && $d_next_upper == 'CHARACTER') {
                        $t_suffix = '_reservedWord';
                    }
                    // Binary as character set
                    if ($d_cur_upper == 'BINARY' && (
                      ($d_bef_prev_upper == 'CHARACTER' && $d_prev_upper == 'SET')
                      || ($d_bef_prev_upper == 'SET' && $d_prev_upper == '=')
                      || ($d_bef_prev_upper == 'CHARSET' && $d_prev_upper == '=')
                      || $d_prev_upper == 'CHARSET'
                      ) && PMA_STR_binarySearchInArr($d_cur, $mysql_charsets, count($mysql_charsets))) {
                        $t_suffix = '_charset';
                    }
                } elseif (PMA_STR_binarySearchInArr($d_cur, $mysql_charsets, $mysql_charsets_count)
                  || PMA_STR_binarySearchInArr($d_cur, $mysql_collations_flat, $mysql_collations_count)