[mysql] base64 encode in MySQL

I want to select a blob col from one table, base64 encode it and insert it into another tables. Is there any way to do this without round tripping the data out of the DB and through my app?

This question is related to mysql base64

The answer is


For those interested, these are the only alternatives so far:

1) Using these Functions:

http://wi-fizzle.com/downloads/base64.sql

2) If you already have the sys_eval UDF, (Linux) you can do this:

sys_eval(CONCAT("echo '",myField,"' | base64"));

The first method is known to be slow. The problem with the second one, is that the encoding is actually happening "outside" MySQL, which can have encoding problems (besides the security risks that you are adding with sys_* functions).

Unfortunately there is no UDF compiled version (which should be faster) nor a native support in MySQL (Posgresql supports it!).

It seems that the MySQL development team are not interested in implement it as this function already exists in other languages, which seems pretty silly to me.


create table encrypt(username varchar(20),password varbinary(200))

insert into encrypt values('raju',aes_encrypt('kumar','key')) select *,cast(aes_decrypt(password,'key') as char(40)) from encrypt where username='raju';


If you need this for < 5.6, I tripped across this UDF which seems to work fine:

https://github.com/y-ken/mysql-udf-base64


create table encrypt(username varchar(20),password varbinary(200))

insert into encrypt values('raju',aes_encrypt('kumar','key')) select *,cast(aes_decrypt(password,'key') as char(40)) from encrypt where username='raju';


For those interested, these are the only alternatives so far:

1) Using these Functions:

http://wi-fizzle.com/downloads/base64.sql

2) If you already have the sys_eval UDF, (Linux) you can do this:

sys_eval(CONCAT("echo '",myField,"' | base64"));

The first method is known to be slow. The problem with the second one, is that the encoding is actually happening "outside" MySQL, which can have encoding problems (besides the security risks that you are adding with sys_* functions).

Unfortunately there is no UDF compiled version (which should be faster) nor a native support in MySQL (Posgresql supports it!).

It seems that the MySQL development team are not interested in implement it as this function already exists in other languages, which seems pretty silly to me.


Yet another custom implementation that doesn't require support table:

drop function if exists base64_encode;
create function base64_encode(_data blob)
returns text
begin
    declare _alphabet char(64) default 'ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789+/';
    declare _lim int unsigned default length(_data);
    declare _i int unsigned default 0;
    declare _chk3 char(6) default '';
    declare _chk3int int default 0;
    declare _enc text default '';

    while _i < _lim do
        set _chk3 = rpad(hex(binary substr(_data, _i + 1, 3)), 6, '0');
        set _chk3int = conv(_chk3, 16, 10);
        set _enc = concat(
            _enc
            ,                  substr(_alphabet, ((_chk3int >> 18) & 63) + 1, 1)
            , if (_lim-_i > 0, substr(_alphabet, ((_chk3int >> 12) & 63) + 1, 1), '=')
            , if (_lim-_i > 1, substr(_alphabet, ((_chk3int >>  6) & 63) + 1, 1), '=')
            , if (_lim-_i > 2, substr(_alphabet, ((_chk3int >>  0) & 63) + 1, 1), '=')
        );
        set _i = _i + 3;
    end while;

    return _enc;
end;

drop function if exists base64_decode;
create function base64_decode(_enc text)
returns blob
begin
    declare _alphabet char(64) default 'ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789+/';
    declare _lim int unsigned default 0;
    declare _i int unsigned default 0;
    declare _chr1byte tinyint default 0;
    declare _chk4int int default 0;
    declare _chk4int_bits tinyint default 0;
    declare _dec blob default '';
    declare _rem tinyint default 0;
    set _enc = trim(_enc);
    set _rem = if(right(_enc, 3) = '===', 3, if(right(_enc, 2) = '==', 2, if(right(_enc, 1) = '=', 1, 0)));
    set _lim = length(_enc) - _rem;

    while _i < _lim
    do
        set _chr1byte = locate(substr(_enc, _i + 1, 1), binary _alphabet) - 1;
        if (_chr1byte > -1)
        then
            set _chk4int = (_chk4int << 6) | _chr1byte;
            set _chk4int_bits = _chk4int_bits + 6;

            if (_chk4int_bits = 24 or _i = _lim-1)
            then
                if (_i = _lim-1 and _chk4int_bits != 24)
                then
                    set _chk4int = _chk4int << 0;
                end if;

                set _dec = concat(
                    _dec
                    ,                        char((_chk4int >> (_chk4int_bits -  8)) & 0xff)
                    , if(_chk4int_bits >  8, char((_chk4int >> (_chk4int_bits - 16)) & 0xff), '\0')
                    , if(_chk4int_bits > 16, char((_chk4int >> (_chk4int_bits - 24)) & 0xff), '\0')
                );
                set _chk4int = 0;
                set _chk4int_bits = 0;
            end if;
        end if;
        set _i = _i + 1;
    end while;

    return substr(_dec, 1, length(_dec) - _rem);
end;

Gist

You should convert charset after decoding: convert(base64_decode(base64_encode('????')) using utf8)


Functions from http://wi-fizzle.com/downloads/base64.sql contain some error when in encoded string are 32-byte (space), ex BASE64_ENCODE(CONCAT(CHAR(15), CHAR(32))). Here is corrected function

DELIMITER $$

USE `YOUR DATABASE`$$

DROP TABLE IF EXISTS core_base64_data$$
CREATE TABLE core_base64_data (c CHAR(1) BINARY, val TINYINT)$$
INSERT INTO core_base64_data VALUES 
('A',0), ('B',1), ('C',2), ('D',3), ('E',4), ('F',5), ('G',6), ('H',7), ('I',8), ('J',9),
('K',10), ('L',11), ('M',12), ('N',13), ('O',14), ('P',15), ('Q',16), ('R',17), ('S',18), ('T',19),
('U',20), ('V',21), ('W',22), ('X',23), ('Y',24), ('Z',25), ('a',26), ('b',27), ('c',28), ('d',29),
('e',30), ('f',31), ('g',32), ('h',33), ('i',34), ('j',35), ('k',36), ('l',37), ('m',38), ('n',39),
('o',40), ('p',41), ('q',42), ('r',43), ('s',44), ('t',45), ('u',46), ('v',47), ('w',48), ('x',49),
('y',50), ('z',51), ('0',52), ('1',53), ('2',54), ('3',55), ('4',56), ('5',57), ('6',58), ('7',59),
('8',60), ('9',61), ('+',62), ('/',63), ('=',0) $$

DROP FUNCTION IF EXISTS `BASE64_ENCODE`$$

CREATE DEFINER=`YOUR DATABASE`@`%` FUNCTION `BASE64_ENCODE`(input BLOB) RETURNS BLOB
    DETERMINISTIC
    SQL SECURITY INVOKER
BEGIN
    DECLARE ret BLOB DEFAULT '';
    DECLARE done TINYINT DEFAULT 0;
    IF input IS NULL THEN
        RETURN NULL;
    END IF;
each_block:
    WHILE NOT done DO BEGIN
        DECLARE accum_value BIGINT UNSIGNED DEFAULT 0;
        DECLARE in_count TINYINT DEFAULT 0;
        DECLARE out_count TINYINT;
each_input_char:
        WHILE in_count < 3 DO BEGIN
            DECLARE first_char BLOB(1);

            IF LENGTH(input) = 0 THEN
                SET done = 1;
                SET accum_value = accum_value << (8 * (3 - in_count));
                LEAVE each_input_char;
            END IF;

            SET first_char = SUBSTRING(input,1,1);
            SET input = SUBSTRING(input,2);

            SET accum_value = (accum_value << 8) + ASCII(first_char);
            SET in_count = in_count + 1;
        END; END WHILE;

        -- We've now accumulated 24 bits; deaccumulate into base64 characters
        -- We have to work from the left, so use the third byte position and shift left
        CASE
            WHEN in_count = 3 THEN SET out_count = 4;
            WHEN in_count = 2 THEN SET out_count = 3;
            WHEN in_count = 1 THEN SET out_count = 2;
            ELSE RETURN ret;
        END CASE;

        WHILE out_count > 0 DO BEGIN
            BEGIN
                DECLARE out_char CHAR(1);
                DECLARE base64_getval CURSOR FOR SELECT c FROM core_base64_data WHERE val = (accum_value >> 18);
                OPEN base64_getval;
                FETCH base64_getval INTO out_char;
                CLOSE base64_getval;
                SET ret = CONCAT(ret,out_char);
                SET out_count = out_count - 1;
                SET accum_value = accum_value << 6 & 0xffffff;
            END;
        END; END WHILE;
        CASE
            WHEN in_count = 2 THEN SET ret = CONCAT(ret,'=');
            WHEN in_count = 1 THEN SET ret = CONCAT(ret,'==');
            ELSE BEGIN END;
        END CASE;

    END; END WHILE;
    RETURN ret;
END$$

DELIMITER ;

Looks like no, though it was requested, and there’s a UDF for it.

Edit: Or there’s… this. Ugh.


SELECT `id`,`name`, TO_BASE64(content) FROM `db`.`upload`

this will convert the blob value from content column to base64 string. Then you can do with this string whatever you want even insert it into another table


Functions from http://wi-fizzle.com/downloads/base64.sql contain some error when in encoded string are 32-byte (space), ex BASE64_ENCODE(CONCAT(CHAR(15), CHAR(32))). Here is corrected function

DELIMITER $$

USE `YOUR DATABASE`$$

DROP TABLE IF EXISTS core_base64_data$$
CREATE TABLE core_base64_data (c CHAR(1) BINARY, val TINYINT)$$
INSERT INTO core_base64_data VALUES 
('A',0), ('B',1), ('C',2), ('D',3), ('E',4), ('F',5), ('G',6), ('H',7), ('I',8), ('J',9),
('K',10), ('L',11), ('M',12), ('N',13), ('O',14), ('P',15), ('Q',16), ('R',17), ('S',18), ('T',19),
('U',20), ('V',21), ('W',22), ('X',23), ('Y',24), ('Z',25), ('a',26), ('b',27), ('c',28), ('d',29),
('e',30), ('f',31), ('g',32), ('h',33), ('i',34), ('j',35), ('k',36), ('l',37), ('m',38), ('n',39),
('o',40), ('p',41), ('q',42), ('r',43), ('s',44), ('t',45), ('u',46), ('v',47), ('w',48), ('x',49),
('y',50), ('z',51), ('0',52), ('1',53), ('2',54), ('3',55), ('4',56), ('5',57), ('6',58), ('7',59),
('8',60), ('9',61), ('+',62), ('/',63), ('=',0) $$

DROP FUNCTION IF EXISTS `BASE64_ENCODE`$$

CREATE DEFINER=`YOUR DATABASE`@`%` FUNCTION `BASE64_ENCODE`(input BLOB) RETURNS BLOB
    DETERMINISTIC
    SQL SECURITY INVOKER
BEGIN
    DECLARE ret BLOB DEFAULT '';
    DECLARE done TINYINT DEFAULT 0;
    IF input IS NULL THEN
        RETURN NULL;
    END IF;
each_block:
    WHILE NOT done DO BEGIN
        DECLARE accum_value BIGINT UNSIGNED DEFAULT 0;
        DECLARE in_count TINYINT DEFAULT 0;
        DECLARE out_count TINYINT;
each_input_char:
        WHILE in_count < 3 DO BEGIN
            DECLARE first_char BLOB(1);

            IF LENGTH(input) = 0 THEN
                SET done = 1;
                SET accum_value = accum_value << (8 * (3 - in_count));
                LEAVE each_input_char;
            END IF;

            SET first_char = SUBSTRING(input,1,1);
            SET input = SUBSTRING(input,2);

            SET accum_value = (accum_value << 8) + ASCII(first_char);
            SET in_count = in_count + 1;
        END; END WHILE;

        -- We've now accumulated 24 bits; deaccumulate into base64 characters
        -- We have to work from the left, so use the third byte position and shift left
        CASE
            WHEN in_count = 3 THEN SET out_count = 4;
            WHEN in_count = 2 THEN SET out_count = 3;
            WHEN in_count = 1 THEN SET out_count = 2;
            ELSE RETURN ret;
        END CASE;

        WHILE out_count > 0 DO BEGIN
            BEGIN
                DECLARE out_char CHAR(1);
                DECLARE base64_getval CURSOR FOR SELECT c FROM core_base64_data WHERE val = (accum_value >> 18);
                OPEN base64_getval;
                FETCH base64_getval INTO out_char;
                CLOSE base64_getval;
                SET ret = CONCAT(ret,out_char);
                SET out_count = out_count - 1;
                SET accum_value = accum_value << 6 & 0xffffff;
            END;
        END; END WHILE;
        CASE
            WHEN in_count = 2 THEN SET ret = CONCAT(ret,'=');
            WHEN in_count = 1 THEN SET ret = CONCAT(ret,'==');
            ELSE BEGIN END;
        END CASE;

    END; END WHILE;
    RETURN ret;
END$$

DELIMITER ;

Looks like no, though it was requested, and there’s a UDF for it.

Edit: Or there’s… this. Ugh.


SELECT `id`,`name`, TO_BASE64(content) FROM `db`.`upload`

this will convert the blob value from content column to base64 string. Then you can do with this string whatever you want even insert it into another table


Looks like no, though it was requested, and there’s a UDF for it.

Edit: Or there’s… this. Ugh.


Yet another custom implementation that doesn't require support table:

drop function if exists base64_encode;
create function base64_encode(_data blob)
returns text
begin
    declare _alphabet char(64) default 'ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789+/';
    declare _lim int unsigned default length(_data);
    declare _i int unsigned default 0;
    declare _chk3 char(6) default '';
    declare _chk3int int default 0;
    declare _enc text default '';

    while _i < _lim do
        set _chk3 = rpad(hex(binary substr(_data, _i + 1, 3)), 6, '0');
        set _chk3int = conv(_chk3, 16, 10);
        set _enc = concat(
            _enc
            ,                  substr(_alphabet, ((_chk3int >> 18) & 63) + 1, 1)
            , if (_lim-_i > 0, substr(_alphabet, ((_chk3int >> 12) & 63) + 1, 1), '=')
            , if (_lim-_i > 1, substr(_alphabet, ((_chk3int >>  6) & 63) + 1, 1), '=')
            , if (_lim-_i > 2, substr(_alphabet, ((_chk3int >>  0) & 63) + 1, 1), '=')
        );
        set _i = _i + 3;
    end while;

    return _enc;
end;

drop function if exists base64_decode;
create function base64_decode(_enc text)
returns blob
begin
    declare _alphabet char(64) default 'ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789+/';
    declare _lim int unsigned default 0;
    declare _i int unsigned default 0;
    declare _chr1byte tinyint default 0;
    declare _chk4int int default 0;
    declare _chk4int_bits tinyint default 0;
    declare _dec blob default '';
    declare _rem tinyint default 0;
    set _enc = trim(_enc);
    set _rem = if(right(_enc, 3) = '===', 3, if(right(_enc, 2) = '==', 2, if(right(_enc, 1) = '=', 1, 0)));
    set _lim = length(_enc) - _rem;

    while _i < _lim
    do
        set _chr1byte = locate(substr(_enc, _i + 1, 1), binary _alphabet) - 1;
        if (_chr1byte > -1)
        then
            set _chk4int = (_chk4int << 6) | _chr1byte;
            set _chk4int_bits = _chk4int_bits + 6;

            if (_chk4int_bits = 24 or _i = _lim-1)
            then
                if (_i = _lim-1 and _chk4int_bits != 24)
                then
                    set _chk4int = _chk4int << 0;
                end if;

                set _dec = concat(
                    _dec
                    ,                        char((_chk4int >> (_chk4int_bits -  8)) & 0xff)
                    , if(_chk4int_bits >  8, char((_chk4int >> (_chk4int_bits - 16)) & 0xff), '\0')
                    , if(_chk4int_bits > 16, char((_chk4int >> (_chk4int_bits - 24)) & 0xff), '\0')
                );
                set _chk4int = 0;
                set _chk4int_bits = 0;
            end if;
        end if;
        set _i = _i + 1;
    end while;

    return substr(_dec, 1, length(_dec) - _rem);
end;

Gist

You should convert charset after decoding: convert(base64_decode(base64_encode('????')) using utf8)


If you need this for < 5.6, I tripped across this UDF which seems to work fine:

https://github.com/y-ken/mysql-udf-base64