MySQL: User Defined Function Tutorial

This tutorial explains what a User Defined Function (UDF) is, what it does, and why/when it is useful.

1. What Is a User Defined Function?

Basically, a User Defined Function (UDF) is a piece of code that extends the functionality of a MySQL server by adding a new function that behaves just like a native (built-in) MySQL function, such as abs() or concat(). UDFs are written in C (or C++ if you really need to). Maybe there is a way to write them in BASIC, .NET, or whatever but I don't see why anybody would want to do that.

2. Why/When Are UDFs Useful?

As implied by the name, UDFs are useful when you need to extend the functionality of your MySQL server. This little table should make it clear which method is best for a given situation:

Method Speed Language Development
Stored Procedures slow SQL ~minutes (for small functions)
UDF fast C ~hour
Native Function fast C major pain

And by "slow," I mean "slower than the others!" Stored procedures are still much faster then normal SQL statements!

A little explanation on native functions: The code you have to write here is essentially the same as the one for a UDF, but you have to write it in the MySQL source code and recompile the whole thing. This will (believe me) be a lot of work because you have to do it again and again with every new version of MySQL.

3. How to Use UDFs

This part is really easy. When you have your UDF finished, you just use it like every other native function. For example:

SELECT MyFunction(data1, data2) FROM table

4. Writing the UDF

Now, you can get started writing your first UDF. Just follow these steps:

Step 1: Create a new shared-library project

In the example, I used VC++ 6.0 with a standard DLL).

Step 2: Create some headers

These headers are either standard library headers or from the MySQL Server's include directory.

/* STANDARD is defined. Don't use any MySQL functions */
#include <stdlib.h>
#include <stdio.h>
#include <string.h>
#ifdef __WIN__
typedef unsigned __int64 ulonglong;     /* Microsoft's 64 bit types */
typedef __int64 longlong;
typedef unsigned long long ulonglong;
typedef long long longlong;
#endif /*__WIN__*/
#include <my_global.h>
#include <my_sys.h>
#include <mysql.h>
#include <ctype.h>
static pthread_mutex_t LOCK_hostname;

Step 3: Decide what kind of function you want

There are essentially two choices to be made:

  • Is the function an aggregate function or not? (You will learn more about aggregate functions later.)
  • Which type of return value should the function return? Here, you have four options:
  • Type Description
    STRING A string literal. Translates to char* in C
    INTEGER A normal integer. Translates to a 64-bit integer in C
    REAL A floating point number. Translates to double in C
    DECIMAL This one isn't really finished at this time. MySQL treats it as STRING

Step 4: Assign non-aggregate functions

Now, you have to declare and implement some functions the MySQL server needs to use your UDF. But, first some structs you'll need for that:

  • Type Name Description
    my_bool maybe_null 1 if function can return NULL
    unsigned int decimals for REAL functions
    unsigned long max_length For string functions
    char * ptr free pointer for function data
    my_bool const_item 0 if result is independent of argument

  • Type Name Description
    unsigned int arg_count Number of argument
    enum Item_result * arg_type Array containing the types of the arguments
    char ** args Array of pointer to the arguments
    unsigned long * lengths Array of the argument's lengths (only needed for strings)
    char * maybe_null Array of "maybe_null" flags (1 if argument maybe null)
    char ** attributes Array of pointers to the arguments' attributes
    unsigned long * attribute_lengths Array of attributes' lengths

Step 5: Examine the functions


extern "C" my_bool MyTest_init(UDF_INIT *initid, UDF_ARGS *args,
                               char *message)
   // The most important thing to do here is to set up the memory
   // you need...
   // Say you need a lonlong type variable to keep a checksum
   // although you do not need one in this case
   longlong* i = new longlong;    // create the variable
   *i = 0;                        // set it to a value

   // store it as a char pointer in the pointer variable
   // Make sure that you don't run into typecasting troubles later!!
   initid->ptr = (char*)i;

   // check the arguments format
   if (args->arg_count != 1)
      strcpy(message,"MyTest() requires one arguments");
      return 1;

   if (args->arg_type[0] != INT_RESULT)
      strcpy(message,"MyTest() requires an integer");
      return 1;
   return 0;

extern "C" void MyTest_deinit(UDF_INIT *initid)
   // Here you have to free the memory you allocated in the
   // initialization function
   delete (longlong*)initid->ptr;

The actual function:

extern "C" longlong MyTest(UDF_INIT *initid, UDF_ARGS *args,
                           char *is_null, char *error)
   /* So, finally, this is the part were you do the real work.
   This function is called for every record and the current value(s)
   or better pointers to the current values are stroed in the
   UDF_ARGS variable. You have to get the values, do your
   calculation, and return the result.
   NOTE: You can access the memory allocated in MyTest_init
   through the UDF_INIT variable.
   In this example, you will just add 5 to every value...*/
   return *((longlong*)args->args[0])+5;

Step 6: All done!

Now, you have to compile the library and copy it to a directory where your OS can find it. On Windows, that would be anywhere the PATH System variable says. Personally, I use the MySQL server's bin directory. You have to make sure that the library is in one of those directories; otherwise, MySQL can't use it! Also, make sure to export all the functions MySQL needs!

Step 7: Tell MySQL about it

This is really straightforward: Just execute the following SQL command:

SONAME the_libraries_exact_name

Now, you can use it like any other function.

5. Aggregate Functions

Now, some words about aggregate functions. When your UDF is an aggregate function, you have to add some more functions and some functions are used in a different way. The calling sequence is:

  1. Call MyTest_init to allocate memory (just like a normal UDF).
  2. MySQL sorts the table according to the GROUP BY statement.
  3. Call MyTest_clear for the first row in each group.
  4. Call MyTest_add for each row that belongs to the same group.
  5. Call MyTest to get the result when the group changes or the last row has been processed.
  6. Repeat Steps 3 to 5 until all rows have been processed.
  7. Call MyTest_deinit to free any used memory.

Now, look at the new functions needed for the aggregate function. In this example, you'll simply add up all the values (like the native SUM function).

void MyTest_clear(UDF_INIT *initid, char *is_null, char *error)
   /* The clear function resets the sum to 0 for each new group
   Of course, you have to allocate a longlong variable in the init
   function and assign it to the pointer as seen above */
   *((longlong*)initid->ptr) = 0;

void MyTest_add(UDF_INIT *initid, UDF_ARGS *args, char *is_null,
                char *error)
   // For each row, the current value is added to the sum
   *((longlong*)initid->ptr) = *((longlong*)initid->ptr)
                             + *((longlong*)args->args[0]);

longlong MyTest(UDF_INIT *initid, UDF_ARGS *args, char *is_null,
                char *error)
   // And, in the end, the sum is returned
   return *((longlong*)initid->ptr);


6. Advanced Topics

Here are some things you should know when you write more complex UDFs:

  • A string function should return a pointer to the result and set result and length to the contents and length of the return value. For example:
  • memcpy(result, "result string", 13); *length = 13;

    The result buffer that is passed to the MyTest function is 255 bytes long. If your result fits in this, you don't have to worry about memory allocation for results.

    If your string function needs to return a string longer than 255 bytes, you must allocate the space for it with malloc() or new in your MyTest_init() function or your MyTest() function and free it in your MyTest_deinit() function. You can store the allocated memory in the ptr slot in the UDF_INIT structure for reuse by future MyTest() calls.

  • To indicate an error return in the main function, set *error to 1: If MyTest() sets *error to 1 for any row, the function value is NULL for the current row and for any subsequent rows processed by the statement in which MyTest() was invoked.
  • For more information, see the MySQL Online Manual.

7. Some Guidelines

Here are some guidelines that you should follow if you want to make sure your UDF runs smoothly.

  • Do not call any other applications or processes inside a UDF!
  • Do not store any information locally! (This goes for shared libraries in general.)
  • Do not allocate any global or static variables!
  • Always check the type of your arguments. As you can see, MySQL converts everything to char pointers. This can lead to major troubles if you convert a string literal to a integer pointer and so on.
  • Be extra careful with the memory allocation! If you have memory leaks, you can bring down the sever in no time.

8. Debugging UDFs

Debugging a UDF can be pretty nerve wracking because, every time your UDF crashes, it takes down the whole MySQL server along with it. So, I wrote a little command-line tool to work around that problem. Just execute it after compilation and it does the rest. It emulates a call to the function by calling an "SELECT" command and then passing the results to the library and printing out the result on the command line. So, when the UDF produces some serious errors, only the little helper goes down and not the whole server. It is still in "beta," so don't expect to much....

9. Sources

MySQL Online Manual



  • There are no comments yet. Be the first to comment!

Leave a Comment
  • Your email address will not be published. All fields are required.

Top White Papers and Webcasts

  • Live Event Date: November 20, 2014 @ 2:00 p.m. ET / 11:00 a.m. PT Are you wanting to target two or more platforms such as iOS, Android, and/or Windows? You are not alone. 90% of enterprises today are targeting two or more platforms. Attend this eSeminar to discover how mobile app developers can rely on one IDE to create applications across platforms and approaches (web, native, and/or hybrid), saving time, money, and effort and introducing apps to market faster. You'll learn the trade-offs for gaining long …

  • IBM Worklight is a mobile application development platform that lets you extend your business to mobile devices. It is designed to provide an open, comprehensive platform to build, run and manage HTML5, hybrid and native mobile apps.

Most Popular Programming Stories

More for Developers

Latest Developer Headlines

RSS Feeds