making a generic function wrapper

[markww]

Hi,

I have a wrapper around some 3rd party database library function. The
pseudo code looks like the following - it is meant to open a table in a
database, extract values from a table, then copy it into my own user
defined structures. Since the process of opening and retrieving data
from the database is exactly the same for all struct types, and the
only part that's different is how the data is copied into the diff
structs, I was wondering if there was a way I could have one generic
wrapper, then somehow pass a callback or something to handle the
copying over specific stuff: This is what I have now, an entire wrapper
which reproduces the entire open table and copy to struct code for
every different type of structure I have:


bool MyDatabaseWrapper(structure A,
structure B,
structure N)
{
// Generic open stuff
// Generic open stuff
// Generic open stuff

while (...) {

// Generic execute stuff
// Generic execute stuff

// Now data is copied from the retrieved record into
// my specific passed structure.
if (struct type A) {
a.income = data.col(0);
}
else if (struct type B) {
b.name = data.col(0);
b.age = data.col(1);
}
// and so on for many different struct types.
}

return true;
}


See I would rather have it like:

bool MyGenericDatabaseWrapper(void **structType, CopyFunctionPtr)
{
// Generic open stuff

while (...) {

// Generic execute stuff

// Ok now call the function pointer to handle copying the
specifics
// into the unknown struct type.
CopyFunctionPtr(structType, data);
}
return true;
}

void CopyFunctionPtrStructA(struct &A, &data)
{
A.income = data.col(0);
}

void CopyFunctionPtrStructB(struct &B, &data)
{
B.name = data.col(0);
B.age = data.col(1);
}

Ok I hope that was clear. This whole wrapper thing is to make sure this
database lib I'm using is thread safe, and in the function there are
like 30 lines of code that I don't want to recopy for every type of
struct retrieval I have to do from the database.

Thanks,
Mark

 

[mlimber]

Hi,

I have a wrapper around some 3rd party database library function. The
pseudo code looks like the following - it is meant to open a table in a
database, extract values from a table, then copy it into my own user
defined structures. Since the process of opening and retrieving data
from the database is exactly the same for all struct types, and the
only part that's different is how the data is copied into the diff
structs, I was wondering if there was a way I could have one generic
wrapper, then somehow pass a callback or something to handle the
copying over specific stuff: This is what I have now, an entire wrapper
which reproduces the entire open table and copy to struct code for
every different type of structure I have:


bool MyDatabaseWrapper(structure A,
structure B,
structure N)
{
// Generic open stuff
// Generic open stuff
// Generic open stuff

while (...) {

// Generic execute stuff
// Generic execute stuff

// Now data is copied from the retrieved record into
// my specific passed structure.
if (struct type A) {
a.income = data.col(0);
}
else if (struct type B) {
b.name = data.col(0);
b.age = data.col(1);
}
// and so on for many different struct types.
}

return true;
}


See I would rather have it like:

bool MyGenericDatabaseWrapper(void **structType, CopyFunctionPtr)
{
// Generic open stuff

while (...) {

// Generic execute stuff

// Ok now call the function pointer to handle copying the
specifics
// into the unknown struct type.
CopyFunctionPtr(structType, data);
}
return true;
}

void CopyFunctionPtrStructA(struct &A, &data)
{
A.income = data.col(0);
}

void CopyFunctionPtrStructB(struct &B, &data)
{
B.name = data.col(0);
B.age = data.col(1);
}

Ok I hope that was clear. This whole wrapper thing is to make sure this
database lib I'm using is thread safe, and in the function there are
like 30 lines of code that I don't want to recopy for every type of
struct retrieval I have to do from the database.

You might be able to do something like the second one, but you'll still
have to have logic to determine what type to copy somewhere. In your
second pseudo-code snippet above, that logic disappeared altogether,
which makes me wonder if you had enough information to select the copy
function outside the database wrapper function. If so, then you could
use templates to come up with a solution, and if the different types
have a common base class, you might consider using an object factory
(see chapter 8 of _Modern C++ Design_ by Alexandrescu; download the
accompanying library from http://sourceforge.net/projects/loki-lib).

Cheers! --M

 

[markww]

Hi mlimber,

Thanks for replying. I think my example is overly complicated and I may
have explained it incorrectly. Here's a much clearer example of what
I'm trying to do - I'm really new with function pointers and the like,
but if you have any suggestions with the following example that'd be
great. In the meantime I'll do the reading you suggested,

Thanks


struct TYPE_A {
string name, incoming;
};
struct TYPE_B {
int age;
};

int main()
{
TYPE_A a;
TYPE_B b;

ReadData(CopyStructA, a);

ReadData(CopyStructB, b);

return 0;
}

void ReadData(void *pStruct, void *fnPtr)
{
// Read the data into 'data'.
fnPtr(data, pStruct);
}

void CopyStructA(data, TYPE_A *a)
{
a.name = data.col(0);
a.incoming = data.col(1);
}

void CopyStructB(data, TYPE_B *b)
{
b.age = data.col(0);
}

 

[mlimber]

Hi mlimber,

Hi. Please don't top-post here. Put your reply inline or below the post
you are responding to.

Thanks for replying. I think my example is overly complicated and I may
have explained it incorrectly. Here's a much clearer example of what
I'm trying to do - I'm really new with function pointers and the like,
but if you have any suggestions with the following example that'd be
great. In the meantime I'll do the reading you suggested,

Thanks


struct TYPE_A {
string name, incoming;
};

N.B., this is not a POD struct (see
http://www.parashift.com/c++-faq-lite/intrinsic-types.html#faq-26.7).
Referring to it with a void pointer is not recommended.

struct TYPE_B {
int age;
};

int main()
{
TYPE_A a;
TYPE_B b;

ReadData(CopyStructA, a);

ReadData(CopyStructB, b);

I think you meant:

ReadData( &a, CopyStructA );
ReadData( &b, CopyStructB );

You'll also need CopyStructA and CopyStructB defined or have a
prototype above.

return 0;
}

void ReadData(void *pStruct, void *fnPtr)

I think you meant something like:

void ReadData(void *pStruct, void (*fnPtr)(const Data&, void*) )

{
// Read the data into 'data'.
fnPtr(data, pStruct);
}

Void pointers are generally discouraged in C++ since it throws away
type information. Use templates instead:

template<class T, class FN>
void ReadData( T& t, FN fn )
{
const Data data = /* get data somehow */;

// Convert data into the given struct
fn( data, t );
}

void CopyStructA( data, TYPE_A *a)
{
a.name = data.col(0);
a.incoming = data.col(1);
}

void CopyStructB(data, TYPE_B *b)
{
b.age = data.col(0);
}

What type is data? A POD struct? An object? What is the return type of
data.col()?

You might consider writing the data to a std::stringstream and then
extracting it with your custom std:stream extraction operators for
each type.

Try writing a minimal but *complete* example that demonstrates what you
are trying to do. Then we can help you further. See
http://parashift.com/c++-faq-lite/how-to-post.html#faq-5.8 on how to
post code here.

Cheers! --M