Object-orientation in C

What would be a set of nifty preprocessor hacks (ANSI C89/ISO C90 compatible) which enable some kind of ugly (but usable) object-orientation in C?

I am familiar with a few different object-oriented languages, so please don't respond with answers like "Learn C++!". I have read "[Object-Oriented Programming With ANSI C][1]" (beware: PDF format) and several other interesting solutions, but I'm mostly interested in yours :-)!

---

See also [Can you write object oriented code in C?][2]

[1]: http://www.planetpdf.com/codecuts/pdfs/ooc.pdf "Object-Oriented Programming With ANSI C" [2]: https://stackoverflow.com/questions/351733 "Can you write object oriented code in C?"

I would advise against preprocessor (ab)use to try and make C syntax more like that of another more object-oriented language. At the most basic level, you just use plain structs as objects and pass them around by pointers:

struct monkey
{
    float age;
    bool is_male;
    int happiness;
};

void monkey_dance(struct monkey *monkey)
{
    /* do a little dance */
}

To get things like inheritance and polymorphism, you have to work a little harder. You can do manual inheritance by having the first member of a structure be an instance of the superclass, and then you can cast around pointers to base and derived classes freely:

struct base
{
    /* base class members */
};

struct derived
{
    struct base super;
    /* derived class members */
};

struct derived d;
struct base *base_ptr = (struct base *)&d;  // upcast
struct derived *derived_ptr = (struct derived *)base_ptr;  // downcast

To get polymorphism (i.e. virtual functions), you use function pointers, and optionally function pointer tables, also known as virtual tables or vtables:

struct base;
struct base_vtable
{
    void (*dance)(struct base *);
    void (*jump)(struct base *, int how_high);
};

struct base
{
    struct base_vtable *vtable;
    /* base members */
};

void base_dance(struct base *b)
{
    b->vtable->dance(b);
}

void base_jump(struct base *b, int how_high)
{
    b->vtable->jump(b, how_high);
}

struct derived1
{
    struct base super;
    /* derived1 members */
};

void derived1_dance(struct derived1 *d)
{
    /* implementation of derived1's dance function */
}

void derived1_jump(struct derived1 *d, int how_high)
{
    /* implementation of derived 1's jump function */
}

/* global vtable for derived1 */
struct base_vtable derived1_vtable =
{
    &derived1_dance, /* you might get a warning here about incompatible pointer types */
    &derived1_jump   /* you can ignore it, or perform a cast to get rid of it */
};

void derived1_init(struct derived1 *d)
{
    d->super.vtable = &derived1_vtable;
    /* init base members d->super.foo */
    /* init derived1 members d->foo */
}

struct derived2
{
    struct base super;
    /* derived2 members */
};

void derived2_dance(struct derived2 *d)
{
    /* implementation of derived2's dance function */
}

void derived2_jump(struct derived2 *d, int how_high)
{
    /* implementation of derived2's jump function */
}

struct base_vtable derived2_vtable =
{
   &derived2_dance,
   &derived2_jump
};

void derived2_init(struct derived2 *d)
{
    d->super.vtable = &derived2_vtable;
    /* init base members d->super.foo */
    /* init derived1 members d->foo */
}

int main(void)
{
    /* OK!  We're done with our declarations, now we can finally do some
       polymorphism in C */
    struct derived1 d1;
    derived1_init(&d1);

    struct derived2 d2;
    derived2_init(&d2);

    struct base *b1_ptr = (struct base *)&d1;
    struct base *b2_ptr = (struct base *)&d2;

    base_dance(b1_ptr);  /* calls derived1_dance */
    base_dance(b2_ptr);  /* calls derived2_dance */

    base_jump(b1_ptr, 42);  /* calls derived1_jump */
    base_jump(b2_ptr, 42);  /* calls derived2_jump */

    return 0;
}

And that's how you do polymorphism in C. It ain't pretty, but it does the job. There are some sticky issues involving pointer casts between base and derived classes, which are safe as long as the base class is the first member of the derived class. Multiple inheritance is much harder - in that case, in order to case between base classes other than the first, you need to manually adjust your pointers based on the proper offsets, which is really tricky and error-prone.

Another (tricky) thing you can do is change the dynamic type of an object at runtime! You just reassign it a new vtable pointer. You can even selectively change some of the virtual functions while keeping others, creating new hybrid types. Just be careful to create a new vtable instead of modifying the global vtable, otherwise you'll accidentally affect all objects of a given type.

I once worked with a C library that was implemented in a way that struck me as quite elegant. They had written, in C, a way to define objects, then inherit from them so that they were as extensible as a C++ object. The basic idea was this:

• Each object had its own file
• Public functions and variables are defined in the .h file for an object
• Private variables and functions were only located in the .c file
• To "inherit" a new struct is created with the first member of the struct being the object to inherit from

Inheriting is difficult to describe, but basically it was this:

struct vehicle {
   int power;
   int weight;
}

Then in another file:

struct van {
   struct vehicle base;
   int cubic_size;
}

Then you could have a van created in memory, and being used by code that only knew about vehicles:

struct van my_van;
struct vehicle *something = &my_van;
vehicle_function( something );

It worked beautifully, and the .h files defined exactly what you should be able to do with each object.

C Object System (COS) sounds promising (it's still in alpha version). It tries to keep minimal the available concepts for the sake of simplicity and flexibility: uniform object oriented programming including open classes, metaclasses, property metaclasses, generics, multimethods, delegation, ownership, exceptions, contracts and closures. There is a draft paper (PDF) that describes it.

Exception in C is a C89 implementation of the TRY-CATCH-FINALLY found in other OO languages. It comes with a testsuite and some examples.

Both by Laurent Deniau, which is working a lot on OOP in C.

The GNOME desktop for Linux is written in object-oriented C, and it has an object model called "GObject" which supports properties, inheritance, polymorphism, as well as some other goodies like references, event handling (called "signals"), runtime typing, private data, etc.

It includes preprocessor hacks to do things like typecasting around in the class hierarchy, etc. Here's an example class I wrote for GNOME (things like gchar are typedefs):

Class Source

Class Header

Inside the GObject structure there's a GType integer which is used as a magic number for GLib's dynamic typing system (you can cast the entire struct to a "GType" to find it's type).

Slightly off-topic, but the original C++ compiler, Cfront, compiled C++ to C and then to assembler.

Preserved here.

If you think of methods called on objects as static methods that pass an implicit 'this' into the function it can make thinking OO in C easier.

For example:

String s = "hi";
System.out.println(s.length());

becomes:

string s = "hi";
printf(length(s)); // pass in s, as an implicit this

Or something like that.

I used to do this kind of thing in C, before I knew what OOP was.

Following is an example, which implements a data-buffer which grows on demand, given a minimum size, increment and maximum size. This particular implementation was "element" based, which is to say it was designed to allow a list-like collection of any C type, not just a variable length byte-buffer.

The idea is that the object is instantiated using the xxx_crt() and deleted using xxx_dlt(). Each of the "member" methods takes a specifically typed pointer to operate on.

I implemented a linked list, cyclic buffer, and a number of other things in this manner.

I must confess, I have never given any thought on how to implement inheritance with this approach. I imagine that some blend of that offered by Kieveli might be a good path.

dtb.c:

#include <limits.h>
#include <string.h>
#include <stdlib.h>

static void dtb_xlt(void *dst, const void *src, vint len, const byte *tbl);

DTABUF *dtb_crt(vint minsiz,vint incsiz,vint maxsiz) {
    DTABUF          *dbp;

    if(!minsiz) { return NULL; }
    if(!incsiz)                  { incsiz=minsiz;        }
    if(!maxsiz || maxsiz<minsiz) { maxsiz=minsiz;        }
    if(minsiz+incsiz>maxsiz)     { incsiz=maxsiz-minsiz; }
    if((dbp=(DTABUF*)malloc(sizeof(*dbp))) == NULL) { return NULL; }
    memset(dbp,0,sizeof(*dbp));
    dbp->min=minsiz;
    dbp->inc=incsiz;
    dbp->max=maxsiz;
    dbp->siz=minsiz;
    dbp->cur=0;
    if((dbp->dta=(byte*)malloc((vuns)minsiz)) == NULL) { free(dbp); return NULL; }
    return dbp;
    }

DTABUF *dtb_dlt(DTABUF *dbp) {
    if(dbp) {
        free(dbp->dta);
        free(dbp);
        }
    return NULL;
    }

vint dtb_adddta(DTABUF *dbp,const byte *xlt256,const void *dtaptr,vint dtalen) {
    if(!dbp) { errno=EINVAL; return -1; }
    if(dtalen==-1) { dtalen=(vint)strlen((byte*)dtaptr); }
    if((dbp->cur + dtalen) > dbp->siz) {
        void        *newdta;
        vint        newsiz;

        if((dbp->siz+dbp->inc)>=(dbp->cur+dtalen)) { newsiz=dbp->siz+dbp->inc; }
        else                                       { newsiz=dbp->cur+dtalen;   }
        if(newsiz>dbp->max) { errno=ETRUNC; return -1; }
        if((newdta=realloc(dbp->dta,(vuns)newsiz))==NULL) { return -1; }
        dbp->dta=newdta; dbp->siz=newsiz;
        }
    if(dtalen) {
        if(xlt256) { dtb_xlt(((byte*)dbp->dta+dbp->cur),dtaptr,dtalen,xlt256); }
        else       { memcpy(((byte*)dbp->dta+dbp->cur),dtaptr,(vuns)dtalen);   }
        dbp->cur+=dtalen;
        }
    return 0;
    }

static void dtb_xlt(void *dst,const void *src,vint len,const byte *tbl) {
    byte            *sp,*dp;

    for(sp=(byte*)src,dp=(byte*)dst; len; len--,sp++,dp++) { *dp=tbl[*sp]; }
    }

vint dtb_addtxt(DTABUF *dbp,const byte *xlt256,const byte *format,...) {
    byte            textÝ501¨;
    va_list         ap;
    vint            len;

    va_start(ap,format); len=sprintf_len(format,ap)-1; va_end(ap);
    if(len<0 || len>=sizeof(text)) { sprintf_safe(text,sizeof(text),"STRTOOLNG: %s",format); len=(int)strlen(text); }
    else                           { va_start(ap,format); vsprintf(text,format,ap); va_end(ap);                     }
    return dtb_adddta(dbp,xlt256,text,len);
    }

vint dtb_rmvdta(DTABUF *dbp,vint len) {
    if(!dbp) { errno=EINVAL; return -1; }
    if(len > dbp->cur) { len=dbp->cur; }
    dbp->cur-=len;
    return 0;
    }

vint dtb_reset(DTABUF *dbp) {
    if(!dbp) { errno=EINVAL; return -1; }
    dbp->cur=0;
    if(dbp->siz > dbp->min) {
        byte *newdta;
        if((newdta=(byte*)realloc(dbp->dta,(vuns)dbp->min))==NULL) {
            free(dbp->dta); dbp->dta=null; dbp->siz=0;
            return -1;
            }
        dbp->dta=newdta; dbp->siz=dbp->min;
        }
    return 0;
    }

void *dtb_elmptr(DTABUF *dbp,vint elmidx,vint elmlen) {
    if(!elmlen || (elmidx*elmlen)>=dbp->cur) { return NULL; }
    return ((byte*)dbp->dta+(elmidx*elmlen));
    }

dtb.h

typedef _Packed struct {
    vint            min;                /* initial size                       */
    vint            inc;                /* increment size                     */
    vint            max;                /* maximum size                       */
    vint            siz;                /* current size                       */
    vint            cur;                /* current data length                */
    void            *dta;               /* data pointer                       */
    } DTABUF;

#define dtb_dtaptr(mDBP)                (mDBP->dta)
#define dtb_dtalen(mDBP)                (mDBP->cur)

DTABUF              *dtb_crt(vint minsiz,vint incsiz,vint maxsiz);
DTABUF              *dtb_dlt(DTABUF *dbp);
vint                dtb_adddta(DTABUF *dbp,const byte *xlt256,const void *dtaptr,vint dtalen);
vint                dtb_addtxt(DTABUF *dbp,const byte *xlt256,const byte *format,...);
vint                dtb_rmvdta(DTABUF *dbp,vint len);
vint                dtb_reset(DTABUF *dbp);
void                *dtb_elmptr(DTABUF *dbp,vint elmidx,vint elmlen);

PS: vint was simply a typedef of int - I used it to remind me that it's length was variable from platform to platform (for porting).

I think what Adam Rosenfield posted is the correct way of doing OOP in C. I'd like to add that what he shows is the implementation of the object. In other words the actual implementation would be put in the .c file, while the interface would be put in the header .h file. For example, using the monkey example above:

The interface would look like:

//monkey.h

    struct _monkey;
    
    typedef struct _monkey monkey;
    
    //memory management
    monkey * monkey_new();
    int monkey_delete(monkey *thisobj);
    //methods
    void monkey_dance(monkey *thisobj);

You can see in the interface .h file you are only defining prototypes. You can then compile the implementation part " .c file" into a static or dynamic library. This creates encapsulation and also you can change the implementation at will. The user of your object needs to know almost nothing about the implementation of it. This also places focus on the overall design of the object.

It's my personal belief that oop is a way of conceptualizing your code structure and reusability and has really nothing to do with those other things that are added to c++ like overloading or templates. Yes those are very nice useful features but they are not representative of what object oriented programming really is.

http://www.ffmpeg.org/">ffmpeg</a> (a toolkit for video processing) is written in straight C (and assembly language), but using an object-oriented style. It's full of structs with function pointers. There are a set of factory functions that initialize the structs with the appropriate "method" pointers.

If you really thinks catefully, even standard C library use OOP - consider FILE * as an example: fopen() initializes an FILE * object, and you use it use member methods fscanf(), fprintf(), fread(), fwrite() and others, and eventually finalize it with fclose().

You can also go with the pseudo-Objective-C way which is not difficult as well:

typedef void *Class;

typedef struct __class_Foo
{
    Class isa;
    int ivar;
} Foo;

typedef struct __meta_Foo
{
    Foo *(*alloc)(void);
    Foo *(*init)(Foo *self);
    int (*ivar)(Foo *self);
    void (*setIvar)(Foo *self);
} meta_Foo;

meta_Foo *class_Foo;

void __meta_Foo_init(void) __attribute__((constructor));
void __meta_Foo_init(void)
{
    class_Foo = malloc(sizeof(meta_Foo));
    if (class_Foo)
    {
        class_Foo = {__imp_Foo_alloc, __imp_Foo_init, __imp_Foo_ivar, __imp_Foo_setIvar};
    }
}

Foo *__imp_Foo_alloc(void)
{
    Foo *foo = malloc(sizeof(Foo));
    if (foo)
    {
        memset(foo, 0, sizeof(Foo));
        foo->isa = class_Foo;
    }
    return foo;
}

Foo *__imp_Foo_init(Foo *self)
{
    if (self)
    {
        self->ivar = 42;
    }
    return self;
}
// ...

To use:

int main(void)
{
    Foo *foo = (class_Foo->init)((class_Foo->alloc)());
    printf("%d\n", (foo->isa->ivar)(foo)); // 42
    foo->isa->setIvar(foo, 60);
    printf("%d\n", (foo->isa->ivar)(foo)); // 60
    free(foo);
}

This is what may be resulted from some Objective-C code like this, if a pretty-old Objective-C-to-C translator is used:

@interface Foo : NSObject
{
    int ivar;
}
- (int)ivar;
- (void)setIvar:(int)ivar;
@end

@implementation Foo
- (id)init
{
    if (self = [super init])
    {
        ivar = 42;
    }
    return self;
}
@end

int main(void)
{
    Foo *foo = [[Foo alloc] init];
    printf("%d\n", [foo ivar]);
    [foo setIvar:60];
    printf("%d\n", [foo ivar]);
    [foo release];
}

My recommendation: keep it simple. One of the biggest issues I have is maintaining older software (sometimes over 10 years old). If the code is not simple, it can be difficult. Yes, one can write very useful OOP with polymorphism in C, but it can be difficult to read.

I prefer simple objects that encapsulate some well-defined functionality. A great example of this is GLIB2, for example a hash table:

GHastTable* my_hash = g_hash_table_new(g_str_hash, g_str_equal);
int size = g_hash_table_size(my_hash);
...

g_hash_table_remove(my_hash, some_key);

The keys are:

1. Simple architecture and design pattern
2. Achieves basic OOP encapsulation.
3. Easy to implement, read, understand, and maintain

I'm a bit late to the party here but I like to avoid both macro extremes - too many or too much obfuscates code, but a couple obvious macros can make the OOP code easier to develop and read:

/*
 * OOP in C
 *
 * gcc -o oop oop.c
 */

#include <stdio.h>
#include <stdlib.h>
#include <math.h>

struct obj2d {
    float x;                            // object center x
    float y;                            // object center y
    float (* area)(void *);
};

#define X(obj)          (obj)->b1.x
#define Y(obj)          (obj)->b1.y
#define AREA(obj)       (obj)->b1.area(obj)

void *
_new_obj2d(int size, void * areafn)
{
    struct obj2d * x = calloc(1, size);
    x->area = areafn;
    // obj2d constructor code ...
    return x;
}

// --------------------------------------------------------

struct rectangle {
    struct obj2d b1;        // base class
    float width;
    float height;
    float rotation;
};

#define WIDTH(obj)      (obj)->width
#define HEIGHT(obj)     (obj)->height

float rectangle_area(struct rectangle * self)
{
    return self->width * self->height;
}

#define NEW_rectangle()  _new_obj2d(sizeof(struct rectangle), rectangle_area)

// --------------------------------------------------------

struct triangle {
    struct obj2d b1;
    // deliberately unfinished to test error messages
};

#define NEW_triangle()  _new_obj2d(sizeof(struct triangle), triangle_area)

// --------------------------------------------------------

struct circle {
    struct obj2d b1;
    float radius;
};

#define RADIUS(obj)     (obj)->radius

float circle_area(struct circle * self)
{
    return M_PI * self->radius * self->radius;
}

#define NEW_circle()     _new_obj2d(sizeof(struct circle), circle_area)

// --------------------------------------------------------

#define NEW(objname)            (struct objname *) NEW_##objname()


int
main(int ac, char * av[])
{
    struct rectangle * obj1 = NEW(rectangle);
    struct circle    * obj2 = NEW(circle);

    X(obj1) = 1;
    Y(obj1) = 1;

    // your decision as to which of these is clearer, but note above that
    // macros also hide the fact that a member is in the base class

    WIDTH(obj1)  = 2;
    obj1->height = 3;

    printf("obj1 position (%f,%f) area %f\n", X(obj1), Y(obj1), AREA(obj1));

    X(obj2) = 10;
    Y(obj2) = 10;
    RADIUS(obj2) = 1.5;
    printf("obj2 position (%f,%f) area %f\n", X(obj2), Y(obj2), AREA(obj2));

    // WIDTH(obj2)  = 2;                                // error: struct circle has no member named width
    // struct triangle  * obj3 = NEW(triangle);         // error: triangle_area undefined
}

I think this has a good balance, and the errors it generates (at least with default gcc 6.3 options) for some of the more likely mistakes are helpful instead of confusing. The whole point is to improve programmer productivity no?

If I were going to write OOP in C I would probably go with a pseudo-Pimpl design. Instead of passing pointers to structs, you end up passing pointers to pointers to structs. This makes the content opaque and facilitates polymorphism and inheritance.

The real problem with OOP in C is what happens when variables exit scope. There are no compiler-generated destructors and that can cause issues. Macros can possibly help, but it is always going to be ugly to look at.

#include "triangle.h"
#include "rectangle.h"
#include "polygon.h"

#include <stdio.h>

int main()
{
	Triangle tr1= CTriangle->new();
	Rectangle rc1= CRectangle->new();

	tr1->width= rc1->width= 3.2;
	tr1->height= rc1->height= 4.1;

	CPolygon->printArea((Polygon)tr1);

	printf("\n");

	CPolygon->printArea((Polygon)rc1);
}
    

Output:

6.56
13.12

Here is a show of what is OO programming with C.

This is real, pure C, no preprocessor macros. We have inheritance, polymorphism and data encapsulation (including data private to classes or objects). There is no chance for protected qualifier equivalent, that is, private data is private down the innheritance chain too. But this is not an inconvenience because I don't think it is necessary.

CPolygon is not instantiated because we only use it to manipulate objects of down the innheritance chain that have common aspects but different implementation of them (Polymorphism).

I'm also working on this based on a macro solution. So it is for the bravest only, I guess ;-) But it is quite nice already, and I'm already working on a few projects on top of it. It works so that you first define a separate header file for each class. Like this:

#define CLASS Point
#define BUILD_JSON

#define Point__define                            \
    METHOD(Point,public,int,move_up,(int steps)) \
    METHOD(Point,public,void,draw)               \
                                                 \
    VAR(read,int,x,JSON(json_int))               \
    VAR(read,int,y,JSON(json_int))               \

To implement the class, you create a header file for it and a C file where you implement the methods:

METHOD(Point,public,void,draw)
{
    printf("point at %d,%d\n", self->x, self->y);
}

In the header you created for the class, you include other headers you need and define types etc. related to the class. In both the class header and in the C file you include the class specification file (see the first code example) and an X-macro. These X-macros (1,2,3 etc.) will expand the code to the actual class structs and other declarations.

To inherit a class, #define SUPER supername and add supername__define \ as the first line in the class definition. Both must be there. There is also JSON support, signals, abstract classes, etc.

To create an object, just use W_NEW(classname, .x=1, .y=2,...). The initialization is based on struct initialization introduced in C11. It works nicely and everything not listed is set to zero.

To call a method, use W_CALL(o,method)(1,2,3). It looks like a higher order function call but it is just a macro. It expands to ((o)->klass->method(o,1,2,3)) which is a really nice hack.

See Documentation and the code itself.

Since the framework needs some boilerplate code, I wrote a Perl script (wobject) that does the job. If you use that, you can just write

class Point
    public int move_up(int steps)
    public void draw()
    read int x
    read int y

and it will create the class specification file, class header, and a C file, which includes Point_impl.c where you implement the class. It saves quite a lot of work, if you have many simple classes but still everything is in C. wobject is a very simple regular expression based scanner which is easy to adapt to specific needs, or to be rewritten from scratch.

Another way to program in an object oriented style with C is to use a code generator which transforms a domain specific language to C. As it's done with TypeScript and JavaScript to bring OOP to js.

You can try out COOP, a programmer-friendly framework for OOP in C, features Classes, Exceptions, Polymorphism, and Memory Management (important for Embedded code). It's a relatively lightweight syntax, see the tutorial in the Wiki there.

For me object orientation in C should have these features:

1. Encapsulation and data hiding (can be achieved using structs/opaque pointers)

2. Inheritance and support for polymorphism (single inheritance can be achieved using structs - make sure the abstract base is not instantiable)

3. Constructor and destructor functionality (not easy to achieve)

4. Type checking (at least for user-defined types as C doesn't enforce any)

5. Reference counting (or something to implement RAII)

6. Limited support for exception handling (setjmp and longjmp)

On top of the above it should rely on ANSI/ISO specifications and should not rely on compiler-specific functionality.

@Adam Rosenfield has a very good explanation of how to achieve OOP with C

Besides, I would recommend you to read

1. pjsip

A very good C library for VoIP. You can learn how it achieves OOP though structs and function pointer tables

2. iOS Runtime

Learn how iOS Runtime powers Objective C. It achieves OOP through isa pointer, meta class

Look at http://ldeniau.web.cern.ch/ldeniau/html/oopc/oopc.html. If nothing else reading through the documentation is an enlightening experience.

If you need to write a little code try this: https://github.com/fulminati/class-framework

#include "class-framework.h"

CLASS (People) {
    int age;
};

int main()
{
    People *p = NEW (People);

    p->age = 10;

    printf("%d\n", p->age);
}

The open-source Dynace project does exactly that. It's at https://github.com/blakemcbride/Dynace