Awestruck by C Structs

Awestruck by C Structs:

 

C is an amusing language, because there are some people who are used to writing in assembly or (I assume) knitting ones and zeros together by hand that refer to it as a “high-level language.” And it is, to a point! It has concepts about memory management, variable types, functions, complex branching and control constructs, and more! But, at the same time, there are those who are used to even higher-level languages like Python, Java, PHP, Ruby, JavaScript, and similar languages that look at C like some sort of low-level, bare-metal secret wizard magic impenetrable forbidden knowledge. And there’s no shame in that either. Until very recently, when I forced myself to buckle down and just. write. a butt. ton. of C. And Google the things I didn’t know. And trust me: it was (is) a lot of things.

This post is one of what I hope are a few, providing some clarity into some things that I didn’t understand about C at first. They are the pieces that allow higher-level logic and bundling of data into more manageable packages. They are the closest thing that C has to objects in an Object-Oriented Programming mindset. They are Structs.

What is a Struct?

Struct or “structure” is a way of declaring that several fields of data are related. They look like this in the simplest form:

struct Fraction 
{
  int numerator;
  int denominator;
};
Enter fullscreen mode Exit fullscreen mode

That’s it. We’ve taken two concepts, integers named “numerator” and “denominator,” and we’ve denoted that they can be bundled up together in one concept called a “fraction.”

Then you can use them like this:

#include 

int main() 
{
  struct Fraction f;
  f.numerator = 3;
  f.denominator = 4;

  printf("%d/%d\n", f.numerator, f.denominator);
  //=> 3/4
}
Enter fullscreen mode Exit fullscreen mode

The full name of the type you give to the variable f is struct Fraction, which is essentially saying, this is a struct, but what kind of pre-defined struct is it? A fraction. With what we’ve got here, you cannot simply say:

  Fraction f;
Enter fullscreen mode Exit fullscreen mode

But that possibility is within reach… and we’ll get there.

You’ll notice that the above example is a little verbose. You can speed up the initialization with some braces.

#include 

int main() 
{
  struct Fraction f = {3, 4};
  // ...
}
Enter fullscreen mode Exit fullscreen mode

This assigns values to the fields in the order that they were originally defined. And, if you want to be a little more explicit, depending on your compiler version:

  struct Fraction f = {.numerator = 3, .denominator = 4};
Enter fullscreen mode Exit fullscreen mode

Or, a little eeeevil

  struct Fraction f = {.denominator = 4, .numerator = 3};
Enter fullscreen mode Exit fullscreen mode

typedef Aliases

Let’s be real. It’s a real drag to type struct before each of these variable declarations. Lucky for you, there’s a way around this: typedef. You can define a type alias in the normal namespace. More on that in a second, but let’s see it in action first.

typedef struct Date 
{
  int year;
  int month;
  int day;
} Date;

int main() 
{
  Date today = {2020, 3, 13};

  // Using the struct name still works:
  struct Date tomorrow = {2020, 3, 14};
}
Enter fullscreen mode Exit fullscreen mode

The first Date is the name of the struct. The second one is the alias in the main namespace. You can do some crazy things here. You can make the alias different than the struct name:

typedef struct Jerome 
{
  int year;
  int month;
  int day;
} Date;
Enter fullscreen mode Exit fullscreen mode

You can even omit the struct name completely, making it so you can onlyreference this struct via its alias. This is the option that makes the rest of your code the least verbose.

typedef struct 
{
  int year;
  int month;
  int day;
} Date;
Enter fullscreen mode Exit fullscreen mode

But watch out. Creating aliases has a bit of a gotcha attached: since aliases are in the main namespace, the same place that function names live, you can run into crashes:

// This is OK, since struct names are in their own
// namespace, separate from function names
struct love 
{
  int magnitude;
};

void love() 
{
  printf("I love you!\n");
}

int main()
{
  struct love me_n_wife = {1000000}; //  I love you!
}


// This would cause an error:
typedef struct
{
  int red;
  int green;
  int blue;
} color;

void color()
{
  printf("Coloring is fun!\n");
}

int main()
{
  color bada55 = {186, 218, 85};
  color();
  //=> error: redefinition of 'color' as different kind of symbol
}
Enter fullscreen mode Exit fullscreen mode

Using Structs with Functions

Passing by Value (The Default)

Have no fear. Even though structs are a little bit fancier than regular variables, they can still be used with functions with no extra fuss.

#include 

struct Crate
{
  int length;
  int width;
  int height;
};

int volume(struct Crate);

int main()
{
  struct Crate c = {5, 12, 13};
  printf("The volume is %d.\n", volume(c));
  //=> The volume is 780.
}

int volume(struct Crate crate)
{
  return crate.length * crate.width * crate.height;
}
Enter fullscreen mode Exit fullscreen mode

Doing it this way is called Passing by Value. That’s how C handles all arguments to functions unless you explicitly give it a reference to something. More on that later. The only downside of this is that C makes an entire copy of your struct and provides that to be used inside the function. This uses more memory, especially if your structs are giant. The upside of that, is that any changes the function makes to that struct don’t affect the original. The downside of that is that the function can’t make any changes to the original.

#include 

struct Fruit
{
  char name[10];
  int seeds;
  int age;
};

void forgetToEat(struct Fruit);

int main()
{
  struct Fruit banana = {
    .name = "banana",
    .seeds = 9,
    .age = 1,
  };

  forgetToEat(banana);
  printf("Oh, no! I forgot to eat the %s and now it's %d days old!\n", banana.name, banana.age);
  //=> Oh, no! I forgot to eat the banana and now it's 1 days old!
}

void forgetToEat(struct Fruit fruit)
{
  fruit.age = 1000;
}
Enter fullscreen mode Exit fullscreen mode

See? The banana didn’t age! IT’S AN IMMORTAL BANANA. JK, we just haven’t learned about Pass by Reference yet.

Passing by Reference

Passing arguments by reference means that you don’t provide a copy of your argument to the function; you provide a reference (or pointer to its memory location). Then, the function can access the argument’s memory directly, and manipulate it in ways that show up outside of the function after its execution has finished.

Here’s how that looks. This is the same example as above, except it works.

#include 

struct Fruit
{
  char name[10];
  int seeds;
  int age;
};

void forgetToEat(struct Fruit *); // 1

int main()
{
  struct Fruit banana = {
    .name = "banana",
    .seeds = 9,
    .age = 1,
  };

  forgetToEat(&banana); // 2
  printf("Oh, no! I forgot to eat the %s and now it's %d days old!\n", banana.name, banana.age);
  //=> Oh, no! I forgot to eat the banana and now it's 1000 days old!
}

void forgetToEat(struct Fruit *fruit) // 3
{
  (*fruit).age = 1000; // 4
}
Enter fullscreen mode Exit fullscreen mode

Here are the main changes:

  1. We update the function to expect, rather than a Fruit struct, a pointer to a Fruit struct.
  2. We call the function and, instead of passing it the struct directly, to make a copy, we use the & operator to send it the memory address of the banana.
  3. Same as #1, our function expects a pointer to a Fruit, not an actual Fruit.
  4. We use the dereferencing operator to get at the value that lives at that memory address. Then we update that struct’s age.

This concept of reaching through a pointer to read and update fields of a referenced struct is so common that they have an additional convenience operator: ->. We can update the function body above thus.

void forgetToEat(struct Fruit *fruit)
{
  fruit->age = 1000;
}
Enter fullscreen mode Exit fullscreen mode

Note how it replaces the parentheses, the dereferencing operator, and the dot operator.

Going OOP with Structs (Kind Of)

Now, the OOP purists will probably have thoughts on this, but, if you squint and cock your head to the side a little, you can begin to encapsulate some data and provide some operations for your structs using these techniques. This is a little long for a single code example, so we’ll break it into steps.

We are going to create a Student type and the operations for assigning them a letter grade, failing them wholesale, and calculating their GPA. We’ll also need a constructor to create Students, and, because it’s C, we’ll need a destroyer to clean up our objects when we’re done with them.

The Struct Definition

Let’s get things started. Our student will have fields for their name and an array of letter grades for each of their classes.

#define NUM_CLASSES 10
#define MAX_NAME_LENGTH 20

typedef struct
{
  char name[MAX_NAME_LENGTH + 1];
  char grades[NUM_CLASSES];
} Student;
Enter fullscreen mode Exit fullscreen mode

Note that we define macros for the magic numbers we use because we’re good C citizens who don’t like to hide hard-coded values.

The Constructor

Now we need a way to create new students. Yes, we could do this by hand each time, but that’s error prone, and we might forget an important step sometime. Better to write it once and make sure it’s right. Or, at least, we’ll have all the bugs written in one place.

#include 
#include 
// ... snip ...

Student *makeStudent(char name[])
{
  if (strlen(name) > MAX_NAME_LENGTH)
  {
    return NULL;
  }

  Student *s = (Student *)malloc(sizeof(Student));
  strcpy(s->name, name);
  s->name[MAX_NAME_LENGTH] = '\0';

  for (int i = 0; i grades[i] = 'A';
  }

  return s;
}
Enter fullscreen mode Exit fullscreen mode

Our constructor is actually allocating a Student on the heap and returning a pointer, which you can see by the type of the function’s return value. It takes a string and stores that in the Student’s name. We also do a sanity check to make sure the name is small enough to fit. Returning NULL from a function that encounters an issue creating some sort of object is reasonably common. Lots of pieces of the C standard library does it, and you’ll start constantly finding yourself checking if something is NULL.

We also need to include  in order to use malloc and string.h in order to use strcpy. Because we’re benevolent school administrators, we also initialize our students’ grades to all A’s.

Because it would be lame to write a bunch of code and not have it work until the end, let’s write some main logic to exercise our code and find any warnings. We’ll also need  to print output.

#include 
// ... snip ...

int main()
{
  Student *s = makeStudent("[your name]");
  printf("%s is my name.\n", s->name);
  return 0;
}
Enter fullscreen mode Exit fullscreen mode

Give it a compile:

$ gcc student.c -Wall -o student
$ ./student
[your name] is my name.

Enter fullscreen mode Exit fullscreen mode

Everything working well? Great! Let’s move on.

The Destructor

As it stands right now, we either have to clean up any students we create by hand or let the operating system clean them up when our program ends. This is fine for our tiny example, but if we started writing a bigger application that was constantly creating students, we’d want to make sure they were getting cleaned up properly to minimize the chances of a memory leak or pointer errors.

void destroyStudent(Student *s)
{
  free(s);
  s = NULL;
}
Enter fullscreen mode Exit fullscreen mode

Since our student doesn’t own any other allocated values (all of its members are just arrays of characters, not pointers), it doesn’t have to do anything fancy. We can just free it right away. The pointer still exists and points to that memory, however, so it’s considered good practice to update the pointer to point to NULL. This helps find memory bugs down the road.

Even though it won’t change our functionality, let’s go ahead and clean up our student in the main function.

int main()
{
  Student *s = makeStudent("Ryan");
  printf("%s is my name.\n", s->name);
  destroyStudent(s);  // <===
  return 0;
}
Enter fullscreen mode Exit fullscreen mode

OK! With housekeeping complete, we can move onto the fun stuff! Let’s do the GPA calculation first, so we can have something interesting to do with the other functions.

Calculating the GPA

Let’s jump right into the code. The most important thing to notice as we start writing functions that operate on our “objects” is that they all have an argument that accepts a pointer to a Student. This is just about the closest we can get in C to making “methods” for our “classes.”

float gpa(Student *s)
{
  float total = 0;
  for (int i = 0; i grades[i])
    {
    case 'A':
      total += 4;
      break;
    case 'B':
      total += 3;
      break;
    case 'C':
      total += 2;
      break;
    case 'D':
      total += 1;
      break;
    default:
      break;
    }
  }

  return total / NUM_CLASSES;
}

Enter fullscreen mode Exit fullscreen mode

A GPA is really just a weighted average, with letter grades weighted certain amounts. One common US system is to make A’s worth 4 points, B’s worth 3, C’s worth 2, D’s worth 1, and award no points for failure.

Our gpa() function runs through our grades, awards adds each score up, and returns the average by dividing by the number of scores.

Since our students start out with all A’s, we’d expect that to work out to a GPA of 4.0. Let’s exercise our new function:

int main()
{
  Student *s = makeStudent("Ryan");
  printf("%s is my name.\n", s->name);
  printf("My initial GPA is %0.02f.\n", gpa(s));
  s->grades[2] = 'C';
  printf("Rough semester for history. Now I'm at %0.02f.\n", gpa(s));
  destroyStudent(s);
  return 0;
}
Enter fullscreen mode Exit fullscreen mode

And give it the ol’ run-aroony:

$ gcc student.c -Wall -o student
$ ./student
[your name] is my name.
My initial GPA is 4.00.
Rough semester for history.  Now I'm at 3.80.
Enter fullscreen mode Exit fullscreen mode

Yay! The next step is to add functionality for teachers to award scores to students. We can already do that now manually, but, for the sake of example, and to add some error checking, we’ll do a function.

Assigning Letter Grades

Here’s the next function:

void assign(Student *s, Class class, char grade)
{
  // Don't allow assignment to a class that doesn't exist.
  if (class >= NUM_CLASSES)
  {
    return;
  }

  // Only allow A-F grading.
  if (grade  'F')
  {
    return;
  }

  s->grades[class] = grade;
}
Enter fullscreen mode Exit fullscreen mode

For this, it seems OK to let grade assignments fail silently. If a teacher can’t figure out how to assign an A-F to a class that exists, then they don’t deserve to change the student’s grade from an A.

We’re going to make one other change here for some added readability. You’ll notice that the grades[] array is indexible via integers, so you could give a student a B in class 4. But how lame is that? What class is that? Let’s create an enum that defines names for our grades. I don’t want to dive into the nitty gritty of enums in this article, but they’re one of my favorite things ever, and I definitely want to make that article in the coming weeks. Right now, just trust me that we’re basically giving names to the numbers 0-9 (our 10 classes).

typedef enum
{
  MATH,
  SCIENCE,
  HISTORY,
  ENGLISH,
  SPANISH,
  PHYSICS,
  PHILOSOPHY,
  GOVERNMENT,
  COMPUTER_SCIENCE,
  SHOP,
} Class;
Enter fullscreen mode Exit fullscreen mode

So now, we can update our main() function accordingly.

int main()
{
  Student *s = makeStudent("Ryan");
  printf("%s is my name.\n", s->name);
  printf("My initial GPA is %0.02f.\n", gpa(s));
  assign(s, HISTORY, 'B');
  printf("Rough semester for history. Now I'm at %0.02f.\n", gpa(s));
  destroyStudent(s);
  return 0;
}
Enter fullscreen mode Exit fullscreen mode

Awesome. One more function to go: the SUPER FAIL!

Fail all Classes

Is it over-the-top, vindictive overkill? You betcha! Let’s write it.

void fail(Student *s)
{
  for (int i = 0; i < NUM_CLASSES; i++)
  {
    assign(s, i, 'F');
  }
}
Enter fullscreen mode Exit fullscreen mode

And the code to exercise it:

int main()
{
  Student *s = makeStudent("Ryan");
  printf("%s is my name.\n", s->name);
  printf("My initial GPA is %0.02f.\n", gpa(s));
  assign(s, HISTORY, 'B');
  printf("Rough semester for history. Now I'm at %0.02f.\n", gpa(s));

  printf("\n Hehehe, I'm going to vandalize school property...\n");
  fail(s);

  printf("Oh no. Now my GPA is %0.02f.\n", gpa(s));
  destroyStudent(s);
  return 0;
}
Enter fullscreen mode Exit fullscreen mode

When we run it:

$ gcc student.c -Wall -o student
$ ./student
[your name] is my name.
My initial GPA is 4.00.
Rough semester for history.  Now I'm at 3.90.
Hehehe, I'm going to vandalize school property...
Oh no.  Now my GPA is 0.00.
Enter fullscreen mode Exit fullscreen mode

Final Notes on this Example

This isn’t necessarily exactly the same as your ideal Object-Oriented scenario. Right now, all members and functions are public. You can use header files for some privacy, and that might be a topic for the future. But hopefully, this helps make C a little more accessible to someone used to working with actual classes, methods, and properties.

Struct LYFFFFEEEE

So there you are. Zero to more than you maybe thought you wanted to know about C Structures in almost 2800 words. No longer are you bound to simple scalar datatypes with one value each. No, now you can have like two or three values tied up together in one! Go forth and conquer! And definitely reach out if you have any questions.

from Tumblr https://generouspiratequeen.tumblr.com/post/634668112335224832

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s