Creating a dynamic custom array of nested structs in C - c

Question

I'm trying to create the following array:
"Fruits", 25, {
{"Apple", 2},
{"Grapes", 13},
{"Melon", 10}
}
"Meats", 40, {
{"Beef", 9},
{"Chicken", 27},
{"Pork", 4}
}
...
Feels like there's a more elegant way of doing what I got so far. Any feedback/samples on how to create this structure more efficient given the input struct would be appreciated.
#include <stdlib.h>
#include <stdio.h>
#include <string.h>
typedef struct Product {
char *name;
int qty;
} Prods;
typedef struct Category {
char *name;
int qty;
int prods_count;
Prods *prod;
} Cats;
typedef struct Inventory {
Cats *cat;
int cats_count;
} Inv;
struct tmp_input {
char name[12];
int qty;
char cat[12];
};
// return index if found
int in_array(Inv *inv, char *k) {
int i;
if (inv->cats_count == 0)
return -1;
for (i = 0; i < inv->cats_count; i++) {
if (strcmp (k, inv->cat[i].name) == 0) return i;
}
return -1;
}
int main () {
int i, j, exists = 0;
// temp struct.
struct tmp_input items[] = {
{"Apple", 2, "Fruit"}, {"Coke", 10, "Drink"}, {"Pork", 4, "Meat"},
{"Beef", 9, "Meat"}, {"Chicken", 27, "Meat"}, {"Melon", 10, "Fruit"},
{"Tea", 3, "Drink"}, {"Coffee", 20, "Drink"}, {"Grapes", 13, "Fruit"}
};
size_t len = sizeof (items) / sizeof (struct tmp_input);
Inv *inven = malloc(sizeof(Inv));
inven->cats_count = 0;
inven->cat = calloc(1, sizeof(Cats));
for (i = 0; i < len; i++) {
exists = in_array(inven, items[i].cat);
// category does not exist
if (exists == -1) {
inven->cat = realloc(inven->cat, sizeof(Cats) * (inven->cats_count + 1));
inven->cat[inven->cats_count].name = strdup(items[i].cat);
inven->cat[inven->cats_count].qty += items[i].qty;
inven->cat[inven->cats_count].prods_count = 1;
inven->cat[inven->cats_count].prod = calloc (1, sizeof (Prods));
inven->cat[inven->cats_count].prod->name = strdup (items[i].name);
inven->cat[inven->cats_count].prod->qty = items[i].qty;
inven->cats_count++;
}
// category found
else {
inven->cat[exists].qty += items[i].qty;
int size = inven->cat[exists].prods_count + 1;
inven->cat[exists].prod = realloc(inven->cat[exists].prod, sizeof(Prods) * (size));
inven->cat[exists].prod[size - 1].name = strdup (items[i].name);
inven->cat[exists].prod[size - 1].qty= items[i].qty;
inven->cat[exists].prods_count++;
}
}
for (i = 0; i < inven->cats_count; i++) {
printf("%3d %s\n", inven->cat[i].qty, inven->cat[i].name);
for (j = 0; j < inven->cat[i].prods_count; j++) {
printf("%3d %s\n", inven->cat[i].prod[j].qty, inven->cat[i].prod[j].name);
}
}
return 0;
}

Answer 1

You aren't allocating any memory for the Prod array.
Something like
...
if (exists == -1) {
inven->cat = realloc(inven->cat, sizeof(Cats) * (inven->cats_count + 1));
inven->cat[inven->cats_count].name = items[i].cat;
inven->cat[inven->cats_count].qty += items[i].qty;
// Allocate memory for 1 product
inven->cat[inven->cats_count].prods_count = 1;
inven->cat[inven->cats_count].prod = malloc (sizeof (Prods));
// Now allocate space and copy the name
inven->cat[inven->cats_count].prod->name = strdup (items[i].name + 1);
inven->cats_count++;
}
...
I will leave it to you to handle the case where there are more than 1 product in a category, where you'll need to reallocate the memory again.
Another error is that you need to allocate and copy the category name
inven->cat[inven->cats_count].name = items[i].cat;
should be replaced by
inven->cat[inven->cats_count].name = strdup (items[i].cat);
This is because the items array does not exist outside of this function, so if you just do
inven->cat[inven->cats_count].name = items[i].cat;
then after you leave this function, invent->cat[inven->cats_count].name will point to garbage memory.
A final suggestion would be to split each structure into a function that handles creation of it, just to clean up the code.
--- edit to add comments on Abstract Data Types
Arrays are useful if you have data that you know you will be accessing via indices. If you don't know the index of the item you want (as in this case), an array is less useful.
Unlike other comments, I don't think using a Linked List really gives you anything useful. Linked Lists are useful when you need to walk sequentially through all the items, without really caring where they are in the list. It seems to me that the most common use case for a system like you are creating is searching: Do we have any Fruit in stock? Add 10 cases of Coke to the inventory... those sorts of things.
Also, you only want a single entry for each category/product. You don't want 3 Fruit categories in the data. Both arrays and linked lists don't really have any restrictions on adding the same structure multiple times. This means that every time you'll need to check the whole list to see if you need to add the new structure.
For that reason, I'd certainly make both the Categories and products arrays into hashtables (or called a dictionary in some languages) that map name -> structure. This will speed up your search as you don't have to search the entire dataset every time and will prevent you from adding the same structure multiple times.
Wikipedia article on Hashtables: http://en.wikipedia.org/wiki/Hashtable

Answer 2

Here's an example of how to set up the structures dynamically (simpler than a linked list, but not as flexible).
typedef struct Product {
char *name;
int qty;
} Prods;
typedef struct Category {
char *name;
int qty;
int prods_count;
Prods *prod; // dynamic array of Products
} Cats;
The structures as they were.
struct Category categ[10];
An arbitrary number of categories, for now take categ[0] for 'Fruits'.
Next dynamically create an array of 10 product structures:
Prods *prod_array = malloc(sizeof(Prods) * 10); // i.e. prod_array[0] to [9]
Now just store the array in the category structure:
categ[0].prod = prod_array;
categ[0].prods_count = 10;
If you need to access the product name, it's just: categ[i].prod[j].name
Now, if you need another 10 products, you can just use realloc to increase the size of the array, and update its count.
Put all these things in functions, and the code isn't too complex.