http://www.cdrom.com/pub/infozip/zlib/

If you are talking about compressed files, this usually involves
encoding and decoding schemes.

Packed files are usually just a container to hold the individual
files unmodified (they can compressed or uncompressed). Which is
good if you want to create some sort of file packer program to
quickly add and remove files without touching them.

------------------
Dance with me......

If your index file is limited to having only one type of data in that is defined as the following:

typedef struct _packedIndex
{
int resource_id; // unique id
char name[20]; // name of resource
unsigned long offset; // offset from start of file
unsigned long length; // length of data segment
}

The index file would just contain all the structures pointing to the data in the packed file. That way you can search much more quickly for specific data either by name, or by unique ID. The unique ID works well if your packing program is able to output an RESOURCE_INDEX.H file or some such that you can then refer to certain files by #define. In my reosurce packing utility I have a text input file like this:

# each line is a seperate entry
#
# resource name, resource path
INTRO_SCREEN1, C:\data\images\intro8bit.bmp
INTRO_SCREEN2, C:\data\images\intro24bit.bmp
GOLD_MINE, C:\data\images\sprites\goldmine8.bmp
..

Once the program reads this and puts together the resource.pak and resource.idx files, it ouputs an include filecalled "packed_index.h" that ,looks like this:

...

// resource IDs

#define INTRO_SCREEN1 0
#define INTRO_SCREEN2 1

and then I can load it in my function as

ret = LoadPakResource(INTRO_SCREEN1)..

my LoadPakResource then looks in the resource.idx for entry 0 (that's what INTRO_SCREEN1 is defined as), reads that and finds the offset in resource.pak .. it then opens resource.pak, seeks to the offset and loads the length variable worth of data..