Creating an Index File

NOTE:

These are a little sketchy...I'll fill in details 'soon'.

What is an Index File?

when one file contains offsets into another file as its data, it is said to index the second file or to be the index file of the second
example:
- Data File:
```
Jason James
Tammy James
George Freedman
Smith Barney
Mr. Bubble
```
- Index File:
```
0 12 24 40 53
```
- each element of the index file is the offset of a name from the data file
example:
- Data File:
```
x = 4
y = 9
z = 8
x=-3
z=-9
y=-4
y=5
z=9
y=8
x=12
x=3
y=7
z=7
```
- Index File:
```
0 18 33 41 50
```
- each element of the index file is the offset of a point from the data file (note that there are missing labels but block structure requires we stop)
when doing this by hand remember that there is a newline at the end of each line which counts as a byte in the offset

Creating Index Files

you must create index files rather than transfer them (i.e. via ftp) because newline characters change between platforms (i.e. WinDOS uses a carriage-return/line-feed pair -- 2 bytes -- whereas Unix uses simply a line-feed -- 1 byte)
by reading through the data once and using a tellg before each data group's read to tell you the correct offset, you can form an index file

    // open data file for input
    // skip any comments and line up at position of first data item
    // open index file for output -- we are creating it
    first_data = data_loc = datafile.tellg();
    datafile.peek();
    while (!datafile.eof())
    {
        data.input(datafile);
        indexfile << data_loc << ' ';
        data_loc = datafile.tellg();
        datafile.peek();
    }
    indexfile.close();    // done with written version -- now use
                          // to read
    datafile.clear();              // clear eof flag
    datafile.seekg(first_data);    // reset to beginning

note we reset to the beginning of the data file instead of closing it. we'll normally need to re-process the file using the index file as positions to seek to

Benefits

once an index file has been created, we can discover the position of data items with relative ease
reading through the index file is often faster than reading through the original data file to find, say, the 19th item
if there were inter-item comments, we won't have to read through them anymore, either -- we simply seek to the start position of each item we want to read
index files become more useful the larger the data blocks are (less useful for a set of simple numbers, more useful for a database of all information about students at Harper)