Say I write in a file
Mesh: 1
Vertices: 345
Indices: 123
V: 1,3,4 1,4,5 ..
Mesh: 2
Vertices: 456
Indices: 42
etc.
How do I go about seeking at any position? E.g. I want to go to Vertices: of Mesh 2 or V: of Mesh 3 etc.
What's the proper way to go about these things?
You would normally use a binary format. One way would be to allocate a certain amount of space as a header in the file. Here, you put the mesh numbers, vertex and index counts, and an offset into the file where the vertex data begins. You read the header when loading the file, then seek to the appropriate place to read the data you want.
Open the file for reading, and read a line until End-of-File (EOF) is reache. For each read line, check if the line matches with your query. If match, report and return. Otherwise, move on to the next line.
The main cognitive work is checking for matches. Have a well defined format and easily parse-able format for the lines to make your job easy.
There is no efficient way of random seeking in text file formats. This is because you cannot know the right offset in the file without reading all the contents before. The only way of processing these is sequential - from beginning to end.
So read and parse the entire file into some data structure in memory. Then use this structure instead of the file as needed.
If the file is too large to keep everything in memory (these days it's highly improbable), read through the file without storing everything in memory - instead store just file offsets to the beginning of each Mesh in an array. Then you can easily seek to the right place.
As other answers have pointed out, C can only seek to byte offsets within files.
However, if your "Mesh" objects are always stored in the file in numerical order, then you do not have to read the entire file sequentially to find the Mesh that you are after. You can instead perform a binary search on the file - whenever you seek to a position in the file, scan ahead to find the next Mesh.