How do you seeking a position of a filestream without knowing it?

[bounty-hunter]

Hey guys, I'm basically writing untyped data into a single file and adding them sequentially.
For example
Writebuffer(VersionID,Sizeof(versionID); // will write the version id of the file
len1 := lengthof(string1);
Writebuffer(len1,sizeof(len1)); // will write the size of the length of the string so we can read it
Writebuffer(string1,len1); // will write the string1 to the file

// so if I happen to add multiple file types say I add bitmaps and videofiles and other data types, to read them would be easy I would just read the size of it first then read the datatype but for this to happen I would have to read the file sequentially for example :

Readbuffer(versionid,sizeof(versionID); // reads the first thing in the file since we know the structure
the position moves 4 bytes because versionID is an Int which is 4bytes of data so there fore the filstream.position = 4 right now.

So what I'm wondering is it possible to skip to a particular part of the filestream without having to read everything in a sequencing order? But for that we need to know the size in bytes right? what would be the easiest way to do this ?

 

[Adult]

Well, the easiest way would be to simply read everything to that point, then disregard it. In general though, you should be using Seekg.

You must be registered to see links

Could be a lazy mofo, not sure if it would work in your case, and run a is ignore.

istream& ignore (streamsize n = 1, int delim = EOF);

You must be registered to see links

But as said, I think SeekG is more what you're looking for.

Good luck!

 

[GHOST107]

if you hve multiple big data types, add a header to your file that will specify the file poisition of the data in the file.

 

[Negata]

If possible, you could also first gather all your data and write their positions-to-be at the beginning of the file and then continuously write the actual data at the end. So the offsets at the beginning would be just like memory pointers, but pointers to locations in a file. Then you wouldn't need to sequentially read all of the file but only the pointers section.

Actually this would also work in reverse: keep writing into the file first but save the starting offset for each of the writes as you go. When closing the file, write the pointers section at the end of the file.

 

[TheAJ]

Tracking the index by adding it to the beginning (best if you do not alter the file, as altering the header requires rewriting the whole file) or end of the file (as mentioned above). Alternatively (if you're always changing the file), you can keep track of the indexes in a separate file dedicated solely for indexing purposes.

If you don't want to do the above, add a string (name of the file?) along with the length before the file data. Once you've done that, start reading the file, read the string (name), if it matches the one you want, read the length and then read the data. If it doesn't match, read the length, and seek/skip that many bytes, and read the next string.

 

[bounty-hunter]

Tracking the index by adding it to the beginning (best if you do not alter the file, as altering the header requires rewriting the whole file) or end of the file (as mentioned above). Alternatively (if you're always changing the file), you can keep track of the indexes in a separate file dedicated solely for indexing purposes.

If you don't want to do the above, add a string (name of the file?) along with the length before the file data. Once you've done that, start reading the file, read the string (name), if it matches the one you want, read the length and then read the data. If it doesn't match, read the length, and seek/skip that many bytes, and read the next string.

The reading string and length and skip, until matches is exactly what im looking for. I'm surprised I didn't think of this, its genius haha. But instead i'm going to use integers because they are a fixed size of 4 bytes, that way i can id each file by 1,2,3,4,5 since for my case im loading images. Unfortunately the strings do not have a fixed size so I'd have to write the length of the string as well. Thanks guys for the suggestions !

 

[TheAJ]

The reading string and length and skip, until matches is exactly what im looking for. I'm surprised I didn't think of this, its genius haha. But instead i'm going to use integers because they are a fixed size of 4 bytes, that way i can id each file by 1,2,3,4,5 since for my case im loading images. Unfortunately the strings do not have a fixed size so I'd have to write the length of the string as well. Thanks guys for the suggestions !

an integer is much more efficient than a variable-sized string. Even a 1-byte integer is good if you have less than 257 entries!

 

[bounty-hunter]

Ok now I have another problem, Say I have 6 images in the one file, if I want to delete the Image and Index 2 how would I go about doing that?

 

[שเ๒єtгเ๒є]

Ok now I have another problem, Say I have 6 images in the one file, if I want to delete the Image and Index 2 how would I go about doing that?

Unfortunately changing the file length (except for appending data on the end) results in rewriting the whole thing. You could, however, replace the affected data by 0's, that can be ignored by your seek function.