Printable View
Hi
I'd like to edit a large file (~1GB) such that I append a small block of bytes at the beginning of the file. The only way I know of doing that is to create a new file, write the new bytes and then the large file into the stream.
Under Windows, is there a better solution?
best regards
Sebastian
What are you prepending to a 1GB file? A jump instruction for a virus routine? ;-)
I'm afraid afaik, you will either need to read in the entire file and write it out again or write your bytes to a new file, then copy byte-by-byte the old file. Then move new to old. Presto.
Also, what language are you using?
QuantumPete
The only other way is to do it "in-place" using CreateFileMapping and MapViewOfFile. However, this method is complicated due to the size of the file.
When doing this the "traditional" way as you describe, you may gain a little performance by using FILE_FLAG_SEQUENTIAL_SCAN and FILE_FLAG_NO_BUFFERING and performing your reads and writes in multiples of the sectors size. See CreateFile API for more info.
gg
Hi
I'm editing a medical image and want to edit header information. Unfortunately, the header can vary in length but the actual picture data remains the same.
I'll give it a try with the CreateFile and Caching modes for sequential access.
Thanks again
Sebastian
>> What are you prepending to a 1GB file? A jump instruction for a virus routine? ;-)
Don't be giving people any ideas. :D
> I'm editing a medical image and want to edit header information. Unfortunately,
> the header can vary in length but the actual picture data remains the same.
So like you're adding some kind of comment information (like the patient name) to the image?
Do you have any control over how the file gets created in the first place?
If you do, you might want to 'pre-pad' the header information when the file is first created with a useful maximum size. Then all you have to do (most of the time) is just overwrite the header with your new data and the rest of the file remains unchanged.
Only in rare cases (when you need to add more than your max guess) would you need to recopy the entire file.
If the header varies in length then it should have a data member that indicates its size. If not how would any other program know where the image data started since the header length could vary from file to file? It will either have a member that indicates the header size or an offset into the file where the image data begins.Quote:
I'm editing a medical image and want to edit header information. Unfortunately, the header can vary in length but the actual picture data remains the same.
A third possibility is that the header could be relying on the presence of some byte pattern that marks the end of the header/start of the image data. In which case you'd (or any program) would have to scan through the file first, searching for this pattern. Probably not the best of header designs, but it is a possibility.