Scanning a file with getBytes:range:? - objective-c

Question

I'm currently scanning a binary file with the following code:
while (offset < headerLength) {
memset(buffer, 0, sizeof(buffer));
[bookData getBytes:buffer range:NSMakeRange(0 + offset, 4)];
NSString *output = [NSString stringWithCString:buffer encoding:NSUTF8StringEncoding];
if ([output isEqualToString:#"EXTH"]) {
...
}
offset++;
}
However, I suspect that's not the correct way, because it is quite slow. How to search for a specific string then?
GOOD BUT SLOW:
while (offset < headerLength) {
memset(buffer, 0, sizeof(buffer));
[bookData getBytes:buffer range:NSMakeRange(0 + offset, 4)];
if (buffer[0] == 'E' && buffer[1] == 'X' && buffer[2] == 'T' && buffer[3] == 'H') {
//run my stuff
break;
}
offset++;
}
memset(buffer, 0, sizeof(buffer));
Doesn't run my stuff
FILE * handle = fopen ([path UTF8String], "r");
while (1) {
fseek(handle, ++offset, SEEK_SET);
if (fgets(buffer, 4, handle)) {
if (buffer[0] == 'E' && buffer[1] == 'X' && buffer[2] == 'T' && buffer[3] == 'H') {
//doesn't run my stuff
break;
}
}
}
fclose(handle);

Answer 1

IO is slow, so I'd read in large chunks and avoid reading small chunks from a stream inside a loop. getBytes: makes a copy from the NSData object's bytes into your buffer, so you want to avoid doing that frequently too. Here is some untested pseudocode, assuming you have your NSData object to start off with:
const char *bytes = [bookData bytes];
while (offset < headerLengh - 4)
{
if (bytes[offset] == 'E' && bytes[offset + 1] == 'X' && bytes[offset + 2] == 'T' && bytes[offset + 3] == 'H')
{
//...
}
offset++;
}
I will note that there exist faster/more complicated algorithms such as boyer-moore for this type of search, but this should do for now.
I'd also advise against creating many NSString objects inside the loop. If you want to, you can create a NSString from your NSData object, and retrieve substrings inside the loop appropriately.

Answer 2

Instead of creating a full-fledged NSString on each iteration, you could just compare the raw bytes directly:
memset(buffer, 0, sizeof(buffer));
while (offset < headerLength) {
[bookData getBytes:buffer range:NSMakeRange(0 + offset, 4)];
if (buffer[0] == 'E' && buffer[1] == 'X' && buffer[2] == 'T' && buffer[3] == 'H') {
...
}
offset++;
}
This is assuming your buffer holds 4 chars, of course.
Also, I don't believe it's needed to clear the memory on each iteration with memset. Just once before the loop will do.

Answer 3

It's been a while since I practiced plain old C-style file reading, but basically, there it goes...
file handle = fopen('/var/mobile/...', 'r');
char *buf = malloc(4 * sizeof(char));
long offset = 0;
while (1) {
fseek(handle, ++offset);
ret = fgets(handle, 4, &buf); // returns NULL on EOF
if (ret) {
if (buf[0] == 'E' && ...) {
// Your logic goes here
break;
}
}
else
break;
}
fclose(handle);
Of course this is all pseudo-code, feel free to investigate or correct me.