Hi guys

I would like to know whats the fastest (general purpose) way of parsing and handling strings.

I need basic operations (find, replace, etc) - Tried one of the libs I've found in this forums but it wasn't cutting it (against a benchmark with a hl language)...

Anyone know if its possible to read by chunks rather than going from one char to the other?, I'm working with ASCII data right now but I'd love to support UNICODE as well.

Thx!

reading by chunks is preferred.
The only problem is you need to make sure that your allocated string buffer has enough padding at the end.

Say you have a string "ABC" stored in a 3 byte long block of memory, IF you are reading 4 bytes at a time you're going to go outside of the buffer.

I optimized C library would work just fine for you, no sense in reinventing the wheel. I recall a few threads on this forum about an optimized StrLen function, but I'm not sure about a fully optimized String parsing library with unicode support.

SomeoneNew: there are sooo many things to take into account. One major problem is whether string is static and dynamic. Static strings CAN be faster (often aren't in practice because of implementation), but they are way harder to use and have limited size. Dynamic strings are easy to use, not limited, but generally slower when enlarging. Note that nothing i mentioned here is nescessary, and in both cases it can be circumvented, problem is just it usually isn't.

So the question i would really care about is: Is the string handling bottleneck of my application?


For unicode support try ICU. Of course, the more complete unicode support you want, more HUNDREDS OF KILOBYTES are wasted.

hi thanks
yes vid, it is a bottleneck. since i parse comma delimited text files and html!