Parsing Text

Okay, I just want to be able to easily parse a string, but I have -no- idea how... I know someone else will though! :D

write a simple finite state machine

-beats his head on his keyboard-
Huh?

a vague question deserves a vague answer

Use stringstream or find() and substr().

Use a regular expression.

rofl rocky

Quote:

---

Originally Posted by Daved

Use stringstream or find() and substr().

---

sounds slow, a FSM would be much faster, do things in one pass is always better than multiple find calls

>> a FSM would be much faster, do things in one pass is always better than multiple find calls

I disagree. It depends on the situation, and as you noted this situation is quite vague. There's no reason to spend time or effort coding an FSM when a simple solution will work just as well. Premature optimization and all that.

fsms are simple to write, one for loop, check a few states. usually when someone says they want to "parse text", they have something like "field1;field2;field3;field4;...;field20;". In this case it would be simple to write an fsm.

If the input is like rockytriton describes, you could just use a stringstream object to parse the text in one pass.

Code:

---

#include <iostream> 
#include <string>   
#include <sstream> 
using namespace std;

int main(int argc, char *argv[])
{
  string text = "field1;field2;field3;field4;field20;" ;
 
  stringstream ss(text);
 
  while( ss )
  {
    string field ;
    getline( ss, field, ';' );
    cout << field << "\n" ;
  }
 
  cin.get(); 
  return 0 ;
}

---

Output:

Code:

---

field1
field2
field3
field4
field20

---

there you go, even better

>> In this case it would be simple to write an fsm.

Maybe I'm not understanding what you mean. Could you convert Dante Shamest's program to use a finite state machine?

>> even better

That uses a stringstream, so I don't understand your problem with my original suggestion.

ok, that was a bad example, now that I started writing an fsm for it, I relaized it doesn't need more than 1 state. it would make more sense for something that would parse data out of html tags, that's the last thing I did that I used an fsm to implement. In this case, it's still pretty simple and would be much faster than running through multiple finds for '<' and ">' or '/>'.

>> That uses a stringstream, so I don't understand your problem with my original suggestion.

no problem really, I didn't know about the changing the delimiter feature of getline(), it was really the find() and substr() part, it just makes me think of looping through with multiple finds.

Quote:

---

it would make more sense for something that would parse data out of html tags, that's the last thing I did that I used an fsm to implement.

---

Ah yes. If the input was HTML, a FSM would work very well.