Stop parsing on pattern match

[jocutajar]

Hi, I'd like to stop parsing the string input when a given pattern is matched. I'll give example for SMTP:

helo abc
mail from:<[email protected]>
rcpt to:<[email protected]>
data
$%!...?^ö

The desire is that I could parse all the commands until including the data command. There is no point parsing further as there may be binary data there.

You could add a #quit instruction to terminate further processing:

pub cmd_data -> SmtpCommand
= "data"i NL #quit
{ SmtpCommand::Data }

FYI: the full SMTP grammar on github. Btw, thanks a lot for PEG.

[kevinmehall]

Right now the generated wrapper function checks that the match reaches the end of input, and reports a expected EOF if it does not. This ensures that a top-level rule like statement* does not ignore statements that fail to match at the end of file. In your case, the "data" serves the same role. So we just need to add an optional attribute that disables that check.

[kevinmehall]

#[no_eof] was added in 0.7 - #233

[Mingun]

For someone who looking for concrete solution:

peg::parser! {
  pub grammar stream() for str {
    pub rule event() -> T = ...;

    pub rule event_list() -> Vec<T> = event()*;

    #[no_eof] // Do not check for EOF after parsing rule
    pub rule event_iter() -> (T, usize)
      // Return position after parsing
      = r:event() p:position!() { (r, p) };
  }
}

let mut content = "...";
while !content.is_empty() {
  let (event, pos) = stream::event_iter(content)?;
  content = &content[pos..];
  // process event...
}

We add a new *_iter rule which calls our target rule and returns result together with the position after parsing, so we know where to start parsing of the next piece. It is equivalent to event_list() rule except that we get events one-by-one.