Parser throws error if text to parse doesn't end with a new line.

[RobRoseKnows]

This was a problem I came across in a grammar I was writing, but I was able to reproduce it with some of the examples. I don't know if this is Lark working as designed and just grammars are defined incorrectly, or if this is a problem with Lark.

If you start a grammar with a line like: start: (_item | _NL)* and then try to read a file that doesn't end with a new line, it will throw a bad parse error. Reproduce by using lark.lark as the grammar and then try to parse it without a \n at the end of the file.

The easy solution to this is, of course, just add a newline at the end of any string you want to parse. But this doesn't feel like a very good solution. Any thoughts of how to fix?

[erezsh]

The reason that lark.lark requires files to end in a newline is because of the way its rules are written: You can see that rule, token and statement all require _NL at the end. The reason I wrote it like this is that making _NL optional would make it harder to parse with LALR(1).

But now that you mention it, I have a different idea. I can add an EOF symbol, something like:

rule: RULE priority? ":" expansions (_NL|$)

[RobRoseKnows]

Yeah, I think adding an EOF symbol would be a good idea. Especially since Windows is now using Unix line endings in Notepad by default.

[nomorepanic]

I can see the usefulness of an EOF token too, so you should definitely consider adding it

[erezsh]

Closing. This idea of adding EOF continues in issue #237