Parser failes with Context: <no context> about lark

[erezsh]

Hi, you should really just use a regular expression for that.
Something like

import re
re.findall('parsley\s+\d+', some_string)

Lark is meant for recognizing well-defined structure in text.

[DuyguA]

Hello,

I gave tiny lang as an example to say what happens with LALR parser, I'm an indeed parsing time expressions (context free and more complicated of course:) ) I can't disclose the grammar here. Something like this:

Can we meet next year, late March / early April?

or

We can meet tomorrow 12.00-14.00 or Wednesday 15.00-16.00.

My grammar looks like

sentence: (time_expr|rest)+
time_expr: .....
rest: REST
REST: /[^\s]+/

So, I want to both extract the strings, also parse the expression at the same time (then make entity resolution). Is there a way to extract parsed strings from the outer string, that was my question.

[erezsh]

Yes, it's possible, but not recommended in terms of performance.

I recommend that you use a two stage approach:

1. Use a regular expression to capture every phrase that might be relevant (As a whole string)

2. Parse each matched string (with Earley, not LALR), and throw away those that fail to parse.

Another suggestion is to use PyParsing for this. While Lark is generally better, PyParsing has the "scan()" function which seems to do what you want (maybe?).

[DuyguA]

Thanks for the prompt answer!

• I'm unhappy with performance of PyParsing in general. Input strings to this parser might be "long", so I'm not sure really, I'll give it a try anyway.

• Why am I not surprised that I'll fall back onto good, old regex at the end of the day? 🥇

Thanks for suggestions, issue can be closed.