neg 1.1.0 2013-02-28 home comments

Neg 1.1.0 is out. Neg is a small PEG parser library, very un-awesome and naive. One could say it’s a small brother of Parslet.

The bulk of the work from 1.0.0 to 1.1.0 was the removal of the “UnconsumedInputError”.

Consider this parser:

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16

class BlockParser < Neg::Parser

  parser do

    blocks == nl? + block + (nl + block) * 0 + nl?

    block == sp? + `begin` + sp + _('a-z') + nl + body + sp? + `end`
    body  == ((line | block) + nl) * 0
    line  == sp + `line`

    nl  == _("\s\n") * 1
    nl? == _("\s\n") * 0
    sp  == _(" ") * 1
    sp? == _(" ") * 0
  end
end

It’s meant for parsing begin blocks, nested or given in succession:

# simple
begin a
end

# nested
begin a
  begin b
  end
end

# sequence
begin a
end
begin b
end

So far, so good. But the parser becomes kind of unhelpful when it’s fed something like:

begin a
  begin b
  end
end
begin c
  begin d
end

Neg 1.0.0, would just raise an UnconsumedInputError and say something like “I could parse that, but there’s trailing input I couldn’t parse, it starts at ‘begin c’”.

Neg 1.1.0, will instead raise a ParseError stating it couldn’t find the end somewhere after “begin c”.

This 1.1.0 also introduce a #flattened_results helper for the Translator. It’s used in the JSON sample, the arith sample and the scheme sample. It comes in handy when translating for a parse rule like:

  array == `[` + (value + (`,` + value) * 0) * 0 + `]`

It flattens alls the values into a single array.

  on(:array) { |n| n.flattened_results }

 

© 2011-2013 John Mettraux