What's wrong with my Treetop grammar?

Question

I have the grammar file alexa_scrape.tt:

grammar AlexaScrape
  rule document
    category_listing*
  end
  rule category_listing
    category_line url_line*
  end
  rule category_line
    category "\n"
  end
  rule category
    ("/" [^/]+)+
  end
  rule url_line
    [0-9]+ ". " url "\n"
  end
  rule url
    [^\n]*
  end
end

I have a ruby file which attempts to make use of it:

#!/usr/bin/env ruby -I .
require 'rubygems'
require 'polyglot'
require 'treetop'
require 'alexa_scrape.tt'

parser = AlexaScrapeParser.new
p( parser.parse("") || parser.failure_reason )
p( parser.parse("/x\n") || parser.failure_reason )

But I'm not getting the results I expected:

SyntaxNode offset=0, ""
"Expected one of /, \n at line 2, column 1 (byte 4) after /x\n"

It parses the empty string properly (as the trivial match for document, zero category_listings), but fails to parse "/x\n" (as the document containing a single category_listing that itself has zero url_lines).

What am I doing wrong?

Answer 1

It looks like the regex in category is advancing through the white space needed to match category_line ... do this:

  rule category
    ("/" [^/\s]+)+    # or perhaps ("/" [^/\n]+)+
  end

(And, wow, a Treetop question. This is number 47 in the history of SO and its 4 million total questions. One in 87,000 SO questions are tagged Treetop).