Parsing Paragraphs in Peg.JS

Question

I am trying to learn peg.js and want to parse simple "blocks" of text, but am struggling with how to group sequential lines without getting a "possible infinite loop" error from my syntax.

Goal:

line 1

line 3
line 4

line 6

When parsed would become:

{
   "type": "root",
   "children": [
      { type: "para", content: "line 1" },
      { type: "para", content: "line 3\nline 4" },
      { type: "para", content: "line 6" },
   ]
}

In other words:

• line one is a paragraph of its own because it is followed by a blank line
• lines three and four are a paragraph because they're followed by a blank line
• line six is a paragraph because it's the last line(s) (one ore more)

I can write a grammar that matches lines and blank lines (see http://peg.arcanis.fr/4f4NdP/), but anything I do to try to get multiple consecutive lines followed by a blank line (or EOF) turned into a paragraph ends up with recursion errors. I feel like this is a really simple n00b thing that I'm just missing because I haven't used a PEG before.

I know I could write a global function in the initializer block and track the last element and make it contextual, but I feel like that's not really using the grammar like I should be.

Answer 1

You know those weeks where you struggle with something for a day or so and then finally give up, swallow your pride and post a question to stack overflow ... and then ten minutes later figure out the answer? Yep! That's my week. I think the process of writing out the question makes you think about the problem in a different way and your synapses start firing again or something ...

Anyway, here's the solution: http://peg.arcanis.fr/4f4NdP/2/

Grammar for posterity:

start = head:Para tail:(newline Para)*
   {
      var t;

      t = tail.reduce(function(memo, element) {
         return memo.concat(element[1]);
      }, []);

      return {
         type: 'root',
         children: [ head ].concat(t),
      }
   }

Para = text:LineOfText+
   { return { type: 'para', content: text.join('\n') } }

LineOfText = text:$(char+) EOL
   { return text }

char = [^\n\r]
newline = '\n' / '\r' '\n'?
EOL = newline / !.

Input:

line 1

line 3
line 4

line 6

Output:

{
   "type": "root",
   "children": [
      {
         "type": "para",
         "content": "line 1"
      },
      {
         "type": "para",
         "content": "line 3
line 4"
      },
      {
         "type": "para",
         "content": "line 6"
      }
   ]
}