To show how to create a grammar with its rules and semantic actions, and to see what Pegged can manage, I added an /examples/ directory containing, as of this writing, the following grammars:

These grammars have about a dozen rules (say, from 5 to 20 rules):

• arithmetic.d. Parses arithmetic expression, like 1*(2+ x/3) - ( x * y * y -4).

• numbers.d. Parses different kind of numbers: integral, floating point, hexadecimal.

• strings.d. Parses double-quoted strings. An example of how to do escaping and looking for an end marker.

• parameterized.d. Examples of parameterized rules, as described on the Parameterized Rules page.

• csv.d. Comma-separated values, in files.

• json.d. The standard JSON grammar, taken from json.org. Only 12 rules!

• PEG.d the basic PEG grammar, as described by Ford.

• peggedgrammar.d the somewhat extended PEG grammar used by Pegged itself, as described in Extended PEG Syntax.

These are intermediate-size grammars. They have about 20 - 100 rules.

• markdown.d. The Markdown language grammar. The PEG file comes from this Github project. You can find the official description of Markdown syntax here. TODO: add unit tests.

• c.d. A simple C grammar, most probably full of bugs. I'll change some rule structure as time permit (for example, JumpStatement is too wide for my taste). Btw, that's about 50 rules.

• oberon2.d. The Oberon-2 programming language grammar. This grammar was written by Bjoern Lietz-Spendig. You can find the Oberon-2 description here.

• xml.d. OK, I'm cheating, it's a 5-rules long XML grammar, but it show how to use semantic actions to validate an input. See Semantic Actions.

• xml2.d The entire XML grammar is described here by the W3C. An 80-rules long grammar. No semantic action to validate nodes yet. I plan to fuse the two ASAP.

Grammars that have hundreds of rules:

• dgrammar.d. It's a first pass through the https://dlang.org online spec. It's about 500 rules long! Pegged generates the module in a few seconds alright, but the resulting module is a bit slow to load while linking (as in, 1-2 secs). It's probably buggy, though I corrected dozens of errors.

• The XML grammar (xml2.d) also contains a DTD grammar. The goal would be to get the following chain : DTD -> XML validator -> Validated XML. I think Pegged can be used to create a compile-time static validator: using the type system to accept only documents following the given DTD.

• The Javascript grammar, probably (I'm not so sure about this one).

• printf/writef format grammar: a nice example of a small DSL.

• A .INI grammar?