Rules:

• non-terminals must be enclosed in brackets <nonterminal>
• Terminals must be surrounded in "quotes".
• At least one space is necessary between tokens. For example, <nonterminal>::=... is INVALID. You must have at least one space surrounding '::=' i.e. <nonterminal> ::= ...
• Comments: Only c++ multi-line style comments are currently supported: /* interesting comment about something here */

Syntactic Sugar:

• Single digit and letter intervals are allowed. For instance, instead of writing: <e> ::= "1" | "2" | "3", you can write <e> ::= [1-3]
• Basic EBNF is supported. Specifically:
  • "+" can be used to mean "one or more of the previous." For example: <e> ::= [a-z]+ can produce any non-zero length sequence of lower case letters.
  • "*" can be used to mean "zero or more of the previous."
  • "?" can be used to mean "zero or one occurances of the previous."
  • "(" ")" can be used to group elements. For example, <e> ::= ([1-9] [a-z])+ allows strings such as: 1a, 4f, 4g3f9d, etc.
  • Inside parentheticals, you may also indicate choice by "|". For example, <e> ::= ([1-9] | [a-z])+ allows strings such as: 2, 3553, 1ffvv2, ggg, etc.
  • Note: EBNF symbols must touch the element(s) they are grouping. valid:( [1-9]? | [a-z] )+ invalid: ( [1-9] || [a-z] ) +.
  • Check out the short Real Numbers example to see all these EBNF tricks in action.

Life Hacks:

• Basic textual auto-complete is available via ctrl+space.

Once you compile your BNF, you can test strings for validity by typing them in the box above. Whatever the FIRST definition of your BNF is is what will be tested. You can easily test different terms of your BNF by setting the top line to: <totest> ::= <termToTest>. Green indicates a valid string; red is invalid.

You can also generate random strings for any term in your definition. Note that the validitity indicator (green or red background) still only indicates the validity of the topmost definition.