A flexible cross-language parser generator with support for languages without explicit tokenization rules (like Gherkin).

It can be installed from NuGet. The executable is within the tools/net471 folder inside the package.

• generates parser for it's own grammar (the "hello world" for parser generators), see Berp Grammar
• does not generate a lexer/tokenizer, so ideal for languages where tokenization is easy or anyway not really possible
• simple, BNF-like grammar definition
• supports multiple target languages (currently C#, Java, Ruby, JavaScript, Go, Python) with the same grammar (the language generation is specified in template files)
• allows building AST, with AST-building hooks
• supports streamed token reading (tokens can be kept attached to the input stream to avoid unnecessary data transfer and object creation)
• supports context-sensitive tokens, also possible to change the tokenization rules during parsing (e.g. when a #language: no is encountered)
• supports a special "other" token, that matches to the "anything-else" case, when there is no better match
• support for recursive grammar rules is limited (it parses them up to a certain level only)
• simple, look-ahead rules can be specified
• rules can be marked as production rules to be represented in AST
• allows capturing ignored content tokens (e.g. comments)

• Berp Grammar
• Gherkin Grammar

• C# - CSharp.razor
• Java - Java.razor
• Ruby - Ruby.razor
• JavaScript (TypeScript) - TypeScript.razor
• Go - Go.razor
• Python - Python.razor

• Go: separate line from token
• Go: support for SimpleTokenMatcher
• Go: support for MaxCollectedError
• Import Perl from Gherkin
• Import Objective-C from Gherkin header & implementation
• Import C from Gherkin header & implementation