testrig | Magento installation automation - A tool | Ecommerce library

the normal grun alias takes the grammarName and startRule as parameters and expects the input from stdin:

You should make sure that:

• the classes CPP14Lexer.java and CPP14Parser.java are generated
• the generated CPP14Lexer.java and CPP14Parser.java are compiled
• the compiled CPP14Lexer and CPP14Parser classes are added to the classpath

So, assuming CPP14Lexer and CPP14Parser are in the default package (and reside in the same folder as the antlr-4.7.2-complete.jar file), then your command should look like this:

Shouldn't that be:

Since ANTLR 4.6.5-beta001 release of ANTLR4 for C#, it is possible to use ANTLR4 directly within VS to precompile grammar files to C#, then compile and run your app. All you need is install the nuget package. Today, you have several choices

1. "ANTLR4 Standard" version
2. "ANTLR4CS" optimized version

It works out of the box. Update your grammar, save it, it gets precompiled. Hit F5 and you are running it! More info in this answer

If you need to fiddle the inside workings of the build, check this link

To solve your path problem, the code generator installed by the AntlrCS package is antlr4.exe: jar is stored inside, so no more path problems. I have a simple setup that precompiles grammars outside VS. Working from the Antlr4Dy folder, I generate C# code in the src sub folder. The Code generator package is downloaded into the given folder, along with Antlr4.exe

"C:\Users...\source\repos\Antlr4Dy\packages\Antlr4.CodeGenerator.4.6.6\tools\net45\Antlr4.exe" C:\Users...\source\repos\Antlr4Dy\Speak.g4 -o src -Dlanguage=CSharp -package Antlr4x -no-listener -visitor

It appears that the invalid range 9-_ in your character class [a-zA-Z0-9-_] is the culprit. If you properly escape the -, the exception isn't thrown:

Your problem is the caused by the : in the filename,
because the configparser has it's default delimiters set as delimiters=("=", ":")
- see https://docs.python.org/3/library/configparser.html#configparser-objects for more.

Explictly setting the delimiters should work for you.

Wow... The solution was, of course, an easy solution... My problem was that I had cd'd to the root folder of my test project, and I just had to cd to (and call grun from) the directory where the compiled files are.

Edit:
It seems like I still have an issue with grun. See edited question.

Edit2:
From what I can tell, grun doesn't cooperate very well with antlr4's -package option.
So my solution was that I had to set up my project as a Gradle project, because apparently that works better.

That grammar really is ambiguous, because the grammar allows two interpretations of derp foo derp foo derp:

You seem to miss the loop operator in your consideration. The lexer rule WORD can match a sequence of any LOWERCASE, UPPERCASE and DIGIT, which means it is totally valid to match asd90 by using 5 runs of that loop (matching the lower case rule 3 times and the digits rule twice).

The example from the book has no loop, so it can only match once, either a type or void.

A few more details: fragment rules are kinda private rules (and in fact in older ANLTR versions there was the keyword private, which later was renamed to fragment). However, they are not available in the parser and do not appear in the list of lexer rules. Your WORD rule is stored in the ATN like so:

which shows that the 3 fragment rules are called like any other rule.

But there's another difference (and that's what Pavel mentions). Fragment lexer rules are handled a bit differently when it comes to ambiquity resolution.

Normally the resolution is like this: The rule which matches the longest input wins. If two rules match the same input then the one which comes first in the grammar wins. This holds true for both parser and lexer rules. However, not for fragment rules. Even if a fragment rule that matches the same input as another (non-fragment) rule appears first in the grammar, the non-fragment rule still wins.

You can prove that by changing your grammar a bit: