tokens section in combined grammar files can't work

[exKAZUu]

I modified your Java.g4 to the following file to make it suitable for the 4.1-opt release of ANTLR.
https://github.com/exKAZUu/Code2Xml/blob/92f3abaf173ebe31b5755c1d3cba7b3906e2d0ec/Code2Xml.Languages.ANTLRv4/Processors/Java/Java.g4

However, IntegerLiteral that is declared in the tokens section appears wrongly in JavaParser class instead of JavaLexer class.
Currently, both ANTLR 4.1 Java and C# targets generate such classes.

Note that the following issue shows the background to find this bug.
https://github.com/sharwell/antlr4cs/issues/28

[sharwell]

More specifically, the tokens{} block does work in a combined grammar, but it only adds the specified tokens to the generated parser. This prevents the lexer from using the tokens in a ->type() command.

[sharwell]

Apparently making this work is non-trivial. If you need access to items from the tokens{} block within your lexer rules (e.g. for a ->type command), you'll need to use separate lexer and parser grammars for now.