If statement and "=>"

[Champii]

If/Else statements cannot have a "=>" into quotes anywhere if they are multilines because of the preprocessor that adds ';' after the predicate expression in order to disambiguate with the following block

if someStr == "=>" // That fails
  doSomething()

Temporary workaround: Put the incriminated string into a variable outside of the if/else statement

Even if its a very specific issue, we need to ignore what's in the quotes when doing that kind of changes

[skull-squadron]

This project is awesome. 👏😎

Most lexers change their token state (Anltr calls it "mode"; start lexing using different rules, returning only after reaching the matching ending terminal) once they encounter such a terminal. In this way, some lexing rules which apply to the starting lexing state don't apply in the string lexing state. It would mean factoring out some token recognition to a separate lexer-only grammar, but it's unfortunately necessary since ANTLR doesn't allow lexical modes in mixed lexer-parsers.

See Lexical Modes in https://github.com/antlr/antlr4/blob/master/doc/lexer-rules.md

Examples:

https://github.com/antlr/grammars-v4/blob/master/xml/XMLLexer.g4

https://stackoverflow.com/q/18243283/246672

Pseudocode:

1. See a ".
2. Push lexer state to become a string lexer.
3. Lex inside of a string using string lexing rules until ".
4. See an ending ".
5. Pop lexer state back to whatever it was.

[Champii]

Hi !

Actually it's more a dirty hack, as the preprocessor don't tokenize the input but rather naively analyze each line as raw text to match some input. I need to fix that using a more powerful parsing technic, obviously :)

Thank you for that piece of information !

[skull-squadron]

Hey. No prob. Check out the ANTLR XML, SmallTalk and esp. Python example parsers for the bits needed to do less hacks. Good luck, my friend.

[Champii]

That's great material, thank you !
I might use Antlr's semantic predicate to mitigate this issue, also my grammar definition is poorly designed so I need to review all that to produce an efficient and reliable parser for these specific cases.