David542
0
Let's say I have the following statement:
SELECT "hi\n
there";
Notice there is a literal newline in there, and the escape \n. The string that antlr4 picks up for me is:
String_Literal: "hi\n\nthere"
In other words, not differentiating between the literal newline and the \n one. Is there a way to differentiate the two, or what's the usual process to do that?
My guess is that the output you pasted into your question comes from a call to the Antlr4 runtime method tree.toStringTree(parser) (or equivalent in whatever target language you've chosen).
That function calls escapeWhitespace in the utilities class/module/file, and that function does what it's name suggests: it converts (some) whitespace characters to C-like backslash escape sequences. (Specifically, it handles newline, carriage return, and tab characters.) It does not escape backslash characters, which makes its output ambiguous; there's no way to distinguish between the two character escape sequence \n and the escaped conversion of a newline character in the message.
They are different in the actual character string, because the Antlr4 lexer does not transform the string value of the matched token in any way. That's your responsibility.
In computing, it is very often the case that what you see is not what you got. What you see is just what you see, and a lot of computational power has gone into creating that vision for you. By the same token, nothing guarantees that the vision is an unambiguous, or even useful, representation of the actual values. The best you can say for it is that it's probably more useful than trying to read the data as individual bits. (And, indeed, the individual bits are not physical objects either; despite the common refrain, you could completely disassemble a computer and examine it with an arbitrarily powerful microscope, and you will not see a single 1 or 0.)
That might seem like irrelevant philosophizing, but it has a real consequence: when you're debugging and you see something that makes you think, "that looks wrong", you need to consider two possibilities: maybe the underlying data is incorrect, but may it's the process which rendered the representation which is at fault. In this case, I'd say that the failure of escapeWhitespace to convert backslash characters into pairs of backslashes is a bug, but that's a value judgement on my part. Anyway, the function is not critical to the operation of Antlr4, and you could easily replace it.