Escape reserved words during grammar generation (@parrt version)

[KvanTTT]

fixes #1070

Always escape x to x_, X to X_, XContext to X_Context (@parrt version, see #1070 (comment)). It does not relate to languages with reserved words escaping (C#, Swift).

Deprecate USE_OF_BAD_WORD

[KvanTTT]

@kaby76 or @studentmain could you provide updated lists of reserved words (both keywords and conflicting symbols) for all runtimes while I'm completing the fixes?

[kaby76]

@kaby76 or @studentmain could you provide updated lists of reserved words (both keywords and conflicting symbols) for all runtimes while I'm completing the fixes?

As we have an undocumented requirement in grammars-v4 that an Antlr g4 should be target agnostic, the "bad word list" is merged for all targets, not a target-specific. There is no differentiation between keyword vs a conflicting symbol --all bad words are off limits regardless. The initial version was culled from the Antlr tool sources, discussed here. But, you already are modifying that code.

To get an updated, combined list, the easiest would be to use trxgrep and look for anything that ends in an underscore. But, that would be a merged list for all targets. I would have to write a grammar against each bad word, and test it against each of the targets, easily doable but I can't come up with that quickly.

[KvanTTT]

There is no differentiation between keyword vs a conflicting symbol --all bad words are off limits regardless

Differentiation is worth only for languages that support keywords escaping (C#, Swift). But if this info is unknown, a word is treated as a conflicting symbol.

To get an updated, combined list, the easiest would be to use trxgrep and look for anything that ends in an underscore. But, that would be a merged list for all targets. I would have to write a grammar against each bad word, and test it against each of the targets, easily doable but I can't come up with that quickly.

It would be great. No rush, especially since I haven't finished yet.

[KvanTTT]

I'm closing it in favor of #3451 Let's continue there.

[kaby76]

Here is the aggregated list of "bad symbols" (with an underscore at the end).

all_underscore.txt

It was computed using the following script on a clean git clone of grammars-v4.

get-all-unames.txt