Get original text that a rule matched in visitor?

[dhowe]

For debugging I'd like to access the text of a matched element in my visitor -- something roughly equivalent to ANTLRs getText()... Is this possible ?

[bd82]

No, you would have to implement it, e.g:

1. flatten the tree of tokens in the CST
2. find the first and last offset of these tokens
3. substring from the original text according to first/last offsets found earlier.

You could theoretically re-build the text from the CST

• that is what prettier plugins implemented using Chevrotain do.

However, this second approach would be less accurate as it does not account for skipped tokens (e.g whitespace).

[dhowe]

hmmm -- any thoughts about adding this to the API? or possibly accepting it as a PR ?

[bd82]

I am not 100% sure about making it part of the root api namespace because:

• How would it work on partial CSTs (error recovery)
• Different ways to implement this (e.g CSTParser vs EmbeddedActionsParser)

But the debugging scenario does not require a productive quality level of utilities...
So a PR implementing such debugging utilities in a new separate sub-package (@chevrotain/debug)
would be favorably reviewed.

[bd82]

a related issue in debugging context:

• Implement Debug helper method #874