Repository navigation
Exponential grow for case insensitive input. #37
Description
Hi,
Thanks for this tool. This is awesome!
I noticed some generator produce all the possibles combinaisons for case incentive input. It means this generate a list of item growing exponentially relative to the input length.
For instance if you use: (input: abnf)
test = "test"
Then you get your get. (output: "rrta")
add('test', Diagram(
Choice(0,
Terminal("test"),
Terminal("Test"),
Terminal("tEst"),
Terminal("TEst"),
Terminal("teSt"),
Termina
8000
l("TeSt"),
Terminal("tESt"),
Terminal("TESt"),
Terminal("tesT"),
Terminal("TesT"),
Terminal("tEsT"),
Terminal("TEsT"),
Terminal("teST"),
Terminal("TeST"),
Terminal("tEST"),
Terminal("TEST"))));
I was curious to know if this is something you want to keep or fix later.
If you want to fix it later, would you prefer I try to submit PR?
I tried all the generator online using this tool https://arthursonzogni.com/Diagon/#code_area
exponential output:
- rrparcon
- rrl
- rrta
non exponential output:
- rrutf8
- rrtext
- svg
- html5
- xhtml5
- ebnfhtml5
- ebnfxhtml5
- blab
- abnf
- dot
- rdot
- rrdump
- rrtdump
Can't say (reached unimplemented)
- bnf
- wsn
- iso-ebnf
- rbnf
- sid
Reacted by Katherine
Activity
[-]Expone[/-][+]Exponential grows for case insensitive input.[/+][-]Exponential grows for case insensitive input.[/-][+]Exponential grow for case insensitive input.[/+]katef commentedon Nov 26, 2020
This is pretty terrible, I agree. I was trying to think about better ways to handle this, when I wrote it. Do you have any suggestions please?
The problem is that those dialects don't support case insensitive terminals, so we have to expand out every spelling.
ArthurSonzogni commentedon Nov 29, 2020
I was thinking: instead of outputting every combination, to rewrite it using a sequence of alternatives:
Input:
output (current)
output (proposed;)
Obviously, this is a very naive idea from me. I guess there are reasons this can't be done. I just found out this is implemented by
permute_case. I will take a look.katef commentedon Dec 8, 2020
Yes, unfortunately that has a different meaning. For a grammar where there could be whitespace between tokens (depending on how lexing is defined), here you're introducing the possibility for space between each letter, and to produce multiple tokens for output. The original grammar there has only one token for the entire string.