CIF2 grammar and `triple-quoted-string` and `quoted-string`

[rowlesmr]

I've been putting together a CIF parser and I have another question. I may also be totally off-base, as I'm translating to a PEG, and it does have different characteristics to ENBF.

I think the triple-quoted-strings should be tested before quoted-strings, as opposed to the current ENBF grammar:

https://github.com/COMCIFS/cif_core/blob/02b65fa614a4d087a428caa0e3e96892a6c50b8b/CIF2-EBNF.txt#L183-L187 and https://github.com/COMCIFS/cif_core/blob/02b65fa614a4d087a428caa0e3e96892a6c50b8b/CIF2-EBNF.txt#L164-L165

quoted-string and triple-quoted-string are defined as: (just taking "-delimited quotes to make things simpler, but the same argument applies to ')

quoted-string =  quote-delim, quote-content, quote-delim ;
quote-content = { char - quote-delim } ;
quote-delim = '"' ;

triple-quoted-string = quote3-delim, quote3-content, quote3-delim  ;
quote3-delim = '"""' ;
quote3-content = { [ '"', [ '"' ] ], not-quote, { not-quote } } ;
not-quote = allchars - '"' ;

Valid quoted-strings are "hello", "", "hi 'there' you" Valid triple-quoted-strings are """hello""", """""", """hi 'there' you \n"how" \n are ""you""?"""

If I try to parse """hello""" as a quoted-string first, as per nospace-value and table-entry, it spits out "", "hello", and "", so where I'm expecting to get a single string, I've got three: triple-quoted-strings should be tested for first.

or is this just how ENBF works...?

[vaitkus]

Unlike in PEG, the order of alternatives in EBNF is unimportant, therefore:

 nospace-value = 
     quoted-string 
   | triple-quoted-string 
   | list 
   | table ; 

is equivalent to

 nospace-value =
   triple-quoted-string 
   | quoted-string 
   | list 
   | table ; 

I guess in the case of your program in makes sense switching them up.

[rowlesmr]

the order of alternatives in EBNF is unimportant

Roger. PEG is an ordered choice; first match wins, even if there are other matches further down.

Now, on to figuring out why my text fields are broken.