Regex expression of RInChI

[kiddr]

Fairsharing.org have asked if we have a regular expression to match a RInChI String, as they include on their records (https://fairsharing.org/FAIRsharing.58b6f9).

The InChI regex is:
^InChI=1S?/[A-Za-z0-9.]+(+[0-9]+)?(/[cnpqbtmsih][A-Za-z0-9-+(),/?;.]+)*$

[Artoria2e5]

Would be simple to make one based on the inchi one, although it won't be short with all the optional parts. Might be a good idea to just match the prefix ^RInChI=1\.[0-9]{2}\1.S?/ and call it a day.

In the tradition of internet-standards i specify the format in ABNF:

; "Everything can be omitted, except the characters in bold [i.e. the head]"
; "separators are omitted if they have nothing to separate."
rinchi = rinchi-head ["/" inchi-list *2("<>" inchi-list)] ["/" layer5] ["/" layer6]
; may want to relax the .00 part to any two numbers, with the assumption being 1.xx all look similar
rinchi-head = "RInChI=1.00.1" ["S"] 
inchi-list = inchi-body *("!" inchi-body)
layer5 = "d" ("+" / "-" / "=")
layer6 = "u" 1*DIGIT *("-" 1*DIGIT)

with inchi-body being the regex we have above with the head chopped off. regex isn't good at deliminator-separated repetition, so inchi-list would already be an ugly (remove all whitespace):

[A-Za-z0-9.]+(+[0-9]+)?(/[cnpqbtmsih][A-Za-z0-9-+(),/?;.]+)*
  (![A-Za-z0-9.]+(+[0-9]+)?(/[cnpqbtmsih][A-Za-z0-9-+(),/?;.]+)*)*

so the full regex would be... ugh... (remove the whitespace)

^
RInChI=1\.[0-9]{2}\1.S?
(
  /[A-Za-z0-9.]+(+[0-9]+)?(/[cnpqbtmsih][A-Za-z0-9-+(),/?;.]+)*
    (![A-Za-z0-9.]+(+[0-9]+)?(/[cnpqbtmsih][A-Za-z0-9-+(),/?;.]+)*)*
  (<>[A-Za-z0-9.]+(+[0-9]+)?(/[cnpqbtmsih][A-Za-z0-9-+(),/?;.]+)*
    (![A-Za-z0-9.]+(+[0-9]+)?(/[cnpqbtmsih][A-Za-z0-9-+(),/?;.]+)*)*){,2}
)?
(/d[-+=])?
(/u[0-9]+(-[0-9]+){,2})?
$

Not as bad as e-mail, but, well.

---

You know, this actually brings up a tangential. ABNF is great for describing InChI... and you can probably use it to build a really accurate regex for InChI out of it!