Stream: implementers
Topic: Regex Patten for dateTime
Dave Barnet (Feb 15 2019 at 11:18):
Its been brought to my attention that the regex pattern for dateTime differs from that declared in the schema files to that on the data type definition pages
http://hl7.org/fhir/STU3/datatypes.html#dateTime
Regex:
-?[0-9]{4}(-(0[1-9]|1[0-2])(-(0[0-9]|[1-2][0-9]|3[0-1])(T([01][0-9]|2[0-3]):[0-5][0-9]:[0-5]0-9?(Z|(\+|-)((0[0-9]|1[0-3]):[0-5][0-9]|14:00)))?)?)?
In HL7 fhir schema
-?[0-9]{4}(-(0[1-9]|1[0-2])(-(0[0-9]|[1-2][0-9]|3[0-1])(T([01][0-9]|2[0-3]):[0-5][0-9]:[0-5]0-9?(Z|(\+|-)((0[0-9]|1[0-3]):[0-5][0-9]|14:00))?)?)?)?
Which is correct (I'm assuming the schema?)
John Silva (Feb 15 2019 at 14:11):
@Dave Barnet - I don't know the answer but I used https://text-compare.com/ to see that the difference is the fourth question mark from the end. ?)?)?)?
Maybe this RegEx online parser page would be useful to test out which of the above is 'correct' with sample FHIR date/times: https://regex101.com/
Marcus Fearnett (Feb 15 2019 at 15:03):
@Dave Barnet
Top one from the spec is correct.
In FHIR spec, 'If hours and minutes are specified, a time zone SHALL be populated' so it's:
-?[0-9]{4}(-(0[1-9]|1[0-2])(-(0[0-9]|[1-2][0-9]|3[0-1])(T([01][0-9]|2[0-3]):[0-5][0-9]:[0-5]0-9?(Z|(\+|-)((0[0-9]|1[0-3]):[0-5][0-9]|14:00)))?)?)?
-?[0-9]{4}(-(0[1-9]|1[0-2])(-(0[0-9]|[1-2][0-9]|3[0-1])(T([01][0-9]|2[0-3]):[0-5][0-9]:[0-5]0-9?(Z|(\+|-)((0[0-9]|1[0-3]):[0-5][0-9]|14:00))?)?)?)?
leaves the offset part of a dateTime optional by lazy matching the group starting (Z|...)
p.s. It's only the STU3 schema that's off. The R4 version has got it correct.
Lloyd McKenzie (Feb 15 2019 at 15:41):
Can you submit a change request? We're doing an R3 patch soon and we might be able to fix this.
Simone Heckmann (Feb 17 2019 at 13:24):
@Alexander Zautke
Patrick Werner (Feb 17 2019 at 16:49):
looking @ R4:
what about the regex for time: ([01][0-9]|2[0-3]):[0-5][0-9]:([0-5][0-9]|60)(\.[0-9]+)?
Patrick Werner (Feb 17 2019 at 16:51):
i think the last capture group is wrong? What is the purpose of it?
Now something like: 08:23:12.2234323434534646234485767843756872980 would be valid
Grahame Grieve (Feb 17 2019 at 16:57):
not sure what the limits are on decimal places. I support up to 9
Patrick Werner (Feb 17 2019 at 16:58):
spec says: A time during the day, in the format hh:mm:ss whereas the regex allows infinite milliseconds
Grahame Grieve (Feb 17 2019 at 16:58):
fractions are allowed
Patrick Werner (Feb 17 2019 at 17:00):
ok. Couldn't find it in the spec here: http://build.fhir.org/datatypes.html#time
Patrick Werner (Feb 17 2019 at 17:01):
would be nice to have it explicitly stated to avoid misunderstandings
John Silva (Feb 17 2019 at 21:34):
Wow, where is that time machine that supports 'infinite milliseconds' ? :-)
Patrick Werner (Feb 18 2019 at 14:27):
i made a ticket: GF#20400]
Elliot Silver (Feb 19 2019 at 19:47):
What are infinite milliseconds? There are either milliseconds ([0-9]{1,3}) or infinite precision ([0-9]+).
Patrick Werner (Feb 19 2019 at 20:08):
please excuse my fuzzy language, i meant infinitely accurate fractions of a second = infinite precision
Last updated: Apr 12 2022 at 19:14 UTC
