1

I dont (need to) care about performance!

My regex matches the date format dd.mm.yyyy

((([0][1-9]|[12][\d])|[3][01])[./]([0][13578]|[1][02])[./][1-9]\d\d\d)|((([0][1-9]|[12][\d])|[3][0])[./]([0][13456789]|[1][012])[./][1-9]\d\d\d)|(([0][1-9]|[12][\d])[-/][0][2][./][1-9]\d([02468][048]|[13579][26]))|(([0][1-9]|[12][0-8])[./][0][2][./][1-9]\d\d\d)

Here are the dates my regex does not match yet. Any help appreciated.

09. Juni 1997
01.Aug.1995
27.06. 1997
29.02.1996
21. 01. 1999
28.05. 1996
07..4..1995
20:03:1998
9.4.1997
14 .03 - 1995

I started out by trying to add the month letters but failed (probably because of the whitespaces between them)

here is a regex that validates the months' letter order (Januar, Februar, März, April, Mai, Juni, August, September, Oktober, November, Dezember)

(?:J(anuar|u(n|li))|Februar|Mä(rz|i)|A(pril|ugust)|(((Sept|Nov|Dez)em)|Okto)ber)

I found this on the internet, which focusses on the issue if only 3 letters of the months are availiable

(((([1-9])|([0][1-9])|([1-2][0-9])|(30))\-([A,a][P,p][R,r]|[J,j][U,u][N,n]|[S,s][E,e][P,p]|[N,n][O,o][V,v]))|((([1-9])|([0][1-9])|([1-2][0-9])|([3][0-1]))\-([J,j][A,a][N,n]|[M,m][A,a][R,r]|[M,m][A,a][Y,y]|[J,j][U,u][L,l]|[A,a][U,u][G,g]|[O,o][C,c][T,t]|[D,d][E,e][C,c])))\-[0-9]{4}$)|(^(([1-9])|([0][1-9])|([1][0-9])|([2][0-8]))\-([F,f][E,e][B,b])\-[0-9]{2}(([02468][1235679])|([13579][01345789]))$)|(^(([1-9])|([0][1-9])|([1][0-9])|([2][0-9]))\-([F,f][E,e][B,b])\-[0-9]{2}(([02468][048])|([13579][26]))
id345678
  • 97
  • 1
  • 3
  • 21

1 Answers1

2

You can use

pattern = r"""(?x)(?<!d)(?:
  (?:(?:0?[1-9]|[12]\d)|3[01])\s?[./:-][\s.]?(?:0?[13578]|1[02]|J(?:an(?:uar)?|uli?)|M(?:ärz?|ai)|Aug(?:ust)?|Dez(?:ember)?|Okt(?:ober)?)\s?(?:[./:-][\s.]?)?[1-9]\d\d\d|
  (?:(?:0?[1-9]|[12]\d)|30)\s?[./:-][\s.]?(?:0?[13-9]|1[012]|J(?:an(?:uar)?|u[nl]i?)|M(?:ärz?|ai)|A(?:pr(?:il)?|ug(?:ust)?)|Sep(?:tember)?|(?:Nov|Dez)(?:ember)?|Okt(?:ober)?)\s?(?:[./:-][\s.]?)?[1-9]\d\d\d|
  (?:0?[1-9]|[12]\d)\s?[./:-][\s.]?(?:0?2|Fe(?:b(?:ruar)?)?)\s?(?:[./:-][\s.]?)?[1-9]\d(?:[02468][048]|[13579][26])|
  (?:0?[1-9]|[12][0-8])\s?[./:-][\s.]?(?:0?2|Fe(?:b(?:ruar)?)?)\s?(?:[./:-][\s.]?)?[1-9]\d\d\d
)(?!\d)"""

See the regex demo.

Main POIs:

  • The month regex is (?:J(?:an(?:uar)?|u[nl]i?)|Fe(?:b(?:ruar)?)?|M(?:ärz?|ai)|A(?:pr(?:il)?|ug(?:ust)?)|Sep(?:tember)?|(?:Nov|Dez)(?:ember)?|Okt(?:ober)?) and it is tested here. Adjust for shortenings as you see fit.
  • Febraury pattern is used separately for the last two alternations (they are specifically for Februrary) and is subtracted from the month pattern for the rest of the alternatives
  • From the first alternation, for 31-day months, February, April, June, September and November months are removed
  • Leading zeros in days and months is made optional by adding ? quantifier after 0
  • The separator between days and months is changed to \s?[./:-][\s.]?: an optional whitespace, a char from ./:- char set, and then an optional whitespace or .
  • The separator between months and years is changed to \s?(?:[./:-][\s.]?)?: an optional whitespace and then an optional sequence of a char from ./:- char set and then an optional whitespace or ..

The numeric boundaries, (?<!\d) / (?!\d), are added on both ends to make sure there are no other digits on both ends of the match.

Wiktor Stribiżew
  • 607,720
  • 39
  • 448
  • 563
  • It looks like you made every single group a non-capturing one -- why? Wouldn't it work just as well with regular groups? The `?:` seem to do nothing except adding additional clutter to this already extremely complex regex. (Very impressive nonetheless!) – fsimonjetz Oct 25 '21 at 15:37
  • 1
    @fsimonjetz See [Are non-capturing groups redundant?](https://stackoverflow.com/questions/31500422/are-non-capturing-groups-redundant) Caprturing groups need extra memory allocation for the substrings captured, and that slows it all down. – Wiktor Stribiżew Oct 25 '21 at 16:05