Variable length Lookbehinds RegEx don't works in Python

Question

I want extract the text between ':' and '|' characters, but in second and third data there are a space after the ':'

The intput:

Referencia:22726| Referencia Cliente Ak: 233726 | Referencia histórica: 256726 | Suelo | AGOLADA (Pontevedra) -  CARPAZO O PE#A LONJA [EXTRACT]
Referencia:39766| Referencia Cliente Ak: 39767 | Referencia histórica: 39768 | Garaje | MOJACAR (Almería) -  URB.VILLA MIRADOR DEL MAR - MOD. # [EXTRACT]
Referencia:397A5| Referencia Cliente Ak: 397B5 | Referencia histórica: 397C5 | Garaje | MOJACAR (Almería) -  VILLA MIRADOR DEL MAR-MODULO #-PLAZA 4 [EXTRACT]
Referencia:AA39803| Referencia Cliente Ak: P_39803 | Referencia histórica: 200_39803 | Garaje | MOJACAR (Almería) -  VILLA MIRADOR DEL MAR - MODULO [EXTRACT]

Output desired:

My first pattern: (?<=:)(\w{5,12}) This matches only the first column.

My second pattern: (?<=:\s)(\w{5,12}) This matches the second and third columns

So I believed that my third pattern was the correct one: (?<=:\s?)(\w{5,12}) That pattern don't works.

You can use the `regex` module, which supports variable lengths in lookaheads and lookbehinds. It is planned to replace the standard `re` module, they say. You can install it with Pip. (`pip install regex`). — Brōtsyorfuzthrāx, Aug 12 '14 at 23:26

score 2 · Accepted Answer · answered Feb 26 '14 at 09:36

2

a lookbehind can't be variable length in python. A way to solve this:

(?:(?<=:\s)|(?<=:))(\w{5,12})

But since you use a capturing group, a lookbehind is useless:

:\s?(\w{5,12})

answered Feb 26 '14 at 09:36

Casimir et Hippolyte

88,009
5
94
125

Oh, thanks. I didn't know that a lookbehind can't be variable length in python. I imagine that neither a lookahead. But, is it thereby only in python or in other languages too. – Trimax Feb 26 '14 at 09:45
a lookahead can be variable length. – Casimir et Hippolyte Feb 26 '14 at 09:48
With .net you can use a variable length lookbehind. Java allows to use this quantifier `{n,m}` and PHP allows alternations. – Casimir et Hippolyte Feb 26 '14 at 09:50
That new solution is more clean. Why a lookahead can be variable length but not either lookbehind? – Trimax Feb 26 '14 at 09:51
I don't have a definitive answer about this, as you can see limitations are different between regex engines. People says:"Since a regex engine works from left to right, it is hard to implement a variable length lookbehind." (to repeat three times before sleeping) – Casimir et Hippolyte Feb 26 '14 at 09:57

score 1 · Answer 2 · answered Feb 26 '14 at 09:33

1

Remove the ? character in the Lookbehind move the \s to the matches

(?<=:)(\s?\w{5,12})

answered Feb 26 '14 at 09:33

Nambi

11,944
3
37
49

Thanks, elegant solution! – Trimax Feb 26 '14 at 09:46

Variable length Lookbehinds RegEx don't works in Python

2 Answers2