I'm trying to write a regex pattern capturing four different groups, the first groups ends either when we encounter either _ne or _re or a dot, The second groups is an optional one, it captures the re or ne if encountered, otherwise it's empty, the third and fourth group are a bit easier to capture as they are just words proceeded by a dot. here is a code snippet to get a sample data:

import pandas as pd 
sample = pd.Series(["abc_ne.c.d", "kc_E5_re.c.d", "kc_E5_re13.c.d", "kc_E5.c.d"]).rename('raw')

using the following pattern (\w )(?:_(ne|re)\d*)\.(\w*)\.(\w*) I can capture most cases

the exception is when the second group is absent, in which case it fails: I tried making it optional (\w )(?:_(ne|re)\d*)?\\.(\w*)\.(\w*) but it captures everything in the first groups up to the dot.

This snippet could be used to capture groups with pandas if needed:

pattern = r'(\w )(?:_(ne|re)\d*)?\.(\\w*)\.(\w*)'
sample.to_frame().join(sample.str.extract(pattern))

The expected output is:

Can anyone help me get the pattern right ?

Thanks in advance.

CodePudding user response：

I'd say you probably want the 2nd group in an optional non-capture group and make the characters captured by the 1st group lazy:

^(\w ?)(?:_([nr]e\d*))?\.(\w )\.(\w )$

See an online demo

• ^ - Start-line anchor;
• (\w ?) - 1st Capture group to catch 1 (Lazy) word-characters (thus including underscore);
• (?:_([nr]e\d*))? - Optional non-capture group to match an underscore and an nested 2nd capture group to match both 're' or 'ne' followed by 0 digits;
• \.(\w )\.(\w ) - Match both the 3rd and 4th capture group in succession inbetween literal dots;
• $ - End-line anchor.