df and regex, NaN and df.concat

[buran]

here is nice site to test your regex patterns.
https://regex101.com/

you will see that your patterns do not match anything.

---

here is first pattern with some changes
https://regex101.com/r/dER5xH/1

note that using named groups could help and will make your code more readble

---

Actually, because you just want the year, pattern can be updated to catch all possible cases from your example
r"199\d{1}|20\d{2}|\d{2}(?![/,\-\w ])"
https://regex101.com/r/naVdDj/1
if you wish you may add named group, as is the entire match is what you need