Additional nickname delimiter specifications

[aikimark]

I've got a large file of names/addresses and decided to try the nameparser library with my name field. I was able to get more of my names parsed correctly by adding two regex patterns and adding these regex patterns to the list.
In regexes.py, added:
('double_apos', re.compile(r"''(.?)''", re.U)),
('apos', re.compile(r"'(.?)'", re.U)),

In parser.py, added:
re_apos = self.C.regexes.apos
re_double_apos = self.C.regexes.double_apos

In parser.py, changed:
for _re in (re_quoted_word, re_double_quotes, re_parenthesis, re_double_apos, re_apos):

[derek73]

This seems like a good idea and not too hard to do.

[derek73]

Thanks for the detailed writeup! Rather than patching regexes.py/parser.py directly (which meant re-forking on every upgrade), as of v1.3.0 there's a supported extension point for exactly this: Constants.extra_nickname_delimiters, added in #190. It's a dict of extra regex patterns that parse_nicknames() runs after the three built-in delimiters (quotes, double quotes, parens), so you can register your patterns at runtime without touching the library source:

import re
from nameparser import HumanName

hn = HumanName("John ''Big J'' Smith", constants=None)
hn.C.extra_nickname_delimiters['double_apos'] = re.compile(r"''(.*?)''")
hn.C.extra_nickname_delimiters['apos'] = re.compile(r"'(.*?)'")
hn.parse_full_name()

hn.nickname  # 'Big J'

Note constants=None gives this HumanName instance its own Constants copy so the change doesn't leak into other instances/threads; if you want it applied globally, register on nameparser.config.CONSTANTS instead.

These custom patterns go through the same suffix-vs-nickname disambiguation as the built-ins, so something like (Ret) still won't get swallowed as a nickname.

This shipped in #190 and will be part of the v1.3.0 release. Closing this out — let me know if the new hook doesn't cover your case.