def cleanup_keywords(keywords):
"""
Normalize the keywords: convert to titlecase, strip whitespace, split if
necessary, remove duplicates, remove empty strings, sort.
"""
cleaned_keywords = set()
for kw in keywords:
kw = kw.strip()
if not ALPHANUM_RE.match(kw):
continue
# If the keyword contains commas or semicolons we split it and
# re-process
if u";" in kw:
cleaned_keywords.update(cleanup_keywords(kw.split(u";")))
continue
if u"," in kw:
cleaned_keywords.update(cleanup_keywords(kw.split(u",")))
continue
if u"|" in kw:
cleaned_keywords.update(cleanup_keywords(kw.split(u"|")))
continue
# If the keyword contains one and only one quote char we remove it
if kw.count(u'"') == 1:
kw = kw.replace(u'"', u'')
if not ALL_CAPS.match(kw):
kw = titlecase(kw)
cleaned_keywords.add(kw)
return sorted(cleaned_keywords)
def test_all_caps_regex():
"""Test - all capitals regex"""
from titlecase import ALL_CAPS
assert bool(ALL_CAPS.match('THIS IS ALL CAPS')) is True
def test_all_caps_regex():
"""Test - all capitals regex"""
from titlecase import ALL_CAPS
assert bool(ALL_CAPS.match('THIS IS ALL CAPS')) is True
def test_all_caps_regex_2():
"""Test - all capitals regex with numer"""
from titlecase import ALL_CAPS
assert bool(ALL_CAPS.match('THIS IS ALL CAPS JOHHNY7')) is True