32 lines
837 B
Python
32 lines
837 B
Python
# Copyright 2019 (c) Anna Schumaker.
|
|
import unicodedata
|
|
import os
|
|
|
|
def bisect(lst, lhs, *args):
|
|
begin = 0
|
|
end = len(lst)
|
|
|
|
while end - begin > 0:
|
|
pos = (end + begin) // 2
|
|
if lhs == (rhs := lst[pos].sort_key(*args)):
|
|
return (pos, lst[pos])
|
|
elif lhs < rhs:
|
|
end = pos
|
|
else:
|
|
begin = pos + 1
|
|
return (begin, None)
|
|
|
|
def key(text):
|
|
if os.path.exists(text):
|
|
return text.strip("/").split("/")
|
|
words = normalize(text).lower().split()
|
|
words = [ ''.join(filter(str.isalnum, w)) for w in words ]
|
|
words = [ w for w in words if w != "" ]
|
|
if words[0] in [ "a", "the" ]:
|
|
return words[1:]
|
|
return words
|
|
|
|
def normalize(text):
|
|
decode = unicodedata.normalize("NFD", text)
|
|
return decode.encode("ascii", "ignore").decode("utf8")
|