mirror of
https://github.com/moparisthebest/SickRage
synced 2024-11-13 21:05:11 -05:00
222 lines
6.0 KiB
Python
222 lines
6.0 KiB
Python
|
#!/usr/bin/env python2
|
||
|
# -*- coding: utf-8 -*-
|
||
|
#
|
||
|
# Smewt - A smart collection manager
|
||
|
# Copyright (c) 2008-2012 Nicolas Wack <wackou@gmail.com>
|
||
|
#
|
||
|
# Smewt is free software; you can redistribute it and/or modify
|
||
|
# it under the terms of the GNU General Public License as published by
|
||
|
# the Free Software Foundation; either version 3 of the License, or
|
||
|
# (at your option) any later version.
|
||
|
#
|
||
|
# Smewt is distributed in the hope that it will be useful,
|
||
|
# but WITHOUT ANY WARRANTY; without even the implied warranty of
|
||
|
# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
|
||
|
# GNU General Public License for more details.
|
||
|
#
|
||
|
# You should have received a copy of the GNU General Public License
|
||
|
# along with this program. If not, see <http://www.gnu.org/licenses/>.
|
||
|
#
|
||
|
|
||
|
from __future__ import unicode_literals
|
||
|
from guessit import s
|
||
|
from guessit.patterns import sep
|
||
|
import functools
|
||
|
import unicodedata
|
||
|
import re
|
||
|
|
||
|
# string-related functions
|
||
|
|
||
|
def normalize_unicode(s):
|
||
|
return unicodedata.normalize('NFC', s)
|
||
|
|
||
|
|
||
|
def strip_brackets(s):
|
||
|
if not s:
|
||
|
return s
|
||
|
|
||
|
if ((s[0] == '[' and s[-1] == ']') or
|
||
|
(s[0] == '(' and s[-1] == ')') or
|
||
|
(s[0] == '{' and s[-1] == '}')):
|
||
|
return s[1:-1]
|
||
|
|
||
|
return s
|
||
|
|
||
|
|
||
|
def clean_string(s):
|
||
|
for c in sep[:-2]: # do not remove dashes ('-')
|
||
|
s = s.replace(c, ' ')
|
||
|
parts = s.split()
|
||
|
result = ' '.join(p for p in parts if p != '')
|
||
|
|
||
|
# now also remove dashes on the outer part of the string
|
||
|
while result and result[0] in sep:
|
||
|
result = result[1:]
|
||
|
while result and result[-1] in sep:
|
||
|
result = result[:-1]
|
||
|
|
||
|
return result
|
||
|
|
||
|
|
||
|
_words_rexp = re.compile('\w+', re.UNICODE)
|
||
|
|
||
|
def find_words(s):
|
||
|
return _words_rexp.findall(s.replace('_', ' '))
|
||
|
|
||
|
|
||
|
def reorder_title(title):
|
||
|
ltitle = title.lower()
|
||
|
if ltitle[-4:] == ',the':
|
||
|
return title[-3:] + ' ' + title[:-4]
|
||
|
if ltitle[-5:] == ', the':
|
||
|
return title[-3:] + ' ' + title[:-5]
|
||
|
return title
|
||
|
|
||
|
|
||
|
def str_replace(string, pos, c):
|
||
|
return string[:pos] + c + string[pos+1:]
|
||
|
|
||
|
|
||
|
def str_fill(string, region, c):
|
||
|
start, end = region
|
||
|
return string[:start] + c * (end - start) + string[end:]
|
||
|
|
||
|
|
||
|
|
||
|
def levenshtein(a, b):
|
||
|
if not a:
|
||
|
return len(b)
|
||
|
if not b:
|
||
|
return len(a)
|
||
|
|
||
|
m = len(a)
|
||
|
n = len(b)
|
||
|
d = []
|
||
|
for i in range(m+1):
|
||
|
d.append([0] * (n+1))
|
||
|
|
||
|
for i in range(m+1):
|
||
|
d[i][0] = i
|
||
|
|
||
|
for j in range(n+1):
|
||
|
d[0][j] = j
|
||
|
|
||
|
for i in range(1, m+1):
|
||
|
for j in range(1, n+1):
|
||
|
if a[i-1] == b[j-1]:
|
||
|
cost = 0
|
||
|
else:
|
||
|
cost = 1
|
||
|
|
||
|
d[i][j] = min(d[i-1][j] + 1, # deletion
|
||
|
d[i][j-1] + 1, # insertion
|
||
|
d[i-1][j-1] + cost # substitution
|
||
|
)
|
||
|
|
||
|
return d[m][n]
|
||
|
|
||
|
|
||
|
# group-related functions
|
||
|
|
||
|
def find_first_level_groups_span(string, enclosing):
|
||
|
"""Return a list of pairs (start, end) for the groups delimited by the given
|
||
|
enclosing characters.
|
||
|
This does not return nested groups, ie: '(ab(c)(d))' will return a single group
|
||
|
containing the whole string.
|
||
|
|
||
|
>>> find_first_level_groups_span('abcd', '()')
|
||
|
[]
|
||
|
|
||
|
>>> find_first_level_groups_span('abc(de)fgh', '()')
|
||
|
[(3, 7)]
|
||
|
|
||
|
>>> find_first_level_groups_span('(ab(c)(d))', '()')
|
||
|
[(0, 10)]
|
||
|
|
||
|
>>> find_first_level_groups_span('ab[c]de[f]gh(i)', '[]')
|
||
|
[(2, 5), (7, 10)]
|
||
|
"""
|
||
|
opening, closing = enclosing
|
||
|
depth = [] # depth is a stack of indices where we opened a group
|
||
|
result = []
|
||
|
for i, c, in enumerate(string):
|
||
|
if c == opening:
|
||
|
depth.append(i)
|
||
|
elif c == closing:
|
||
|
try:
|
||
|
start = depth.pop()
|
||
|
end = i
|
||
|
if not depth:
|
||
|
# we emptied our stack, so we have a 1st level group
|
||
|
result.append((start, end+1))
|
||
|
except IndexError:
|
||
|
# we closed a group which was not opened before
|
||
|
pass
|
||
|
|
||
|
return result
|
||
|
|
||
|
|
||
|
def split_on_groups(string, groups):
|
||
|
"""Split the given string using the different known groups for boundaries.
|
||
|
>>> s(split_on_groups('0123456789', [ (2, 4) ]))
|
||
|
['01', '23', '456789']
|
||
|
|
||
|
>>> s(split_on_groups('0123456789', [ (2, 4), (4, 6) ]))
|
||
|
['01', '23', '45', '6789']
|
||
|
|
||
|
>>> s(split_on_groups('0123456789', [ (5, 7), (2, 4) ]))
|
||
|
['01', '23', '4', '56', '789']
|
||
|
|
||
|
"""
|
||
|
if not groups:
|
||
|
return [ string ]
|
||
|
|
||
|
boundaries = sorted(set(functools.reduce(lambda l, x: l + list(x), groups, [])))
|
||
|
if boundaries[0] != 0:
|
||
|
boundaries.insert(0, 0)
|
||
|
if boundaries[-1] != len(string):
|
||
|
boundaries.append(len(string))
|
||
|
|
||
|
groups = [ string[start:end] for start, end in zip(boundaries[:-1],
|
||
|
boundaries[1:]) ]
|
||
|
|
||
|
return [ g for g in groups if g ] # return only non-empty groups
|
||
|
|
||
|
|
||
|
def find_first_level_groups(string, enclosing, blank_sep=None):
|
||
|
"""Return a list of groups that could be split because of explicit grouping.
|
||
|
The groups are delimited by the given enclosing characters.
|
||
|
|
||
|
You can also specify if you want to blank the separator chars in the returned
|
||
|
list of groups by specifying a character for it. None means it won't be replaced.
|
||
|
|
||
|
This does not return nested groups, ie: '(ab(c)(d))' will return a single group
|
||
|
containing the whole string.
|
||
|
|
||
|
>>> s(find_first_level_groups('', '()'))
|
||
|
['']
|
||
|
|
||
|
>>> s(find_first_level_groups('abcd', '()'))
|
||
|
['abcd']
|
||
|
|
||
|
>>> s(find_first_level_groups('abc(de)fgh', '()'))
|
||
|
['abc', '(de)', 'fgh']
|
||
|
|
||
|
>>> s(find_first_level_groups('(ab(c)(d))', '()', blank_sep = '_'))
|
||
|
['_ab(c)(d)_']
|
||
|
|
||
|
>>> s(find_first_level_groups('ab[c]de[f]gh(i)', '[]'))
|
||
|
['ab', '[c]', 'de', '[f]', 'gh(i)']
|
||
|
|
||
|
>>> s(find_first_level_groups('()[]()', '()', blank_sep = '-'))
|
||
|
['--', '[]', '--']
|
||
|
|
||
|
"""
|
||
|
groups = find_first_level_groups_span(string, enclosing)
|
||
|
if blank_sep:
|
||
|
for start, end in groups:
|
||
|
string = str_replace(string, start, blank_sep)
|
||
|
string = str_replace(string, end-1, blank_sep)
|
||
|
|
||
|
return split_on_groups(string, groups)
|