mirror of
https://github.com/arsenetar/dupeguru.git
synced 2024-11-18 21:19:01 +00:00
41 lines
1.2 KiB
Python
41 lines
1.2 KiB
Python
# -*- coding: utf-8 -*-
|
|
# Created By: Virgil Dupras
|
|
# Created On: 2009-10-23
|
|
# Copyright 2011 Hardcoded Software (http://www.hardcoded.net)
|
|
#
|
|
# This software is licensed under the "BSD" License as described in the "LICENSE" file,
|
|
# which should be included with this package. The terms are also available at
|
|
# http://www.hardcoded.net/licenses/bsd_license
|
|
|
|
from hscommon import io
|
|
from hscommon.path import Path
|
|
|
|
from core.engine import getwords
|
|
from ..scanner import *
|
|
|
|
class NamedObject:
|
|
def __init__(self, name="foobar", size=1):
|
|
self.name = name
|
|
self.size = size
|
|
self.path = Path('')
|
|
self.words = getwords(name)
|
|
|
|
|
|
no = NamedObject
|
|
|
|
def pytest_funcarg__fake_fileexists(request):
|
|
# This is a hack to avoid invalidating all previous tests since the scanner started to test
|
|
# for file existence before doing the match grouping.
|
|
monkeypatch = request.getfuncargvalue('monkeypatch')
|
|
monkeypatch.setattr(io, 'exists', lambda _: True)
|
|
|
|
def test_priorize_me(fake_fileexists):
|
|
# in ScannerME, bitrate goes first (right after is_ref) in priorization
|
|
s = ScannerME()
|
|
o1, o2 = no('foo'), no('foo')
|
|
o1.bitrate = 1
|
|
o2.bitrate = 2
|
|
[group] = s.get_dupe_groups([o1, o2])
|
|
assert group.ref is o2
|
|
|