1
0
mirror of https://github.com/arsenetar/dupeguru.git synced 2024-10-31 22:05:58 +00:00
dupeguru/core_me/app.py

90 lines
3.2 KiB
Python
Raw Normal View History

# Created On: 2011/09/20
# 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.util import format_size, format_time
from core.app import (DupeGuru as DupeGuruBase, format_timestamp,
format_perc, format_words, format_dupe_count, cmp_value)
from . import prioritize
from . import __appname__
from . import scanner, fs
from .result_table import ResultTable
class DupeGuru(DupeGuruBase):
NAME = __appname__
METADATA_TO_READ = ['size', 'mtime', 'duration', 'bitrate', 'samplerate', 'title', 'artist',
'album', 'genre', 'year', 'track', 'comment']
def __init__(self, view, appdata):
DupeGuruBase.__init__(self, view, appdata)
self.scanner = scanner.ScannerME()
self.directories.fileclasses = [fs.MusicFile]
self.result_table = ResultTable(self)
def _get_display_info(self, dupe, group, delta):
size = dupe.size
duration = dupe.duration
bitrate = dupe.bitrate
samplerate = dupe.samplerate
mtime = dupe.mtime
m = group.get_match_of(dupe)
if m:
percentage = m.percentage
dupe_count = 0
if delta:
r = group.ref
size -= r.size
duration -= r.duration
bitrate -= r.bitrate
samplerate -= r.samplerate
mtime -= r.mtime
else:
percentage = group.percentage
dupe_count = len(group.dupes)
return {
'name': dupe.name,
'folder_path': str(dupe.folder_path),
'size': format_size(size, 2, 2, False),
'duration': format_time(duration, with_hours=False),
'bitrate': str(bitrate),
'samplerate': str(samplerate),
'extension': dupe.extension,
'mtime': format_timestamp(mtime,delta and m),
'title': dupe.title,
'artist': dupe.artist,
'album': dupe.album,
'genre': dupe.genre,
'year': dupe.year,
'track': str(dupe.track),
'comment': dupe.comment,
'percentage': format_perc(percentage),
'words': format_words(dupe.words) if hasattr(dupe, 'words') else '',
'dupe_count': format_dupe_count(dupe_count),
}
def _get_dupe_sort_key(self, dupe, get_group, key, delta):
if key == 'percentage':
m = get_group().get_match_of(dupe)
return m.percentage
if key == 'dupe_count':
return 0
r = cmp_value(dupe, key)
if delta and (key in self.result_table.DELTA_COLUMNS):
r -= cmp_value(get_group().ref, key)
return r
def _get_group_sort_key(self, group, key):
if key == 'percentage':
return group.percentage
if key == 'dupe_count':
return len(group)
return cmp_value(group.ref, key)
def _prioritization_categories(self):
return prioritize.all_categories()