openmedialibrary/oml/item/scan.py

285 lines
8.8 KiB
Python
Raw Normal View History

2014-05-04 19:26:43 +02:00
# -*- coding: utf-8 -*-
# vi:si:et:sw=4:sts=4:ts=4
2014-09-03 00:32:44 +02:00
2014-05-04 19:26:43 +02:00
2014-08-12 10:16:57 +02:00
from datetime import datetime
2014-05-04 19:26:43 +02:00
import os
import shutil
import time
2014-05-04 19:26:43 +02:00
import ox
from changelog import Changelog
2016-02-13 17:10:37 +05:30
from item.models import File, Item
2014-08-12 10:16:57 +02:00
from user.models import List
from utils import remove_empty_folders
2014-05-04 19:26:43 +02:00
from websocket import trigger_event
2014-08-12 10:16:57 +02:00
import db
import media
import settings
2014-05-17 02:14:15 +02:00
import state
2014-05-19 01:24:04 +02:00
import logging
2015-11-29 15:56:38 +01:00
logger = logging.getLogger(__name__)
2014-05-04 19:26:43 +02:00
2015-03-14 13:05:15 +05:30
extensions = ['epub', 'pdf', 'txt', 'cbr', 'cbz']
2014-05-16 16:30:16 +02:00
2016-02-13 17:10:37 +05:30
def remove_missing(books=None):
2014-05-04 19:26:43 +02:00
dirty = False
2016-02-13 17:10:37 +05:30
logger.debug('remove missing')
2016-02-14 14:25:54 +05:30
prefix = get_prefix()
2016-02-13 17:10:37 +05:30
if books is None:
2016-02-14 14:25:54 +05:30
books = collect_books(prefix)
2014-08-09 18:14:14 +02:00
with db.session():
2014-05-20 12:30:53 +02:00
if os.path.exists(prefix):
2016-02-13 17:10:37 +05:30
logger.debug('scan for removed files')
db_paths = []
items = {}
2014-05-20 12:30:53 +02:00
for f in File.query:
if state.shutdown:
2015-11-18 01:27:53 +01:00
return
2016-02-13 17:10:37 +05:30
path = f.fullpath()
db_paths.append(path)
items[path] = f.sha1
removed = set(db_paths) - set(books)
if removed:
logger.debug('%s files removed', len(removed))
ids = [items[path] for path in removed]
orphaned = set(ids)
for i in Item.query.filter(Item.id.in_(ids)):
i.remove_file()
orphaned.remove(i.id)
dirty = True
if orphaned:
logger.debug('%s files orphaned', len(orphaned))
for f in File.query.filter(File.sha1.in_(orphaned)):
state.db.session.delete(f)
2014-05-20 12:30:53 +02:00
dirty = True
if dirty:
2014-08-09 18:14:14 +02:00
state.db.session.commit()
state.cache.clear('group:')
2016-02-13 17:10:37 +05:30
logger.debug('update filenames')
2016-01-14 16:14:11 +05:30
for f in File.query:
if state.shutdown:
2016-01-14 16:14:11 +05:30
return
f.move()
2016-02-13 17:10:37 +05:30
logger.debug('remove empty folders')
remove_empty_folders(prefix, True)
2016-02-13 17:10:37 +05:30
logger.debug('remove missing done')
2014-05-04 19:26:43 +02:00
2016-02-11 11:45:17 +05:30
def add_file(id, f, prefix, from_=None, commit=True):
2014-05-19 01:24:04 +02:00
user = state.user()
path = f[len(prefix):]
2016-01-24 19:08:07 +05:30
logger.debug('%s extract metadata %s', id, path)
data = media.metadata(f, from_)
2016-01-24 19:08:07 +05:30
logger.debug('%s create file %s', id, path)
2014-05-19 01:24:04 +02:00
file = File.get_or_create(id, data, path)
item = file.item
item.add_user(user)
2014-05-20 02:43:54 +02:00
item.added = datetime.utcnow()
2016-01-24 19:08:07 +05:30
logger.debug('%s load metadata %s', id, path)
item.load_metadata()
Changelog.record(user, 'additem', item.id, file.info)
2016-01-05 21:46:50 +05:30
Changelog.record(user, 'edititem', item.id, item.meta)
2016-01-24 19:08:07 +05:30
logger.debug('%s extract icons %s', id, path)
2014-05-21 02:02:21 +02:00
item.update_icons()
2015-03-07 21:54:07 +05:30
item.modified = datetime.utcnow()
2016-01-24 19:08:07 +05:30
logger.debug('%s save item', id)
2016-02-11 11:45:17 +05:30
item.update(commit=commit)
2016-01-24 19:08:07 +05:30
logger.debug('%s added', id)
2014-05-19 01:24:04 +02:00
return file
2016-02-14 14:02:07 +05:30
def get_prefix():
2015-12-24 20:41:47 +05:30
prefs = settings.preferences
2016-01-04 15:19:14 +05:30
prefix = os.path.join(os.path.expanduser(prefs['libraryPath']), 'Books' + os.sep)
if not prefix[-1] == os.sep:
prefix += os.sep
2015-12-24 20:41:47 +05:30
assert isinstance(prefix, str)
2016-02-14 14:02:07 +05:30
return prefix
2016-02-14 14:25:54 +05:30
def collect_books(prefix, status=None):
2016-02-13 17:10:37 +05:30
logger.debug('collect books')
2015-12-24 20:41:47 +05:30
books = []
2016-02-14 14:25:54 +05:30
count = 0
2015-12-24 20:41:47 +05:30
for root, folders, files in os.walk(prefix):
for f in files:
if state.shutdown:
2016-02-14 14:25:54 +05:30
return []
2015-12-24 20:41:47 +05:30
if f.startswith('.'):
continue
f = os.path.join(root, f)
2016-02-13 15:58:06 +05:30
ext = f.split('.')[-1].lower()
if ext == 'kepub':
ext = 'epub'
2015-12-24 20:41:47 +05:30
if ext in extensions:
books.append(f)
2016-02-14 14:25:54 +05:30
count += 1
if status and not status(count):
return None
2016-02-13 17:10:37 +05:30
logger.debug('found %s books', len(books))
return books
def run_scan():
logger.debug('run_scan')
2016-02-14 14:02:07 +05:30
prefix = get_prefix()
2016-02-14 14:25:54 +05:30
books = collect_books(prefix)
2016-02-13 17:10:37 +05:30
remove_missing(books)
2015-12-24 20:41:47 +05:30
added = 0
2016-02-18 19:22:53 +05:30
with db.session():
for f in ox.sorted_strings(books):
if state.shutdown:
break
if os.path.exists(f):
id = media.get_id(f)
2016-02-03 01:00:40 +05:30
file = File.get(id)
if not file:
file = add_file(id, f, prefix, f)
added += 1
2016-02-14 14:25:54 +05:30
if added:
trigger_event('change', {})
logger.debug('imported %s unknown books', added)
2014-05-04 19:26:43 +02:00
def change_path(old, new):
new_books = os.path.join(new, 'Books')
if not os.path.exists(new_books):
ox.makedirs(new)
shutil.move(os.path.join(old, 'Books'), new_books)
remove_empty_folders(old)
else:
ox.makedirs(new_books)
run_scan()
trigger_event('change', {})
2014-05-16 16:30:16 +02:00
def run_import(options=None):
options = options or {}
2015-12-24 20:41:47 +05:30
logger.debug('run_import')
2016-01-23 18:07:25 +05:30
if state.activity.get('cancel'):
logger.debug('import canceled')
state.activity = {}
return
state.activity = {}
2015-12-24 20:41:47 +05:30
prefs = settings.preferences
prefix = os.path.expanduser(options.get('path', prefs['importPath']))
if os.path.islink(prefix):
prefix = os.path.realpath(prefix)
2016-01-04 15:19:14 +05:30
if not prefix[-1] == os.sep:
prefix += os.sep
2016-02-14 14:02:07 +05:30
prefix_books = get_prefix()
prefix_imported = os.path.join(prefix_books, '.import' + os.sep)
2015-12-24 20:41:47 +05:30
if prefix_books.startswith(prefix) or prefix.startswith(prefix_books):
error = 'invalid path'
elif not os.path.exists(prefix):
error = 'path not found'
elif not os.path.isdir(prefix):
error = 'path must be a folder'
else:
error = None
if error:
trigger_event('activity', {
2014-05-17 02:14:15 +02:00
'activity': 'import',
2015-12-24 20:41:47 +05:30
'progress': [0, 0],
'status': {'code': 404, 'text': error}
})
state.activity = {}
return
listname = options.get('list')
if listname:
listitems = []
assert isinstance(prefix, str)
books = []
2016-02-14 14:25:54 +05:30
def activity(count):
if count % 100 == 0:
state.activity = {
'activity': 'import',
'path': prefix,
'progress': [0, count],
}
trigger_event('activity', state.activity)
if state.activity.get('cancel'):
logger.debug('active import canceled')
state.activity = {}
return False
return True
books = collect_books(prefix, status=activity)
if books is None:
return
2015-12-24 20:41:47 +05:30
state.activity = {
'activity': 'import',
'path': prefix,
'progress': [0, len(books)],
}
trigger_event('activity', state.activity)
position = 0
added = 0
last = 0
for f in ox.sorted_strings(books):
position += 1
if not os.path.exists(f):
continue
with db.session():
2014-05-04 19:26:43 +02:00
id = media.get_id(f)
file = File.get(id)
f_import = f
2014-05-04 19:26:43 +02:00
if not file:
f = f.replace(prefix, prefix_imported)
ox.makedirs(os.path.dirname(f))
2014-05-16 16:30:16 +02:00
if options.get('mode') == 'move':
try:
shutil.move(f_import, f)
except:
shutil.copy2(f_import, f)
2014-05-16 16:30:16 +02:00
else:
shutil.copy2(f_import, f)
file = add_file(id, f, prefix_books, f_import)
2014-05-17 13:45:57 +02:00
file.move()
2014-05-04 19:26:43 +02:00
added += 1
elif options.get('mode') == 'move':
os.unlink(f_import)
if listname:
listitems.append(file.item.id)
if state.activity.get('cancel'):
state.activity = {}
return
if state.shutdown:
return
2015-12-24 20:41:47 +05:30
if time.time() - last > 5:
last = time.time()
state.activity = {
'activity': 'import',
'progress': [position, len(books)],
'path': prefix,
'added': added,
}
trigger_event('activity', state.activity)
2014-05-26 12:41:01 +02:00
2016-02-14 14:25:54 +05:30
if listname and listitems:
with db.session():
2014-05-19 01:24:04 +02:00
l = List.get(settings.USER_ID, listname)
if l:
l.add_items(listitems)
2015-12-24 20:41:47 +05:30
trigger_event('activity', {
'activity': 'import',
'progress': [position, len(books)],
'path': prefix,
'status': {'code': 200, 'text': ''},
'added': added,
})
state.activity = {}
remove_empty_folders(prefix_books)
if options.get('mode') == 'move':
remove_empty_folders(prefix, True)
def import_folder():
2016-01-25 12:18:24 +05:30
if not (state.activity and state.activity.get('activity') == 'import'):
import_path = settings.preferences['importPath']
logger.debug('scan importPath %s', import_path)
if os.path.exists(import_path):
run_import({
'path': import_path,
'mode': 'move'
})
remove_empty_folders(import_path, True)
if state.main:
state.main.call_later(10*60, lambda: state.tasks.queue('scanimport'))