class SnapshoterTest(unittest.TestCase):
'''
Тестирование класса для создания снимка
'''
def setUp(self):
self.account = Account(LOGIN, PASSWORD)
self.account.auth()
self.snapshoter = GroupSnapshoter(GID, API(self.account.oauth(APP_ID, PERMISSIONS)))
def test_count(self):
self.assertGreater(self.snapshoter.get_wall_count(), 0)
self.assertGreater(self.snapshoter.get_videos_count(), 0)
self.assertGreater(self.snapshoter.get_photos_count(), 0)
def fetch_likes(self, item, change, shares=False):
'''
Возвращает список пользователей, совершивших действие
"поставил лайк" пользователей для item.
change -- [<начало списка лайков>, <конец списка лайков>]
shares -- если True, то обрабатывает пользователей, которые сделали
репост объекта
'''
sequence = GroupSnapshoter.create_requests_sequence(
change[0],
change[1],
GroupSnapshoter.LIKES_COUNT
)
retval = []
for offset, count in sequence:
try:
likes = self.api.likes.getList(
type=item.type,
owner_id=item.owner,
item_id=item.data.internal_id,
offset=offset,
count=count,
filter='likes' if not shares else 'copies'
)
retval.extend(likes['users'])
except APIError:
continue
return retval
def setUp(self):
self.account = Account(LOGIN, PASSWORD)
self.account.auth()
self.snapshoter = GroupSnapshoter(GID, API(self.account.oauth(APP_ID, PERMISSIONS)))
def run(self):
'''
Запускает работу кроулера
'''
self.account = self.choose_account()
if self.account is None:
raise Exception, "No valid account for parsing"
self.api = API(self.account.access_token)
self.board = Board(self.gid, api=self.api, resolve_names=False)
# Если True, то для для всех (новых и обновленных) объектов будут
# записываться действия пользователей, иначе только для
# обновленных
self.save_actions = len(Item.objects.all()) != 0 or CREATE_ACTIONS_WHEN_EMPTY
vk_group = VkGroup.objects.get(gid=self.gid)
logging.info("Start crawling for group %s", self.gid)
logging.info("Save actions = %s", self.save_actions)
logging.info("Wall Scanning Depth = %s", vk_group.wall_max)
logging.info("Video Scanning Depth = %s", vk_group.video_max)
logging.info("Photo Scanning Depth = %s", vk_group.photo_max)
logging.info("Board Scanning Depth = %s", vk_group.board_max)
snapshoter = GroupSnapshoter(self.gid, self.api)
logging.info("Old Wall Posts Count = %s", vk_group.wall_count)
logging.info("Old Photos Count = %s", vk_group.photos_count)
logging.info("Old Videos Count = %s", vk_group.videos_count)
logging.info("Old Board Topics Count = %s", vk_group.board_count)
wall_count = snapshoter.get_wall_count()
photos_count = snapshoter.get_photos_count()
videos_count = snapshoter.get_videos_count()
board_count = self.board.get_count()
'''
Рассчитываем разницу в количестве объектов, на которую надо
увеличить глубину сканирования, чтобы собрать все новые объекты и
обновить старые
'''
new_posts = max(wall_count - vk_group.wall_count, 0)
new_videos = max(videos_count - vk_group.videos_count, 0)
new_photos = max(photos_count - vk_group.photos_count, 0)
new_topics = max(board_count - vk_group.board_count, 0)
'''
'''
vk_group.wall_count = wall_count
vk_group.photos_count = photos_count
vk_group.videos_count = videos_count
vk_group.board_count = board_count
logging.info("New Wall Posts Count = %s", wall_count)
logging.info("New Photos Count = %s", photos_count)
logging.info("New Videos Count = %s", videos_count)
logging.info("New Board Topics Count = %s", board_count)
snapshot = []
snapshot = snapshoter.make_for_wall(size=vk_group.wall_max + new_posts)
snapshot.extend(snapshoter.make_for_videos(size=vk_group.video_max + new_videos))
snapshot.extend(snapshoter.make_for_photos(size=vk_group.photo_max + new_photos))
snapshot.extend(snapshoter.make_for_board(size=vk_group.board_max + new_topics))
logging.info("Crawling finished")
logging.info("%s items are at snapshot", len(snapshot))
logging.info("Creating tasks for updating")
tasks = self.create_update_task(snapshot)
logging.info("Created tasks for updating")
logging.info("%s items are to be added", len(filter(lambda task: task.action=='add', tasks)))
logging.info("%s items are to be updated", len(filter(lambda task: task.action=='update', tasks)))
new_comments, new_likes, new_shares = 0, 0, 0
for task in tasks:
if task.comments_change:
new_comments += task.comments_change[1] - task.comments_change[0] + 1
if task.likes_change:
new_likes += task.likes_change[1] - task.likes_change[0] + 1
if task.shares_change:
new_shares += task.shares_change[1] - task.shares_change[0] + 1
logging.info("New comments: %s", new_comments)
logging.info("New likes: %s", new_likes)
logging.info("New shares: %s", new_shares)
logging.info("Start fetching updates")
tasks = self.fetch_updates(tasks)
logging.info("Completed fetching updates")
new_comments, new_likes, new_shares = 0, 0, 0
for task in tasks:
if task.comments_change:
new_comments += len(task.comments)
if task.likes_change:
new_likes += len(task.likes)
if task.shares_change:
new_shares += len(task.shares)
logging.info("New comments are fetched: %s", new_comments)
logging.info("New likes are fetched: %s", new_likes)
logging.info("New shares are fetched: %s", new_shares)
logging.info("Start fixing updates")
self.fix_updates(tasks)
logging.info("Completed fixing updates")
def fetch_comments(self, item, change):
'''
Возвращает список комментариев для item.
change -- [<начало списка лайков>, <конец списка лайков>]
'''
if item.type == 'post':
sequence = GroupSnapshoter.create_requests_sequence(
change[0],
change[1],
GroupSnapshoter.WALL_COUNT
)
get = lambda offset, count: self.api.wall.getComments(
owner_id=item.owner,
post_id=item.data.internal_id,
offset=offset,
count=count
)[1:]
elif item.type == 'topic':
sequence = GroupSnapshoter.create_requests_sequence(
change[0],
change[1],
GroupSnapshoter.BOARD_COMMENT_COUNT
)
get = lambda offset, count: self.board.get_comments(
topic_id=item.data.internal_id,
offset=offset,
)
elif item.type == 'photo':
sequence = GroupSnapshoter.create_requests_sequence(
change[0],
change[1],
GroupSnapshoter.PHOTOS_COUNT
)
get = lambda offset, count: self.api.photos.getComments(
owner_id=item.owner,
pid=item.data.internal_id,
offset=offset,
count=count
)[1:]
elif item.type == 'video':
sequence = GroupSnapshoter.create_requests_sequence(
change[0],
change[1],
GroupSnapshoter.VIDEOS_COUNT
)
get = lambda offset, count: self.api.video.getComments(
owner_id=item.owner,
vid=item.data.internal_id,
offset=offset,
count=count
)[1:]
retval = []
for offset, count in sequence:
try:
comments = get(offset, count)
except APIError:
continue
for comment in comments:
if 'uid' in comment:
comment['from_id'] = comment['uid']
if 'message' in comment:
comment['text'] = comment['message']
if isinstance(comment['date'], basestring) or isinstance(comment['date'], int):
try:
comment['date'] = date.fromtimestamp(int(comment['date']))
except (ValueError, KeyError):
comment['date'] = None
retval.extend(comments)
return retval