def puts(cls, posts):
posts = list(posts)
tagnames = set(chain.from_iterable([post['tags'] for post in posts]))
tags = {name: Tag.get_or_add_bi_name(name) for name in tagnames}
def inner(posts):
for kargs in posts:
entry = Entry.get_or_add_bi_md5(md5=kargs['md5'], ctime=kargs['ctime'])
kargs = {k: v for k, v in kargs.items() if k != 'tags'}
try:
post = cls.query.filter_by(post_url=kargs['post_url']).options(db.joinedload(cls.entry)).one()
if post.entry.md5 != kargs['md5']:
raise Exception('md5 changed %s -> %s' % (post.md5, kargs['md5']))
post.from_dict(kargs)
except NoResultFound:
post = cls(entry=entry)
post.from_dict(kargs)
db.session.add(post)
yield post
flushed = False
seen = set()
for kargs, post in zip(posts, list(inner(posts))):
if post.id is None:
if not flushed:
db.session.flush()
flushed = True
db.session.expire(post, ['id'])
else:
if post.id in seen:
continue
seen.add(post.id)
for name in set(kargs['tags']):
tag = tags[name]
db.session.merge(Tagged(post_id=post.id, tag_id=tag.id, entry_id=post.entry.id))
def _update_images(begin=None, limit=65536):
start = time()
sources = [make(dict) for make in makesources()]
from concurrent.futures import ThreadPoolExecutor as Ex
with Ex(len(sources)) as ex:
posts = list(chain.from_iterable(
ex.map(partial(fetch_posts, begin, limit), sources)
))
log.info(
'fetch posts done, {} fetched, take {}',
len(posts),
time() - start
)
start = time()
current_app.store.Post.puts(posts=posts)
log.info('put posts done, take {}', time() - start)