def run(cls, args):
try:
# print(args)
args = cls.args_schema.validate(args)
except SchemaError as e:
sys.exit(e)
session = Session()
# make sure lucene be inited
lucene.initVM()
lucene.getVMEnv().attachCurrentThread()
if args['--index'] is True:
configure_logging(
'lucene.index', console_level=args['--console-log-level'])
mgid = get_or_create_m(
session,
MetaInfo,
data=dict(
name='article_group_id_lucene_index',
value='0',
value_type='int',
description='article.group_id used for lucene index'),
fb_uk='name')
if args['--mode'] == 'create':
mgid.set_value(0)
session.commit()
logger.debug('Indexing started.. Getting articles..')
q = """
SELECT DISTINCT ON (a.group_id) a.id, a.group_id,
a.canonical_url,
a.title, a.meta, a.content,
coalesce(a.date_published, a.date_captured) AS pd,
s.domain, s.site_type
FROM article AS a
JOIN site AS s ON s.id=a.site_id
WHERE a.site_id IS NOT NULL AND s.is_enabled IS TRUE
AND a.group_id>:gid
ORDER BY group_id, pd ASC
"""
articles_iter = session.execute(
sqlalchemy.text(q).bindparams(gid=mgid.get_value()))
cls.index(session, args['--mode'], articles_iter, mgid)
elif args['--search'] is True:
configure_logging(
'lucene.search', console_level=args['--console-log-level'])
cls.search(args['--query'], args['--top'])
else:
print("Unrecognized command!")
sys.exit(2)
def run(cls, args):
"""Overriding method as the entry point of this command."""
session = Session()
cls.init(session, args['--force-drop'],
ignore_inactive=args['--ignore-inactive'],
force_inactive=args['--force-inactive'],
ignore_redirected=args['--ignore-redirected'])
def run(cls, args):
"""Overriding method as the entry point of this command."""
session = Session(expire_on_commit=False)
if args['--twitter-streaming'] is True:
configure_logging('twitter.streaming',
console_level=args['--console-log-level'])
cls.twitter_stream(session, args)
def db_query_top_spreaders(engine, upper_day, most_recent=False):
"""Query top 20 spreaders in the 30 days window.
Parameters
----------
engine : object
A SQLAlchemy connection, e.g., engine or session.
upper_day : datetime
The right edge of the 30 days window.
most_recent : bool
When no results for the `upper_day`, whether result the
most recent available results.
Returns
-------
pandas.DataFrame
Columns of the dataframe are ['upper_day', 'user_id',
'user_raw_id', 'user_screen_name', 'site_type',
'spreading_type', 'number_of_tweets', 'bot_score']
"""
q0 = """
SELECT upper_day, user_id, user_raw_id, user_screen_name, site_type,
spreading_type, number_of_tweets, bot_or_not
FROM top20_spreader_monthly WHERE upper_day=:upper_day
ORDER BY site_type, spreading_type, number_of_tweets DESC
"""
q = text(q0).bindparams(upper_day=upper_day)
rp = engine.execute(q)
df = pd.DataFrame(iter(rp), columns=rp.keys())
if len(df) == 0 and most_recent is True:
session = Session()
upper_day = get_max(session, Top20SpreaderMonthly.upper_day)
if upper_day is None:
raise APINoResultError
else:
q = text(q0).bindparams(upper_day=upper_day)
rp = engine.execute(q)
df = pd.DataFrame(iter(rp), columns=rp.keys())
df['user_raw_id'] = df.user_raw_id.astype(str)
def get_bot_score(bon):
if bon is None:
return None
elif 'score' in bon:
return bon['score']
elif 'scores' in bon:
return bon['scores'].get('universal')
else:
return None
df['bot_score'] = df.bot_or_not.apply(get_bot_score)
df = df.drop('bot_or_not', axis=1)
return df
def saver_queue(q2, number_of_workers):
engine = create_engine(URL(**CONF['database']['connect_args']),
pool_size=1,
pool_recycle=CONF['database']['pool_recycle'],
client_encoding='utf8')
Session = scoped_session(sessionmaker(bind=engine))
session = Session()
workers_status = [1 for i in range(number_of_workers)]
while True:
pid, status, uusers = q2.get()
if status == 'STOP':
logger.info(
'Saver process: STOP sign of worker process %s received from q2',
pid)
workers_status[pid] = 0
if sum(workers_status) == 0:
logger.warning('All STOP signs received from q2.')
logger.warning('Data saving task done!')
break
else:
logger.info('Saver process: size of uusers is %s', len(uusers))
stmt_do_nothing = insert(TwitterUserUnion).values(
uusers).on_conflict_do_nothing(index_elements=['raw_id'])
session.execute(stmt_do_nothing)
session.commit()
def run(cls, args):
"""Overriding method as the entry point of this command."""
try:
args = cls.args_schema.validate(args)
except SchemaError as e:
raise SystemExit('\n' + e + '\n')
session = Session(expire_on_commit=False)
if args['--twitter-streaming'] is True:
configure_logging('twitter.streaming')
cls.twitter_stream(session, args)
elif args['--load-tweets'] is True:
configure_logging('twitter.load-tweets')
cls.load_tweets(session, args)
elif args['--reparse-db-tweets'] is True:
configure_logging('twitter.reparse-db', file_level='WARNING')
cls._test_table_names(session, args)
cls.reparse_db(session, args)
def db_query_top_articles(engine, upper_day, most_recent=False):
"""Query top 20 articles in the 30 days window.
Parameters
----------
engine : object
A SQLAlchemy connection, e.g., engine or session.
upper_day : datetime
The right edge of the 30 days window.
most_recent : bool
When no results for the `upper_day`, whether result the
most recent available results.
Returns
-------
pandas.DataFrame
Columns of the dataframe are ['upper_day', 'date_captured',
'title', 'canonical_url', 'site_type', 'number_of_tweets'].
"""
q0 = """
SELECT upper_day, date_captured, title, canonical_url, site_type,
number_of_tweets
FROM top20_article_monthly WHERE upper_day=:upper_day
ORDER BY site_type, number_of_tweets DESC
"""
q = text(q0).bindparams(upper_day=upper_day)
rp = engine.execute(q)
df = pd.DataFrame(iter(rp), columns=rp.keys())
if len(df) == 0 and most_recent is True:
session = Session()
upper_day = get_max(session, Top20ArticleMonthly.upper_day)
if upper_day is None:
raise APINoResultError
else:
q = text(q0).bindparams(upper_day=upper_day)
rp = engine.execute(q)
df = pd.DataFrame(iter(rp), columns=rp.keys())
return df
def _monitor(self):
"""Monitor the queue for tweet, and use function parse to parse it.
This method runs on a separate, internal thread.
The thread will terminate if it sees a sentinel object in the queue.
"""
# scoped_session
# Session itself is not thread safe, use scoped_session
# each thread use only one scoped_session object
# We never delete anything from database, and we rely much
# on the `id` of existed object to build relaship
# set expire_on_commit=False
# to avoid re-fetch of these existed objects
session = Session(expire_on_commit=False)
parser = Parser(session, self.platform_id, **self.p_kwargs)
q = self.queue
has_task_done = hasattr(q, 'task_done')
while not self._stop.isSet():
# Server down, hold on
if self._hold_on is True:
logger.info('qsize is %s', q.qsize())
time.sleep(self._hold_on_unit)
self._hold_on_counter += self._hold_on_unit
if self._hold_on_counter >= self._hold_on_max:
return
logger.info('Hold on, keep tring to connect SQL server...')
logger.info('Elapsed %s seconds, since recent server down',
self._hold_on_counter)
if self._test_connection(session):
self._hold_on = False
self._hold_on_counter = 0
continue
try:
jd = self.dequeue(True)
if jd is self._sentinel:
break
self._counter += 1
if self._counter % self._window_size == 0:
logger.info('qsize is %s', q.qsize())
parser.parse(jd)
if has_task_done:
q.task_done()
except Queue.Empty:
break
except Exception as e:
logger.error('Exception %s when parsing %s', e, jd)
if isinstance(e, SQLAlchemyError):
session.rollback()
if isinstance(e, OperationalError):
# if 'could not connect to server' in str(e):
logger.error('Hold on until SQL service back! %s', e)
self._hold_on = True
# There might still be records in the queue.
while True:
try:
jd = self.dequeue(False)
if jd is self._sentinel:
break
parser.parse(jd)
if has_task_done:
q.task_done()
except Queue.Empty:
break
except Exception as e:
logger.error('Exception %s when parsing %s', e, jd)
if isinstance(e, SQLAlchemyError):
session.rollback()
if isinstance(e, OperationalError):
return
def run(cls, args):
"""Overriding method as the entry point of this command."""
session = Session()
if args['--volume'] is True:
configure_logging('report.volume',
console_level=args['--console-log-level'],
file_level='WARNING')
table_names = ['tweet', 'url', 'article']
table = args['--table']
if table not in table_names:
logger.critical('Available tables are: %s', table_names)
sys.exit(2)
interval_names = [
'minute', 'hour', 'day', 'week', 'month', 'quarter', 'year'
]
interval = args['--interval']
if interval not in interval_names:
logger.critical('Available intervals are: %s', interval_names)
sys.exit(2)
limit = args['--limit']
if int(limit) <= 0:
logger.critical('%r should larger than 0', limit)
sys.exit(2)
sql = """
SELECT count(id) as agg_num,
date_trunc(:interval, created_at) as interval
FROM %s
GROUP BY interval
ORDER BY interval DESC
LIMIT :limit""" % table
stmt = text(sql).bindparams(interval=interval, limit=limit)
strf = '%Y-%m-%d %H:%M:%S'
with ENGINE.connect() as conn:
result = conn.execute(stmt).fetchall()
print(('-' * 35))
print(('{0:^20s} | {1:12s}'.format('Timeline (%s)' % interval,
'Aggregation')))
print(('-' * 35))
for v, t in result:
print(('{0:^20s} | {1:8d}'.format(t.strftime(strf), v)))
print(('-' * 35))
elif args['--status']:
configure_logging('report.streaming-status',
console_level=args['--console-log-level'])
table_name = None
if args['--status'] == 'twitter':
table_name = 'tweet'
if table_name is None:
logger.critical('SNS %r has not been implemented!',
args['--status'])
sys.exit(2)
sql = 'SELECT created_at FROM {} ORDER BY id DESC LIMIT 1'.format(
'tweet')
with ENGINE.connect() as conn:
most_recent, = conn.execute(text(sql)).fetchone()
delta_minutes = 30
delta = timedelta(minutes=delta_minutes)
current_utc = datetime.utcnow()
if current_utc - most_recent > delta:
logger.critical(
'No %s streaming update in the past %s minutes!',
args['--status'], delta_minutes)
else:
logger.info('Most recent %s streaming update is %s',
args['--status'],
str(most_recent) + ' (UTC)')
elif args['--top-spreader'] is True:
configure_logging('report.top-spreaders',
console_level=args['--console-log-level'],
file_level='WARNING')
# try to create table
if (Top20SpreaderMonthly.__table__.exists(bind=ENGINE)) is False:
Top20SpreaderMonthly.__table__.create(bind=ENGINE)
if args['--force-today'] is True:
upper_day = datetime.utcnow().date()
elif args['--upper-day'] is None:
upper_day = datetime.utcnow().date() - timedelta(days=1)
else:
try:
upper_day = parse(args['--upper-day']).date()
except Exception:
raise ValueError('Invalid date: %s', args['--upper-day'])
if args['--generate'] is True:
logger.warning(
'Generating top spreaders for uppder_day=%r ...',
upper_day)
cls.generate_top_spreaders(session, upper_day)
elif args['--look-up'] is True:
cls.look_up_top_spreaders(session, upper_day,
args['--most-recent'])
elif args['--top-article'] is True:
configure_logging('report.top-article',
console_level=args['--console-log-level'],
file_level='WARNING')
# try to create table
if (Top20ArticleMonthly.__table__.exists(bind=ENGINE)) is False:
Top20ArticleMonthly.__table__.create(bind=ENGINE)
if args['--force-today'] is True:
upper_day = datetime.utcnow().date()
elif args['--upper-day'] is None:
upper_day = datetime.utcnow().date() - timedelta(days=1)
else:
try:
upper_day = parse(args['--upper-day']).date()
except Exception:
raise ValueError('Invalid date: %s', args['--upper-day'])
if args['--generate'] is True:
logger.warning('Generating top articles for uppder_day=%r ...',
upper_day)
cls.generate_top_articles(session, upper_day)
elif args['--look-up'] is True:
cls.look_up_top_articles(session, upper_day,
args['--most-recent'])
session.close()
def run(cls, args):
"""Overriding method as the entry point of this command."""
try:
args = cls.args_schema.validate(args)
except SchemaError as e:
raise SystemExit(e)
session = Session(expire_on_commit=False)
# session = Session()
where_expr = args['--where-expr']
ob_expr = args.get('--order-by', 'asc')
limit = args['--limit']
# --fetch-url
if args['--fetch-url'] is True:
configure_logging('crawl.fetch-url',
console_level='DEBUG',
file_level='WARNING')
purpose = 'update' if args['--update'] is True else 'archive'
if where_expr is None:
where_expr = [text(DEFAULT_WHERE_EXPR_FETCH_URL)]
else:
where_expr = [text(where_expr)]
ob_expr = Site.id.asc() if ob_expr == 'asc' else Site.id.desc()
msites = get_msites(session,
f_expr=where_expr,
ob_expr=ob_expr,
limit=limit)
if len(msites) == 0:
logger.warning("None sites you queried found in DB!")
raise SystemExit(2)
platform_id = get_platform_id(session, name=N_PLATFORM_WEB)
# detach msites and mplatform from session,
# since they definitely would not be modified in session
for ms in msites:
session.expunge(ms)
logger.warning('Starting crawling process to fetch URL update ...')
cls.fetch_url(session, msites, platform_id, purpose)
elif args['--fetch-html'] is True:
configure_logging('crawl.fetch-html',
console_level='DEBUG',
file_level='WARNING')
if not session.query(Site.id).count() > 0:
raise SystemExit('Your site table is empty!')
q = session.query(Url.id, Url.raw)
if where_expr is None:
where_expr = [text(DEFAULT_WHERE_EXPR_FETCH_HTML)]
else:
where_expr = [text(where_expr)]
ob_expr = Url.id.asc() if ob_expr == 'asc' else Url.id.desc()
q = q.filter(*where_expr).order_by(ob_expr)
if limit is not None:
q = q.limit(limit)
logger.info(
q.statement.compile(compile_kwargs={"literal_binds": True}))
url_tuples = q.all()
if not url_tuples:
logger.warning('No such URLs in DB!')
raise SystemExit(2)
logger.warning('Staring crawling process to fetch HTML ...')
cls.fetch_html(session, url_tuples)
# --parse-article
elif args['--parse-article'] is True:
configure_logging('crawl.parse-article',
console_level='DEBUG',
file_level='WARNING')
q = session.query(Url.id, Url.created_at, Url.date_published,
Url.canonical, Url.site_id)
if where_expr is None:
where_expr = [text(DEFAULT_WHERE_EXPR_PARSE_ARTICLE)]
else:
where_expr = [text(where_expr)]
ob_expr = Url.id.asc() if ob_expr == 'asc' else Url.id.desc()
q = q.filter(*where_expr).order_by(ob_expr)
if limit is not None:
q = q.limit(limit)
logger.info(
q.statement.compile(compile_kwargs={"literal_binds": True}))
url_tuples = q.all()
if not url_tuples:
logger.warning('No URLs found from DB!')
raise SystemExit(2)
logger.warning('Starting crawling process to parse article ...')
cls.parse_article(session, url_tuples)
session.close()
break
# q = """
# SELECT tw.id, tw.json_data, u.id, u.raw
# FROM tweet AS tw
# LEFT JOIN ass_tweet_url AS atu ON atu.tweet_id=tw.id
# LEFT JOIN url AS u ON u.id=atu.url_id
# WHERE tw.raw_id in (894686360900177920)
# """
for tw_id, jd, url_id, url in engine.execute(
text(q).bindparams(l=w_open_left, r=w_close_right)):
# text(q)):
jds[tw_id] = jd
if url_id is not None:
g_urls_map[url] = url_id
w_open_left = w_close_right
w_close_right += window_size
# import pdb; pdb.set_trace()
parser.bulk_parse_and_save(session=session, jds=jds,
existed_tweets=True,
g_urls_map=g_urls_map)
# break
if __name__ == '__main__':
# setting sqlalchemy logging
# logging.getLogger('sqlalchemy.engine').setLevel(logging.INFO)
logger = logging.getLogger()
logging.basicConfig(level='INFO')
session = Session()
main_test(ENGINE, session, min_id=0, window_size=10000, drop_first=True)
# main_test(ENGINE, session, window_size=1000, drop_first=False)
def run(cls, args):
"""Overriding method as the entry point of this command."""
session = Session(expire_on_commit=False)
# session = Session()
# expand user home for the file
if args['<file>'] is not None:
args['<file>'] = os.path.expanduser(args['<file>'])
# --load-domains commands
if args['--load-domains'] is True:
configure_logging(
'site.load-domains',
console_level=args['--console-log-level'],
file_level='WARNING')
fn = args.get('<file>', join(HOAXY_HOME, 'domains.txt'))
logger.info('Loading data from file %r', fn)
cls.load_domains(
session,
fn,
site_type=args['--site-type'],
ignore_inactive=args['--ignore-inactive'],
force_inactive=args['--force-inactive'],
ignore_redirected=args['--ignore-redirected'])
# --load-sites commands
elif args['--load-sites'] is True:
configure_logging(
'site.load-sites',
console_level=args['--console-log-level'],
file_level='WARNING')
fn = args.get('<file>', join(HOAXY_HOME, 'sites.yaml'))
logger.info('Loading data from file %r', fn)
cls.load_sites(
session,
fn,
ignore_inactive=args['--ignore-inactive'],
force_inactive=args['--force-inactive'],
ignore_redirected=args['--ignore-redirected'])
# --add commands
elif args['--add'] is True:
configure_logging(
'site.add',
console_level=args['--console-log-level'],
file_level='WARNING')
msite = qquery_msite(session, domain=args['--domain'])
if msite is not None:
logger.warning('Site %s already exists!', args['--domain'])
else:
cls.add_site(
session,
domain=args['--domain'],
site_type=args['--site-type'],
name=args['--name'],
tag_source=args['--tag-source'],
site_tags=args['--site-tag'],
alternate_domains=args['--alternate-domain'],
ignore_inactive=args['--ignore-inactive'],
force_inactive=args['--force-inactive'],
ignore_redirected=args['--ignore-redirected'])
# --add-site-tags
elif args['--add-site-tags'] is True:
configure_logging(
'site.add-site-tags',
console_level=args['--console-log-level'],
file_level='WARNING')
if args['--name'] is not None:
site_identity = args['--name']
else:
site_identity = args['--domain']
msite = qquery_msite(
session, name=args['--name'], domain=args['--domain'])
if msite is None:
logger.warning('Site %s does not exist!', site_identity)
else:
cls.add_site_tags(session, msite, args['--tag-source'],
args['--site-tag'])
# --replace-site-tags
elif args['--replace-site-tags'] is True:
configure_logging(
'site.repalce-site-tags',
console_level=args['--console-log-level'],
file_level='WARNING')
if args['--name'] is not None:
site_identity = args['--name']
else:
site_identity = args['--domain']
msite = qquery_msite(
session, name=args['--name'], domain=args['--domain'])
if msite is None:
logger.warning('Site %s does not exist!', site_identity)
else:
cls.replace_site_tags(session, msite, args['--tag-source'],
args['--site-tag'])
# --add-alternate-domains
elif args['--add-alternate-domains'] is True:
configure_logging(
'site.add-alternate-domains',
console_level=args['--console-log-level'],
file_level='WARNING')
if args['--name'] is not None:
site_identity = args['--name']
else:
site_identity = args['--domain']
msite = qquery_msite(
session, name=args['--name'], domain=args['--domain'])
if msite is None:
logger.warning('Site %s does not exist!', site_identity)
else:
cls.add_alternate_domains(session, msite,
args['--alternate-domain'])
# --replace-alternate-domains
elif args['--replace-alternate-domains'] is True:
configure_logging(
'site.replace-alternate-domains',
console_level=args['--console-log-level'],
file_level='WARNING')
if args['--name'] is not None:
site_identity = args['--name']
else:
site_identity = args['--domain']
msite = qquery_msite(
session, name=args['--name'], domain=args['--domain'])
if msite is None:
logger.warning('Site %s does not exist!', site_identity)
else:
cls.replace_alternate_domains(session, msite,
args['--alternate-domain'])
elif args['--disable'] is True:
configure_logging(
'site.disable',
console_level=args['--console-log-level'],
file_level='WARNING')
if args['--name'] is not None:
site_identity = args['--name']
else:
site_identity = args['--domain']
msite = qquery_msite(
session, name=args['--name'], domain=args['--domain'])
if msite is None:
logger.warning('Site %s does not exist!', site_identity)
else:
cls.disable_site(session, msite)
elif args['--enable'] is True:
configure_logging(
'site.enable',
console_level=args['--console-log-level'],
file_level='WARNING')
if args['--name'] is not None:
site_identity = args['--name']
else:
site_identity = args['--domain']
msite = qquery_msite(
session, name=args['--name'], domain=args['--domain'])
if msite is None:
logger.warning('Site %s does not exist!', site_identity)
else:
cls.enable_site(session, msite)
# --status
elif args['--status'] is True:
configure_logging(
'site.status',
console_level=args['--console-log-level'],
file_level='WARNING')
if args['--include-disabled'] is True:
cls.site_status(session, True)
else:
cls.site_status(session, False)
# --dump
elif args['--dump'] is True:
configure_logging(
'site.status',
console_level=args['--console-log-level'],
file_level='INFO')
cls.dump(session, args['<file>'])
session.close()
def workers_queue(pid, q1, q2):
"""Receiving parameters from q1, then computing and finally put results
into q2
"""
engine = create_engine(URL(**CONF['database']['connect_args']),
pool_size=1,
pool_recycle=CONF['database']['pool_recycle'],
client_encoding='utf8')
Session = scoped_session(sessionmaker(bind=engine))
session = Session()
parser = BulkParser(platform_id=1, save_none_url_tweet=True)
while True:
try:
data = q1.get(timeout=1)
except Empty:
logger.info('Worker process %s: queue is empty for 1 seconds', pid)
q2.put((pid, 'STOP', None))
break
if data == 'STOP':
logger.info('Worker process %s: STOP sign received from q1!', pid)
q1.put('STOP')
q2.put((pid, 'STOP', None))
break
else:
logger.info('Worker process %s: data=%s received', pid, data)
w_open_left, w_close_right = data
jds = dict()
g_urls_map = dict()
query = """
SELECT tw.id, tw.json_data, u.id, u.raw
FROM tweet AS tw
LEFT JOIN ass_tweet_url AS atu ON atu.tweet_id=tw.id
LEFT JOIN url AS u ON u.id=atu.url_id
WHERE tw.id>:l AND tw.id<=:r
"""
for tw_id, jd, url_id, url in engine.execute(
text(query).bindparams(l=w_open_left, r=w_close_right)):
jds[tw_id] = jd
if url_id is not None:
g_urls_map[url] = url_id
g_uusers_set = set()
g_edges_set = set()
for tw_id, jd in jds.items():
parser.parse_existed_one(tw_id,
jd,
session,
g_urls_map=g_urls_map,
g_uusers_set=g_uusers_set,
g_edges_set=g_edges_set)
edges = [
dict(tweet_raw_id=t0,
from_raw_id=t1,
to_raw_id=t2,
url_id=t3,
is_quoted_url=t4,
is_mention=t5,
tweet_type=t6) for t0, t1, t2, t3, t4, t5, t6 in g_edges_set
if t3 != -1
]
uusers = [dict(raw_id=t1, screen_name=t2) for t1, t2 in g_uusers_set]
# session.bulk_insert_mappings(TwitterNetworkEdge, edges)
stmt_do_nothing = insert(TwitterNetworkEdge).values(
edges).on_conflict_do_nothing(index_elements=[
'tweet_raw_id', 'from_raw_id', 'to_raw_id', 'url_id',
'is_quoted_url', 'is_mention', 'tweet_type'
])
session.execute(stmt_do_nothing)
session.commit()
q2.put((pid, 'RUN', uusers))
logger.info('Worker process %s: tweets from %s to %s done', pid,
w_open_left + 1, w_close_right)
def _monitor(self):
"""Monitor the queue for tweet incoming and then parse and save it into
the database.
This method runs on a separate, internal thread.
The thread will terminate if it sees a sentinel object in the queue.
"""
# scoped_session
# Session itself is not thread safe, we use scoped_session.
# Each thread uses only one scoped_session object
# We never delete anything from database in this function.
# set expire_on_commit=False to avoid re-fetch of these existed objects
session = Session(expire_on_commit=False)
parser = Parser(**self.parser_kwargs)
platform_id = get_platform_id(session, name=N_PLATFORM_TWITTER)
has_task_done = hasattr(self.queue, 'task_done')
while not self._stop.isSet():
if self.is_connection_failed is True:
self.on_db_server_down(session)
continue
# normal bulk insert process
try:
# fill the bucket
for i in range(self.bucket_size):
# dequeue with block=True
jd = self.queue.get(True)
if has_task_done is True:
self.queue.task_done()
if jd is not self._sentinel:
self.global_counter += 1
self.bucket.append(jd)
else:
break
# consume this bucket
self.consume_this_bucket(parser, session, platform_id)
self.bucket = []
# database is shutdown unexpectedly
except OperationalError as err:
session.rollback()
if 'server closed the connection unexpectedly' in repr(
err) or 'could not connect to server' in repr(err):
logger.critical('Causion: database server is down!')
self.is_connection_failed = True
else:
logger.error(err)
self.on_db_bulk_save_error()
except SQLAlchemyError as err:
session.rollback()
logger.exception(err)
self.on_db_bulk_save_error()
except BaseException as err:
# unexpected exception, logging (will exit)
logger.exception(err)
raise
# There might still be records in the queue.
while True:
try:
jd = self.queue.get(False)
if has_task_done:
self.queue.task_done()
if jd is self._sentinel:
break
self.bucket.append(jd)
except queue.Empty:
break
if self.bucket:
try:
self.consume_this_bucket(parser, session, platform_id)
self.bucket = []
except SQLAlchemyError as err:
session.rollback()
logger.exception('Consumer thread: %s', err)
self.on_db_bulk_save_error()
if self._fp_db_down is not None:
self._fp_db_down.close()
if self._fp_db_bulk_save is not None:
self._fp_db_bulk_save.close()