def update_metrics_all(self,
db_conn,
nonbib_conn,
row_view_schema='ingest',
start_offset=1,
end_offset=-1):
"""update all elements in the metrics database between the passed id offsets"""
# we request one block of rows from the database at a time
start_time = time.time()
step_size = 1000
count = 0
offset = start_offset
max_rows = self.config['MAX_ROWS']
sql_sync = nonbib.NonBib(row_view_schema)
Session = sessionmaker(bind=nonbib_conn)
session = Session()
session.execute('set search_path to {}'.format(row_view_schema))
for current_row in session.query(models.NonBibTable).yield_per(100):
metrics_dict = self.row_view_to_metrics(current_row, nonbib_conn,
row_view_schema)
self.save(db_conn, metrics_dict)
count += 1
if max_rows > 0 and count > max_rows:
break
if count % 1000 == 0:
self.logger.debug(
'metrics.py, metrics count = {}'.format(count))
self.flush(db_conn)
end_time = time.time()
def update_metrics_bibcode(
self,
bibcode,
db_conn,
nonbib_conn,
row_view_schema='nonbib'): #, delta_schema='delta'):
"""changed bibcodes are in sql table, for each we update metrics record"""
Metrics_Session = sessionmaker()
metrics_sess = Metrics_Session(bind=db_conn)
metrics_sess.execute('set search_path to {}'.format('metrics'))
sql_sync = nonbib.NonBib(row_view_schema)
row = sql_sync.get_by_bibcode(nonbib_conn, bibcode)
metrics_old = metrics_sess.query(models.MetricsTable).filter(
models.MetricsTable.bibcode == bibcode).first()
metrics_new = self.row_view_to_metrics(row, nonbib_conn,
row_view_schema, metrics_old)
if metrics_old:
metrics_sess.merge(metrics_new)
else:
metrics_sess.add(metrics_new)
metrics_sess.commit()
metrics_sess.close()
self.flush(db_conn)
def update_metrics_changed(self,
db_conn,
nonbib_conn,
row_view_schema='ingest'):
"""changed bibcodes are in sql table, for each we update metrics record"""
Nonbib_Session = sessionmaker(bind=nonbib_conn)
nonbib_sess = Nonbib_Session()
nonbib_sess.execute('set search_path to {}'.format(row_view_schema))
Metrics_Session = sessionmaker()
metrics_sess = Metrics_Session(bind=db_conn)
metrics_sess.execute('set search_path to {}'.format('metrics'))
sql_sync = nonbib.NonBib(row_view_schema)
query = nonbib_sess.query(models.NonBibTable)
count = 0
for delta_row in nonbib_sess.query(
models.NonBibDeltaTable).yield_per(100):
row = sql_sync.get_by_bibcode(nonbib_conn, delta_row.bibcode)
metrics_old = metrics_sess.query(models.MetricsTable).filter(
models.MetricsTable.bibcode == delta_row.bibcode).first()
metrics_new = self.row_view_to_metrics(row, nonbib_conn,
row_view_schema,
metrics_old)
if metrics_old:
metrics_sess.merge(metrics_new)
else:
metrics_sess.add(metrics_new)
metrics_sess.commit()
if (count % 10000) == 0:
self.logger.debug('delta count = {}, bibcode = {}'.format(
count, delta_row.bibcode))
count += 1
nonbib_sess.close()
metrics_sess.close()
self.flush(db_conn)
def update_metrics_test(self, bibcode, row_view_schema='ingest'):
sql_sync = nonbib.NonBib(row_view_schema)
row_view_bibcode = sql_sync.get_by_bibcode(bibcode)
metrics_dict = self.row_view_to_metrics(row_view_bibcode, sql_sync)
self.save(metrics_dict)
self.flush()
