def _get_revisions(args):
""" Retrieve total set of revision records for users within timeframe """
um.log_pool_worker_start(__name__, _get_revisions.__name__, args[0], args[1])
users = args[0]
state = args[1]
metric_params = um.UserMetric._unpack_params(state)
query_args_type = namedtuple('QueryArgs', 'date_start date_end namespace')
revs = list()
umpd_obj = UMP_MAP[metric_params.group](users, metric_params)
try:
for t in umpd_obj:
revs += \
list(query_mod.rev_query(t.user, metric_params.project,
query_args_type(t.start, t.end,
metric_params.namespace)))
except query_mod.UMQueryCallError as e:
logging.error('{0}:: {1}. PID={2}'.format(__name__,
e.message, os.getpid()))
return []
um.log_pool_worker_end(__name__, _process_help.__name__)
return revs
def _process_help(args):
"""
Determine the bytes added over a number of revisions for user(s). The
parameter *user_handle* can be either a string or an integer or a list
of these types. When the *user_handle* type is integer it is
interpreted as a user id, and as a user_name for string input. If a
list of users is passed to the *process* method then a dict object
with edit rates keyed by user handles is returned.
The flow of the request is as follows:
#. Get all revisions for the specified users in the given
timeframe
#. For each parent revision get its length
#. Compute the difference in length between each revision and its
parent
#. Record edit count, raw bytes added (with sign and absolute),
amount of positive bytes added, amount of negative bytes added
- Parameters:
- **user_handle** - String or Integer (optionally lists). Value
or list of values representing user handle(s).
- Return:
- Dictionary. key(string): user handle, value(Float): edit counts
"""
um.log_pool_worker_start(__name__, _process_help.__name__, args[0], args[1])
revs = args[0]
state = args[1]
metric_params = um.UserMetric._unpack_params(state)
bytes_added = dict()
# Get the difference for each revision length from the parent
# to compute bytes added
row_count = 1
missed_records = 0
total_rows = len(revs)
for row in revs:
try:
user = str(row[0])
rev_len_total = int(row[1])
parent_rev_id = row[2]
except IndexError:
missed_records += 1
continue
except TypeError:
missed_records += 1
continue
# Produce the revision length of the parent. In case of a new
# article, parent_rev_id = 0, no record in the db
if parent_rev_id == 0:
parent_rev_len = 0
else:
try:
parent_rev_len = query_mod.rev_len_query(parent_rev_id,
metric_params.project)
except query_mod.UMQueryCallError:
missed_records += 1
logging.error(__name__ +
'::Could not produce rev diff for %s on '
'rev_id %s.' % (user, str(parent_rev_id)))
continue
# Update the bytes added hash - ignore revision if either rev length
# is undetermined
try:
bytes_added_bit = int(rev_len_total) - int(parent_rev_len)
except TypeError:
missed_records += 1
continue
try:
# Exception where the user does not exist. Handle this by
# creating the key
bytes_added[user][0] += bytes_added_bit
except KeyError:
bytes_added[user] = [0] * 5
bytes_added[user][0] += bytes_added_bit
pass
bytes_added[user][1] += abs(bytes_added_bit)
if bytes_added_bit > 0:
bytes_added[user][2] += bytes_added_bit
else:
bytes_added[user][3] += bytes_added_bit
bytes_added[user][4] += 1
row_count += 1
results = [[user] + bytes_added[user] for user in bytes_added]
extra = 'Processed {0} out of {1} records.'.\
format(total_rows - missed_records, total_rows)
um.log_pool_worker_end(__name__, _process_help.__name__, extra=extra)
return results
