def run(dbcache, users, revert_cutoff, revert_radius, no_headers):
if no_headers:
headers=None
else:
headers=HEADERS
output = tsv.Writer(sys.stdout, headers=headers)
for user in users:
sys.stderr.write("{0}, {1}: ".format(user.wiki, user.user_id))
db = dbcache.get_db(user.wiki)
day_revisions = 0
day_deleted_revisions = 0
day_main_revisions = 0
day_deleted_main_revisions = 0
day_reverted_main_revisions = 0
day_productive_edits = 0
week_revisions = 0
week_deleted_revisions = 0
week_main_revisions = 0
week_deleted_main_revisions = 0
week_reverted_main_revisions = 0
week_productive_edits = 0
week_sessions = 0
week_session_seconds = 0
registration = Timestamp(user.user_registration)
end_of_life = Timestamp(int(registration) + 60*60*24*7) # One week
user_revisions = db.all_revisions.query(
user_id=user.user_id,
direction="newer",
before=end_of_life,
include_page=True
)
user_events = chain(
[(user.user_id, registration, ('registration', registration, None))],
(
(
rev['rev_user'],
rev['rev_timestamp'],
('revision', Timestamp(rev['rev_timestamp']), rev)
)
for rev in user_revisions
)
)
for _, events in sessions.sessions(user_events):
for event_type, timestamp, payload in events:
if event_type == "revision":
rev = payload
day = Timestamp(rev['rev_timestamp']) - registration <= 60*60*24 # one day
week_revisions += 1
day_revisions += day
week_deleted_revisions += rev['archived']
day_deleted_revisions += rev['archived'] * day
if rev['page_namespace'] == 0:
week_main_revisions += 1
day_main_revisions += day
rev_timestamp = Timestamp(rev['rev_timestamp'])
cutoff_timestamp = Timestamp(int(rev_timestamp) + revert_cutoff)
if rev['archived']:
week_deleted_main_revisions += 1
day_deleted_main_revisions += day
sys.stderr.write("a")
else:
revert = reverts.database.check_row(db, rev,
radius=revert_radius,
before=cutoff_timestamp)
if revert != None: # Reverted edit!
week_reverted_main_revisions += 1
day_reverted_main_revisions += day
sys.stderr.write("r")
else:
day_productive_edits += day
week_productive_edits += 1
sys.stderr.write(".")
else:
sys.stderr.write("_")
week_sessions += 1
week_session_seconds += events[-1][1] - events[0][1]
sys.stderr.write("\n")
output.write([
user.wiki,
user.bucket,
user.first_event,
user.user_id,
user.user_registration,
day_revisions,
day_main_revisions,
day_reverted_main_revisions,
day_productive_edits,
week_revisions,
week_main_revisions,
week_reverted_main_revisions,
week_sessions,
week_session_seconds,
week_productive_edits
])
def run(users, revert_cutoff, revert_radius, dbuser, host, defaults_file):
output = tsv.Writer(sys.stdout, headers=HEADERS)
for wiki, users in groupby(users, lambda u:u.wiki):
db = database.DB(connection(wiki, host, dbuser, defaults_file))
for user in users:
sys.stderr.write("{0}, {1}: ".format(wiki, user.user_id))
day_revisions = 0
day_deleted_revisions = 0
day_main_revisions = 0
day_deleted_main_revisions = 0
day_reverted_main_revisions = 0
day_productive_edits = 0
week_revisions = 0
week_deleted_revisions = 0
week_main_revisions = 0
week_deleted_main_revisions = 0
week_reverted_main_revisions = 0
week_productive_edits = 0
week_sessions = 0
week_session_seconds = 0
registration = Timestamp(user.user_registration)
end_of_life = registration + 60*60*24*7 # One week after reg.
user_revisions = db.all_revisions.query(
user_id=user.user_id,
direction="newer",
before=end_of_life,
include_page=True
)
user_events = chain(
[
(
user.user_id,
registration,
('registration', registration, None)
)
],
(
(
rev['rev_user'],
rev['rev_timestamp'],
('revision', Timestamp(rev['rev_timestamp']), rev)
)
for rev in user_revisions
)
)
for _, events in sessions.sessions(user_events):
for event_type, timestamp, payload in events:
if event_type == "revision":
rev = payload
day = Timestamp(rev['rev_timestamp']) - registration <= 60*60*24 # one day
week_revisions += 1
day_revisions += day
week_deleted_revisions += rev['archived']
day_deleted_revisions += rev['archived'] * day
if rev['page_namespace'] == 0:
week_main_revisions += 1
day_main_revisions += day
rev_timestamp = Timestamp(rev['rev_timestamp'])
cutoff_timestamp = Timestamp(int(rev_timestamp) + revert_cutoff)
if rev['archived']:
week_deleted_main_revisions += 1
day_deleted_main_revisions += day
sys.stderr.write("a")
else:
revert = reverts.database.check(
db, rev_id=rev['rev_id'],
page_id=rev['page_id'],
radius=revert_radius,
before=int(Timestamp(rev['rev_timestamp'])) + revert_cutoff
)
if revert != None: # Reverted edit!
week_reverted_main_revisions += 1
day_reverted_main_revisions += day
sys.stderr.write("r")
else:
day_productive_edits += day
week_productive_edits += 1
sys.stderr.write(".")
else:
sys.stderr.write("_")
week_sessions += 1
week_session_seconds += events[-1][1] - events[0][1]
sys.stderr.write("\n")
output.write([
wiki,
user.user_id,
day_revisions,
day_deleted_revisions,
day_main_revisions,
day_deleted_main_revisions,
day_reverted_main_revisions,
day_productive_edits,
week_revisions,
week_deleted_revisions,
week_main_revisions,
week_deleted_main_revisions,
week_reverted_main_revisions,
week_productive_edits,
week_sessions,
week_session_seconds
])