def __init__(self, name, **kwargs):
self.name = name
self.projects = []
self.time = data_converter.create_datacontainer(datatype='list')
self.verified = True
for kw in kwargs:
setattr(self, kw, kwargs[kw])
def determine_article_count(articles_edited, first_year, final_year):
'''
This function counts the number of unique articles by year edited by a
particular editor.
'''
dc = data_converter.create_datacontainer(first_year, final_year)
dc = data_converter.add_months_to_datacontainer(dc, 'dict')
for year in articles_edited:
for month in articles_edited[year]:
for ns in articles_edited[year][month]:
dc[year][month][ns] = len(articles_edited[year][month][ns])
dc = cleanup_datacontainer(dc, {})
return dc
def determine_last_edit_by_year(edits, first_year, final_year):
'''
This function determines the date of the last edit in a given year for a
given editor.
'''
dc = data_converter.create_datacontainer(first_year, final_year, 0)
for edit in edits:
date = str(edit['date'].year)
if dc[date] == 0:
dc[date] = edit['date']
elif dc[date] < edit['date']:
dc[date] = edit['date']
return dc
def determine_number_edits(edits, first_year, final_year):
'''
This function counts the number of edits per namespace per month per year.
'''
dc = data_converter.create_datacontainer(first_year, final_year)
dc = data_converter.add_months_to_datacontainer(dc, 'dict')
for edit in edits:
ns = str(edit['ns'])
year = str(edit['date'].year)
month = str(edit['date'].month)
dc[year][month].setdefault(ns, 0)
dc[year][month][ns] += 1
dc = cleanup_datacontainer(dc, {})
return dc
def determine_number_reverts(edits, first_year, final_year):
'''
This function counts the number of times an edit was reverted in a given
month/year.
'''
dc = data_converter.create_datacontainer(first_year, final_year)
dc = data_converter.add_months_to_datacontainer(dc, 'dict')
for edit in edits:
year = str(edit['date'].year)
month = str(edit['date'].month)
ns = str(edit['ns'])
if edit['revert']:
dc[year][month].setdefault(ns, 0)
dc[year][month][ns] += 1
dc = cleanup_datacontainer(dc, {})
return dc
def determine_namespaces_workedon(edits, first_year, final_year):
'''
This function creates a list of namespaces that an editor has worked on in
a given month/year.
'''
dc = data_converter.create_datacontainer(first_year, final_year)
dc = data_converter.add_months_to_datacontainer(dc, 'set')
for edit in edits:
year = str(edit['date'].year)
month = str(edit['date'].month)
dc[year][month].add(edit['ns'])
for year in dc:
for month in dc[year]:
dc[year][month] = list(dc[year][month])
dc = cleanup_datacontainer(dc, [])
return dc
def add(self, key, value):
if value == 'NEXT':
self.n += 1
edits = self.drop_years_no_obs(self.editors[key]['edits'])
self.insert(key, edits, self.editors[key]['username'])
del self.editors[key]
else:
if key not in self.editors:
self.editors[key] = {}
self.editors[key]['obs'] = 0
self.editors[key][
'edits'] = data_converter.create_datacontainer(
2001, self.final_year, 'list')
self.editors[key]['username'] = value.pop('username')
else:
value.pop('username')
year = str(value['date'].year)
self.editors[key]['edits'][year].append(value)
self.editors[key]['obs'] += 1
def determine_edit_volume(edits, first_year, final_year):
'''
This function counts the number of characters added and remove by year
by month by namespace for a particular editor.
'''
dc = data_converter.create_datacontainer(first_year, final_year)
dc = data_converter.add_months_to_datacontainer(dc, 'dict')
for edit in edits:
year = str(edit['date'].year)
month = str(edit['date'].month)
ns = str(edit['ns'])
dc[year][month].setdefault(ns, {})
if edit['delta'] < 0:
dc[year][month][ns].setdefault('removed', 0)
dc[year][month][ns]['removed'] += edit['delta']
elif edit['delta'] > 0:
dc[year][month][ns].setdefault('added', 0)
dc[year][month][ns]['added'] += edit['delta']
dc = cleanup_datacontainer(dc, {})
return dc
def determine_articles_workedon(edits, first_year, final_year):
'''
This function creates a list of article_ids that an editor has worked on in
a given month/year.
'''
dc = data_converter.create_datacontainer(first_year, final_year)
dc = data_converter.add_months_to_datacontainer(dc, 'dict')
for edit in edits:
year = str(edit['date'].year)
month = str(edit['date'].month)
ns = str(edit['ns'])
dc[year][month].setdefault(ns, set())
dc[year][month][ns].add(edit['article_id'])
#convert the set to a list as mongo cannot store sets.
for year in dc:
for month in dc[year]:
for ns in dc[year][month]:
dc[year][month][ns] = list(dc[year][month][ns])
dc = cleanup_datacontainer(dc, {})
return dc
def __call__(self):
if self.edits == []:
return
first_year, final_year = determine_year_range(self.edits)
last_edit_by_year = determine_last_edit_by_year(
self.edits, first_year, final_year)
articles_edited = determine_articles_workedon(self.edits, first_year,
final_year)
article_count = determine_article_count(articles_edited, first_year,
final_year)
namespaces_edited = determine_namespaces_workedon(
self.edits, first_year, final_year)
character_count = determine_edit_volume(self.edits, first_year,
final_year)
revert_count = determine_number_reverts(self.edits, first_year,
final_year)
edit_count = determine_number_edits(self.edits, first_year, final_year)
totals = {}
counts = data_converter.create_datacontainer(first_year, final_year)
totals = calculate_totals(totals, counts, character_count,
'character_count')
totals = calculate_totals(totals, counts, revert_count, 'revert_count')
totals = calculate_totals(totals, counts, article_count,
'article_count')
totals = calculate_totals(totals, counts, edit_count, 'edit_count')
cum_edit_count_main_ns, cum_edit_count_other_ns = calculate_cum_edits(
self.edits)
edits = sort_edits(self.edits)
if len(edits) > self.cutoff:
new_wikipedian = edits[self.cutoff]['date']
else:
new_wikipedian = False
first_edit = edits[0]['date']
final_edit = edits[-1]['date']
data = {
'user_id': self.editor_id,
'username': self.username,
'new_wikipedian': new_wikipedian,
'cum_edit_count_main_ns': cum_edit_count_main_ns,
'cum_edit_count_other_ns': cum_edit_count_other_ns,
'final_edit': final_edit,
'first_edit': first_edit,
'last_edit_by_year': last_edit_by_year,
'articles_edited': articles_edited,
'edit_count': edit_count,
'namespaces_edited': namespaces_edited,
'article_count': article_count,
'character_count': character_count,
'revert_count': revert_count,
'totals': totals,
}
self.db_dataset.insert(data)