pageviews: Relatives au nombre de vues total/jour
This commit is contained in:
parent
9f3459fd24
commit
0faad66aa2
|
@ -5,20 +5,27 @@ import numpy
|
|||
from scipy import stats
|
||||
|
||||
# Chemin racine pour les API Wikimedia
|
||||
wikimedia_base_path = 'https://wikimedia.org/api/rest_v1'
|
||||
pageviews_root = 'https://wikimedia.org/api/rest_v1'
|
||||
|
||||
# Patron d’accès à l’API pageviews de Wikimedia
|
||||
wikimedia_pageviews_path = '/' + '/'.join([
|
||||
# Patron d’accès à l’API pageviews pour un article
|
||||
pageviews_article_endpoint = '/' + '/'.join([
|
||||
'metrics', 'pageviews', 'per-article', '{project}',
|
||||
'{access}', '{agent}', '{article}', '{granularity}',
|
||||
'{start}', '{end}'
|
||||
])
|
||||
|
||||
# Chemin d’accès à l’API pageviews pour un projet complet
|
||||
pageviews_project_endpoint = '/' + '/'.join([
|
||||
'metrics', 'pageviews', 'aggregate', '{project}',
|
||||
'{access}', '{agent}', '{granularity}',
|
||||
'{start}', '{end}'
|
||||
])
|
||||
|
||||
# Format de dates utilisée pour l’API Wikimedia
|
||||
wikimedia_date_format = '%Y%m%d'
|
||||
pageviews_date_format = '%Y%m%d'
|
||||
|
||||
# Date de première disponibilité des pageviews sur l’API Wikimedia
|
||||
wikimedia_pageviews_start = datetime(2015, 7, 1)
|
||||
pageviews_first_data = datetime(2015, 7, 1)
|
||||
|
||||
# Tableau contenant tous les jours de l’année de 1 à 365
|
||||
year_all_days = numpy.arange(1, 366)
|
||||
|
@ -68,23 +75,56 @@ def smooth(views, scale):
|
|||
return pdf_matrix.dot(views)
|
||||
|
||||
|
||||
def get(project, article):
|
||||
def get_aggregate(project):
|
||||
"""
|
||||
Obtient le nombre de visites sur une page Wikipédia par jour.
|
||||
Obtient le nombre de visites sur Wikipédia par jour.
|
||||
|
||||
:param project: Projet Wikipédia ciblé.
|
||||
:return: Compteur associant chaque jour à son nombre de visites.
|
||||
"""
|
||||
res = session.get(pageviews_root + pageviews_project_endpoint.format(
|
||||
project=project,
|
||||
access='all-access',
|
||||
agent='user',
|
||||
granularity='daily',
|
||||
start=pageviews_first_data.strftime(pageviews_date_format),
|
||||
end=datetime.today().strftime(pageviews_date_format)
|
||||
))
|
||||
|
||||
data = res.json()
|
||||
|
||||
# Vérifie que la réponse reçue indique un succès
|
||||
if res.status_code != 200:
|
||||
if 'detail' in data:
|
||||
detail = data['detail']
|
||||
message = ', '.join(detail) if type(detail) == list else detail
|
||||
raise Exception(message)
|
||||
else:
|
||||
raise Exception('Erreur {}'.format(res.status_code))
|
||||
|
||||
# Construit le dictionnaire résultant
|
||||
return collections.Counter(dict(
|
||||
(record['timestamp'][:8], record['views'])
|
||||
for record in data['items']
|
||||
))
|
||||
|
||||
|
||||
def get_article(project, article):
|
||||
"""
|
||||
Obtient le nombre de visites sur un article Wikipédia par jour.
|
||||
|
||||
:param project: Projet Wikipédia ciblé.
|
||||
:param article: Article ciblé dans le site.
|
||||
:return: Compteur associant chaque jour à son nombre de visites.
|
||||
"""
|
||||
# Soumet une requête à l’API REST pour obtenir les vues de l’article
|
||||
res = session.get(wikimedia_base_path + wikimedia_pageviews_path.format(
|
||||
res = session.get(pageviews_root + pageviews_article_endpoint.format(
|
||||
project=project,
|
||||
article=article,
|
||||
access='all-access',
|
||||
agent='user',
|
||||
granularity='daily',
|
||||
start=wikimedia_pageviews_start.strftime(wikimedia_date_format),
|
||||
end=datetime.today().strftime(wikimedia_date_format)
|
||||
start=pageviews_first_data.strftime(pageviews_date_format),
|
||||
end=datetime.today().strftime(pageviews_date_format)
|
||||
))
|
||||
|
||||
data = res.json()
|
||||
|
@ -123,7 +163,7 @@ def mean(views):
|
|||
accumulator[datemonth] = []
|
||||
|
||||
for date_str, views in views.items():
|
||||
date = datetime.strptime(date_str, wikimedia_date_format)
|
||||
date = datetime.strptime(date_str, pageviews_date_format)
|
||||
|
||||
if not (date.month == 2 and date.day == 29):
|
||||
datemonth = date.strftime(datemonth_format)
|
||||
|
|
|
@ -35,6 +35,8 @@ output = sys.argv[2]
|
|||
articles = sys.argv[3:]
|
||||
|
||||
site = mediawiki_api.instanciate(project)
|
||||
project_views = wikipedia_pageviews.get_aggregate(project)
|
||||
|
||||
output_to_file = output != '-'
|
||||
|
||||
if output_to_file:
|
||||
|
@ -49,14 +51,22 @@ if output_to_file:
|
|||
fig = pyplot.figure(figsize=(4.7, 3.3))
|
||||
ax = fig.add_subplot(111)
|
||||
|
||||
# Configuration de l’abscisse pour afficher les jours de l’année
|
||||
ax.set_xlabel('Jours de l’année')
|
||||
ax.set_xticks([
|
||||
datetime(1, month, 1).toordinal()
|
||||
for month in range(1, 13)
|
||||
])
|
||||
|
||||
ax.set_xticklabels(calendar.month_abbr[1:13])
|
||||
|
||||
# Configuration de l’ordonnée pour être affichée en pourcentage
|
||||
ax.set_ylabel('Proportion de vues par jour')
|
||||
ax.yaxis.set_major_formatter(
|
||||
matplotlib.ticker.FuncFormatter(
|
||||
lambda y, _: '{:.5} ‰'.format(y * 1000).replace('.', ',')
|
||||
)
|
||||
)
|
||||
|
||||
for article in articles:
|
||||
canonical = mediawiki_api.article_canonical(site, article)
|
||||
|
||||
|
@ -66,21 +76,24 @@ for article in articles:
|
|||
.format(article, canonical)
|
||||
)
|
||||
|
||||
del article
|
||||
article = canonical
|
||||
del canonical
|
||||
|
||||
redirects = mediawiki_api.article_redirects(site, canonical)
|
||||
mean_views = wikipedia_pageviews.mean(sum(
|
||||
(wikipedia_pageviews.get(project, page)
|
||||
for page in redirects + [canonical]),
|
||||
redirects = mediawiki_api.article_redirects(site, article)
|
||||
total_views = sum(
|
||||
(wikipedia_pageviews.get_article(project, page)
|
||||
for page in redirects + [article]),
|
||||
start=collections.Counter()
|
||||
))
|
||||
|
||||
ax.plot(
|
||||
wikipedia_pageviews.smooth(mean_views, 10),
|
||||
label=canonical
|
||||
)
|
||||
|
||||
ax.set_ylabel('Vues par jour')
|
||||
relative_views = dict((
|
||||
(date, total_view / project_views[date])
|
||||
for date, total_view in total_views.items()
|
||||
))
|
||||
|
||||
mean_views = wikipedia_pageviews.mean(relative_views)
|
||||
smoothed_views = wikipedia_pageviews.smooth(mean_views, 10)
|
||||
ax.plot(smoothed_views, label=article)
|
||||
|
||||
fig.legend(framealpha=1)
|
||||
fig.autofmt_xdate()
|
||||
|
|
Loading…
Reference in New Issue