def update_visitors_metric():
"""Get new visitor data from webtrends and save."""
# Start updating the day after the last updated.
latest_metric = _get_latest_metric(VISITORS_METRIC_CODE)
if latest_metric is not None:
latest_metric_date = latest_metric.start
else:
latest_metric_date = date(2011, 01, 01)
start = latest_metric_date + timedelta(days=1)
# Collect up until yesterday
end = date.today() - timedelta(days=1)
# Get the visitor data from webtrends.
visitors = Webtrends.visits(start, end)
# Create the metrics.
metric_kind = MetricKind.objects.get(code=VISITORS_METRIC_CODE)
for date_str, visits in visitors.items():
day = datetime.strptime(date_str, '%Y-%m-%d').date()
Metric.objects.create(
kind=metric_kind,
start=day,
end=day + timedelta(days=1),
value=visits)
def update_visitors_metric():
"""Get new visitor data from webtrends and save."""
try:
# Get the latest metric value.
last_metric = Metric.objects.filter(
kind__code=VISITORS_METRIC_CODE).order_by('-start')[0]
# Start updating the day after the last updated.
start = last_metric.start + timedelta(days=1)
except IndexError:
# There are no metrics yet, start from 2011-01-01
start = date(2011, 01, 01)
# Collect up until yesterday
end = date.today() - timedelta(days=1)
# Get the visitor data from webtrends.
visitors = Webtrends.visits(start, end)
# Create the metrics.
metric_kind = MetricKind.objects.get(code=VISITORS_METRIC_CODE)
for date_str, visits in visitors.items():
day = datetime.strptime(date_str,"%Y-%m-%d").date()
Metric.objects.create(
kind=metric_kind,
start=day,
end=day + timedelta(days=1),
value=visits)
def test_visits(self, key_metrics):
"""Test Webtrends.visits()."""
key_metrics.return_value = KEY_METRICS_JSON_RESPONSE
visits = Webtrends.visits(date(2012, 01, 01), date(2012, 01, 07))
eq_(7, len(visits))
eq_(495974, visits['2012-01-01'])
eq_(529301, visits['2012-01-07'])
def test_visits_by_locale(self, request):
"""Test Webtrends.visits_by_locale()."""
request.return_value = L10N_METRICS_JSON_RESPONSE
visits = Webtrends.visits_by_locale(date(2012, 02, 11), date(2012, 03, 11))
eq_(77, len(visits))
eq_(7561779.0, visits['en-US'])
eq_(815609.0, visits['es'])
def update_l10n_metric():
"""Calculate new l10n coverage numbers and save.
L10n coverage is a measure of the amount of translations that are
up to date, weighted by the number of visits for each locale.
The "algorithm" (see Bug 727084):
SUMO visits = Total non-en-US SUMO visits for the last 3 months;
Total translated = 0;
For each locale {
Total up to date = Total up to date +
((Number of up to date articles from the en-US top 50 visited)/50 ) *
(Visitors for that locale / SUMO visits));
}
An up to date article is any of the following:
* An en-US article (by definition it is always up to date)
* The latest en-US revision has been translated
* There are only new revisions with TYPO_SIGNIFICANCE not translated
* There is only one revision of MEDIUM_SIGNIFICANCE not translated
"""
# Get the top 60 visited articles. We will only use the top 50
# but a handful aren't localizable so we get some extras.
top_60_docs = _get_top_docs(60)
# Get the visits to each locale in the last 30 days.
end = date.today() - timedelta(days=1) # yesterday
start = end - timedelta(days=30)
locale_visits = Webtrends.visits_by_locale(start, end)
# Total visits.
total_visits = sum(locale_visits.itervalues())
# Calculate the coverage.
coverage = 0
for locale, visits in locale_visits.iteritems():
if locale == settings.WIKI_DEFAULT_LANGUAGE:
num_docs = MAX_DOCS_UP_TO_DATE
up_to_date_docs = MAX_DOCS_UP_TO_DATE
else:
up_to_date_docs, num_docs = _get_up_to_date_count(
top_60_docs, locale)
if num_docs and total_visits:
coverage += ((float(up_to_date_docs) / num_docs) *
(float(visits) / total_visits))
# Save the value to Metric table.
metric_kind = MetricKind.objects.get(code=L10N_METRIC_CODE)
day = date.today()
Metric.objects.create(
kind=metric_kind,
start=day,
end=day + timedelta(days=1),
value=int(coverage * 100)) # Store as a % int.
def update_visitors_metric():
"""Get new visitor data from webtrends and save."""
# Start updating the day after the last updated.
latest_metric = _get_latest_metric(VISITORS_METRIC_CODE)
if latest_metric is not None:
latest_metric_date = latest_metric.start
else:
latest_metric_date = date(2011, 01, 01)
start = latest_metric_date + timedelta(days=1)
# Collect up until yesterday
end = date.today() - timedelta(days=1)
# Get the visitor data from webtrends.
visitors = Webtrends.visits(start, end)
# Create the metrics.
metric_kind = MetricKind.objects.get(code=VISITORS_METRIC_CODE)
for date_str, visits in visitors.items():
day = datetime.strptime(date_str, '%Y-%m-%d').date()
Metric.objects.create(kind=metric_kind,
start=day,
end=day + timedelta(days=1),
value=visits)
def update_l10n_metric():
"""Calculate new l10n coverage numbers and save.
L10n coverage is a measure of the amount of translations that are
up to date, weighted by the number of visits for each locale.
The "algorithm" (see Bug 727084):
SUMO visits = Total non-en-US SUMO visits for the last 3 months;
Total translated = 0;
For each locale different to en-US {
Total translated = Total Translated +
((Number of updated articles from the en-US top 50 visited)/50 ) *
(Visitors for that locale / SUMO visits));
}
"""
# Get the top 60 visited articles. We will only use the top 50
# but a handful aren't localizable so we get some extras.
top_60_docs = _get_top_docs(60)
# Get the visits to each locale in the last 90 days.
end = date.today() - timedelta(days=1) # yesterday
start = end - timedelta(days=90)
locale_visits = Webtrends.visits_by_locale(start, end)
# Discard en-US.
locale_visits.pop('en-US')
# Total non en-US visits.
total_visits = sum(locale_visits.itervalues())
# Calculate the coverage.
coverage = 0
for locale, visits in locale_visits.iteritems():
up_to_date_docs = 0
num_docs = 0
for doc in top_60_docs:
if num_docs == 50:
# Stop at 50 documents.
break
if not doc.is_localizable:
# Skip non localizable documents.
continue
num_docs += 1
cur_rev_id = doc.latest_localizable_revision_id
translation = doc.translated_to(locale)
if (translation and translation.current_revision_id and
translation.current_revision_id >= cur_rev_id):
up_to_date_docs += 1
if num_docs and total_visits:
coverage += ((float(up_to_date_docs) / num_docs) *
(float(visits) / total_visits))
# Save the value to Metric table.
metric_kind = MetricKind.objects.get(code=L10N_METRIC_CODE)
day = date.today()
Metric.objects.create(
kind=metric_kind,
start=day,
end=day + timedelta(days=1),
value=int(coverage * 100)) # Store as a % int.
def json_for(cls, period):
"""Return the JSON-formatted WebTrends stats for the given period."""
start, end = period_dates()[period]
return Webtrends.wiki_report(start, end)
def update_l10n_metric():
"""Calculate new l10n coverage numbers and save.
L10n coverage is a measure of the amount of translations that are
up to date, weighted by the number of visits for each locale.
The "algorithm" (see Bug 727084):
SUMO visits = Total non-en-US SUMO visits for the last 3 months;
Total translated = 0;
For each locale different to en-US {
Total translated = Total Translated +
((Number of updated articles from the en-US top 50 visited)/50 ) *
(Visitors for that locale / SUMO visits));
}
"""
# Get the top 60 visited articles. We will only use the top 50
# but a handful aren't localizable so we get some extras.
top_60_docs = _get_top_docs(60)
# Get the visits to each locale in the last 90 days.
end = date.today() - timedelta(days=1) # yesterday
start = end - timedelta(days=90)
locale_visits = Webtrends.visits_by_locale(start, end)
# Discard en-US.
locale_visits.pop('en-US')
# Total non en-US visits.
total_visits = sum(locale_visits.itervalues())
# Calculate the coverage.
coverage = 0
for locale, visits in locale_visits.iteritems():
up_to_date_docs = 0
num_docs = 0
for doc in top_60_docs:
if num_docs == 50:
# Stop at 50 documents.
break
if not doc.is_localizable:
# Skip non localizable documents.
continue
num_docs += 1
cur_rev_id = doc.latest_localizable_revision_id
translation = doc.translated_to(locale)
if (translation and translation.current_revision_id
and translation.current_revision_id >= cur_rev_id):
up_to_date_docs += 1
if num_docs and total_visits:
coverage += ((float(up_to_date_docs) / num_docs) *
(float(visits) / total_visits))
# Save the value to Metric table.
metric_kind = MetricKind.objects.get(code=L10N_METRIC_CODE)
day = date.today()
Metric.objects.create(kind=metric_kind,
start=day,
end=day + timedelta(days=1),
value=int(coverage * 100)) # Store as a % int.
