def summarize_ranks(collection, labels, min_prop=None):
"""
:param Collection collection:
:param list labels:
:param float min_prop:
:return tuple:
"""
summary = {}
if min_prop:
total_count = collection.find(verified_mongo).count()
min_count = min_prop * float(total_count)
labels = [
label
for label in labels
if collection.find(
{'tags.label': label}
).count() > min_count
]
query = {
'tags.label': {'$in': labels}
}
query.update(verified_mongo)
cursor = collection.find(
query,
{'tags': True}
)
selectors = {
label: {'label': label}
for label in labels
}
for article in cursor:
positions, _ = compute_positions(article['tags'], selectors)
if not positions:
continue
summary[article['_id']] = compute_ranks(positions)
return summary, labels
def _filter_first(tags):
positions, _ = compute_positions(tags, _filter_motreg_selectors)
return positions['filter'] < positions['motreg']