def test_sorted_by_popularity(self):
assert jautils.sorted_by_popularity(
[u'山', u'田', u'xxx', u'yyy', u'zzz']) == \
[u'xxx', u'yyy', u'zzz', u'山', u'田']
assert jautils.sorted_by_popularity(
[u'山', u'田', u'はなこ']) == \
[u'はなこ', u'山', u'田']
assert jautils.sorted_by_popularity(
[u'山', u'田', u'龍', u'太', u'郎']) == \
[u'龍', u'太', u'郎', u'山', u'田']
def test_sorted_by_popularity(self):
assert jautils.sorted_by_popularity(
[u'山', u'田', u'xxx', u'yyy', u'zzz']) == \
[u'xxx', u'yyy', u'zzz', u'山', u'田']
assert jautils.sorted_by_popularity(
[u'山', u'田', u'はなこ']) == \
[u'はなこ', u'山', u'田']
assert jautils.sorted_by_popularity(
[u'山', u'田', u'龍', u'太', u'郎']) == \
[u'龍', u'太', u'郎', u'山', u'田']
def sort_query_words(query_words):
"""Sort query_words so that the query filters created from query_words are
more effective and consistent when truncated due to NeedIndexError, and
return the sorted list."""
# (1) Sort them lexicographically so that we return consistent search
# results for query 'AA BB CC DD' and 'DD AA BB CC' even when filters
# are truncated.
sorted_query_words = sorted(query_words)
# (2) Sort them according to popularity so that less popular query words,
# which are usually more effective filters, come first.
sorted_query_words = jautils.sorted_by_popularity(sorted_query_words)
# (3) Sort them according to the lengths so that longer query words,
# which are usually more effective filters, come first.
return sorted(sorted_query_words, key=len, reverse=True)
def sort_query_words(query_words):
"""Sort query_words so that the query filters created from query_words are
more effective and consistent when truncated due to NeedIndexError, and
return the sorted list."""
# (1) Sort them lexicographically so that we return consistent search
# results for query 'AA BB CC DD' and 'DD AA BB CC' even when filters
# are truncated.
sorted_query_words = sorted(query_words)
# (2) Sort them according to popularity so that less popular query words,
# which are usually more effective filters, come first.
sorted_query_words = jautils.sorted_by_popularity(sorted_query_words)
# (3) Sort them according to the lengths so that longer query words,
# which are usually more effective filters, come first.
return sorted(sorted_query_words, key=len, reverse=True)
