def query(self, q, ranked=True):
"""Return a ranked list of matching `Document` instances."""
qq = Query.parse(q)
res = self.discodex_client.query(self.spec.invindex_name, qq)
res = map(TfIdf.undemux, res)
if not res:
return []
pageranks = None
if ranked:
scoredb = ScoreDB(self.spec.scoredb_path)
uris = [e[0] for e in res]
pageranks = dict(scoredb.rank(uris))
if not pageranks:
raise Exception("no ranks available")
docs = []
for uri,scores in res:
doc = self.docset.get(uri)
doc.score = Score(**scores)
if pageranks:
doc.score['pagerank'] = pageranks[uri]
doc.excerpt = doc.excerpt(qq)
docs.append(doc)
return docs
def test_expected_ranking(self):
if self.__class__.__name__ == 'IntegrationTestCase':
return
if not hasattr(self, 'expected_ranking'):
return
scoredb = ScoreDB(self.fqclient.spec.scoredb_path)
actual = scoredb.rank()
self.assertResultsSimilar(self.expected_ranking,
[Document(uri,scores=dict(pagerank=pr)) \
for uri,pr in actual])
def show_scores(program, spec):
"""Usage: <spec>
Shows the scores of documents in the specified `spec` ScoreDB file.
"""
from freequery.index.scoredb import ScoreDB
from freequery.client.client import Spec
spec = Spec(spec)
scoredb = ScoreDB(spec.scoredb_path)
for uri, score in scoredb.items():
print "%.8f\t%s" % (score, uri)
class TestScoreDB(unittest.TestCase):
sample_scores = {
"http://example.com/": 0.05,
"http://example.com/a.html": 0.04,
"http://stanford.edu/": 0.02,
"http://zzz.com/": 0.02,
}
def __write_fixture(self):
dbw = ScoreDBWriter(TEST_SCOREDB_PATH)
dbw.set_scores(self.sample_scores)
dbw.save_and_close()
def setUp(self):
self.__write_fixture()
self.scoredb = ScoreDB(TEST_SCOREDB_PATH)
def test_reads_writes(self):
self.assertAlmostEqual(0.05, self.scoredb.get_one("http://example.com/"))
self.assertAlmostEqual(0.04, self.scoredb.get_one("http://example.com/a.html"))
self.assertAlmostEqual(0.02, self.scoredb.get_one("http://stanford.edu/"))
self.assertAlmostEqual(0.02, self.scoredb.get_one("http://zzz.com/"))
def test_iterates(self):
self.assertEquals(sorted(self.sample_scores.items()), sorted(self.scoredb.items()))
def test_rank(self):
self.assertEquals(
[
("http://example.com/", 0.05),
("http://example.com/a.html", 0.04),
("http://zzz.com/", 0.02),
("http://stanford.edu/", 0.02),
],
self.scoredb.rank(),
)
self.assertEquals(
[("http://example.com/", 0.05), ("http://stanford.edu/", 0.02)],
self.scoredb.rank(["http://stanford.edu/", "http://example.com/"]),
)
def setUp(self):
self.__write_fixture()
self.scoredb = ScoreDB(TEST_SCOREDB_PATH)