def close(self):
"""compute and display similarities on closing (i.e. end of parsing)"""
total = sum([len(lineset) for lineset in self.linesets])
duplicated = 0
stats = self.stats
for num, couples in self._compute_sims():
msg = []
for lineset, idx in couples:
msg.append("==%s:%s" % (lineset.name, idx))
msg.sort()
# pylint: disable-msg=W0631
for line in lineset._real_lines[idx:idx+num]:
msg.append(line.rstrip())
self.add_message('R0801', args=(len(couples), '\n'.join(msg)))
duplicated += num * (len(couples) - 1)
stats['nb_duplicated_lines'] = duplicated
stats['percent_duplicated_lines'] = total and duplicated * 100. / total
def close(self):
"""compute and display similarities on closing (i.e. end of parsing)"""
total = sum([len(lineset) for lineset in self.linesets])
duplicated = 0
stats = self.stats
for num, couples in self._compute_sims():
msg = []
for lineset, idx in couples:
msg.append("==%s:%s" % (lineset.name, idx))
msg.sort()
# pylint: disable-msg=W0631
for line in lineset._real_lines[idx:idx + num]:
msg.append(line.rstrip())
self.add_message('R0801', args=(len(couples), '\n'.join(msg)))
duplicated += num * (len(couples) - 1)
stats['nb_duplicated_lines'] = duplicated
stats['percent_duplicated_lines'] = total and duplicated * 100. / total
def _display_sims(self, sims):
"""display computed similarities on stdout"""
nb_lignes_dupliquees = 0
for num, couples in sims:
print
print num, "similar lines in", len(couples), "files"
couples = list(couples)
couples.sort()
for lineset, idx in couples:
print "==%s:%s" % (lineset.name, idx)
# pylint: disable-msg=W0631
for line in lineset._real_lines[idx:idx+num]:
print " ", line,
nb_lignes_dupliquees += num * (len(couples)-1)
nb_total_lignes = sum([len(lineset) for lineset in self.linesets])
print "TOTAL lines=%s duplicates=%s percent=%s" \
% (nb_total_lignes, nb_lignes_dupliquees,
nb_lignes_dupliquees*1. / nb_total_lignes)
def _display_sims(self, sims):
"""display computed similarities on stdout"""
nb_lignes_dupliquees = 0
for num, couples in sims:
print
print num, "similar lines in", len(couples), "files"
couples = list(couples)
couples.sort()
for lineset, idx in couples:
print "==%s:%s" % (lineset.name, idx)
# pylint: disable-msg=W0631
for line in lineset._real_lines[idx:idx + num]:
print " ", line,
nb_lignes_dupliquees += num * (len(couples) - 1)
nb_total_lignes = sum([len(lineset) for lineset in self.linesets])
print "TOTAL lines=%s duplicates=%s percent=%s" \
% (nb_total_lignes, nb_lignes_dupliquees,
nb_lignes_dupliquees*1. / nb_total_lignes)
def test_sum(self):
from logilab.common.compat import sum
self.assertEquals(sum(range(5)), 10)
self.assertRaises(TypeError, sum, 'abc')
def test_sum(self):
from logilab.common.compat import sum
self.assertEquals(sum(range(5)), 10)
self.assertRaises(TypeError, sum, 'abc')
