def __init__(self, files=None, morphs=None, lmin=1):
self.name = "Reuters Performance"
c = reuters.get_keywords()
d = set()
for a in c:
if len(a) >= lmin: d.add(a)
for b in morphs or []:
e = b(a)
if len(e) >= lmin: d.add(e)
print len(d)
self.patterns = list(d)
c = ""
a = reuters.get_data_files()
for b in files or [0]:
c += a[b].read()
self.text = c
def __init__(self, files=None, morphs=None, lmin=1):
self.name = "Reuters Performance"
c = reuters.get_keywords()
d = set()
for a in c:
if len(a) >= lmin: d.add(a)
for b in morphs or []:
e = b(a)
if len(e) >= lmin: d.add(e)
print len(d)
self.patterns = list(d)
c = ""
a = reuters.get_data_files()
for b in files or [0]:
c += a[b].read()
self.text = c
def testReuters(self):
machine = SBOM(reuters.get_keywords())
self.hits = 0
machine(reuters.get_data_files()[0].read(), self.reuters_cb)
self.assertEqual(16193, self.hits)
def testReuters(self):
machine = SBOM(reuters.get_keywords())
self.hits = 0
machine(reuters.get_data_files()[0].read(), self.reuters_cb)
self.assertEqual(16193, self.hits)