def get_tagger():
"""
Return a tagging function given some app settings.
`Settings` is the settings module of an app.
The returned value is a function that receives a unicode string and returns
a list of `Word` instances.
"""
if settings.USE_FREELING:
from quepy.freeling import run_freeling
tagger_function = lambda x: run_freeling(x, settings.FREELING_CMD)
else:
from quepy.nltktagger import run_nltktagger
tagger_function = lambda x: run_nltktagger(x, settings.NLTK_DATA_PATH)
def wrapper(string):
assert_valid_encoding(string)
words = tagger_function(string)
for word in words:
if word.pos not in PENN_TAGSET:
logger.warning("Tagger emmited a non-penn " "POS tag {!r}".format(word.pos))
return words
return wrapper
def test_run_freeling(self):
class FakeStringIO(StringIO):
name = "some_name"
class FakeExecutionCtx(object):
runcmd_called = False
def __init__(self, *args, **kwargs):
pass
def runcmd(self, cmd, stdin=None):
self.cmd = cmd
FakeExecutionCtx.runcmd_called = True
return (StringIO(), StringIO())
def tmpfile(self, name):
return FakeStringIO()
class FakeFunction(object):
def __init__(self):
self.called = False
self.args = None
self.kwargs = None
def __call__(self, *args, **kwargs):
self.called = True
self.args = args
self.kwargs = kwargs
bkp_ctx = sysutils.ExecutionContext
bkp_parse_output = freeling._parse_freeling_output
sysutils.ExecutionContext = FakeExecutionCtx
fake_parse_output = FakeFunction()
freeling._parse_freeling_output = fake_parse_output
freeling.run_freeling(u"who is Tom Cruise?", FREELING_CMD)
self.assertTrue(FakeExecutionCtx.runcmd_called)
self.assertTrue(fake_parse_output.called)
sysutils.ExecutionContext = bkp_ctx
freeling._parse_freeling_output = bkp_parse_output
def test_run_freeling(self):
class FakeStringIO(StringIO):
name = "some_name"
class FakeExecutionCtx(object):
runcmd_called = False
def __init__(self, *args, **kwargs):
pass
def runcmd(self, cmd, stdin=None):
self.cmd = cmd
FakeExecutionCtx.runcmd_called = True
return (StringIO(), StringIO())
def tmpfile(self, name):
return FakeStringIO()
class FakeFunction(object):
def __init__(self):
self.called = False
self.args = None
self.kwargs = None
def __call__(self, *args, **kwargs):
self.called = True
self.args = args
self.kwargs = kwargs
bkp_ctx = sysutils.ExecutionContext
bkp_parse_output = freeling._parse_freeling_output
sysutils.ExecutionContext = FakeExecutionCtx
fake_parse_output = FakeFunction()
freeling._parse_freeling_output = fake_parse_output
freeling.run_freeling(u"who is Tom Cruise?", FREELING_CMD)
self.assertTrue(FakeExecutionCtx.runcmd_called)
self.assertTrue(fake_parse_output.called)
sysutils.ExecutionContext = bkp_ctx
freeling._parse_freeling_output = bkp_parse_output
def test_real_run(self):
out = freeling.run_freeling(u"who is Tom Cruise?", FREELING_CMD)
out = list(out)
expected_pos = {
u"who": u"WP",
u"is": u"VBZ",
u"Tom Cruise": u"NNP",
u"?": u".",
}
for word in out:
self.assertIsInstance(word, freeling.Word)
self.assertEqual(word.pos, expected_pos[word.token])
def test_real_run(self):
out = freeling.run_freeling(u"who is Tom Cruise?", FREELING_CMD)
out = list(out)
expected_pos = {
u"who": u"WP",
u"is": u"VBZ",
u"Tom Cruise": u"NNP",
u"?": u".",
}
for word in out:
self.assertIsInstance(word, freeling.Word)
self.assertEqual(word.pos, expected_pos[word.token])
def get_tagger():
"""
Return a tagging function given some app settings.
`Settings` is the settings module of an app.
The returned value is a function that receives a unicode string and returns
a list of `Word` instances.
"""
if settings.USE_FREELING:
from quepy.freeling import run_freeling
tagger_function = lambda x: run_freeling(x, settings.FREELING_CMD)
else:
from quepy.nltktagger import run_nltktagger
tagger_function = lambda x: run_nltktagger(x, settings.NLTK_DATA_PATH)
def wrapper(string):
assert_valid_encoding(string)
words = tagger_function(string)
for word in words:
if word.pos not in PENN_TAGSET:
logger.warning("Tagger emmited a non-penn "
"POS tag {!r}".format(word.pos))
return words
return wrapper
