def test_plot_isotonic_regression(self):
data = os.path.join(os.path.dirname(__file__), "data")
plot = os.path.join(data, "plot_isotonic_regression.py")
try:
verify_script(plot, try_onnx=False)
except NameError as e:
# Issues with local variable in comprehension list.
self.assertIn("'y'", str(e))
def test_plot_kernel_ridge_regression(self):
data = os.path.join(os.path.dirname(__file__), "data")
plot = os.path.join(data, "plot_kernel_ridge_regression.py")
res = verify_script(plot, try_onnx=False)
self.assertIsInstance(res, dict)
loc = res['locals']
self.assertNotEmpty(filter(lambda n: n.endswith('_onnx'), loc))
def test_plot_anomaly_comparison(self):
data = os.path.join(os.path.dirname(__file__), "data")
plot = os.path.join(data, "plot_anomaly_comparison.py")
res = verify_script(plot, try_onnx=False)
self.assertIsInstance(res, dict)
loc = res['locals']
self.assertIn('algorithm', loc)
def test_plot_iris_logistic(self):
data = os.path.join(os.path.dirname(__file__), "data")
plot = os.path.join(data, "plot_iris_logistic.py")
res = verify_script(plot)
self.assertIsInstance(res, dict)
loc = res['locals']
self.assertIn('X', loc)
self.assertIn('logreg', loc)
self.assertIn('logreg_onnx', loc)
def test_plot_examples(self):
fLOG(__file__, self._testMethodName, OutputPrint=__name__ == "__main__")
datas = ["git"]
no_onnx = {}
noconv = {}
skipped = {}
issues = {}
has_onnx = {}
for data in datas: # pylint: disable=R1702
if data == "git":
temp_git = get_temp_folder(
__file__, "temp_sklearn", clean=False)
examples = os.path.join(temp_git, "examples")
if not os.path.exists(examples):
fLOG('cloning scikit-learn...')
clone(temp_git, "github.com", "scikit-learn",
"scikit-learn", fLOG=fLOG)
fLOG('done.')
data = examples
fold = os.path.join(os.path.dirname(__file__), data)
if not os.path.exists(fold):
continue
for ind_root, (root, _, files) in enumerate(os.walk(fold)):
last = os.path.split(root)[-1]
if last in TestLONGSklearnExample.skipped_folder:
continue
for ind, nfile in enumerate(files):
full_ind = ind_root * 1000 + ind
if full_ind < TestLONGSklearnExample.begin:
skipped[nfile] = None
continue
if (full_ind >= TestLONGSklearnExample.end and
TestLONGSklearnExample.end >= 0):
skipped[nfile] = None
continue
nfile = nfile.replace("\\", "/")
if nfile in TestLONGSklearnExample.skipped_examples:
skipped[nfile] = None
continue
ext = os.path.splitext(nfile)[-1]
if ext != '.py':
continue
name = os.path.split(nfile)[-1]
if not name.startswith('plot_'):
continue
fLOG("verify {}/{}:{} - '{}'".format(
ind + 1, len(files), full_ind, nfile))
plot = os.path.join(root, nfile)
try:
res = verify_script(
plot, existing_loc=TestLONGSklearnExample.existing_loc)
except NotFittedError as e:
fLOG(' model was not trained',
str(e).split('\n')[0])
noconv[nfile] = e
continue
except MissingShapeCalculator as e:
fLOG(' missing converter', str(e).split('\n')[0])
noconv[nfile] = e
continue
except MissingVariableError as e:
issues[nfile] = e
fLOG(' missing variable', str(e).split('\n')[0])
continue
except ValueError as e:
if skl_version == "0.22.2.post1":
fLOG(' value error', str(e).split('\n')[0])
continue
raise e
except (KeyError, NameError, RuntimeError, TypeError,
ImportError, AttributeError) as e:
issues[nfile] = e
fLOG(' local function', str(e).split('\n')[0])
continue
if res is not None:
if any(filter(lambda n: n.endswith('_onnx'), res['locals'])):
fLOG(' ONNX ok')
has_onnx[nfile] = res['onx_info']
else:
no_onnx[nfile] = res['locals']
fLOG(' no onnx')
else:
fLOG(' issue')
if len(has_onnx) == 0:
raise RuntimeError("Unable to find any example in\n{}".format(
"\n".join(datas)))
rows = []
for n, d in [('no_onnx', no_onnx),
('noconv', noconv),
('issues', issues),
]:
for k, v in d.items():
sv = str(v).replace("\r", "").replace("\n", " ")
rows.append(dict(name=k, result=sv, kind=n))
for k, v in has_onnx.items():
rows.append(dict(name=k, result='OK', kind='ONNX'))
temp = get_temp_folder(__file__, 'temp_plot_examples')
stats = os.path.join(temp, "stats.csv")
pandas.DataFrame(rows).to_csv(stats, index=False)
for k, v in has_onnx.items():
pkl = os.path.join(temp, k + '.pkl')
with open(pkl, 'wb') as f:
pickle.dump(v, f)