Merge pull request #65 from UVA-MachineLearningBioinformatics/test

Add dataset tests

pytest.ini

@@ -0,0 +1,5 @@
+# content of pytest.ini
+# (or tox.ini or setup.cfg)
+[pytest]
+addopts = -ra -q
+testpaths = test

requirements.txt

@@ -3,7 +3,7 @@ filelock
 language_check
 lru-dict
 nltk
-numpy<1.17
+numpy
 pandas
 scikit-learn
 scipy
@@ -16,3 +16,4 @@ tensorflow_hub
 terminaltables
 tqdm
 visdom
+pytest

test/test_models.py

@@ -0,0 +1,104 @@
+import textattack.datasets as datasets
+
+def test_ag_news_load():
+
+    # Expected
+    expected_label = 0
+    expected_text = (
+        "Thirst, Fear and Bribes on Desert Escape from Africa AGADEZ, Niger (Reuters) - Customs officers in this dusty "
+        "Saharan town turned a blind eye as yet another creaking truck piled with grain, smuggled cigarettes and dozens "
+        "of migrants heading for Europe rumbled off into the desert."
+    )
+
+    # Actual
+    data = datasets.classification.AGNews().__next__()
+    print(data[1])
+    print(expected_text)
+    actual_label = data[0]
+    actual_text = data[1]
+
+    # Test
+    assert expected_label == actual_label
+    assert expected_text == actual_text
+
+
+def test_imdb_load():
+
+    # Expected 
+    expected_label = 0
+    expected_text = (
+        "I bought this film on DVD so I could get an episode of Mystery Science Theater 3000. " 
+        "Thankfully, Mike, Crow, and Tom Servo are watchable, because the film itself is not. " 
+        "Although there is a plot, a story one can follow, and a few actors that can act, there isn't " 
+        "anything else. The movie was so boring, I have firmly confirmed that I will never watch it again " 
+        "without Tom, Crow and Mike. As summarized above, however, it was better than the film featured in "
+        "the MST3K episode that preceded it; Mitchell."
+    )   
+
+    # Actual
+    data = datasets.classification.IMDBSentiment().__next__()
+    actual_label = data[0]
+    actual_text = data[1]
+
+    # Test
+    assert expected_label == actual_label
+    assert expected_text == actual_text
+
+
+def test_movie_review_load():
+
+    # Expected
+    expected_label = 0
+    expected_text = "possibly the most irresponsible picture ever released by a major film studio ."
+    
+    # Actual
+    data = datasets.classification.MovieReviewSentiment().__next__()
+    actual_label = data[0]
+    actual_text = data[1]
+
+    assert expected_label == actual_label
+    assert expected_text == actual_text
+
+
+def test_yelp_load():
+
+    # Expected
+    expected_label = 1
+    expected_text = (
+        "Mmmm nothing better than a late night Filly-B run! Had three shredded beef tacos. "
+        "High Quality with all of the ingredients that you love is what you get here! I "
+        "know they are bad for me but they taste so good..nom..nom..nom.."
+    )
+
+    # Actual
+    data = datasets.classification.YelpSentiment().__next__()
+    actual_label = data[0]
+    actual_text = data[1]
+
+    assert expected_label == actual_label
+    assert expected_text == actual_text
+
+
+def test_MNLI_load():
+
+    # Expected 
+    expected_label = 0
+    expected_text = (
+        "In Temple Bar , the bookshop at the Gallery of Photography carries a large selection "
+        "of photographic publications , and the Flying Pig is a secondhand bookshop .>>>>"
+        "There is a bookshop at the gallery ."
+    )
+
+    # Actual
+    data = datasets.entailment.MNLI().__next__()
+    actual_label = data[0]
+    actual_text = data[1]
+
+    assert expected_label == actual_label
+    assert expected_text == actual_text
+
+
+
+
+
+

textattack/test/test.py

@@ -1,9 +0,0 @@
-
-""" @TODO: 
-    - Test downloading models from online
-    - Load yelp model and test accuracy
-    - Test a greedy attack
-    - Test a greedy attack w/ word constraint
-    - Test a greedy attack w/ sentence constraint
-    - Test a greedy WIR attack
-"""